The Archive

A year in review: 2025

Wed, 18 Feb 2026 00:00:00 +0000

2025 flew by, I don’t feel like it’s really over. There were not enough days. Luckily I do feel it was a successful year, a year in which I grew as a person and engaged in new habits.

Let’s get bragging!

From passive consumer to active reading and listening #

In 2025 I decided to get rid of my ‘mindless media consumption habits’.

From passive to active reading 🎉 #

One such habit was the pointless compulsory scrolling on sites like nu.nl, tweakers.net, linkedin.com in unguarded moments. Every 10s of unfilled time got me to whip out my phone. Getting rid of deeply-nested habits is difficult (especially for someone as easily addicted to distraction as me!), but I managed. Incredibly proud of that.

I created a small script on my phone that shows a black screen in front of the website with a count-down timer. This tiny blockade is enough to stop me from visiting said websites.

Instead, I got a subscription to the NRC newspaper and the VPRO-gids. Both give me more depth and quality, and (even though both leftist) engage me more with topics outside my normal comfort-zone. Quality above quantity.

I also enjoy reading from many RSS-feeds.

The scrolling is not gone completely. Getting rid of LinkedIn is not possible in my work, I’m stuck a bit longer with that one.

From passive binging to mindful watching 🍿 #

We used to watch what’s ‘available’ on Netflix, Apple TV, Disney+. We did not get happy anymore from watching, and it definitely didn’t live up to the ever-increasing pricing. More than once we decided to turn off the TV after 10 minutes of searching for ‘something good’.

In 2025 I decided to get rid of these streaming-services. It was an easy decision.

We kept NPO and discovered many fun things from the Dutch public services. In particular the kids-shows are more fun (Klokhuis!).

I also hunt for movies on Apple Movies (through Cheapcharts) based on tips from the VPRO-gids. And I bought a (gasp!) DVD-player for Disney movies.

All-in-all, I’m extremely happy with this move.

From a wallpaper sound to active listening 💿 #

I rediscovered Plex as a way of listening to music. Instead of turning on ‘infinite stream of music’ on Spotify I now again choose the albums from my own collection.

I used to love Spotify, I discovered new artists, and I was hooked on the weekly. I can honestly say I looked forward to the weekly even.

But lately I listen to the same music. Most recommendations feel stale, or are commercials in disguise. The music quality is inconsistent, the UI overcomplicated with podcasts and books, and the pricing is ever-increasing.

I buy music from Bandcamp, or (gasp again!) actual CDs.

Digital sovereignty 🇪🇺 #

Part of my digital passive consumerism was an over-reliance on USA-based cloud-companies. I picked these companies out of habit (Google, Microsoft, Apple), not because I thought of any consequences.

In light of the global shift (and especially due to Trump and his cronies’ middle-fingers to Europe), I decided to rethink my automatic allegiance to American cloud services.

I moved from Google to Fastmail, Whatsapp to Threema, ChatGPT to Mistral, and am experimenting with TomTom instead of Apple Maps. I got rid of Hubspot completely.

My blog and other websites are now hosted at the Dutch mijn.host

Physical health 💪 #

I started sporting at Muscleminded. Love it!

Working as a freelancer #

I took my freelance-company more seriously and experimented with different ways of ‘being a company’. Most of those experiments didn’t see the light-of-day, but they helped me think about myself as a business.

I joined a Business Open (a club of entrepreneurs who eat breakfast together every two weeks), and I enjoy sharing my business there. It also lead to my first Product-Management role at Fizyr, a robot-vision company.

I’m the co-applicant on a subsidy project for improving the sustainability in the sheep-sector through data-predictions.

I restarted the meetup-club of The Hague Lean UX. We organized 6 meetups, and the club grew with 313 people to 1576 members.

This year I hope to finally release a product of my own.

Vibe-coding #

I experimented a lot with different tools in AI this year. I find vibe-coding is especially useful for small prototypes, custom data-analysis tools and demos. Larger projects and functional programming have become possible near the end of 2025 as well.

Piano-lessons 🎹 #

Although I started in 2024 already, the piano-lessons were a big part of 2025. I even did a public performance of Yann Tiersen’s Comptine d’un Autre été from the movie Amelie.

Family and projects #

My family had some illness, but overall we created many beautiful memories.

The kids-parties were crazy fun again (I hosted science-experiments for my daughter, which was hilarious).
We visited loads of museums. We went indoor skydiving.
I helped out a lot at my childrens’ school.
They kids grew tremendously. I’m so proud of them.
I turned 40. We went to a beautiful home in The Veluwe.
We went tot Istanbul for holidays.
We bought a great amplifier with speakers. Our music sounds awesome now.

I organized my traditional Easter puzzle. And I handwrote a physical newspaper (‘de opwelling’), which I sent out three times. I built a flow (with AI) that gives me all activities in Rijswijk, Rotterdam and The Hague each weekend.

There were not too many big maintenance-projects this year in our house.

Closing thoughts #

2025 was a year of changing allegiances. Connecting more with people, and severing unhealthy habits.

Finding a fancy CD-player

Fri, 09 Jan 2026 00:00:00 +0000

We bought a Bluesound Powernode Edge last year with a professional pair of speakers, and I love the way our music sounds now. Next step is extending this system with a CD-player. Yes, streaming is the future, but I’m going back in time.

Our new Bluesound system is amazing. We coupled it with the beautiful walnut-colored Dali Oberon 1 speakers that look fantastic in our living room. We easily upped our music-listening fivefold. The sound! I hear new details in my favorite albums, making me discover them all anew. My wife discovered her favorite song contains two singers instead of one. Unbelievable the emotion we get from the music now.

The four downsides of streaming #

When Spotify first launched it changed everything. I dropped my old collection in a heartbeat and went all in. Listen to everything there ever was, for a low fixed price. It sounded too good to be true, and I’ve since experienced the downsides.

1. Sound quality #

The first downside is that ‘sounding too good to be true’ is merely a saying. The sound quality of Spotify is worse than CDs. I investigated streaming quality before on this blog.

2. Mindlessness #

Another thing I don’t like about Spotify is that it feels transient. It is streaming music. Press play on ‘Spotify’ and you get an endless stream of music. Like diving ever further in the infinity pools of endless scrolling on Instagram and Facebook, music is running the risk of mindlessness. A ‘set and forget’ background noise.

3. Out of touch #

Digital music ‘streams’ are ethereal. There’s nothing to touch, nothing to browse, no tactile experience. I need to grab the distraction-machine that is my phone to operate the music. A small and simple rectangular piece of glass. Boring!

Contrast this with records, CDs and cassettes. A mindful, tactile experience.

And as extra bonus, they were prominently on display. Beautifully designed pieces of art that show my taste in music. For myself, and for others. Whenever I visit other people’s homes, I love looking at their book- and music-collection. These are conversation-starters, “oh you read this?” and “I love this band”. Ebooks and streaming music killed this.

Before changing to Spotify I had a small collection of albums. Those were sold ten or twenty years ago, but I do remember all of them. My first bought album (Red Hot Chilli Peppers, bought for 20 euros), my first Cradle of Filth album (bootleg copy by a class mate), the Lord of the Rings soundtrack (I loved the cover), the album I bought after a concert… I have digital copies and I can listen on Spotify - but those I “rent” and I’m nostalgic for the ownership.

4. Worth the price? #

And the price. When we started, Spotify felt a nobrainer. Endless music for a small monthly fee. But that fee has since increased to 18 euro per month. The cost of almost two cds. In the 7 years I paid Spotify, I could’ve been the owner of a music collection of 137 cds. I think that amount of discs would cover all my listening habits.

Now what?

Back in time #

If not streaming, what are the alternatives?

We tried records before, but that didn’t stick. In my new plan, I’ve slowly prepared the family for re-introducing physical albums back in the home by scattering second-hand CDs in the house…

And now it’s time for part two of the plan: I’d like to add a CD player to our home system! I’m looking for something small and stylish rather than a full-size “hi-fi beast.” The Bluesound Node attracted us for being small and it’s ability to hide behind some books (finally the Lullos have their Hidden Sound System!).

The Bluesound Edge gives me a few connection options for the CD-player: digital (optical/TOSLINK), line-in, and Bluetooth — with sound quality in that order, as the Powernode’s DAC is good and Bluetooth inevitably means some loss. Update: after ordering the FiiO M13 I found out the Bluetooth that device (and other cd-players) have is different from the Bluetooth used in our phones/laptops. Conclusion is they cannot speak to eachother.

The contenders #

There are several small and stylish options for cd players. I collected a range of devices from stores worldwide. I prefer to buy local, but the renewed interest in CDs is an Asian-specific thing for now and hasn’t reached Europe yet.

There are cheaper knock-offs and “transports” (players that don’t do decoding and merely forward the digital signal) with good aesthetics on Alibaba, but I figure you get what you pay for.

Size is a factor for me. I want the device to be small and not overly present in our room. The majority of these devices is meant as portable players (and thus contain a battery).

1. KM5 CP1 – ~€135 #

The KM5 CP1 are compact Bluetooth CD players from the Japanese KM5 brand. They are portable, easy to operate, and perceived as high quality for casual listening.

What’s neat is that the player can be mounted on the wall, and you can show the album booklet. The player then resembles a Polaroid-frame with the album art.

The newer KM5 CP2 is equally stylish and includes a built-in speaker.

Options:

Bluetooth: ✅
TOSLINK (optical): ❌
External power: ✅
Battery: ✅

2. Tinyl SPLIT (Modular 3-in-1 CD Player) – ~€140–160 #

The Tinyl SPLIT is another cute Japanese CD player. What I like is that it’s pointlessly modular. You can take it apart and use the speaker separately (the cd player then needs to be powered through a usb-cable), it barely makes sense and that makes it an exquisite conversation starter. It’s more of a design object than a high-end audio device, but its ultra-clean aesthetic and small footprint make it attractive.

This device can be wall mounted, or stood up right at an angle with a small Nintendo-switch like stand in the back.

Options:

Bluetooth: ✅
TOSLINK (optical): ❌
External power: ✅
Battery: ✅

3. Shanling EC Zero (AKM) – ~€300–330 #

The Shanling EC Zero is one of the smallest modern portable CD players with audiophile-grade AKM DACs. It offers optical output, battery operation, and compact portability. Praised for its build quality, versatility (it features a standalone cd-ripper and works as external soundcard), and high-quality sound, it is considered the best small CD player for serious portable listening. Tremendous overkill for my usecase.

I think this one looks 80s style funky, especially with the red cd they use in example photos.

Besides the AKM version, Shanling also offers a similar T-model. This has a traditional tube-based amplifier for warmer sound. Even more overkill for my usecase 😅

Options:

Bluetooth: ✅
TOSLINK (optical): ✅
External power: ✅
Battery: ✅

4. Shanling EC Mini – ~€300–350 #

Although the Shanling EC Mini is called mini it is larger than the EC Zero but still compact. It serves as a hybrid CD player, DAC, and digital audio player (DAP), providing multiple outputs and very high sound quality. Portable, battery-powered, and Bluetooth-enabled, it’s one of the most versatile small players on the market.

The aesthetics are a major step down from the Zero: it’s a simple and boring block of aluminum. The advantage is the loading mechanism, cds slide in and the whole thing can (should) be hidden inside a stack of books.

Options:

Bluetooth: ✅
TOSLINK (optical): ✅
External power: ✅
Battery: ✅

5. FiiO DM15 R2R – ~€300–350 #

The FiiO DM15 R2R is small but in-your-face in bright colors. It uses an R-2R DAC design for a more analog-like sound signature.

Options:

Bluetooth: ✅
TOSLINK (optical): ✅
External power: ✅
Battery: ✅

6. Moondrop Discdream 2 – ~€180–220 #

The funky Moondrop Discdream 2 is small and stylish. I like the retro design with greys and the single yellow touch. There’s no screen on it, other than the LED to indicate track number.

Like the Shanlings, this apparatus also doubles as an external soundcard. I had no idea that was still a thing!

Options:

Bluetooth: ✅
TOSLINK (optical): ✅
External power: ✅
Battery: ✅

7. FiiO DM13 BT – ~€130–160 #

The FiiO DM13 BT is FiiO’s entry-level portable CD player. Similar to number 6, the DM15, but digital amps. It’s available in bright colors with a closed top, or with a see-through window so you can see the cd spinning.

Options:

Bluetooth: ✅
TOSLINK (optical): ❌
External power: ✅
Battery: ✅

8. MUJI Wall-Mounted CD Player CPD‑4 – ~€150–280 #

The MUJI CPD‑4 is the original minimalist wall-mounted CD player. Stylish and compact, it also plays MP3/WMA and FM radio. There’s no cover and no buttons. You insert cd and pull the cord.

Connectivity is bad, but I love the design. This was the first cd player I encountered that I thought that was not a traditional black rectangle, and it’s what set off my search.

Options:

Bluetooth: ❌
TOSLINK (optical): ❌
External power: ⚠️ AC only
Battery: ❌

9. SMSL PL200T (CD Transport) – ~€140–180 #

The SMSL PL200T is a compact CD transport designed for external DACs. As a transport it needs to be connected to the Bluesound system through toslink.

This one’s bigger and clunkier than the other options. It looks like an early era StarTrek device. Without a cd playing, I’m sure there will be people questioning its function. It’s a statement.

Options:

Bluetooth: ❌
TOSLINK (optical): ✅
External power: ✅
Battery: ❌

10. Sony D-EJ2000 – ~€200–400 (used) #

The Sony D-EJ2000 is a classic Discman with optical digital output. Classic 90s design with reliable performance, but I’d have to find it second hand.

Options:

Bluetooth: ❌
TOSLINK (optical): ✅
External power: ✅
Battery: ✅

11. Panasonic SL-CT790 – ~€150–300 (used) #

The Panasonic SL-CT790 is another slim vintage Discman.

Options:

Bluetooth: ❌
TOSLINK (optical): ⚠️ region-dependent
External power: ✅
Battery: ✅

Overview of the options #

Player	Bluetooth	TOSLINK	Battery
KM5 CP1 / CP2 (€150–200) - Compact, minimalist, lifestyle design, easy to use	✅	❌	✅
Tinyl SPLIT (3-in-1) (€140–160) - Modular, ultra-compact, Japanese design focus	✅	❌	❌
Shanling EC Zero (€300–330) - Small audiophile player, AKM DAC, portable, optical out	✅	✅	✅
Shanling EC Mini (€300–350) - Hybrid CD/DAP/DAC, portable, versatile, high quality	✅	✅	✅
FiiO DM15 R2R (€300–350) - R-2R DAC, “analog-like” sound, stylish, portable	✅	✅	✅
Moondrop Discdream 2 (€180–220) - Lightweight, modern design, portable, good value	✅	✅	✅
FiiO DM13 BT (€130–160) - Entry-level, compact, Bluetooth, casual listening	✅	❌	✅
MUJI CPD‑4 (€150–280) - Wall-mounted, minimalist, stylish, limited connectivity	❌	❌	❌
SMSL PL200T (€140–180) - Compact CD transport for external DACs, high-quality digital output	❌	✅	❌
Sony D-EJ2000 (€200–400, used) - Classic Discman, optical output, vintage style, portable	❌	✅	✅
Panasonic SL-CT790 (€150–300, used) - Slim vintage Discman, minimalist, portable, optical varies by region	❌	⚠️ region-dependent	✅

Videos #

Reading and photos don’t give a complete overview of _what you’ll experience. Here are three videos highlighting the usecases, design and features of the devices most interesting to me.

The CP1 #

I was enthusiastic about this device. But after seeing the hassle it is to change the cd-cover I’m now hesitant. I honestly do not hink that gimmick will last long for me.

FiiO VS Moondrop #

This video puts the FiiO in a positive light: it’s smaller than the Moondrop and has more features. The amount of connections and buttons is confusing, though.

Multiple players ranked #

This includes the Shanling and the SMSL transport.

Honorable mention: the Burger #

I already know what I will do when the ship of money comes in: buy the Burger!

It’s a foldable, and wireless record player. It’s 100% crazy and cute and beautiful. And it comes in different colors!

Next steps #

As always, ~~the dwarves~~ I delved too greedily and too deep. My research has overwhelmed me with options.

My gut has a preference for a couple of devices, and I will leave this choice simmer in the back of my head for a while. In parallel will come talks with my partner to get her opinion on the matter.

~~more research is necessary~~ To be continued!

Easter puzzle of 2025

Wed, 16 Apr 2025 00:00:00 +0000

It’s that time of year again: I created 21 cryptographic puzzles for you to enjoy!

Play the puzzle now!

These illustrations came out really well

Some tips #

The puzzles are very difficult, so it may take you a long time for each puzzle. That’s by design!
Work together if you can
Print the puzzle (thinking is easier on paper)
Focus on a couple of puzzles and then take ’time off’ to give your mind time to process it
Start ‘outside in’ by writing down words related to the theme and then see if they fit with a puzzle

This year’s puzzle #

I took a theme I’m wholly unfamiliar with: the oceans. 80% of the earth’s surface is covered by water (that’s from memory and may be completely wrong) and for the most part I don’t know what goes on below the waterline.

It was great fun to dive into a new subject that’s out of my depths and see all the creative terminology the experts came up with.

Manual labor #

Another focus point for me was to make the puzzle without the use of generative AI. My goal was to create an authentic, holistic and personal experience. I drew all puzzles on my iPad with an Apple Pencil.

The result is that some drawings look pathetically childish, and others are very nice. So not holistic at all. But maybe the imperfection is the cuteness-factor. I don’t know. I do know I don’t want to redo the illustrations, so what-you-see-is-what-you-get.

I wish you a happy Easter and a lot of fun playing the puzzle!.

Pair-programming with an AI-agent

Fri, 21 Mar 2025 00:00:00 +0000

I moved away from programming because I usually get fed up with it after three days. That’s when frustration sets in, as I get stuck more-and-more. But, I may have found my way out!

The last three months I’ve had fun programming with AI in Cursor. It’s improved significantly over this period, with multiple iterations in its agent-models and interactions. Programming is less “getting stuck” and more “building”.

I should add that I’m not a professional programmer in any shape or form. I program purely for pleasure.

What is Cursor? #

Cursor is a fork of Visual Studio Code (VS Code) with AI integrations. It has some local running capabilities for tab-completions, while larger AI-integrations rely on cloud-integration.

Cursor analyzes you entire codebase, not just the file you are in. It can interact with the terminal, executing commands on its own. It also knows about coding-related tasks such as databases, Docker containers, Git, etc.

The good #

It’s amazing to see your thoughts come to life by programming with Cursor. Just watch your browser do live-reloading as agents set up entire functionalities. Interactivity, database, APIs, unit-tests, documentation, test data… The agent takes a holistic approach and creates functionality end-to-end.

It’s not just programming:

Debugging is much more efficient with an agent, as the chat interface forces you to clearly describe what’s wrong. The distance to code is good in this case, it allows for reflection instead of ‘diving in directly’. Together with the agent one can reason more broadly about the root cause
The agent can also execute terminal commands, so I can do tricky tasks quickly in human language (Git rebases, database re-initialization, renaming functions/variables)
Understanding. I understand the codebase better because of the agent’s explanations. It listens to my suggestions and gives counter-arguments: “why don’t we do this instead?”.

In the best cases, it feels like I’m programming with an overly friendly and articulate developer. Everything is “great!” and explained in long passages.

The bad #

LLM agents love generating. MORE CODE! MORE CHAT! Every edit shows the lines removed and added, and it’s always a net positive. 230 lines removed; 1090 added. This becomes unmangeable in the long run.

The agent randomly generates absurdly complex solutions. Creating custom code for problems simply solved with a standard HTML element (button, dialog), or generating new code instead of reusing existing code. I’ve had plenty of examples where I could reduce code size by a factor of 10 by rewriting it myself.

Agents aren’t neutral. They are eager to overdeliver. You ask for A, and they deliver A+B+Z. You have to pay close attention to ensure they don’t suddenly do things you didn’t ask for. These unexpected additions are buried in long chat responses!

There seems to be a “ghost in the machine.” One agent instance is “better” than another. Sometimes, you have to roll back large chunks of code when an agent gets stuck in its thought process and fails to find a solution. Reset and repeat. A new agent might solve the same problem in a blink. Be on your toes, and commit your code at the right moments.

Setting up a solid project is not something agents excel at. You need a good template to start with. Ultimately, LLM-agents remain predictive algorithms based on input. If you don’t provide the right input, you won’t get a good prediction.

Conclusion: exit developers? #

The generated code is good, but not great. Mindlessly clicking “generate more” doesn’t work; I have to put in the effort. The AI provides suggestions, but I have to stay in control.

But most importantly: software is only partially code. A large part consists of object models, business rules, and interactions. You need to have a clear understanding of these concepts before asking AI to take action.

It seems the experts are safe for now.

Beware the dangers of easy-life #

Working with Cursor is fun and easy. But “fun and easy” comes with a trade-off. At first, I overestimated Cursor and spent too much time “chatting and generating.” Suddenly, hours had passed when I could have solved the problem myself in 30 minutes. Easy choices now, hard times later!

Hubspot ➡️ Bear, contacts, Todoist

Mon, 10 Mar 2025 00:00:00 +0000

I’m moving away from American services in favor of European ones. This time I remove Hubspot from my stack.

I’m a reluctant Hubspot user. I’m not particularly fond of the app: aggressive pricing, bloated features, clunky interface. I knew this going in, why did I chose it still?

Hubspot is overly complete and gave me the idea I could grow in my marketing/sales journey. A sign-up form for collecting email-addresses for a newsletter, integrations with other tools, connection with my inbox, analytics, tracking sales.

Sound good, but I don’t do any sales and marketing. I dislike tracking people (my websites have no analytics). I don’t have the stamina to maintain a steady newsletter.

Really, I don’t need Hubspot.

What do I need? – Requirements #

If I don’t need Hubspot, why did I get started with it then?

I use Hubspot for two things:

keeping notes on conversations with ‘prospects’: in my case usually recruiters or clients
tracking ‘deals’: potential projects

I then proceed to ignore this information.

I realize I get more benefit from processing the information in a structured way, and not so much the tracking or re-reading the information.

Hubspot can do this, but the interface for writing notes is very clunky. It’s a tiny edit-field that feels like an after-thought. Tracking deals is a lot better with its Trello-like board, although that can be replicate in many other tools. On top of this, Hubspot contains a duplication of my contacts-information (it’s also in CardDAV in Fastmail, which syncs to my phone and computer). Maintaining my contacts in two places feels so-2020.

My main advantage is being a solo-entrepreneur. I do not need to work with other people, and I can align a lot of apps to work with my process.

My requirements are:

Single contact-list synced through CardDAV
Multiple notes per contact
Track ‘deals’ and be reminded to follow-up

I have just 30 contacts in Hubspot and 27 ‘deals’. I don’t need a complicated system.

My options #

There are tons of non-USA CRM systems that fit my basic requirements. Even ‘famous’ Dutch players (🇳🇱) AFAS and Exact, and slightly less famous Simplicate, Perfectview and Alexion. I used Brevo (French 🇫🇷) for a while, a straight-up Hubspot replica, and was equally overwhelmed as I was with Hubspot. I have a similar feeling with Odoo (Belgian 🇧🇪). I appreciate the look of the native OSX Daylite-app (Canada 🇨🇦). The downside of all these great tools is that they are overly feature-complete for me, and I end up paying 15-30 euro per month for stuff I don’t need.

Another option is to self-host with Nextcloud (German 🇩🇪), or go for one of the many opensource options.

Decision made #

I reflected on my simple requirements, and I realize I do not need a complicated CRM. I have excellent European note-taking software already (Bear, Italian 🇮🇹), a wonderful European app for tracking my tasks (Todoist, Spanish 🇪🇸), and I use my operating system’s contact-apps which syncs with Fastmail (Australian, 🇦🇺).

My new setup:

basic contact-information goes in the Contacts-app
add a callback-url to a Bear-note if I need to write down information on the contact
track my ‘deals’ in Todoist, with a callback-url to Bear for detailed notes

This might seem like a huge step down from the automated and integrated approach of Hubspot. Instead it feels like a liberation:

best-in-market note-taking and tasks
effortless syncing between laptop and phone
integration with my other workflows: I already work with Bear and Todoist
effortless switching if I’m ever unhappy again (contacts through CardDAV and notes through Markdown-export)
a more human approach: I spend conscious time with my contacts

Why not Obsidian/Roam or Notion or a database? I think I mostly like the current approach as it stays very close to my current modus-operandi. Hoping (fingers-crossed) this will make it easier for me to actually do something with my contacts.

It took me 4 hours of manual transferring contacts, during which I learned a lot about y’all.

I’m happy to report that most of the PII information entrusted to me is no longer in the USA. The only exceptions are iCloud&Photos, and information shared through LinkedIn/Whatsapp/Slack. The latter is in your own control!

as always, if you like to be removed from my systems, just let me know!

Review of auto-auto-bio app

Sat, 08 Mar 2025 00:00:00 +0000

I’m writing speculative futures! I’m imagining future ‘news clippings’ from newspapers or other media. The goal is to be neutral: not overly optimistic, not Black Mirror.

This time the future sent us an app review for the app “auto-auto-bio”, a tool that writes an autobiography for you:

⭐️ one star review: Severe disappointment #

This app promises an exciting look at your life, a thrilling autobiography in the style of Walter Isaacson. Tales of brilliance and perseverance like DaVinci or Steve Jobs.

Full of enthusiasm I gave the app access to my email, calendar, photos, all my life’s data. I did the five one hour interviews with the AI-agent, and my friends gave approval to talk to their AI-agents. That part was smooth and painless.

And sure enough: I got a beautiful book in the mail two days later. It looks amazing: the photography, typesetting, top notch.

But the content!! Big disappointment. I got bored after reading the first two chapters. The writing is like Isaacson, so that’s good, but apparently my life is too plain and unexciting. This was a huge confrontational blow.

Apparently I lack any notable achievements, in work and in private. Nothing exceptional. And little respect from my children, colleagues, friends and family. And all this in beautiful prose. It’s a high-class insult.

I ordered 20 copies, and none of my friends and family are interested. Even my parents can’t bring up the decency or motivation to read it.

This app ruined my life.

If anyone is looking for me, I’m in the corner, crying and reflecting on my worthless life. Thank you auto-auto-bio.

Apple Pay ➡️ Back to plastic

Tue, 04 Mar 2025 00:00:00 +0000

I’m moving away from American services in favor of European ones. After WhatsApp and Apple Maps, Meet and mail, I’m now trying to limit my Apple addiction with an admittedly small step.

As soon as it was available, I enabled Apple Pay so I could pay everywhere with my phone. It’s a simple quality-of-life thing.

But, in my journey to limit my reliance on American tech I’ve decided to use my physical debit-card again in stores - or even paying with cash again.

Baby steps.

Harvest ➡️ Timemator ➡️ Toggl ➡️ Harvest

Mon, 03 Mar 2025 00:00:00 +0000

I’m moving away from American services in favor of European ones. This time I tried to switch from Harvest, my time-tracking software, but I couldn’t find a reliable alternative.

The first time I used Harvest was 12 years ago, when I started working at Angi Studio. I had a love/hate relationship with time-tracking, but it was never because of the software (my problem was forgetting to start timers, and unclarity on what to write as working hours). Harvest simply works, which is why I chose it again when I started freelancing.

Besides time-tracking, Harvest offers forecasting, planning, budgeting, reporting, expenses and invoicing. I’m using Moneybird for expenses and invoicing, and haven’t used the forecasting extensively. This makes switching easy for me: I only require time tracking.

Harvest offers nice integrations and a powerful API (which I used a couple of weeks ago to make a target-dashboard), these extensions are a nice bonus.

The options #

My main focus was finding an app in Europe that is single-pay. The one thing I didn’t like about Harvest were the recurring costs which total 185 euro per year.

A quick search found me Timemator: one-time payment, built in Germany 🇩🇪. Data is stored locally and synced through iCloud (optionally). There is an iPhone app, which is rated quite high.

I also considered Timebird (built in The Netherlands 🇳🇱) as it integrates with Moneybird, but Timemator is easily more feature-complete.

Timemator #

I installed the app, set it up, worked with it. Was great. Wrote a review. And then it lost my notes.

There’s one thing apps shouldn’t do, and that’s lose user data. Sadly, I’m not the only one and it hasn’t been fixed for three months already.

Sadly, Timemator was out.

Toggl #

As the single-pay option didn’t work, I searched for alternative and found Toggl (🇪🇪 Estonia). This is one of the more famous time-tracking apps, and it indeed looks beautiful.

I had the most trouble getting my Harvest data into Toggl, and had to resort to ChatGPT to format the CSV-data for me. Toggl needs certain headers, toggl needs time to be just so, toggl needs an emailaddress, toggl needs… Ugh. Can we talk about what I need?

Onboarding shouldn’t be this hard.

In the end I managed to import everything. Toggl shows my tracked time. Yay.

But now comes the second round of setting up Toggl, which is my billable rates. I like to see cold-hard-cash instead of hours, thank you very much.

I got stuck in setting these rates. My rates differ per project and sometimes within projects (different months).

Here comes the snag about Toggl’s data-model, which is different from my mental model. In Toggl you cannot set tasks as billable, but instead have to set individual time entries as billable. I now need to go through all timesheets and set them as billable. I guess I should’ve done this on import?

I have four options now:

Set time entries to be billable through Toggl. But I can do this only in batches of 50 entries, because Toggl only allows you to make changes for entries on the page instead of “all”
Erase all data, update the CSV with billable rate (it is exported by Harvest), re-import
Screw previous data, start fresh in Toggl
Stop the Toggl-experiment all-together

I’ve decided on the last one. Sorry, this experiment is taking me too much time (I can see how much in Harvest), with setting up Timemator first and then Toggl…

Conclusion #

Toggl is probably a good European alternative to Harvest. But I have other work to do instead of trying it out 😃

Future work #

Toggl has over 5 million users if their marketing material is to be believed. Immediately this triggered my “how hard can it be to create my own”-feeling. If I can make 5000 people switch… 😉

Gmail and Gcal ➡️ Fastmail

Sun, 02 Mar 2025 00:00:00 +0000

I’m moving away from American services in favor of European ones. It’s time to move from Google’s mail and calendar.

I have been using Google mail since 2004. 20 years! And now it’s time to move.

Requirements #

As I have my whole life’s history in Gmail, I’d like to transfer my email instead of starting fresh
My current inbox has 13GB of data, so I need to be ready for the next 20 years. A minimum of 20GB data or the option to grow
I’m using both email and calendar from Google. But not photos or drive
Email needs to be secure and simple. And as you’ll see, its hard to meet these two guidelines
I’d like to setup multiple domain names to the email (either in one inbox, or multiple inboxes) and use mail rules to filter

Competitors #

The possible competitors are Proton (Swiss 🇨🇭), Startmail (Dutch 🇳🇱), Fastmail (Australian 🇦🇺, I know, not European), self hosted (with or without Nextcloud), Mailbox (German 🇩🇪) or Infomaniak (Swiss 🇨🇭).

I have been eyeing Proton for a long time. They combine a total commitment to privacy (with PGP) with user-friendliness. They have a complete offering, with calendar and drive for a fair price. But what is holding me back is that Proton is a walled garden, which makes sense from the encryption point-of-view. I have to use their email and their calendar app. Yes, it’s secure, but… I like the idea of being able to switch: from Apple to Linux, and so on. That’s not possible with Proton.

Startmail has no calendar-offering, but “it’s coming”. I don’t have time to wait for this.

Mailbox is a typical German offering. Overall it looks very decent and strong, but clunky. Transferring email from Gmail is possible, but uses a separate service called Audriga and it feels like a hassle. Mailbox offers other things like conferencing and drive, and Twitter-feeds and RSS-feeds and todo-apps and… It feels like I will need weeks to work out all the kinks with this tooling. I just need a calendar and email, okay you guys?

So I have to do how many steps? And then it still can fail? And all mail from the Archive is not exported? – Doesn’t feel trustworthy Mailbox.org!

I decided to take a look at my “Archived mail” and I wouldn’t be missing much¹.

Infomaniak has an extensive offering. Email, VPS, domain names, document-editing; You name it, they have it. Similarly to Mailbox the offering feels overwhelming. It looks more polished, but… THE NAME! I mean, a rose would smell just as sweet, yada-yada, but how can I trust a company that names itself Maniak? It also names everything with a k: “kmail”, “kmeet”, “kchat”. I know Apple did it with an “i” and Google with the “g”, but it feels cheap. Plus, I get the feeling they are stretching themselves too thin. I rather have a company that does one thing perfectly.

Self hosting (either with or without NextCloud). I love self-hosting of my web content. I’m free from external services. But, self hosting email is another thing. I don’t want to deal with SPIF and DMARC and other technical things. I want my email to work reliable without doing anything. I trust my hosting provider can do this reliably, but I don’t want the hassle.

Fastmail is an Australian operator. Compared to the European offerings this one makes a lot of sense. It’s simple, feels trustworthy and feature-complete. No additional unnecessary cruft, just loads of storage for email and calendars. Easy migration from Google and Workspace. One click and done.

Choice made: Fastmail #

I prefer European services, but I’m not exclusive. The main goal was migrating away from Google. I chose comfort and ease of use over a complex offering, and decided on Fastmail.

Migration was painless and easy. I was expecting Fastmail to take days to import my 107.196 emails from Gmail, but it only took a moment. I don’t even know how long, but less than a couple of hours.

I also imported my work-email and iCloud. Everything is now in one account.

Migrating log-ins (OpenID) #

I can switch e-mail and calendar very easily. But that’s not the only things I used from Google. I also used their “log in with Google account” OpenID functionality for services instead of using a username/password combination.

This one is a PITA. I migrated all my work-logins (as I only had a handful), but migrating all my private logins will take a loooooong time.

Deleting Google Workspace Account #

It’s much easier to create a Google Account than to get rid of one. I had to dig deep into Google Settings and even in the Google Cloud Console to quit an Apps script that was running.

The account is not canceled yet: it takes 30 days for projects to fully delete…

Ah well, 30 more days and my Workspace account is gone.

I’m keeping my personal Gmail account so I can have it forward email until everybody knows I am now using a new email-address.

New mail, new rules #

Now that most of the migration is done I am setting up new rules for my inboxes. I’d like to keep my mailbox fresh and structured this time.

Everything a password. No more OpenID nonsense. I have a password manager and I’m not afraid to use it
You won’t get my private email-address. I have a private email-address and only friends will get it
Business partners get a business email-address.
Your app won’t get my url. So nice of you to have an app that requires an email-address. But your app will get an app-specific email-address from a domain-name I own, and a fake name.

After sorting through the past weeks of mail while setting up my mail-rules, I realize email is more for machines than humans. It’s a low-level notification api; a way for the computer systems to talk to their users. A sad realization, human communication has moved to walled off systems such as WhatsApp, Facebook, LinkedIn, and others.

Looking in Gmail’s archive is not simple. You need the cryptic -in:Sent -in:Draft -in:Inbox has:nouserlabels to find emails exclusively in the Archive. There were 6,625 emails in there but I merely archived spam and other useless stuff. Oh, and all old google-talk chats are in there. Not a big deal. ↩︎

Apple maps ➡️ TomTom Amigo

Thu, 27 Feb 2025 00:00:00 +0000

I’m moving away from American services in favor of European ones. In this article I’m returning to an old favorite: I’ve dusted off TomTom in favor of Apple Maps.

I didn’t realize, but the navigation market is booming. The app stores offer an overload of mapping and navigation apps!

Here in The Netherlands I can safely say we all know Waze, Google Maps, and Apple Maps. But other options include Here (of Nokia fame, now German 🇩🇪), Magic Earth (German 🇩🇪), TomTom (Dutch 🇳🇱) and Sygic (UK 🇬🇧)]. I don’t have time to try them all, and I chose TomTom out of brand familiarity and Dutch nationalism 🌷.

TomTom apps versus Apple Maps #

Two years ago I switched from Waze and Google Maps to Apple Maps because of privacy concerns and a hatred of advertising (Waze is spectacularly horrible with its aggressive and intrusive ads). For this reason I’ll mainly compare TomTom to Apple.

The first thing we have to do when switching to TomTom is figure out their offering. They have two similar apps and I had a hard time understanding the difference.

This is my finding:

TomTom Amigo is most like Apple/Google/Waze: online-only and free. It crowd-sources accidents/events on the road in a similar manner as Waze and Apple.
TomTom Go is the pro offering. It works with offline maps that are updated regularly. It additionally has lane-guidance (“stay in the second lane from the left”). It costs 20 euro per year.

Both have traffic, speed cameras, live rerouting, places (restaurants, gas stations, parking), CarPlay, no advertising, strong privacy. Strangely, the paid TomTom Go seems to miss calendar integration that Amigo does have.

Is it any good? #

I’ve used Amigo for short routes now and the navigation is decent. In my experience it is as reliable as Apple, with the only major downside being the lack of lane guidance. I’m comfortable with an app that says “turn right”, but it would be better if it said “stay in the second lane from the right”.

As you might expect, the app is not as userfriendly and polished as Apple’s offering. The colors and animation feel rougher. The guiding voice is robotic, both in sound and directions. Apple sometimes gives instructions in a human way, such as “at the next traffic light” - making it feel as though a buddy is giving you directions instead of Commander Data.

Amigo is 100% focused on navigation by car. No walking or cycling routes, no “exploring places near me”. Then again, Apple was also not very good at all this compared to Google Maps. Another missing feature is Streetview.

Is it all bad?

Amigo lacks killer-features to set it apart from Apples navigation, but I do like that it shows me both my driving speed and maximum speed in one overview. Apple is not very reliable in maximum speed indication, and the lack of ‘my current speed’ in Apple’s UI is annoying.

Conclusion #

My conclusion is that TomTom is clunky but reliable. It will get my car and me to where I want.

I cannot fully switch from Apple Maps to TomTom, so will revert back to Apple for other map-activities (such as holiday planning and walking). I might try out Here or Magic Earth in the future.

update I missed an exit due to missing lane-assist in TomTom. I find myself reverting back to Apple Maps more and more. Perhaps a good time to try Here or Magic Earth 😅

Google Meet ➡️ kMeet?

Thu, 27 Feb 2025 00:00:00 +0000

I’m moving away from American services in favor of European ones. In this article I’m exploring videoconferencing tools.

Before I became a happy Google Workspace user, I tested the offerings of both Microsoft and Apple. The Apple offering was simple and included in my iCloud-subscription, but only offers videomeeting through Facetime. Microsofts proposition is extensive and confusing: it comes with Office included (or not, or wait, yes, or… ehm only cloud I think), loads of space (except for email) and a weird selection of other cloud-products. All of it felt clunky.

My office-stack now looks as follows:

Fastmail for email and calendar
iCloud for online storage (yes, that too will need to move one day)
Offline apps for text-editing and Excel-work
No app for videomeetings

I’m not a regular video-meeter, but I do like to have the option for client calls or interviews. For the latter it would be great to be able to record so I can transcribe it easier.

Requirements #

The requirements for virtual video calls are straight forward:

it should work reliable
the app should have a professional and trustworthy design. The video meet is in some cases my first introduction to clients
support multiple people
participants can share screen
participants haren’t hassled with unnecessary steps to participate: no app or plugin necessary, without registration
I should be able to record a session
I can generate a specific url for calendar invites
there is a chat with the meeting, as I share urls to a Miroboard in a call
it would be nice to have a booking page
it would be nice to have a desktop/mobile app: with google meet I get tired of the “enable camera”-dance

The contenders #

Oof! Videomeeting-space is a clusterfrack. Propositions are all over the place, with complicated pricing structures and unnecessary features.

The strongest European contenders are:

Jitsy (German 🇩🇪): free, opensource and secure. Feature complete and I’ve used it in large groups reliably. But no recording unless you add a bot, which quickly complicates matters. There are hosted services for Jitsy, like Fairmeeting (which asks all participants for donations, LOL)
NextCloud Talk (German 🇩🇪): NextCloud is a ridiculously complete private server solution. Offering document editing, drive, games (yes), photo-sync, AI-assistants and more. According to the NextCloud documentation you can record talks, but I’ve seen mentions not all servers support this. It also includes an email and calendar client, but no email or calendar hosting. Free when self-hosting.
Infomaniak (Switzerland 🇨🇭): complete business suite including email hosting, drive, document editing. All free. You can even get domain, hosting, VPS and loads more from them. But the name, would you trust a link that says “Infomaniak”? Luckily we can rebrand it for 50 euro per year to our own URLs.
Vectera / Teamleader (Belgium 🇧🇪): has a free tier without recording, but with scheduling. Up to 10 people is 22 euro p/m.
Whereby (Norway 🇳🇴): reliable and simple. I’ve used it in the past. The main selling point of Whereby is reusable rooms (“standup.whereby.com”), which is simple to add to a meeting. But the downside of this is that others can simply guess your meeting-location. Although they cannot join automatically, it feels weird. 10 euro p/m.
Opentalk (Germany 🇩🇪): has a good look and feel. Is from the same people who make mailbox.org. You can get Opentalk as a separate product for 7,50 euro p/m, or as part of Mailbox (for 3 euro p/m, I have no clue how that pricing works). Recording meetings is “planned” or “rolled out” depending on the page you visit…

There are more ‘videoconferencing’ tools made in Europe, but they are targeted more towards large web conferences: DigitalSamba (Spain 🇪🇸 starting at 99 USD/month), ClickMeeting (Poland 🇵🇱 starting at 23 euro/month) and Livestorm (France 🇫🇷 starting at 80 euro/month).

Out of the options listed above, three fit the bill: Whereby, Opentalk and Infomaniak (yuk, that name). Others are too expensive, or require a separate server (which I’d have to find or setup).

I decided to give InfoManiak a go.

Trying Infomaniak kMeet #

Registration requires an address and a phone number (which will need to be confirmed) and you need to validate your location through a separate app. Loads of hurdles.

I then entered into a dashboard with lots of new options. I can buy domain names, get a VPS… Dizzying. It really is an all-in-one solution.

The kMeet tool is built on top of Jitsi and works well when testing it by myself. I can record conversations and there is a desktop app (a simple Chromium wrapper). I can plan meetings with a kMeet-link, but only for calendars managed by ‘kCalendar’.

Colors and logo are customizable, meaning I can get rid of the silly infomaniak naming. Even at my domain name: meet.matthijszwinderman.nl (or maybe kmeet…).

With a name like Infomaniak kMeet I expected the worst. Instead, it works okay. Who knows? I might even decide to transfer my email and calendar one day.

Trying OpenTalk #

I briefly tried Opentalk as well. It has a nicer, friendlier feel than kMeet. But German-clunkiness around some corners, like the meeting-planner, or the sign-up:

I confirm that I am not acting as a consumer, but as an entrepreneur or for a corporation or other organization in the course of its statutory activities

Sigh. That sounds like the most German thing ever.

Conclusion #

I don’t make that many video-meetings, I mainly use the tools of the clients.

For now the decision can be pushed forwards, but when the time arises, I’ll try kMeet.

Whatsapp ➡️ Threema

Wed, 26 Feb 2025 00:00:00 +0000

I’m moving away from American services in favor of European ones. Last week was a huge frenzy in which I tried out multiple non-American services. First step is Whatsapp, which I’m leaving for Threema.

I’m using Whatsapp for my communication with many people. Friends, family, old acquaintances. Plus, it’s used by businesses and the school my kids go to. There is no illusion from my side that I can convince all of them to switch to another app. Luckily, I don’t have to.

Most of my communication happens with a handful of people. I haven’t quantified this, but I think 5 people account for 80%+ of all messages. Especially my wife – our communication easily accounts for 50% of all messages.

Requirements and options #

I am looking for a WhatsApp-replacement that fits these basic requirements:

can reliably send and receive messages
has media sharing (photo and video)
has group chat
is available on Android and iOS
messages are sent encrypted
can make video calls
is made in Europe 🇪🇺

This limits my choices to: Threema (Swiss 🇨🇭), Wire (German 🇩🇪), Olvid (French 🇫🇷), Element (UK 🇬🇧) Session (Australian / Swiss 🇦🇺/🇨🇭). You’ll notice these are not European Union made, but Europe-as-a-continent. That’s a conscious decision, but I do prefer Europe/Swiss if possible.

After a quick initial selection round, both Olvid and Element fell off the boat. Olvid has a paid subscription of 6 euro per month to use video calling: I cannot convince anyone of paying that price, and there’s no point in having a messaging app if no-one else is on it. Element needs you to set up a Matrix server or find one that fits, it’s too much hassle.

Threema, Wire or Session? #

The remaining choices are Threema, Wire and Session.

In the app-stores only Threema gets a decent amount of stars:

Threema: 4,3 ⭐️: nice UI, but sometimes slow server and notifications. Calls don’t always ring.
Session: 3,4-3,7: slow network, notifications not always working
Wire: 3,0-3,4: slow in sending, unreliable in messages received, scammers

I decided to skip Wire based on the reviews, and focus on Threema and Session.

Threema #

Threema has a friendly and polished UI, and works great in messages and videocalling. Those are the absolute basics and they work very well.

Not so good is the onboarding setup. I have to do weird steps like ‘move my finger’ (I know why, I’ve generated RSA-keys in the past, but still), get a cryptic Threema-ID to share with friends [^I do not have to use the Threema-ID to connect to friends. I can also choose to send a hashed version of my address book to Threema. I did this for Meta (who I do not trust), but initially refused for Threema (who I trust more). Why? Because of the scary text that went beside it. After a week I did upload my (hashed) address book and found exactly 1 other person.], have to store passwords… None of that was necessary for Whatsapp

Because Threema prefers privacy over usability, the default-Apple-share-menu and ‘store photos in library’ have to be enabled manually.

Threema relies on in-person verification. That’s where the ‘Three’ in the name comes from: you can be unverified (one red ‘ball’), verified by email or phone (two ‘balls’) or verified by scanning QR-code (three ‘balls’). I get it. But the prominence in the app is a bit annoying. I’m testing it out with a friend who is 5-hours of flight away from me, he’ll have the “one red ball of shame” for a long time.

A lot of the in-app copy is techy: “perfect forward secrecy-session is reset”, and “action required: to avoid missing messages, please complete the update by opening the app”. I have no clue what this means and I do not want to know. Definitely can be more user-friendly by hiring a UX-copywriter.

I like that I can request Threema to send photos in ‘original resolution’, but this still compresses the images and video. I prefer it if my photos are sent (and received) as the original files. That is a privacy hassle for Threema, because of metadata like geolocation. But it would be hugely beneficial for me, as this is the main way I share photos with my wife. Still, Threema delivers a much higher quality than Whatsapp’s “HD”.

The major downside to Threema for adoption is the pricing: a one-time fee of 6 euro. That will be a tough sell for most of my contacts. On the other hand: I prefer to pay for my software instead of being sold myself.

Session #

Session has an even more tech-feel to it. It works on top of a distributed network, and that sometimes shows in the UI.

The onboarding was smoother than Threema. But the unique ID I get is more complicated, instead of Threema’s 8 random characters, I get a long hash-number. That’s just unacceptable.

As messages on Session are routed around a distributed network, they indicate ‘being sent’ much longer than Threema or Whatsapp. Furthermore, I’ve seen the messages ‘could not be sent’ a couple of times. This gets corrected, but it gives off an unreliable feel.

Just like Threema, Session compresses your media before sending it. I cannot adjust the compression, but it seems to be a higher quality than WhatsApps’ HD and lower than Threema’s highest setting.

Session has a bit more focus on in-app privacy. Its screen goes black as soon as you switch apps (like banking apps do) and it notifies people in the chat if you make screenshots. Both these features are irrelevant for me.

Functional differences between Threema and Session #

Both Threema and Session have the same basic functionality. At least the requirements I set in the beginning are all there.

The main differences from the point of view of user-functionality:

Session offers sending GIFs: I am going to miss this in Threema…
Threema offers sending location, although not with continuous updates: very useful, but I can use other apps for this as well

Video calling in Session is in ‘beta’. I couldn’t get it to work at first, and in other occasions it was buggy.

Video calling in Threema is as expected from a modern app. It just works.

Conclusion #

I have tried Threema now for over a week and did multiple video calls. So far so good. I also prefer Threema’s polished look over Session.

My wife and I are now exclusively “apping” in Threema.

Turning down Trump

Mon, 17 Feb 2025 00:00:00 +0000

Trump’s USA is quickly turning the page on decency, friendship, and sensible capitalism. There’s not a lot any individual can do about this, but there is something we can do, we can stop reliance on American Tech. One step at a time.

Today my American friends celebrate Presidents Day. Although celebrate is perhaps not the right word. I am sure that for almost half of the Americans this day has a very bitter aftertaste.

And for us Europeans too. Because although I still see the US and its inhabitants as a friend, it is clear that the current regime is not. In recent weeks, Trump, Vance, Musk and others have clearly shown that Europe has been degraded from good friend to “difficult acquaintance”.

That relationship is lopsided. On the one hand, Europe must “take care of itself and stand up for itself”, on the other hand we get political interference and the demand that we start restricting our own laws and regulations. “Please order more gas and weapons”, but also threats of sanctions and import duties.

The USA has turned into a bully.

I do not participate in that. And even though my impact is small, I can do something.

I have happily used American services for many years, and I prefer to continue doing so. Alas, this is not the time to be lazy and sit back. I am going to help Europe “stand up for itself”, and keep my money in Europe.

This week is Trumpxit (the X is pronounced as “sh”).

Bye Google, Cloudflare, Harvest, Apple Pay, Notion, Waze, Whatsapp. Thanks for everything. Fortunately, there is a European alternative for almost everything.

And to my American friends: enjoy your day off. And hopefully in four years we will have a president you can be proud of.

I found a handy list of alternatives on Tweakers

Yearly review 2025

Sun, 09 Feb 2025 00:00:00 +0000

I look back at a busy year. A first year of freelancing. A year with illness. And with many unfinished projects. But also a year with a lot of fun, feeling better emotionally and enjoying the freedom, responsibility and flexibility of freelancing.

A busy year with loads of stuff

In my head I already had the theme and leading paragraph for this year’s review ready. The theme was going to be “unfinished business” and the leading paragraph would mention “Zeigarnik related anxiety” (the nagging weight of unfinished tasks).

But then I re-read my 2024 review, which I’ll screenshot here for the convenience of my three readers:

That’s what I wrote in 2023 and apparently in 2022 as well. Temet Nosce.

Unfinished is a recurring theme. 2023, 2022, 2021… Every year I write about not having enough time to do all my projects. Which begs the question: am I so bad in planning and finishing, do I simply have too many projects, or is it really outside factors?

Which reminds me of the song “In my Mind” by Amanda Palmer:

And in my mind, in the faraway here and now
I’ve become in control somehow
[…]
And it’s funny how I imagined
That I would be that person now
But it does not seem to have happened
[…]
And when they put me in the ground, I’ll start
Pounding the lid
Saying I haven’t finished yet
I still have a tattoo to get
That says I’m living in the moment
And it’s funny how I imagined that I could win this, win this fight

– Amanda Palmer, In my mind

A reassuring thought. I’m not alone. Everyone has unfinished projects. Everyone thinks they can finish faster than they can. The struggle is real (also close at home–the neighbors have worked on their construction projects for over 8 months, filling the air with constant noise of drilling and sanding).

But quoting lyrics and commiserating with others doesn’t get projects finished. I need to do something, and that’s probably going to need other people get involved to hold me accountable for my planning (i.e. my partner).

Aside from the unfinished personal projects and the illnesses in my family (and The Stuff happening worldwide), overall I’m looking back at a great 2024.

Wins of 2024 #

Like every year, this review is mainly a brag-list to make myself feel good. And that works 🤗

personal growth: this year I felt happier, calmer and more in control. Less emotional outbursts, more understanding.
freelancing is great: I had four different clients and eight projects. I love the freedom I have, that I’m seen as an expert by my clients, that I really help them forward, the diversity in my activities, and that projects are contained with a start- and finish. I reached my financial goal
professional development: I visited multiple work-events, and scheduled “let’s get some coffee”-time with people
reading: I read books! Hooray! I finished a couple of science-fiction classics and business books
blogging: I wrote many articles, a total of 15.545 words. I moved from Jekyll to Hugo. And I rewrote most of the code from the template I used as starter. Pfew. A lot of work.
music: we bought a stereo set! And it’s incredible what it did to our house. We listen so much more music and we really listen to the music. We sit on the couch and listen. It’s great.
more music: I stared piano lessons. A lifelong dream
helping out at school: loads of projects at my children’s school, it’s fun to see more of their world and contribute to it
exploring the world: holiday to Portugal, visiting museums, holidays in The Netherlands, visiting Efteling, visiting my brothers’ family in Germany
celebrations: we celebrated Yalda extensively, I organized fun birthdays for my wife&kids, I had a great birthday organized by my wife, and had a warm Christmas with the whole extended family

Struggles #

Life and lemons 🍋

the book pile: I buy more books than I read. It’s a problem.
project paralysis: I’m starting more projects than I finish. It’s a problem
creative chaos: I have about 50 blurbs for blogposts, all over the place. In Bear, in iA Writer, in Todoist, as ‘draft’ in Hugo. It’s a problem (I see a pattern now…)
disappointment over cancellations: due to Stuff I had to cancel visiting conferences, meet-ups and social gatherings. This bummed me out more than it maybe should have

Work highlights #

2024 was a busy year. I landed two new clients, both startups, and I worked for my existing two clients.

I designed a mobile app, rebuilt an existing concept from the ground up, did user research (with lawyers, tax-professionals, energy-companies and goat- and sheepfarmers), I deepdived into different domains such as Kubernetes, I used large-language models to analyze 1.000+ jurisprudence documents, strategy, roadmapping, opportunity-solution-tree, testing the limits of Miro with huge boards, gave training on innovation and warm-data labs, designed (and partially built) an interactive data-dashboard, product-market-fit workshops, wire framing, prototyping… Jack of all trades 😅

House #

The house, as always, demanded attention:

Our heating broke, we had leaks and electrical issues. All fixed 💪
I plowed and reseeded the grass in our yard (failed, any tips?)
Endless maintenance. Cleaning gutters, washing windows, fixing bikes, car troubles, cleaning shed, cleaning basement
Interior upgrades: bought new cupboards, working with an interior designer, painting outside furniture and inside window frames
A year-long battle with internet providers. Ziggo screwed stuff up for a year, so we switched to Odido. And they promptly started messing stuff up also. Long story. Short version: it’s fixed now

Projects #

I started out with the projects I didn’t finish. But surely, I finished something?

I moved this blog to Hugo, updated all my articles, added new code for series (see below!)
I published my yearly easter puzzle. This year it got picked up by the local radio, who interviewed me for 5 minutes
I bought a home-server and set it up with Plex, n8n and other stuff
I created a website for my father: www.theozwinderman.nl
I cleaned up my photo-library for 2024
I experimented with n8n and AI
I published loads of blogposts
I got a cool clock based on a 1001-post I wrote years ago

Next year #

2025 is already happening. Already an eight of it is gone.

Based on the experience of never finishing anything, I’ve decide to set priorities on two projects:

Penny: a thought-management tool. Finally publish it this year
1001ideas: my ideas-blog needs to move to Hugo so it works again

But above all, I hope to maintain the good vibes, reading, vacations and family-time.

See you next year.

A highschool-level summary to Kubernetes

Wed, 22 Jan 2025 00:00:00 +0000

For a client I had to learn about Kubernetes. I found the excellent O’Reilly book “Kubernetes: up & running” and made a quick summary. I hope it’s useful for others as well.

The basics #

Kubernetes is also called k8s. It’s a way to create an infrastructure that is distributed (consists of many parts), reliable (should work consistently even if a part fails or is updated), scalable (can grow big).

Kubernetes objects are immutable: once on artifact is created it doesn’t change via user modification. To ‘update’ an object you replace it with a new version. This is different from mutable, where you create and then change the artefact to update (install new stuff, run updates, etc). The key benefit of immutable objects is that you have a record how it was created: you can create a new object with the same state incredibly easily.

You should not log into k8s objects to fix stuff. That’s an anti-pattern

Kubernetes is declarative. You describe the desired state of the system, and Kubernetes will fix it for you. This is different from imperative, where you give a sequence of instructions.

Kubernetes is selfhealing. Kubernetes tries to fix the system continuously to match the declarative state (if an object gets stuck, Kubernetes destroys it and creates a new version).

Objects in Kubernetes are decoupled (running independent from each other, micro-services). They have APIs and there are load balancers. To scale a service you just add more instances called replicas (by updating declaration, or letting k8s do it: autoscaling)

Primary objects #

Cluster: collection of nodes with pods with containers. Contains at least 1 worker node and 1 control plane (managing the cluster’s state, API-server, scheduler, CoreDNS service, discovery, and the scheduling of workloads – previously called master node). Kubernetes tries not to run stuff on the control plane.

Node: the thing that runs the pods. Can be a virtual machine, a real machine, a server in cloud. This has own OS&processor&disk&memory&Docker-instance. Nodes have: a name, status (ready, master), age (runtime) and version.

Pods: a group of containers that is a single deployable unit. Also does some DNS to handle traffic. All containers in a pod get to go on the same node (execution environment) and can talk to eachother through localhost. Different pods on same node might as well be on different servers (they communicate through the Kubernetes network model). Containers in a pod can mount the same volume. A pod can define what it needs (this much memory, a gpu, etc) but shouldn’t really care what type of node it runs on. (this is more difficult with LLM and genAI where you need more control over host machine).

Container: is a Docker image. Used to be system-containers (mimics a VM, has ssh, cron, etc) but now prefer application containers that run a singe program. That’s easier for setting budgets and creating pods. Btw: if your container goes over resource-budget, it gets killed and restarted. Data isn’t safe within the container (use persistent storage for that).

Job: creates temporary pods that are destroyed after a succesful termination. Can be part of a ‘work queue’ from a worker-job. Can be a cron-job.

Metadata #

Namespace: for pods (containers get same namespace as pod), services and deployments. Used for isolation and access control (which users&microservices can interact). Nodes cannot have namespaces. Namespaces cannot be nested (but pods can have identical names if they are in separate namespaces, so you can use in API: namespaces/default/pods/podname)

Label: assigned to pods (not containers). Used to filter/select/target pods. Consists of key/value pair: ~[acme.com/app-version](http://acme.com/app-version)~: 1.0.0 or env:prod or ver:2. With this you can do stuff like update all production pods of app blabla with version <2. Nodes can also have labels (gpu=yes).

Annotation: similar to labels key/value pair, but is used to add descriptions (mainly by externtal tools). Can store larger, more freeform data. Example: “I created this pod just for testing, can be destroyed at will” or build-hash, commit-message, etc.

More complex concepts #

Service: is cluster-wide and points to nodes, does stuff like load balancing, naming, discovery (→ to isolate microservices from eachother). Is not something that ‘runs’, is simply an abstraction for exposing pods in a stable and scalable way. They can expose pods to external traffic and maintain stable endpoints even if pods are replaced.

Ingress object: frontend combining multiple microservices into a single externalized API surface area. Manages HTTP(S) traffic, often includes SSL termination.

Replica set: acts as a cluster-wide pod manager (managed by a deployment, see below)

Workload: A workload in Kubernetes refers to an application running on the cluster. Workloads can be comprised of single or multiple pods. They are:

deployments (stateless application like wordpress). It is ensured a specified number of identical pods are running (replicasets)
Statefulsets persistent identity or storage (databases like mysql)
daemonsets (infrastructure-level services like logging or monitoring)
jobs (run a task until completion)
cronjobs (similar to job, but run at set times, example optimize images in wordpress or nightly backups)

Example: my wordpress cluster has a deployment with 3 identical pods. Kubernetes handles load balancing. If one pod fails, traffic gets sent to other 2 pods and failing pod gets destroyed and new one created. Updates to new Wordpress-version happens per pod: terminate old pod, new pod online, repeat until all pods are updated.

User roles in Kubernetes #

There are four standard roles (for humans) within Kubernetes:

cluster-admin (manage entire cluster)
admin (manage a namespace)
edit (can modify namespace)
view (read-only access).

For more detailed and practical information, I can really recommend the book Kubernetes: Up and running on which most of this summary was based. I enhanced it with information from ChatGPT; as Kubernetes has been described extensively online, LLMs can reliably give details about it.

Yearly target as freelancer - Harvest&Sheets

Tue, 21 Jan 2025 00:00:00 +0000

I work as a freelance designer, usually on a per-hour base. This year I want to try something new and set a target for end-of-year earnings. I wrote a little script to create a visualization of the earning goals.

As a freelancer, I track my hours using a tool called Harvest. I’m happy with it, works well enough for me.

Every now and then, I take a look around if I can find better tools, and that’s how I ran into Cushion. I appreciate their communication and openness as a ‘company built by freelancers for freelancers’, and I tried their offer for a while. Sadly, it didn’t ‘click’.

One thing that really attracted me in Cushion was the ability to set yearly goals. A minimum goal, a target and a stretch-goal.

Harvest doesn’t have it, so I built it myself.

Stretch goals visualized in Google Sheets #

Harvest has a really simple-to-use API which works great. I implemented it with my existing Google Workspace tools, and now I have a straightforward dashboard that shows me if I’m on track for my yearly target.

Here you see the visualization in Google Sheets of my actual accumulated income versus my targets. I’ve cutoff the labels so no actual amounts are shown. As you can see, I’m currently behind on reaching even my minimum goal!

How does it work? #

I attached a small piece of javascript to the Google Sheet using Apps Script. It pulls in the latest data from Harvest, accumulates it, and then extrapolates. It then injects the data back into Google Sheets.

Can I use it also? #

You can find the script below. Let me know if you used it, and if you made any improvements!

Read through the code below before going through the steps. Never blindly copy-paste code from the web 😅

In Harvest #

sign into Harvest
go to developers section
create a new ‘personal access token’

In Google Sheets #

create a new Google Sheets document
add a sheet and name it Data
rename the existing sheet Visualization
click ’extensions’ and ‘Apps Script’
this opens a new page in ‘Apps Script’ with an empty editor

In Apps Script #

copy and paste code below into Apps Script
accept the scary terms from Google (“this can read/modify/delete every Sheet in your account”)
run the code and check the output in your Google Sheets
now you have the data, you can create any graphs you like

The code #

As always, it’s MIT-licensed. Have fun.

function fetchHarvestData() {
 const HARVEST_ACCOUNT_ID = 'XXX';
 const TOKEN = 'XXX';
 const START_DATE = '2025-01-01'; // Start of the year
 const END_DATE = new Date().toISOString().slice(0, 10); // Today
 
 // Goal amounts (in Euro)
 const MIN_GOAL = 10000; // Minimum goal (example: €10,000)
 const TARGET_GOAL = 15000; // Target goal (example: €15,000)
 const STRETCH_GOAL = 20000; // Stretch goal (example: €20,000)

// Calculate the total workdays in the entire year (excluding weekends)
 const totalWorkdays = countWorkdays(START_DATE, '2025-12-31'); // Use the whole year range

 // Calculate the daily goal for each goal type
 const minGoalPerDay = MIN_GOAL / totalWorkdays;
 const targetGoalPerDay = TARGET_GOAL / totalWorkdays;
 const stretchGoalPerDay = STRETCH_GOAL / totalWorkdays;

 const url = `https://api.harvestapp.com/v2/time_entries?from=${START_DATE}&to=${END_DATE}`;
 const options = {
 method: 'get',
 headers: {
 'Authorization': `Bearer ${TOKEN}`,
 'Harvest-Account-Id': HARVEST_ACCOUNT_ID
 }
 };

 const response = UrlFetchApp.fetch(url, options);
 const data = JSON.parse(response.getContentText());
 
 const sheet = SpreadsheetApp.getActiveSpreadsheet().getSheetByName('Data');
 sheet.clear();
 sheet.appendRow(['Date', 'Billable Amount', 'Accumulated Amount', 'Accumulated Min Goal', 'Accumulated Target Goal', 'Accumulated Stretch Goal', 'Projected Income']);

 const entries = data.time_entries;
 const dailyTotals = {};
 
 // Calculate daily totals for billable amounts
 entries.forEach(entry => {
 const date = entry.spent_date;
 const hours = entry.hours || 0;
 const rate = entry.billable ? (entry.billable_rate || 0) : 0;
 const amount = hours * rate;

 if (!dailyTotals[date]) {
 dailyTotals[date] = 0;
 }
 dailyTotals[date] += amount;
 });

 // Create an array of all dates in the range (for the whole year)
 const allDates = generateDateRange(START_DATE, '2025-12-31'); // Whole year range

 // Initialize accumulated totals
 let accumulatedAmount = 0;
 let accumulatedMinGoal = 0;
 let accumulatedTargetGoal = 0;
 let accumulatedStretchGoal = 0;
 
 let totalBillableAmount = 0;
 let totalWorkdaysCompleted = 0;

 let averageIncomePerWorkday = 0;
 let projectedIncomeAccumulation = 0;
 let rows = [];

 // Calculate the average income per workday for completed workdays
 allDates.forEach(date => {
 const billableAmount = dailyTotals[date] || 0;

 // Update accumulated amounts only for workdays (not weekends)
 const dayOfWeek = new Date(date).getDay();
 const isWorkday = dayOfWeek !== 0 && dayOfWeek !== 6; // 0 = Sunday, 6 = Saturday
 
 if (isWorkday) {
 accumulatedAmount += billableAmount;
 accumulatedMinGoal += minGoalPerDay;
 accumulatedTargetGoal += targetGoalPerDay;
 accumulatedStretchGoal += stretchGoalPerDay;

 totalBillableAmount += billableAmount;
 totalWorkdaysCompleted++;
 }

 // Calculate projected income
 const isFutureDate = new Date(date) > new Date(END_DATE);
 if (isFutureDate) {
 projectedIncomeAccumulation = accumulatedAmount + averageIncomePerWorkday * countWorkdays(END_DATE, date);
 } else {
 projectedIncomeAccumulation = accumulatedAmount;
 // Update the average income per workday
 averageIncomePerWorkday = totalWorkdaysCompleted > 0 ? totalBillableAmount / totalWorkdaysCompleted : 0;
 }

 // Add the row to the array
 rows.push([
 date,
 billableAmount,
 accumulatedAmount,
 accumulatedMinGoal,
 accumulatedTargetGoal,
 accumulatedStretchGoal,
 projectedIncomeAccumulation
 ]);
 });

 // Write all rows in bulk to the sheet
 sheet.getRange(2, 1, rows.length, rows[0].length).setValues(rows);
}

// Helper function to count workdays between two dates
function countWorkdays(startDate, endDate) {
 const start = new Date(startDate);
 const end = new Date(endDate);
 let count = 0;

 while (start <= end) {
 const dayOfWeek = start.getDay();
 if (dayOfWeek !== 0 && dayOfWeek !== 6) { // Skip weekends (0 = Sunday, 6 = Saturday)
 count++;
 }
 start.setDate(start.getDate() + 1);
 }

 return count;
}

// Helper function to generate all dates between two dates
function generateDateRange(startDate, endDate) {
 const start = new Date(startDate);
 const end = new Date(endDate);
 const dates = [];

 while (start <= end) {
 const dateStr = start.toISOString().split('T')[0]; // YYYY-MM-DD format
 dates.push(dateStr);
 start.setDate(start.getDate() + 1);
 }

 return dates;
}

Exceptional days, exceptional weeks

Mon, 13 Jan 2025 00:00:00 +0000

It’s said life is a string of ordinary days, with intermittent exceptional ones. That’s not how I felt the past weeks!

I often I think: “I didn’t accomplish what I wanted, but that’s because today is not a normal day”. I’ve realized it’s time to shift my mindset: there are no normal days.

Every day has something making it different from the day before. Every week has something making it different from the week before. No week has consistent, regular, predictable days.

I’m dubbing it “the fallacy of the normal week”.

Last week was not a normal week #

Let’s look at last week, what happened?

Monday – I got back to work, the kids had a day off. I had two meetings while they were playing in the background
Tuesday – first day of school madness. Daughters’ headset broke and I fixed it with some soldering. Neighbor’s shed-roof blew off with the wind and I returned it. Cat threw up badly.
Wednesday – the dishwasher had an error code. I fixed it with some intens cleaning with special materials (that I had to go out and buy). Ordered groceries. Piano lessons.
Thursday – 20:30 meeting for work, my son got fever, we slept bad
Friday – son stayed at home with me because of being sick: could barely work. Had to fix Docker-setup because someone somewhere made a code-signing error. Bad sleep again because of sick son
Saturday – son is better, but whole family very tired from sickness (except son, he’s full of energy, thank god). Special event at my daughters’ scouting, where I joined
Sunday – swimming lesson of my son took considerably longer as my daughter also wanted to join and we had lunch there. After that a birthday party

Throughout the week I additionally had the mental load of preparing for my sons’ birthday, planning a holiday to Turkey, and a poster I’m working on.

Is that a normal week? It doesn’t feel like it. Every day was different from “normal”.

But every week is different #

The whole of December was irregular: Sinterklaas, Christmas, new-years’ eve, holidays…

And now January is irregular. There are many one-off events (like my sons’ birthday, holidays, study-days where school is closed, school-outings, hackathon at work).

There is no reason to assume February or March will be different. There is never a normal week or a normal day.

What is normal anyway? #

My original mindset was: most days are ’the same’. Similar and predictable: two hours of morning-crazy (breakfast, getting kids to school on time), eight hours of work, five hours of evening, sleep.

And similar days lead to similar weeks. Four-and-a-half days of work, two-and-a-half-day of weekend.

52 times per year a normal week. With perhaps some exceptions, like Christmas, vacation, holidays, birthday, one-off events like a wedding, maybe sickness… When you write it down like that, you quickly realize that there are a lot of vacations, holidays and birthdays. Apparently the exceptions are the norm and ’normal weeks’ are very rare.

New mindset: every day is exceptional #

I sound like a romantic now, but I’ve realized every day is exceptional. When you look at the details, there is no normal day.

Still learning exactly what this means. But at the very least, I should lower my planning-expectations per day.

I feel I’m terribly late in life on realizing this. Probably most readers will facepalm me for stating the obvious 😅

Now on to my ~normal~¹ workday!

six-month parent-teacher of my son at school at 15:00 breaking up the day, and school-meeting in the evening… ↩︎

Conference summary: Enterprise UX 2023

Wed, 06 Nov 2024 00:00:00 +0000

Last year’s Enterprise UX was a blast with great talks, interesting people and fun discussions. Just in time for the next one, I’m giving a quick recap of 2023’s edition.

Enterprise UX is a conference with a single track. That’s nice, because you don’t have the fear of losing out, and you see talks you might otherwise have skipped. In addition to the talks, there was also an informative visit to a mobile research lab from the Dutch Police, which is used for user research (but also for “real police work” such as requesting information in a neighborhood). Good example of how a physical object (a truck) can help explain an abstract subject (doing user research).

I’ve made a brief summary of all the talks. You can find the full-slides for most talks on the website.

Vitaly Friedman #

Vitaly Friedman gave an impressive talk with an enormous information density. He discussed, among other things, how consumer best practices do not work for enterprise UX (and which best practices do work), and how you can make an impact as a designer and how you can measure that.

consumer best practices do not work for enterprise UX

Vitaly Friedman

For people doing the work #

Enterprise UX lives in complex environments. This means UI will be complicated as well. Typical UI-best-practices (“more whitespace!”) cannot be applied mindlessly.

Enterprise: software for people doing the work

First thing I gave up: the notion of white space. Doctor needs to see anything in one glance, without clicking anywhere

This is the Olympics champion of enterprise UX. “Nothing is complicated for expert users”

I fully agree with Vitaly’s statement. The complex and complicated are what draw me to B2B and enterprise applications.

I 100% agree with Vitaly there is beauty in complicated applications, that’s also why I work as a freelance B2B designer

Working with the business #

First thing you should do when working with a new company is make a chart of the organization.

LinkedIn often says “UX is the center of everything”, but this is not true:

In this slide Vitaly put ’leadership’ in the middle, which theoretically might be closer to reality, but in my mind organizations are complex and dynamic systems, where nothing is really ‘in the middle’

Ouch, this is recognizable. We designers have a bad understanding of the organization.

Frameworks #

Vitaly never worked double diamond. Instead works reverse double diamond.

My thoughts: interesting and thought provoking, but most frameworks are a post-factum thing. It’s never this simple.

Frameworks are basically bullshit. Some companies work very fine with waterfall

Measuring impact #

This quote sounds great, but I’m growing a bit tired of ‘defending design’ quotes in general

I like this list of KPIs, as long as we can keep them as indicators and not targets

How are we doing on the KPI? Simple color coding

NPS is the worst metric, but as everyone uses it, do tap into it.

More people were on Mount Everest than the people who visited the tenth page of your search results.

Vitaly suggests working with the KPI tree. Works great with multiple teams also:

KPIs can/should trickle down according to Vitaly and a lot of other people. Myself, I’m not 100% convinced. This is typical top-down thinking (even if you focus on outcome over output). In a lot of businesses, the sensors of business outcomes are in the bottom. That’s where the data comes in. Those are the most flexible to change also, but with fixed goals they get less flexibility. A difficult puzzle

Work with business analyst. They work top to bottom, designer bottom up. Connect UX work to business objectives.

Best practices for enterprise #

Many “best practices” do not apply to enterprise UX. That doesn’t mean there aren’t any, Vitaly has a large collection of best practices that you can ‘safely’ apply to B2B and enterprise UX:

There is a range of ages using enterprise UX. Make sure you also design for the elderly, but don’t use stereotypes and assumptions about the elderly.

More information on accessiblenumbers.com

60+ will abandon product much faster when encountering problem

Vitaly ended the talk with a satisfyingly distressing video about disappointing interactions:

Marieke van Kouwen - NS #

Marieke van Kouwen gave a wonderfully sincere talk about the design of an application for material controllers of the NS.

The system that aimed to combine information sources into one system, but resulted in yet another system being added. Marieke ironically remarked: “users are so happy that they shout at me”.

It was really nice to see a case study that is not the ideal situation, how Marieke deals with that, and how the project has now got back on track.

“users are so happy that they shout at me”
Marieke (ironically)

There were already a lot of sources of information the users needed to investigate. Instead of replacing some (the original goal of the project), we just added one more… (obligatory XKCD reference inserted here)

That’s me! Here are some generalized user-research findings restated as problems. Powerful stuff

Edo-Jan Meijer - Dutch Police #

Edo-Jan Meijer came on stage with a fantastic opener: the impact of being excluded when something is not accessible (Edo-Jan is in a wheelchair).

Then came another great story about the UX vision of the police: software should be seen as a colleague. Which he elaborated on (what ‘human colleagues’ can expect from that digital colleague). A very impressive story, well-founded.

Nice way to visualize time saved with better UX. ‘40.000 extra hours per week to catch thugs’

44.000 cops, 60 designers. And all have ’their own thing’: tasks, needs, etc.

Software as a colleague #

Edo’s convinced we should see software as a colleague and not an application. He makes a convincing argument for this.

Humans are colleagues of software and the other way around, but they have very different strengths and weaknesses

Humans want to save energy, seek patterns, remember stories, is social. So a good software colleague brings clarity, uses patterns, tells a story and is social

Software should extend the person, be a colleague. This means human-police-values do not translate 1:1 to software values. But they are related

Last slide showing how a software colleague complements the user

Chain of serving: software supports people, supports Dutch Police, supports Dutch society

Edo-Jan uses the following chain: values -> vision -> promise.

The slides tell a very strong, relatable and understandable story. It was a great experience. But looking back, I’m not sure exactly what to do concretely with all this information. Edo-Jan did a great job in looking at the system from the outside and simplifying/abstracting it, but is this usable within the system?

Usability and accessibility #

Edo-Jan was far from finished after these amazing slides and continued effortlessly into another great topic: accessibility.

Police work is difficult and diverse. The software-colleague needs to work reliably in a lot of complex environments.

Edo-Jan made this concrete by showing how different factors complicate matters:

human: all humans share some communalities, limitations and strengths
person: within ‘all humans’ are the many diverse people, all with their own histories, values, capabilities, limitations
role: the person is acting within a certain role, which has again capabilities, limitations, expectations…
activity: what you’re doing within the role (and how: on horse, on bike, etc)
device: the way you communicate with the software-colleague (Edo-Jan shows different screen-sizes, but this can also be an LLM, or a physical device)
situation: mental pressure, circumstances, experience, relationships, location, complexity of situation. Some of these factors are also ‘personal’, but of course people are not stable points, they drift in their emotions, stress-level, etc.

A nice setup for the slam dunk that’s coming in the next slide

This might be one of the best explanations I’ve ever seen of an explosion of complexity through multiplication of factors. Of course software design is difficult if you visualize it like this. It’s a grand finale to a logical and simple story: ‘if this, <listener nods> and that <listener nods again>, then we get this result <listener gasps: oh boy!>

Another anecdote from the talk stuck with me: Edo-Jan wondered out loud why no one knocked on his door during Corona times to ask how to deal with “always sitting inside and not being able to go to the supermarket”. As a wheelchair user, Edo-Jan is an expert in that field, but we never asked him for help. Why?

Harald Lamberts - Essence #

Harald Lamberts gave a talk that took a while for me to get into. It didn’t “click”. My first impression: “a vague story with lots of buzzwords: ‘brand promise’, ‘multi-channel interactions’, ’touch points’, ‘verticals’ and ‘horizontals’”, it sounded like Multilul (Dutch expression for bullshit)…

But eventually it all made sense. The ‘quarter fell’. Harald explained with great clarity, and now these terms are a lot less ‘buzzword’ to me and a lot more “we need these terms to explain our work well to management levels”.

I am left feeling sorry for ’the management layer’ that their work is so complex, that they need to use such vague and abstract words that carry a high risk of ’nobody understands what you’re really talking about). And I’m now more conscious about how the terms we ‘designers’ love are being perceived as multilul by our stakeholders.

With Haralds’ help I no longer turn my mind off when seeing this type of graphics. It makes sense now.

There is Conway’s law! Your fragmented organization leads to a fragmented user experience: recognizable.

I like how this image says aligning and how the arrows go both ways. It’s an iterative process. I’ve seen this as a topdown one-time exercise way too many times: that doesn’t work.

Horizontal alignment across the channels – No longer sounds like something a sailor would say

I’m a skeptical person, and I’m not 100% sold yet on the management part of this. Alignment, yes, but management?

Verticals, horizontals, alignment op je experience strategy, die je multi channel kunt challengen

The employee experience #

As the icing on the cake, Harald showed how they use well-known techniques such as service blueprints and journeys within companies to provide insight to managers.

Impressive blueprint work of the service within a company

Really cool inversion of the pyramid. I 100% support this effort and would like to subscribe to the newsletter. The journey(s) of employees are often ignored, and we look at them from a very small horizon. Are you for instance thinking about off boarding as an opportunity? I like the special attention to the tooling of the employees: I’ve experienced multiple times that this is just bought on basis of ‘cheapest’ or ‘most convincing talk from supplier’

I remember this as interactive tooling, but now these slides look like static images. I think journeys (especially for ‘aligning’) need to be highly interactive working-documents.

Jos Kauling and Rien Buisman - PostNL #

Jos Kauling and Rien Buisman had a great story about the redesign process of the PostNL app for delivery people. Apparently I was so caught up in the story that I didn’t take any notes at all 😅.

Rien had trouble convincing his stakeholders he needed research-insights, and solved this by helping out at the distribution centers himself and interviewing drivers there, and by asking ‘his’ delivery person at the door for help. Those first steps led to a better understanding of the need for research, continuous testing of the app and resulted in a very successful app.

In addition, I remembered that the existing design system was not sufficient for this app, as it was built for consumers. Instead, a specialist design system was created for this app. That’s the kind of flexibility you need for a complex enterprise app!

Ethics and design by Astrid Poot #

The day ended with absolute fireworks from Astrid Poot.

I already knew Astrid’s work from her website goedmaken.org. Her post about Hannah Arendt in particular really appealed to me: I saw so many parallels with my own work and life. I also saw Astrid’s “core and check question” before (core: does it help? And check: does it cause harm?). Powerful stuff.

No wonder I was super enthusiastic to see her talk. And she delivered: she brought tremendous energy and a lot of practical information!

I can’t really do her talk justice, so best to read her slides yourself, or better follow her on LinkedIn and buy her books.

In short:

Astrid explained how everyone is a designer, and that we as designers make choices and carry responsibility
But also that as a designer you can’t improve everything. You also have other obligations: towards your employer, yourself and your family. In that respect, many lists with ’ethical design’ are a slap in the face: how can I ever meet that?
Small steps are also steps. Look at what is within your reach: not everyone can and needs to “change the system”. As Astrid says: “reading something by Mike Monteiro is enough”
Angry people are not bad. See them as useful: they are so involved and invested that they endanger themselves and others. You just have to be able to channel that well. From frustration (destructive), to anger (power, impressive but also intimidating) to positive energy (optimism gives direction). Disobedience is good. That leads to change.
Imposter Syndrome: you project the knowledge of all kinds of different people you know onto all people. “Oh, Henk knows so much about this” and “Suus knows a lot about that”, then I am lagging behind as a designer when apparently all designers know that. I recognized that part of myself enormously (but strangely enough it often has the opposite effect for me: that it does not make me doubt or hold me back, but motivates me to learn all those things too).
The core points of Hannah Arendt. I can’t do justice to that by summarizing what Astrid had already summarized. Still, I’ll try: Amor Mundi (love the world), natality (you can always try again), plurality (everyone is different and valuable, everyone can join in the conversation)
The difference between ‘solving a problem’ and ‘changing the world’. Problem solvers analyze the situation and ‘calculate’ a solution, the solution fits into the space we know. You don’t change the world from research or a fixed process, but from personal involvement and judgment. I also recognize the latter as something ‘Hannah Arendts’: the difference between work and action. Work gives purpose, action gives meaning.

And then there was much more. If I could, I would listen to the talk again.

Be aware #

Astrid was working on preventing sexual disease from spreading, and talked with designers from Axe who joked ‘you fix what we break’. That’s a convincing argument that we should realize: creating something also means breaking something. Is your work worth that?

Similarly to the great Gusteau’s in Pixar’s Ratatouille: ‘Anyone can cook’, everyone is a designer. But that’s metaphorically of course, please keep paying me instead of trying to do my work yourself. Be aware that any change you make (whether it’s conscious or not) has an impact, it’s your choice and responsibility to deal with it.

Philosophers make it seem like it’s difficult to do the right thing, needing to follow very strict rules

And be realistic #

Lol: these kind of lists are a slap in the face. I’m happy to even have a job, and now I should feel bad about not reaching these high lofty goals…

be nice for yourself and the audience for whom you are writing

A small step into being nice is already good. Not everyone has the capability to do amazingly good work.

Beware of impostor syndrome. Don’t project all the world’s knowledge onto all people, nobody knows everything. If you do that, it seems like you are the only person who doesn’t know it all. Simply not true.

Sometimes reading a bit of Mike Monteiro is already enough.

But how, Astrid? #

All ethics-frameworks carry problems, so just apply them all! Check: ‘Am I ashamed to have made this?’, ‘What if everyone would do this?’ and ‘Does it make a lot of people happy?’

Only people on Santa’s naughty list change history

Astrid channeling her inner Yoda: frustration leads to anger, anger leads to positive energy. But not automatically, that needs guidance. Still: angry people are good: they are committed and so invested that they endanger themselves and others. Frustration is destructive. Anger is power: imposing, but also intimidating. If you can give that a direction, it becomes optimism and energy.

Writing the ALT-tag for this one made me realize how much the emphasis on WHAT, HOW, and YOU make this text as powerful as it is. It’s an energizing statement to turn from moping to action

There was a lot of talk of Astrid about Hannah Arendt. Best is to read Astrid’s blog posts, but here’s my short version:

Amor Mundi: love the world. People, raw materials

Natality: can always try again

Plurality: do not strive for uniformity, but to see differences as differences to do them justice. Everyone is different, and of value, everyone can join in the conversation

Hannah Arendt

This. Is. Powerful. Stuff. Real change comes from personal commitment and judgment. Not from research, not from a fixed process.

What is professional disobedience? As a professional, or as a user? It has to start sometime. What better place than here? What better time than now?

Practical example of design by non-designer. A cook saw there was a lot of stress in the room, and he looked what he could do within his reach. Eggheads!

Always ask yourself: am I helping? Am I damaging others? - not just for your intended audience, but also wider and on larger timescale. For example Facebook: it damages society over time.

What is in your reach? Interest, be conscious, change within your work, change your work, create your work, change the system. Start small. Not everyone can change the system (that would be a circus also, lol)

Follow Astrid on LinkedIn. Do it. Now.

That’s it, see you at the 2024 edition!

Alternatives to endless scrum

Mon, 30 Sep 2024 00:00:00 +0000

Scrum is a popular process to develop and deliver digital products. The most common form of scrum is an endless string of subsequent sprints, a draining drudge without room for long-term investments like continuous research. It doesn’t have to be that way. Scrum can give energy instead of taking it.

Let me start with an example:

Usain Bolt is an incredible runner. Seeing him on the 100-meter sprint is comparable to seeing Superman fly. He makes it look easy, almost effortless. If we’d make him run a marathon, he would slash the world record in half! Just keep sprinting, Usain!

Imagine we’d propose that to Mr. Bolt; the reigning world champion would most likely answer: “that’s stupid and dangerous, you will die if you keep sprinting”. In hindsight this is a logical conclusion.

Why is it then that most companies are doing endless sprints?

In this article, I’ll list the reasons why back-to-back sprints are a bad idea, and which three solutions you can try to alleviate the problems.

You will die if you keep sprinting
– Imaginary Usain Bolt, September 2024

1. Endless sprints will kill you #

A popular form of product-development is agile development and the most popular method is scrum. All companies I work with use scrum in one way or another; either following the rules strictly, loosely, or with a complicated framework like SAFe (shudder).

For all the varieties of scrum I’ve experienced, a basic tenet is that the end of one sprint marks the beginning of another. Endless sprints.

Endless sprints lead to several interconnected problems: demotivation in the team, lapses in quality, detachment from ’the business’, and unrealistic estimates.

Do you recognize any of these in your company?

1.1 Unrealistic estimates #

Stories are difficult to estimate: the requirements and acceptance criteria are not clear enough
Ironically, teams complain that requirements and acceptance criteria are too well defined and they cannot use their creativity to create the best solution due to this lack of flexibility
Teams overestimate due to uncertainty. Which leads to stakeholders finding shortcuts, like removing heavily debated requirements

1.2 Distance with the rest of the business #

Business stakeholders find it hard to get stories ‘into development’, it takes a long time to develop features, and the end-result is not what they expected. It’s unclear what to specify
Stakeholders over-ask because it takes a long time before it’s ‘my turn again’ — worsening the problem for others
There is a gap between ‘research’ and ‘development’. Researchers have trouble communicating their findings, while the development team feels they are flying blind without knowing the user needs. Developers have no time to join in on research sessions such as usability tests

1.3 Demotivation #

Tasks spill over from one sprint to another: creating zombie-tickets that follow the team for many sprints
Teams stop celebrating important milestones because they are always working on the next thing
Developers find it difficult to get user-stories into the sprint for important issues like architecture updates, unit-testing, refactoring, improving existing features, or simply ‘time to think’ — because of the apparent focus on ‘shiny new thing’ from ‘the business’
Scrum-team members talk about ‘the grind’ when describing their work 😿

1.4 Lapses in quality #

Releasing features with bugs left unfixed (“we will fix those next sprint”) because of overrunning expectations and end-of-sprint pressure. This leads to rework, aka the-haunting-of-work-of-sprints-past
Once done, it’s difficult to iterate on a feature to incorporate learnings and get the best impact for end-users as the backlog is filled with next-things
Learnings during the sprint are hard to incorporate as developers work on stories (and designs) created months prior

1.5 Procedural arguments and pointing blame #

The problems above lead to procedural arguments and pointing blame. The stories are not small enough, or too small, need more refinement, designs shouldn’t change during a sprint, we need to be more adaptive to change, not enough communication, too many meetings, the ‘business’ doesn’t understand, etcetera.

This can be solved, but first another problem.

2. Some things can’t be rushed (like continuous research) #

Besides the destructive nature of back-to-back sprints, there is another important truth to address: not all activities fit in scrum.

Scrum needs neatly estimated stories. But that requirement can’t be met for open-ended questions in research, design, and IT-architecture. The type of questions that need alignment, ’nights of sleep’, long-term thinking and flexible experimentation. Scrum was not made for that. Sure, we can force these activities into the scrum-process with time-boxed spikes, or by dragging along stories for several sprints, but it rarely really works.

In some companies, the solution to this is doing this type of work outside the development sprints, in a non-scrum process or in parallel sprints of the same cadence. This creates silos and distance in the organization, even though the power of scrum lies in multidisciplinary teams.

We can solve all of this, but first we need one more building block before I introduce my solutions.

3. Fulltime scrum is a lie #

A “three week sprint with 8 FTE” sounds very impressive, and it is: that’s 320 hours of raw development power!

Or is it?

No. Of course not. No one is working on user-stories 40 hours per week every sprint.

People need holidays. They do trainings. Go to conferences. Attend company meetings. Do the scrum-rituals (standup, refinement, retro). Bring their sick cat to the vet. Go to the dentist. Have meetings. Read research reports (just kidding). Take hours to answer ‘quick questions’ in Slack. Have more meetings. And (gasp!) drink coffee and talk about the weather with their colleagues.

None of these things are inherently bad or evil. Most of the things mentioned contribute to a good work-environment. We’re not robots. Not yet.

Scrum promises us that sprint-weeks look like this:

But our in reality, weeks look more like this:

That’s an optimistic simplification: I bet you have more than two meetings per day!

It gets worse. There is no focus if we keep getting interrupted. Our focus-flow is busted by each interruption that leads to context-switching. On average, a person needs up to 15 minutes recovery from an interruption to get back into a concentrated flow. Four 30-minute meetings add another hour of lost focus!

On top of that, interruptions reduce the possibility for collaboration. How can we work together, if you are in meetings half the day, and I’m in meeting the other half of the day.

If we plot team-member agendas on top of each other, it becomes obvious there’s only limited time to work together on tickets:

That looks horrible. But solutions are coming, I promise!

Quick recap of the hole we’ve been digging #

Scrum promises focus, responsibility, continuous improvement, and collaboration. Instead we get:

no collaboration between those ‘in the scrum teams’ and those ‘outside’
no focus because of non-scrum interruptions
no collaboration within the team because team-members have different interruptions
unsuitable activities such as continuous research are either not done, done in a silo or forced into ‘user story’-format
no continuous improvement as the grind makes it difficult to take a broader view
disillusionment and dread from endless-sprints

Enough negativity, let’s crawl out of this hole!

Alternatives to endless scrum #

Yes, I’m reusing the title of this article as a subheading. That is allowed as a writer. Let’s look at three alternatives to endless scrum.

First, a quick introduction of myself. I’m not a scrum-hater. Before I worked as an awesome-strategic-product-designer-currently-available-as-freelancer-contact-me-on-linkedin, I was scrum master on a couple of projects. I have seen the method work.

When scrum works, it can be magical. A multi-functional team with equal members. Full alignment. Full-throttle building. Seeing the product come to life in front of your eyes. Tears of joy streaming down your face as your trust in humanity is restored. Okay, maybe not that last one, but trust me that scrum can work and can even be (gasp) fun.

How do we get there?

Alternative 1: scrum-days #

Nowhere in the scrum-manuals does it say that scrum is a full-time activity. The rituals of the process bring a lot of overhead, so it makes sense in theory to go all-in, every day. In practice it brings the aforementioned problems.

Instead we can separate the week in ‘scrum days’ and ’non scrum days’. Each team-member is ‘in scrum’ for a limited amount of time per week: two or three days.

Three days of scrum per week may sound unworkably low compared to the five days you are used to, but hear me out.

What do ‘scrum days’ look like? #

During the ‘scrum days’, there is a strict ‘scrum only’ policy for people who are ‘in scrum’. People outside the team are not allowed to disturb them with meetings or quick questions, people in the team don’t plan any non-scrum-activities.

That is harsh. But the benefits are overwhelming. Because of the lack of ‘outside disturbance’ the scrum-teamers get three days of concentrated, full-focus work on their stories. No distractions, no context-switching, full-time togetherness of the scrum-team. You get a dedicated, cohesive, multifunctional team during those days.

The rest of the company will need to get used to the information- and meeting-lockdown during the ‘scrum days’. But, paradoxically, they get to work more intensively with the scrum-team during the ’non scrum days’.

The ’non scrum days’ #

‘Non scrum days’ are what regular people would call ’normal days’. Days spent in meetings, quick calls, 1:1s, trainings, knowledge-sharing sessions, collaborate in user research, go to conferences, take time for team-bonding, etcetera.

And now scrum-team members can reap the benefits of these days as well. They can have ample time to do long meetings with deep thinking, and intense but fruitful workshops - without end-of-sprint pressure looming over them.

Researchers can work together with developers. Testers can spend time on improving their automations. Designers can work on long-term deep-dives. Architects can fill up whiteboards and discuss with stakeholders.

Scrum days give teams more time to work on stories and more time to work on other stuff.

Scrum days in practice #

The descriptions of ‘scrum days’ may sound overly optimistic.

Let’s do an example. Look at happens to our previous calendar when we introduce ‘scrum days’:

I only moved stuff around. I did not remove any meetings, didn’t shorten them, all I did was organize the scrum-days. The observant reader will remark that ‘solo work on tickets’ is gone. This has in its entirety been overtaken by collaboration with the team.

If we account for the context-switching, the new ‘scrum days’ have more than 95% of the previous time-spent-on-tickets. Five days of work squeezed into three days. That’s in this theoretical example with a maximum of two meetings per day. The benefits in reality are much larger (are there days when you have less than two meetings?)

When I worked as a scrum-master, this is how we operated. Developers and testers worked three days per week ‘in scrum’, designers, researchers, and product-owner two days per week. I was a strict scrum-master and did not allow any other work than stories, I even physically separated the team from the rest of the organization by working in a different location.

We had a blast. During the scrum-days, designers worked together with developers on the details. Outside the scrum-days developers worked with designers on the larger picture and helped in research. The product-owner was always available to make instant-decisions during scrum-days, and he had ample time to align with the rest of the organization during non-scrum-days.

There was team-cohesion and tremendous speed.

We moved mountains. Scrum days work.

Alternative 2: scrum holidays #

Even when working with no-scrum-days, teams get exhausted from the constant barrage of intense work pressure.

Simply put: no-scrum-days keep teams healthy, but you need a bigger battery-boost to bring out the best.

I introduce to you the scrum-holiday: a week-off from scrum every two or three sprints. Sounds crazy. Crazier even than no-scrum-days. But it works.

Why? Well, scrum works best when teams are fully focused on their tasks and put in their best work. A two-week (or three-week) sprint can deliver massive amounts of product when that happens. Those are the examples the scrum-enthusiasts talk about! But it requires a team that is full of energy and enthusiasm. Not a team beaten down by sprint-after-sprint-after-sprint.

During a scrum-holiday your team can focus on clearing technical debt, investigate the root cause of problems or user research.

Back when I worked as a scrum-master, we’d work in batches of two-sprints. Each sprint had a specific sprint goal that we set together as a team: a big feature or ’epic’, or another commonality between the stories (e.g. a ‘bug hunt and fix’, ‘accessibility push’). Finishing that goal became a healthy obsession. Pushing people to stretch their limits (within working hours, of course, no unhealthy overtime allowed by this scrum master) – and a joyous feeling of victory when we made it. If we made our ambitious targets, I celebrated those victories with food and drinks at the sprint review.

That team delivered. We finished a complicated project in five sprints, with six people. The type of project that would normally take at least a year.

You simply cannot do that with back-to-back sprints.

Taking breaks increases productivity.

(Insert callback to Frederick Winslow Taylor, who proved this in the early 20th century)

Alternative 3: a more flexible thinking 🙈 #

At the beginning of this ~~book~~ article I mentioned scrum is a form of agile working. It’s not the only form.

There’s Kanban, Shape-up (by Basecamp), Lean, Crystal, feature-driven, Dynamic Systems Development Method, and countless others.

Or you can roll-your-own type of process instead of following a prescripted method. Adapt to the team and the needs in your context. Try out new things, keep what works, throw out what doesn’t (it might help to read articles about Demming).

The original agile manifesto was written by four consultants on a ski-trip, as a summary of the similarities between the different processes they took. The manifesto celebrates diversity in approach and flexibility, not trusting in a pre-baked framework:

we have come to value:

Individuals and interactions over processes and tools

Working software over comprehensive documentation

Customer collaboration over contract negotiation

Responding to change over following a plan

Each company operates in a different context, each team within the company has different needs and strengths. No pre-built method will be a perfect fit, and there’s not one method that fits all teams equally.

True agility means learning, understanding, experimenting and subsequently adapting your process (and strategy, and product, and…).

Conclusion #

Scrum can be fun and energizing. Sadly, that’s not how it’s implemented in most companies.

I introduced three simple steps to free yourself from the droning-on of endless-sprints:

introduce scrum-days: bring back focus to the team and expand collaboration with those outside the team
introduce a scrum-holiday to recharge teams and make scrum pack a powerful punch again
be more flexible in your approach to work, instead of stoically following a predefined process. Learn from other methods, try stuff out.

One last reminder: any change you introduce takes time to come into effect and introduces a bit of noise. There is a learning curve. Don’t stop an experiment after a mere two weeks of trying, and don’t introduce too much change at once.

Good luck, and have fun experimenting!

Another blog update

Mon, 09 Sep 2024 00:00:00 +0000

I simplified my blog layout. It’s now simpler.

I simplified the style, removed all unnecessary pages like the about page, and migrated everything from the theme to my own Hugo-folder. I found I was overwriting so much, I practically remade the theme anyway.

The whole thing is now a lot simpler. Which is good. But it doesn’t look all that great.

FIXED! NUC5pcyb hanging with Ubuntu Linux

Wed, 21 Aug 2024 00:00:00 +0000

I bought an old Intel NUC, to have it run as a small server in-house. But even with the latest Ubuntu I had a problem: the machine would run great for about a day, and then would “hang”. The solution? Turn off all Intel power management options in BIOS!

I’m writing this post mainly for other people with the same problem, as I couldn’t find the solution anywhere else online. Hopefully this helps someone some day!

The problem #

My Intel NUC with Ubuntu would run fine, and then suddenly I would found it in a catatonic/zombie state. The blue light is on, external hard drives spinning, but no network connection and no screen output. Newly plugged in keyboards wouldn’t get their lights turned on.

It ran fine otherwise, this hanging only happened sporadically. One hint for the root-problem was that the NUC could run for a couple of days when running long n8n-workflows.

The solution #

I couldn’t find the correct solution online. There were mentions of kernel upgrades and BIOS-upgrades, but I was ahead on all the versions mentioned.

Then, I put the puzzle pieces together myself:

the server worked fine when I used it
there were no weird log entries: logging just “stopped” at a seemingly random point in time
the server didn’t really turn off but went in a half-slumber

That sounds like an energy saving mode triggered in hardware, something Ubuntu couldn’t recover from.

Fixed! #

I turned off all power management in the BIOS - and the NUC has been working for over a week now.

The origin of Lorem ipsum

Fri, 24 May 2024 00:00:00 +0000

I investigated where the famous Lorem Ipsum text came from. And to see if it’s something vulgar, or weird (like a tomato soup recipe). Instead, I found it’s origins lie with the great Cicero – and he even has some good advice for designers!

Most people, and especially designers, will recognize this text:

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

But what does the text mean? #

Lorem Ipsum is used worldwide, what does it mean?

Is Lorem Ipsum something vulgar? Or are we putting an ancient recipe for tomato soup in our designs?

The reality is a bit of a let down (Disappointment warning). While it is derived from Latin, it was intentionally scrambled to not carry any specific meaning.

The first two words are a truncation of “dolorem ipsum”, which means “pain itself”. A weird sentence to paste into your design everywhere 😅.

The rest of the text is nonsensical and improper Latin. The text is so mangled, even Google translate thinks it’s a “pain in the ass” 😅:

The customer is very important, the customer will be followed by the customer. In pursuit, the mass needs the mass of developers, not the investment of the truckload. I chatted in pure but I was chased by trucks and a smartphone. For whom is the bed? For we must drink a cup of pea, or at least some of it. There is no such thing as mass. In order to prepare the smile of life, prepare the homework. Until you need a free port, load the element of the real estate. But the sauce is followed by a pot of protein. Curabitur’s investment was sad with arrows at the gate. It’s a pain in the ass.

The actual text does not mention an ass. Google randomly adds and removes words to make understandable Latin. (“Praesent a luctus tortor, commodo fermentum quam” turns into “Praesent asinum dolor”).

What’s the source of this mumbo jumbo? #

Using random Greek or Latin words to check the layout or other technicalities is a centuries-old practice. It is called “Greeking” and has happened since the 16th century. There is no single “inventor” or original source.

The text we designers use in the modern day does have a source. It is based on a 1914 Loeb Classical Library edition of De finibus bonorum et malorum (“On the Ends of Good and Evil”), a treatise on ethics by Roman statesman and philosopher Cicero from 45 bce. Specifically sections 1.10.32 and 1.10.33.

Okay, so Cicero is the source. But what is that text saying? #

In “De Finibus Bonorum et Malorum” Cicero discusses ethics, moral philosophy and other concepts that are central to the exploration of human values and the pursuit of the good life. Sections 32 and 33 (the source material for Lorem Ipsum) are about the philosophical concepts of pleasure and pain.

What is cool about these old philosophical texts, is that they are still applicable to modern day life.

You can read an exact translation here, but here’s a quick and modern recap:

a bit of Netflix and popcorn is cool, but too much and you’ll get lazy and unhealthy — if you want to be a good and successful person you’ll need to accept “pain” such as work and exercise

Cicero would’ve made a great motivational speaker 😅

What would Cicero have thought of us using Lorem Ipsum this way? #

When we mindlessly use Lorem Ipsum in our designs, we follow what Cicero calls the “simple pleasurable” path. We don’t do the hard work of figuring out the realistic content.

And as Cicero warns us, not doing the hard work can bite us: real content has ridiculously long titles, endless bullet lists, strange explanations, cut-off images, or “no content” where the design expects it. Leaving us with the thought: “If I had known beforehand this was the content, I would have made the design very differently!”

And then we need to do the design all over again. Or more realistically: a non-optimal page or application goes live…

The Future of Lorem Ipsum #

Will designers still use Loren Ipsum in ten years? My prediction is Lorem Ipsum will get replaced by content generated from LLMs (AI-textgenerators like ChatGPT).

And why not? Creating relevant and realistic dummy content is a good usecase for LLM. The AI-results are good, and can even be an inspiring source for the final content.

That’s great news! Using an LLM is easy and fun, and delivers good quality work. You don’t have to do any hard work at all… (you’re likely seeing where this is going) So “by Cicero” we can expect downsides!

The major downside of using an LLM over real content (or even Lorem Ipsum), is that it looks so misleadingly much like real content. This will go wrong. I will bet you money a designers’ LLM-generated-dummy-content goes live somewhere soon, causing confusion for users (or customers) and have real-world negative impact.

The web is a weird place. With one push of a button, billions of people can instantly see your work. At least Lorem Ipsum shows the reader that it is nonsensical.

So maybe the trusty Lorem Ipsum isn’t dead yet!

Summary #

The words Lorem Ipsum mean “pain”, the rest of the text is gibberish. It is however based on real Latin that’s 2000 years old, a philosophy text by Cicero with his version of “no pain, no gain”.

Reading the source text for Lorem Ipsum, we can conclude that a virtuous designer should put in the hard work and notuse Lorem Ipsum or LLM-generated content.

How cool we come full circle!

Published before on my Medium

Blog maintenance

Mon, 06 May 2024 00:00:00 +0000

I recently moved this blog from Jekyll to Hugo. This was a smooth transition, with barely any manual work.

I could have left it at that. I should have left it at that. But, I had to torture myself and manually adjust every post I ever wrote.

What did I do, and why? #

I underwent the megalomaniac Herculean (or Sisyphus) task of going through every post I ever wrote, but why?

I fixed small issues that crept in over the years:

reformatted the weird date format I had to use from way back when that was needed for Hugo (“2023-01-27T00:00:00Z” instead of “2023-01-27”)
use tags instead of categories: I can now use multiple, instead of one
as the previous categories didn’t make sense, I renamed them

And had to make fixes because of the Hugo transition:

fixing shortcodes (for example YouTube embedding)
remove HTML (by default, Hugo disallows raw HTML, it can be enabled, but I like the idea)
figures with caption (I used to put a line of text below an image, but Hugo has a nice figure shortcut)
fixed links (the old links had the category name, new links have the full date)

And then I added upgrades:

moved pictures to same folder as the article: this makes it easier to work with editors (iA Writer) and gives a neater folder architecture. I can do this with Hugo as it has the nice feature “page bundles” (which Hugo doesn’t have)
I added series for posts that belong together (such as my series on Axure tips)
and I improved things “underwater”, making me ready for better html and CSS

How’s it going? #

~~I’m not 100% done yet; there’s four more years to go~~ Done! It’s a slow process as I only do this “maintenance” in my off-hours, when I’m unfocused or have some time to kill.

It’s fun to read (or scan) all my previous articles, and see what interested me in the past. On the other hand, a lot of this work feels like “a chore”. Nobody cares if I added tags to my articles. Nobody cares about the series or the image captions. I’m glad when this whole ordeal is over, and I can go back to writing again 😅

The AI-puzzlesolver: how many Easter puzzles can ChatGPT solve?

Wed, 03 Apr 2024 00:00:00 +0000

Each year I make a puzzle for Easter. This year I used generative AI to create some of the images, and now I wondered: could AI also solve the puzzle? The answer is: not yet, but it’s close!

The Easter puzzle 🐣 #

The puzzle is a yearly tradition since 2020, back when The Thing was a thing and we all had to stay inside. Some people learnt to knit, others read books, I made a series of cryptographic puzzles 😅

The puzzles are based on word and image association, combining world knowledge with language skills and creativity. Each puzzle results in one solution, and the only hints you get are the theme and the amount of letters in the solution:

The theme was candy in 2023, the word we’re looking for has 8 letters: ‘ice cream’. Why? The girl is dreaming of rice, the letter D is a C and we have to ‘remove R’. Easy peasy!

Only a handful of people finish the puzzle each year, and it takes them a long time. People love to say: “It’s complicated!”

Solving with AI #

The puzzles combine interpreting an image with reasoning. Two skills which OpenAI offers in its ChatGPT. Can it solve my cryptographic images?

The AI can do SAT tests and the bar-exam, so I naturally wondered how well it would do on my puzzle. I wrote a Python-script and let OpenAI do it’s magic.

In this post you’ll find my complete experiment, but I’ll spoil the conclusion for you: the AI does not do great as a puzzle solver yet, but it’s not far off.

I can see a future where you right-click on one of my puzzle images and ask a competent puzzle-buddy for hints!

The experiment #

The Easter puzzle consists of 20 cryptic images. Each image is a puzzle that stands alone, only connected to the others by the common theme (‘magic’). My idea then, was to offer each image to the AI separately with the same prompt and check if the answer came up in the response.

This approach puts the AI at a slight disadvantage:

humans learn the ‘way of the puzzle’ by solving the easy ones first. This way they know what goes on in the mind of the puzzle-creator.
humans know which words were used in earlier puzzles, and they know (assume?) that the other puzzles don’t have the same solution
humans see all puzzles at the same time and see similarities or trigger associations from this total-overview
humans try solving the same puzzle multiple times, at separate points in time. You try to solve it in the evening, and then again next morning
humans can work together, combining different insights (and even work together with themselves, by writing their thoughts at one point in time and re-reading later)
humans can work outside-in: make a list of theme-related words and find a puzzle where the words can fit

Still, I think it’s a fair competition as the AI can do something no human can: hold all the knowledge of the entire internet in their working-memory.

Working with the API #

I had to upload 20 images to ChatGPT and ask it the same question for each image. That’s a robotic task, and an ideal candidate for automation! I wrote a small Python script that reads in the images, prompts ChatGPT and spits out the answer. I’m using the latest model as of writing: gpt-4-vision-preview.

I used this prompt at first:

This image is a puzzle, your task is to solve it. The answer is one word related to the theme ‘magic’. Solving the puzzle works best by describing what you see and then do some associations with the words. Give your three best guesses. The answer consists of this amount of letters: 10

But it resulted in ChatGPT refusing to play:

🤖 I’m sorry, but I am unable to provide guesses related to activities such as solving puzzles that require visual identification and manipulation of details within an image that contains text or embedded clues. My capabilities do not include solving visual puzzles or providing speculative answers based on a combination of visual elements and themes. If you have any other questions or if there’s something else I can assist you with, please let me know!

This is a sensible reply, as it stops OpenAI of being used as a solver for Captcha’s. But OpenAI is easily fooled by rewording the sentence a bit and removing the word ‘puzzle’:

This image describes a word in the theme ‘magic’ (for example: wizard, druid or spellbinding). Which word? Give your three best guesses. The word consists of this many letters: 10

Now it mostly works. Mostly, because the API still refused to play in two instances. An inconvenient inconsistent behavior. As I don’t want to spend too much time on this article (spent way too much time on it already) I decided the AI doesn’t get a second chance on the puzzle when it refuses to play.

Results #

The AI gives good reasoning for each puzzle, though the final guesses for the answers are often solely based on the given length of the solution (“the answer has 10 letters”).

Result	Score
✅ Correct	3
❌ Incorrect	15
🤷‍♂️ Doesn’t play	2

Conclusion #

The AI plays enthusiastically but gets it mostly wrong: just three correct answers. This is a score much worse than most humans, on average people get at least 4-5 puzzles correct. And I know ten people finished the entire puzzle.

The result is impressive, however, if you consider the limitations. The AI doesn’t ‘know’ it’s working on a puzzle (I was not allowed to tell it that in the prompt) and it only tried to solve each puzzle once. Plus, the answers it gave are based on brute-force: it gives guesses based on the amount of letters, disconnected from its reasoning. And on top of that, this is a generic AI: if we had trained the AI specifically on this task it would’ve done much better.

I’m willing to bet puzzle-solving is another domain in which AI will quickly outperform humans!

I loved reading through the answers of the AI. The thinking is human-like and colorfully descriptive. It sometimes gets close to the solution, or circles around it. Other times it is way off track and could have used human guidance. I think using ChatGPT for the puzzle would help people solve more of the puzzles, especially in a chat-style where you guide the AI back on track when it does something weird.

Next year I will make a ChatGPT-puzzle-buddy-bot! 😅

The answers per puzzle #

Here are the answers the AI gave per puzzle. I’ve added my own thinking to each answer.

It’s fun to read through the responses and reason if it would have helped you solve the cryptogram. I bet in most cases it would have helped!

✅ Transmutation #

The AI solved this one 🎉, even though the explanation is wrong. It incorrectly thought the “-i” was “-1”, which may have led to the confusion.

🤖 This image depicts a person walking on a train platform. A train is parked on the left side of the platform, and part of the station architecture can be seen on the right. The image has a stylized graphic filter applied to it, giving it a high-contrast, sketch-like appearance. Superimposed over the scene is a large lowercase letter ‘mu’ from the Greek alphabet, accompanied by a small subscript of ‘-1’ at the end of the letter, suggesting an inversion or the mathematical concept of taking the reciprocal.

For the word-association task, given the theme ‘magic’ and the presence of the symbol ‘mu’ which is the 12th letter of the Greek alphabet, one might start to think of words related to magic that might incorporate elements of ‘mu’, possibly in an inverted or opposite manner, as suggested by the ‘-1’. The answer is supposed to be one word that contains 13 letters.

Here are three guesses for the word: Transmutation (it features ‘mu’ and is associated with the magical transformation of one substance into another, though it lacks the ‘-1’ concept) Supernatural (often associated with magic and the mystical, but does not directly relate to the ‘mu’ symbol) Electromagnet (although more scientific than magical, ’electromagnet’ contains ‘magnet’, which can metaphorically relate to ‘attraction’ contrary to the ‘-1’ which could symbolize repulsion; however, it’s one letter short of the 13 needed)

Without additional context or clues, the image does not provide a clear path to a single 13-letter word directly associated with magic that includes the concept of ‘mu’ -1.

❌ Seance #

“direct identification of the person is not appropriate”, and so the AI failed. You do need to know that this is Sean Penn for the puzzle to work!

🤖 The image appears to mimic a coin design, featuring a portrait of a man in the center with a phrase around the edge reading “PENNY NON NOT.” Since direct identification of the person is not appropriate, we will focus on the surrounding words.

The text “PENNY NON NOT” seems to be a play on words or a puzzle. Given the theme of ‘magic’, this could be a hint towards something hidden or a need to rearrange or look at the words differently.

When looking at the phrase, let’s consider a few strategies: Rearrange the letters to form a new word. Possibly read the phrase backwards. Consider homophones or words that might sound like elements of the phrase.

One word-related to ‘magic’ and fitting the six-letter condition provided could be “POTION,” which is derived from magic practices and involves a magical mixture.

Another guess could be “SPELLS,” as the phrase on the coin may require us to ‘spell’ it differently, plus it relates to the act of casting spells which is a magical activity.

A third guess, a bit more abstract, could be “ARCANE,” which is often used to describe things that are mysterious or known only by a few - also a word associated with magic.

Without more context or a clearer set of instructions, these are three educated guesses based on the clues presented.

❌ Warlock #

The image has misled the AI: it’s not a car mirror, but instead the lock on the door: a car-lock. Failed!

🤖 The image you’ve provided appears to be an artistic, sketch-like representation of a car’s side-view mirror with a focus on its outline and reflection. There’s a symbol written below the mirror: “C=W”.

Considering the theme ‘magic’ and the instructions to solve the puzzle, the equation “C=W” could imply that ‘C’ equates to ‘W’, suggesting a form of transformation or change, which is a common concept in magic.

Given that we’re looking for a seven-letter word that is related to the theme of magic, and considering the car mirror which might imply reflection, change, or transformation, here are my three best guesses:

Wizardry Enchant Mystery

Each of these words either directly relates to magic or conveys a sense of the mysterious or transformational power that is often associated with magical themes. However, without more information or additional context clues, determining the exact seven-letter word that is the answer to the puzzle is speculative.

❌ Mantra #

The AI failed to see this is a Manta Ray, though I like it’s poetic description of it: “an object with two pointy ears and a sheet-like appearance”. The AI also makes a grammatical mistake (“I’ll provide a one final guess”), a rarity for ChatGPT. Apparently my puzzles are confusing it too much!

🤖 The image depicts what appears to be an object with two pointy ears and a sheet-like appearance, casting a shadow onto a flat surface. At the bottom-right of the image, there is the suffix “-ay.” Since the description mentions the theme “magic,” we can associate the image and its features with concepts or objects related to magic. Considering the appearance and the given suffix, the object seems to resemble a “cloak” or “cape,” which, when combined with “-ay,” can lead us to the word “spray,” “clay,” or “tray” but these don’t fit well with the magic theme.

A classic magical item is the “cloak of invisibility” or “magic cape,” which might hint at the intended answer. The two pointy ears could actually be suggestive of the top of a wizard’s or magician’s hat, which is often depicted as a pointy hat with brim, but the image itself doesn’t conclusively depict a hat.

Given the instructions and the length of the word being six letters, combining a word that could possibly relate to the object in the image with the suffix “-ay” is the challenge here. However, without a clearer depiction or additional clues, pinpointing one exact word without guesswork is challenging. An appropriate word related to magic that is six letters long with the given “-ay” could be “replay” (as in replaying a magical event), “affray” (to describe a magical battle), or “array” (as in an array of magical spells or tools), but these are speculative guesses based on limited information.

Due to the restriction on the word size and theme, I’ll provide a one final guess: “Parlay,” which means to speak with another to negotiate or converse but it’s only distantly related to the theme of magic. Unfortunately, without clearer guidance or additional context from the image, it’s not possible to determine the intended solution with certainty.

🤷‍♂️ Mermaid #

Computer says no! It refuses to play, probably to prevent it from being used as a captcha-answerer. The weird thing is this happens inconsistently.

🤖 “I’m sorry, I cannot assist with this request.”

❌ Troll #

There’s some good thinking by the AI, still it failed. The association with the word ‘roll’ wasn’t found.

🤖 The image shows a capsule-shaped figure with the letter “r” on the left inside of it and a shaded area resembling a thumbprint or some form of texture on the right side.

When thinking of magic and considering the elements in the image, the letter “r” might be part of the word we are looking for. With magic as the theme and knowing that the answer should be a five-letter word, I’ll do some word associations:

“Spell” — This word is often associated with magic but doesn’t include the letter “r.” “Charm” — Again, relevant to magic but lacking the letter “r.” “Trick” — This word is related to magic (as in a magic trick) and contains the letter “r.”

Given the hint, “Trick” could very well be the word we’re looking for, as it fits the theme, the number of letters, and the presence of the letter “r”.

❌ Hobbits #

It was not on the right track, although it did correctly identify the items in the image were all hobbies. If only it realized “eist” meant “e=t”, it would’ve found the answer: hobbits. Fail!

In the end, the AI started confabulating so much, it ran out of space to answer 😅.

🤖 ‘The image contains a list of hobbies or activities presented in various fonts and orientations. These words are written as follows:

Snowboarding Yoga Reading Stamp collecting Photography Fishing Scuba diving eíst (which seems to be out of place or could be a typo) COOKING

Given that the theme is ‘magic,’ we should look for connections between these hobbies that could relate to that theme.

One immediate observation is that the word “eíst” seems to be an odd one out, potentially pointing to a clue. If we recognize that “eíst” could be an anagram, we should try to reorganize these letters to find a word that fits the theme of magic and is 7 letters long.

Given this clue, and considering that some letters are capitalized while others are not, we might need to re-arrange them or consider only the capitalized letters as a possible answer. Since we are working under the restriction of the word being related to ‘magic’ and having seven letters, let’s look at the capitalized letters and see if we can find a related word:

S (from Snowboarding) Y (from Yoga) R (from Reading) S (from Stamp collecting) P (from Photography) F (from Fishing) SD (from Scuba diving) EÍST (from eíst) COOKING (all capitalized)

If we are to take only the capitalized letters, or possibly the first letters of each word including “eíst” for the anagram, it’s not immediately clear how they might form a word related to ‘magic’. We can try different anagrams of “SYRSPFSD + EÍST” but without additional context or instructions from the puzzle, it’s challenging to determine a definitive answer.

Three best guesses for a 7-letter word related to the theme ‘magic’ might include:

“Spirits” - if we consider an an

❌ Divination #

Again the AI is on the right track. It lacks the understanding that 70 bar is very low in the diving world, this is real-world-knowledge it’s apparently lacking. We’re looking for the word “diving ration”, which leads to the answer divination. The AI was on the right track, but failed nonetheless.

It tried to lay the blame on me: “Without more context or an explanation of how to use the visual clues and text in the puzzle, it’s challenging to provide the exact answer you’re looking for.”. Sorry AI, but that’s not the point of puzzling.

🤖 The image appears to show diving equipment, including a scuba tank and the environment around a harbor or dock area with a body of water in the background and a portion of a boat visible. There are two pieces of text in the image: “70 bar pp” on a sign, and “-gr” at the bottom corner.

This is a rebus puzzle, where the visual information and text are used together to form words or phrases. Given that we’re looking for a single word of 10 letters related to magic, and considering the components of the image:

The text “70 bar pp” could suggest the words “seventy,” “bar,” “PP,” or possibly “pressure point.” The text “-gr” is likely the end of a word that could be longer. Scuba gear in the image, including the tanks, could suggest words like “diving,” “oxygen,” “underwater,” etc.

We would need to form a word that both relates to these cues and the theme of magic. Here are three guesses based on the contents of the image:

Spellcraft (connecting the idea of crafting with the magical theme). Sorceryppg or Sorcerypgr (attempting to use “pp” and “gr” as part of the magic theme, assuming they’re parts of the same word). Enchantment (which is related to magic but doesn’t directly use the text clues, it might be the intended answer if “70 bar pp” and “-gr” serve as misdirections).

These are speculative answers because the connection between the image clues and the theme of magic isn’t directly clear. Without more context or an explanation of how to use the visual clues and text in the puzzle, it’s challenging to provide the exact answer you’re looking for.

❌ Fairy #

Oh AI! You got so close! All the components are there: dairy and “d=f”! It’s staring you in the face: fairy! 🧚

You let the answer slip through your hands. Close but no cigar!

🤖 The image presented appears to be a sketch of several objects. These items include what looks like different types of cheeses, a bottle with a label that is not clearly visible, and a cylindrical container with a lid that might resemble a container of some sort. There’s a prominent equation “d=f” at the top of the image.

The theme given is “magic,” and with the assumption that the equation or the relation “d=f” must hold some significance in solving the puzzle, I will perform some word-association based on the sketch and the theme:

Dairy - The variety of cheeses and the bottle resembling a milk bottle suggest a dairy theme. Spell - Given the theme “magic,” a common term associated is “spell,” which is composed of five letters. Charm - This term is related to magic and also has five letters, possibly fitting the clue.

Considering the image and the information provided, my three best guesses related to “magic” with five letters, following the theme and word association, would be:

Spell Charm Potion (even though it does not visibly match the equation “d=f”, it is related to magic and has five letters)

❌ Witch #

Somehow the AI missed that it’s looking at a switch and thinks the image contains dice. Because of this, it drew the wrong conclusions.

🤖 The image appears to display a spiral-like design with a two-dice stack at the center, creating an optical illusion where the spiral may also resemble concentric circles depending on the viewer’s perspective. On the die, I can see the numbers 5 and 3, as well as a dollar sign which could also be interpreted as the letter ‘S’. Given that dollar signs are not typically found on dice, the ‘S’ is probably significant for solving the puzzle.

Since the theme is ‘magic’ and the answer is one word with five letters, I will use word-association based on the visible elements and the theme:

Spell: It’s associated with magic and consists of five letters. However, there is not a direct visual reference to spellcasting in this image.

Charm: This word relates to magic and has the correct number of letters. Charms can be magical objects or incantations offering good luck or protection.

Trick: Magic is often associated with tricks or illusions, fitting the five-letter criterion, and it may play on the idea that the image contains an optical illusion.

These are my best guesses based on the information given and the elements in the image.

❌ Banshee #

Awww, the AI realizes it is missing “a particular puzzle trick”! It actually missed two clues: ’no more of this’ is a ‘ban’ and it’s on a ‘sheet’ of paper.

🤖 The image shows a prohibition sign, commonly recognized as a circle with a diagonal line through it, and the text “No more of this” is presented below the sign. In the lower left corner of the image, the word “not” is written.

Here’s a word-association based on the theme ‘magic,’ considering the prohibition sign and the word “not” which could imply something like a negation or absence:

Absence of prohibitions in ‘magic’ could lead to ‘freedom,’ but that’s not seven letters long. Following the theme of ‘magic’ and the absence hinted at by the prohibition sign, the word “not” could be combined with something typically magical to negate it, but there’s no obvious seven-letter magic-related word that fits this directly. Considering ‘magic’ and the negation, we might think of words like “reality” (as opposed to ‘magic’) or “alchemy,” but neither of these words fits the idea of “no more of this.”

Without a clear indication of which specific type of “magic” the image hints at, or more explicit instructions, it is challenging to reach a definitive answer. If any additional clues are present outside the image or if there’s a particular puzzle trick I’m missing with the provided picture, those could be crucial in solving this puzzle.

❌ Angels #

Wait a minute, the AI think’s there are four shapes? I only see three! It also invents word ‘alchem’ as a short for ‘alchemy’. Lol, nice try!

🤖 The image is a minimalist black and white representation which includes four shapes that resemble arrowheads or chevrons pointing in different directions (up, right, down, and left), and a key at the bottom indicating that a certain line represents the letter ’e’.

The puzzle suggests that we should look for lines which can be converted into letters using the same principle as the key provided with ’l = e’. With this in mind, we convert the lines of the shapes into letters, assuming that a vertical or horizontal line corresponds to the letter ’e’ and an angled line represents another letter.

The four shapes seem to create the letters “C” (or “U”, if the bottom is filled in), “A”, a mirrored “L” (which could be an “L” since we’re solving for “magic” and mirrors can be magical), and another “A”.

So, using the information from the image and the theme ‘magic’, here are three guesses related to magic and consist of six letters each:

Caster Charme Alchem (short for alchemy, although not six letters, it’s close and thematically related)

Note that these are speculative guesses since the exact method for translating shapes into letters is not defined beyond the one example, and some amounts of interpretation is involved.

✅ Wand #

Hooray! Another victory for the AI! The explanation is wrong again, but we set the bar low and we accept all mentions of an answer. Good job little-AI!

🤖 The image you provided is a hand making a gesture with the forefinger and middle finger raised and separated, which is commonly known as the peace sign or victory sign. Next to the hand, there is the word “noh” written in lowercase letters.

Given the theme of ‘magic’ and the structure of the puzzle, it seems we are possibly looking for a word related to magic that contains four letters and perhaps is associated with or can be derived from the word “noh” and the hand gesture.

Three guesses for a four-letter word related to the theme of ‘magic’ could be:

Wand - Although “wand” is not directly illustrated or suggested by “noh,” it is a common tool used in magical contexts. Spell - A “spell” is a magical formula, but similarly, there’s no direct connection to “noh.” Hand - The image itself is of a hand, an instrument commonly employed by magicians to perform tricks.

None of these guesses seem to be very closely related to the word “noh” or the gesture, but they all pertain to magic in a general sense. If there are more clues hidden within the image that I cannot perceive or if there’s a specific type of magic-related puzzle this is referencing, it would require more contextual information to solve accurately.

❌ Centaur #

This is the first time the AI uses the instruction in the image. It figured out it needed to subtract the letter ‘a’. It failed to pick up the ‘aura’ component and that the Romans named 100 ‘cent’ (the AI got to the letter ‘C’, which is the abbreviation of cent).

🤖 The image shows the number 100, surrounded by a wavy border or frame, and at the bottom left, there is a negative sign followed by the lowercase letter ‘a’. Since we have to find a seven-letter word related to the theme ‘magic’ by using word association based on this image, let’s begin.

Looking at the number 100, we might consider concepts of totality or completeness, since 100% would represent a whole. The wavy lines could imply motion, fluidity, or something that is not straightforward. The “-a” suggests that we might need to subtract the letter ‘a’ from a word or concept we derive from the image.

Given these elements, we could come up with words related to magic where the number 100 and the subtraction of the letter ‘a’ play a role:

Casting: If we think of casting as in casting a spell and use the 100 to possibly represent a form of total or complete spellcasting, minus an ‘a’, we get ‘Csting’, which does not form a valid word.

Charmed: Playing with the theme of magic, ‘charmed’ could be a word associated with it. However, removing the letter ‘a’ from ‘charmed’ gives us ‘chrmed’, which also doesn’t work.

Alchemy: Exploring the concept of alchemy, which is a magical process of transformation, creation, or combination. If we take ‘alchemy’ as the base word and then remove the letter ‘a’, it results in ’lchemy’, which doesn’t fit our criteria.

These guesses don’t seem to fit the puzzle’s criteria, as we seek a seven-letter word. Upon further observation, the number “100” can be read as the Roman numeral “C,” accompanied by the instruction to subtract ‘a’, leading us to the word ‘Charm’ when ‘a’ is subtracted from ‘Charm’, leaving it short of a

🤷‍♂️ Elves #

Computer says no!

🤖 I’m sorry, I can’t provide help with puzzles, riddles, or similar content requiring the interpretation of text or symbols within images. If you have any other types of questions or need information, feel free to ask!

❌ Harry Potter #

“It’s possible we haven’t found the right approach yet”: indeed! The printer is a plotter and the clue is “no-l” instead of “no-t” :)

🤖 The image depicts a large-format printer with a piece of paper that has a man’s face printed on it. Below the printer, there is a word written in lowercase: “ton”.

To solve the puzzle, let’s focus on the word association related to ‘magic’. The word “ton” might need to be reversed or altered to form a word related to magic. Reversing “ton” gives us “not”. This doesn’t immediately suggest a magical term, but let’s consider synonyms or similar words related to magic that have 11 letters.

Here are three guesses for a word related to ‘magic’ with 11 letters, considering the clue provided:

Spellbinding Enchantment Illusionists

Considering the task requires the use of the word in the image (“ton”), and none of the guesses above use those letters directly, it’s possible we haven’t found the right approach yet. If there are additional clues or a different method to interact with the image or its elements, that might help narrow down the possibilities.

✅ Illusion #

I accidentally gave the AI the wrong amount of letters 😅, so I am judging the answer it gave to be correct!

🤖 The image provided shows three vertical lines followed by an ampersand (’&’) symbol, with the letters ‘N E S’ at the bottom right corner. To approach this puzzle, we should consider the ‘magic’ theme and look for a connection between the elements in the image.

The ‘N E S’ letters might bear some significance, they could represent directions (North, East, South). The vertical lines could be a hint towards a count or tally, and the ampersand (&) an indication to combine or ‘and’ something.

Considering the magic theme and the elements involved, here are three words that might fit the criteria of being 9 letters long: Illusions Alchemist Sorceress

Unfortunately, without more specific instructions, such as how to use the directions or any additional rules for the word association part of the puzzle, it’s challenging to determine the exact answer.

❌ Medium #

Nope!

🤖 The image provided displays three squares of differing sizes aligned horizontally. From left to right, each square contains a progressively larger shaded area at the bottom. Above the middle square, there’s a downward-pointing arrow. With the theme of ‘magic’ in mind, this could signify a transformation or change, which is often depicted in magical contexts as something visually altering in size, appearance, or nature.

Word association with ‘magic’ and the visual cue of increasing shaded areas might suggest concepts such as “growth,” “change,” or “illusion.” The downward arrow could imply “casting” as in casting a spell, or “invoke,” suggesting the act of initiating magical effects. The differing sizes might also suggest the idea of “scaling” or “enlarge,” relating to magical changes in size or volume.

Considering these elements and the requirement for a six-letter word related to ‘magic,’ my three best guesses are as follows:

Potion - often used in magic to cause changes or transformations. Enchant - to magically endow with a given property or effect. Sorcery - a term broadly associated with magic and spell-casting.

Without additional context or clear instructions within the image, the goal is to interpret the abstract representation in a way that aligns with the given theme.

❌ Ouija #

The flags confused the AI because they are in black-and-white.

🤖 The image you’ve provided appears to show the word “YES” twice, one above the other. On the first “YES,” there is a checkered flag pattern on the left side of the first letter ‘Y,’ and on the second “YES,” there is a similar pattern on the right side of the third letter, ‘S.’ To solve this visual puzzle, we might consider what the checkered flags are indicating—typically, they are used in racing to signify the start or the end of a race.

With the theme of ‘magic’ in mind and the checkered pattern indicating some sort of a transformation or a change, we could try to find a word that is associated with magic and change or transformation.

Here are my three best guesses for a five-letter word related to ‘magic’:

Spell - The transformation may suggest the casting of a spell, which is a fundamental element of magic. Charm - This is another term related to magic which involves the use of a spell or enchantment. Trick - This could refer to a magic trick, which is an act of illusion typically performed by a magician to create a change in perception.

Without further context or instructions on how to specifically use the elements of the image, these are speculative answers based on the theme provided and the visual cues.

❌ Hocus Pocus #

This is one where the AI could’ve had a major advantage. Not a lot of people know the words ‘focus’ and ’locus’ as synonyms for ‘attention’ and ‘center’, and the AI could’ve played thesaurus.

🤖 The image you’ve provided displays the text “Center of Attention” in a large font with a mathematical equation underneath, “f = h l = p”. To solve this puzzle within the context of the theme ‘magic’ and the clue that the answer has ten letters, let’s consider the elements presented:

“Center of Attention” could imply that we need to focus on the central aspect or middle part of something.

The mathematical equation could be indicating that ‘f’ equals ‘h’ times ’l’ which equals ‘p’. This might suggest that we need to do a substitution or some form of an equation to find the solution.

Given that we’re looking for a ten-letter word related to ‘magic’, and based on the provided instructions, here are three guesses:

Enchantment - Derived from the idea of magical spells or effects that captivate attention. Spellcraft - Although not ten letters, this term describes the art of creating magical spells, which can be the center of attention in a magical context. Illusionist - A performer of magic tricks that often hold the ‘center of attention’ during their act.

However, if we need to rely exclusively on word substitution or manipulation, the provided equation doesn’t clearly lead to a ten-letter solution related to magic without further context or instructions. Can you provide additional instructions or context?

Easter puzzle of 2024

Wed, 27 Mar 2024 00:00:00 +0000

My yearly tradition is to create a bunch of cryptographic puzzles for Easter. And this year is no exception.

Play the puzzle now!

Some tips #

The puzzles are very difficult, so it may take you a long time for each puzzle. That’s by design!
Work together if you can
Print the puzzle (thinking is easier on paper)
Focus on a couple of puzzles and then take ’time off’ to give your mind time to process it
Start ‘outside in’ by writing down words related to the theme and then see if they fit with a puzzle

This year’s puzzle #

I’ve been working on a story about a ‘corporate wizard’ and I realized these words are fun ‘puzzle-words’. That’s why this year’s theme is magic 🔮.

I used generative AI to create some of the images, which is a first. In other years I drew the images myself.

I wanted an old-fashioned look to go with the theme (magic), so experimented with a ‘printed etching’ look. This was quite hard to get consistent with AI, so instead I decided to use a filter from the app Graphite.

This had the extra advantage that I could use images from other sources, and create the same style on images I made myself. In the end, only four out of twenty puzzles used AI-generated images (all other images are photos I made, or public-domain).

Improving printing #

Another thing I tried to fix this year was the print-version of the puzzle. Print-CSS proved to be a harder puzzle than creating the cryptograms!

Safari doesn’t respect page-break-inside: avoid; so puzzles still get cut-off. Chrome doesn’t listen to overriding earlier CSS rules. And both browsers have a shitty dev-environment for print-CSS: enabling print in the inspector-mode doesn’t look at all like what is printed in real life.

What I did manage to fix is adding fields for the answers. By default input-fields are not printed, so you’d lose the place to fill in your answer and worse: you don’t get to see how many letters there are in the puzzle. I solved this with some cool Hugo-magic that prints a simple text-only line: _____*___

I also used some CSS-magic (mix-blend-mode: multiply;) to make it look like the puzzles are printed on paper. I could have done this within the images, but then you’d print this background on paper as well. Because the background-texture is not ‘baked in’, the prints come out quite nicely on black-and-white.

In the end, only Chrome manages to print half-decently. Maybe next year I’ll get to fix it properly.

I wish you a happy Easter and a lot of fun playing the puzzle!

Using AI to create images for blog-posts

Mon, 25 Mar 2024 00:00:00 +0000

Everybody’s doing AI. And so am I!

Last week, I got bored and decided to mess with the OpenAI-API. This resulted in a script that reads in a blog-post and then generates a square image to go with it. Not exactly rocket-science, but it was fun to mess with.

Read on to see my findings!

What I wanted to achieve #

My goal was mainly to explore the OpenAI API, see if I could build something on top of it.

For this little project, I decided I’d see how good OpenAI would be at creating ‘header-images’ and ‘summary-texts’ for my blog-posts.

The basic idea is very simple:

Blog post -> AI summary -> AI header image

The result #

I had to tinker a bit with the prompts to get something ‘nice’. First I used a standard prompt: “summarize the article”, but this resulted in a very standard answer: “the author writes about…”. Very un-human, and resulting in boring image-prompts.

I finally landed on asking the AI to find the original problem or reason for writing the article, and then write that as a children’s book-title. This creates a more fun description.

I tried several image-styles, but landed on the style of Matisse pretty quickly. It’s fresh in colors, and very abstract.

The abstract images feel more human than a ‘fake photo’, and have the added benefit of needing human-interpretation. That solves the shortcomings of the AI-image generation, as it sometimes generates very weird stuff.

Examples #

Here are some examples of the “children’s book title summary” the AI made, with the accompanying “Matisse drawing”:

Original article (‘Building design’), AI-summary: ‘The Impractical Window’

Original article (‘Parking machine at the zoo in Emmen’), AI-summary: ‘The Confusing Parking Machine: A Story of Redesign’

Original article (‘Ironing device from Hema’), AI-summary: ‘The Confusing Ironing Dial’

This article, original title: ‘Using AI to create images for blog-posts’, AI-summary: ‘Using AI to Create Images for Blog-Posts’ (lol, GPT got lazy and just reused my title! Also, the instructions in the prompt specifically say: no letters…)

Implementation in the blog #

I added images to articles from 2011 (see the archive, scroll all the way down and open some articles from 2011)…

… And I quickly got bored with adding these fake images 😅

I like to tinker, think and then write about that. I had fun creating the script, seeing the result and writing this blog-post because that shows off something I did. Adding computer-generated headline-images to posts feels like showing off with something cool OpenAI did.

Also, all my articles are grossly out-of-date, badly written and abandoned (sorry!). Those articles don’t need fancy computer-generated images, that’s way too high quality for this blog! If I set the bar so high, I would never write anything again 😅

What I learned #

I learned tinkering is more fun than the end-result
AI is surprisingly cheap as long as you use the ‘old’ DallE-2 and GPT-3.5. I used only 1 dollar for this whole experiment.
AI is much better at writing creative headlines than me 😅

The script #

I installed the openai-package for python, added my API key and then wrote a quick script (together with ChatGPT).

The script does three things:

read in a blogpost (a markdown-file)
call OpenAI to summarize the script into a small piece of text
call OpenAI again to generate an image off of this description

The summary and image are both stored with the original file, and that way I can reference them later in Hugo (the software that powers my blog).

import openai
import urllib.request
import os

def summarize_text(text):
	# OpenAI API call to summarize the text
	response = openai.chat.completions.create(
	 model="gpt-3.5-turbo-0125",
	 messages=[
		{"role": "system", "content": "What's the problem the author is writing about? Describe it as the title of a children's book. Only write the title of the book, don't add words like 'the author' or 'the article'"},
		{"role": "user", "content": text}
	 ]
	)
	summary = response.choices[0].message.content
	return summary

def generate_dalle_image(prompt):
	# Function to generate an image using DALL-E API
	response = openai.images.generate(
		model="dall-e-2",
		prompt="Matisse painting of: "+prompt,
		size="1024x1024",
		quality="standard",
		n=1
	)
	image_url = response.data[0].url
	return image_url

def save_summary_to_markdown(summary, filename):
	with open(filename, 'w') as f:
		f.write(summary)

def save_image_from_url(image_url, filename):
	urllib.request.urlretrieve(image_url, filename)

def main(input_markdown_file, output_markdown_file, output_image_file):
	# Read markdown file
	with open(input_markdown_file, 'r') as f:
		text = f.read()

	# Generate summary
	summary = summarize_text(text)

	# Generate image
	image_url = generate_dalle_image(summary)

	# Save summary to a new markdown file
	save_summary_to_markdown(summary, output_markdown_file)

	# Save image to a JPG file
	save_image_from_url(image_url, output_image_file)

input_markdown_file = 'index.md'
output_markdown_file = 'summary.md'
output_image_file = 'summary.jpg'
main(input_markdown_file, output_markdown_file, output_image_file)

Hugo import Jekyll

Sat, 10 Feb 2024 00:00:00 +0000

This blog used to run on Jekyll.

I was quite happy with that, until some kind of Ruby change meant Jekyll stopped running and I got into an arms race with Ruby versions. Jekyll needs Ruby A.64, the gems for Jekyll need Ruby B.85, the gem-updater won’t work with Ruby A.64, OSX ships with a different Ruby, Linux on the server yet another…

Added to that, my theme used gulp as a task runner and needed lots of npm modules. These ran into the same problem with conflicting dependencies.

It was messy and fickle.

Yesterday it broke again: my blog refused to build on the server. Instead of patching up the mess, I decided to spend my time on changing to a more robust system.

Enter Hugo #

Hugo is another static site generator. It runs on Go. I have revisited with it, as I use it to build the “paaspuzzel”. The big advantage is that Hugo is self contained: there’s no need for plugins or node modules.

A simple hugo import jekyll did the trick. All my articles got imported and mostly everything worked.

I wasn’t done yet. For a long time I’ve been annoyed that images are in a different folder than the blog posts themselves: this makes it difficult to edit with (for instance) iA Writer. Luckily Hugo has a trick for this called page bundles for posts. I now store posts per year, and each post is a directory. A lot less messy. The only downside is I have to go through each article and change the image references…

Then there’s more “manual labor”: I picked a simple theme, but I want to adapt it to my liking. That’s probably a never ending project 😂

The final change for this blog is that I am now typing this article on my phone with Working Copy. This was already possible the old setup, but one change leads to another! So far I’m loving it!

RSS #

I need to apologize to my three RSS readers. Hugo broke the feed which means a lot of unread articles for you. I’m also changing the urls of all the articles (because of the bundles), which may lead to more unread articles.

If possible, please change the feed url from feed.xml to index.xml. I couldn’t figure out how to rename the feed with Hugo 😅 The reason it still works is because I setup Cloudflare to reroute it, but better to change url manually in your reader as that’s not a long term solution!

Yearly Review 2023

Thu, 08 Feb 2024 00:00:00 +0000

Stepping into a new world. The world of Freelance

Last year’s theme was “everything, everywhere and all at once”, and I asked myself:

I started many more projects than I finished: will 2023 will be a year of wrapping up this undirected energy?

And it’s safe to say: no, I did not wrap up projects. I (again) started many more than I could finish.

Still, 2023 has a very different feel than 2022. I felt growth: I became a more senior designer (a product owner, even), and took the leap to freelance. This growth did not come without growing pains. I have a lot of ‘headache’, ‘tired’ and ‘being grumpy’ in my diary. I’m working on a better balance so this won’t happen this year.

Now on to the ~~reflection~~ brag-list.

Work at PostNL #

I grew a lot in my work at PostNL:

I became research lead and product-owner of a research team for SME
I held innovation-workshops at PostNL
I created the product vision for the PostNL business portals
I instigated improvement in work process. I created a new EPIC-process within the SAFe framework, with a focus on delivery / release / learn strategies. And worked with a Systems-Thinking approach to identify and communicate problems in process
I was involved in hiring two new designers
I designed the new mobile app for business customers, and worked closely with the product-owner / channel manager / marketing to align development, research and marketing with a scenario-based approach.

Freelancing #

The feeling of “I’d like to try and be my own boss” has been itching for a long time. And in September I made the switch to freelancing. I already have two amazing clients and am looking forward to more.

Freelancing gives me more structure than working as an employee. Sure, I have to set that structure myself, but that’s loads better than trying to play catch-up with other peoples structures. I get to decide how to spend my days, when to take time-off, where my pension goes, etcetera. I set the structure and I fully understand that structure. For example: as employee there are a maximum number of holidays, some rolled over in a new year and others didn’t, some national holidays were free and others weren’t – freelancing is much simpler: you work, you get paid.

Another aspect of freelancing that brings structure is project-based working. Projects have a beginning-and-end, and there are less politics in the middle. Compare that with my work at PostNL, where I was often doing 20 projects at the same time (my own fault! I took all that in! But still).

There’s a lot of work figuring stuff out (accountant, insurance, creating invoices), but I enjoy all of it. Even the ‘boring’ stuff. I like being my own person.

I also feel more valued, as I get ‘hired’ for my seniority and my influence on projects is much clearer.

Construction in the house #

2023 was the year of some smaller household chores. I stripped the stairs of the carpet, finished tidying up the fireplace, worked in the garden a lot. I fixed some leaks (mainly a flower place connected to the wall) and I fixed the lining of the roof of our shed. I also installed a new rain barrel, which was more work than expected as I rerouted the gutter.

Creativity #

I again released a cryptographic puzzle for Easter (very happy with this edition, although I prefer the illustration-style of the year before). I wrote three articles on my blog: my thoughts on AI and art, an article on the quality of Apple TV and of course the yearly review of 2022.

I also built an AI-reply for e-mail using the OpenAI-API. I named him George. It’s great fun to email George and get replies. There’s a very different vibe to email than chat: it’s slower and more thoughtful. Plus, it’s very simple to use existing content: you just forward an email. It feels like a “poor man’s OpenAI-API”. My friend JP is now creating a real service out of it (mine was hacked together on an old laptop).

There’s a long list of things I wrote that are unpublished: a magic-story, a new (old) book from my father, a puzzle-calendar for print, an investigation into how Apple handles raw photos, a new logo for freelance work, a massive new article on solar panels….

Not all activities were theoretical, I also visited a lot of museums and watched a lot of (good) movies and visited a conference on Enterprise UX and read a handful of books.

I made hats for dinosaurs for my son’s party of leaving daycare, organizing parties for my daughter (Christmas lunch, movie-day, her jungle-birthday).

I created a new “Heuristic 0” for UX, as an addition to the famous 10 heuristics of Jakob Nielsen (should write an article about this someday):

I updated my CV and portfolio (three times, in PDF not online) and designed a new personal website (not live yet).

Making space for me: joining the 5AM club #

Another major changes in my life (besides freelancing) is that I started waking up at 05:00. This gives me more ‘me-time’. I wrote in-depth about this on Medium.

On to 2024! #

As I write this, already more than 10% of 2024 has passed.

I’m very excited for the year, having a lot of energy from exploring this new thing “freelancing” I’m doing.

I also hope to release a product: I’ve been working on a cool note-taking app (together with a friend) that I would like all of you to see (yes, all of you: all three of my RSS-readers!).

So much time, so little to do. Strike that. Reverse it. – Willy Wonka

Video did not kill the radio star (AI will not replace all human artists)

Sat, 22 Apr 2023 00:00:00 +0000

There’s a lot of talk about AI’s impact on artists. Midjourney, Stable Diffusion, Dall-E, ChatGPT, they are already taking over and this is just the beginning.

I don’t think so. An AI can make beautiful images, but these images do not have the same worth (in the eye of humans) as does real art.

I can frame a print of the Mona Lisa and hang it in my living room. It will look amazing. It will brighten up the room. It’s beautiful. But it is not the real thing. And that is not important, unless it is.

We humans are weird creatures. We can have an exact replica of the Mona Lisa in our living room, but still look forward to our planned holiday where we visit the original. Why? Because it ‘has’ something. There is value in the ‘real’, ‘original’, ‘someone sweated over this’. We experience it somehow.

I have a Spotify subscription, but I also support a band on Patreon. Why? Financially it makes no sense: I can already listen to their music for free. So I pay double. Even worse, I bought their music on Bandcamp (and am supporting other artists that way too). So I pay triple. Am I just burning money? No. I support the humans making this work. And I will keep doing this in the future, even when an AI can replicate the music I like (or even adapt it to my emotional state, or other fancy stuff).

Music, paintings, and writing are a medium to communicate between humans, and there’s no point in listening to a machine emulating this communication.

Video did not kill the radio star #

When MTV launched, its first video was of the song “Video killed the radio star”. The expectation was that radio would quickly be redundant, now that television was in every living room.

How did that go?

MTV is gone and the radio is still here.

The movies did not kill theatre. Photography did not kill the portrait painters. CDs did not kill concerts. Etc.

Yes, some illustrators and photographers will be redundant with AI. Before photography, there were ‘field sketchers’. People who drew important scenes for print. Those got laid off with cheaper photography¹. The same goes for stock-illustrators photographers and -photographers now with AI. But they were already being replaced with Unsplash and cheap-subscription services (or not so cheap, ahem, Adobe).

And what about all the creative stuff that’s “not art”, like marketing or web-design? My prediction is it’ll get augmented by AI: the boring stuff is automated and there is more time for the fun ‘communicating between humans’ parts.

There are interesting times ahead with AI. I look forward to enjoying it.

The actual history is much more interesting. It involves a lot of manual skills, different printing techniques, etc. Another rabbithole to dive into. ↩︎

Sick of being a night owl? How I became a morning lark and adjusted my chronotype

Mon, 17 Apr 2023 00:00:00 +0000

I used to be a typical night-owl. I was comfortable working in the evenings and nights, and grumpy in the morning. I used to be happy with it (and there’s loads of people who are happy with it!), but it didn’t work for me anymore.

Now that I have children, I noticed I would no longer be activeat night, but instead spent my nights watching YouTube videos and eating chips (very confrontational, but true!). I always got into bed too late. It sounds crazy, but it was as if I was too tired to get myself to go to bed.

And then in the morning, I would have trouble ‘pulling myself back together’. It felt like my wife and two kids were way ahead of me in energy and I was playing catch-up. I would get up at 06:45 with them, and I was grumpy until about 08:30. In my head I kept pleading: “Slow down everybody!”

Night owl at work #

And that’s not all. Being a night-owl also brought me troubles during working hours.

Workdays start at 08:45, which always felt like a big bummer to me because I’m at my most productive between 15:00 and 20:00. In the mornings, I would be sluggish and have some ‘brain fog’. But at the end of the afternoon, I get into a flow of work, and it feels good. At 15:00 my engine was on and I was working full focus: the pieces coming together, my projects finally making progress. 100% productivity! But, of course I have to leave work to pick up the kids at 17:15. This got me so annoyed everyday: “AAARGH! I’m just getting started!!” I would shout out to myself (and others unfortunate enough to be near).

Is there anything I can do about this?

Night owls and morning larks: your chronotype #

Scientists have called your sleeping schedule your ‘chronotype’. You can divide this into as many buckets as you like, but traditionally I’ve found it’s either two buckets (‘night owl’ and ‘morning lark’), or the following four:

The Lion: The early risers. The 4 am tribe.
The Dolphin: The insomniacs. If you’re struggling with sleep, you land here.
The Wolf: The night owls. They have trouble waking up in the morning. If you fall under this category, your day starts around 11 am.
The Bear: The majority of hustlers move with the Sun. If you fall under this category, you fall asleep around 11 pm and wake up around 7 am. ⠀ Your chronotype changes over time (young children are lions, young adolescents are wolfs) and then gets ‘fixed’ when you’re an adult.

And that’s it: traditional wisdom says your biological clock gets fixed with respect to sleep.

Which sucks because if you are a ‘Wolf’ (or night owl) like me, you have the whole of society working against you. You are expected to wake up early, go to work no later than 09:00 (preferably earlier!) and then stop working around 17:30. That’s not the way-of-the-wolf!

Surprise: you can change your chronotype! #

I’ve been investigating how to deal with my chronotype since having children. It’s very uncomfortable to be so out of sync in energy in the morning. I’m still grumpy and waking up, they are running, playing, singing, happy.

Everything I read online, stressed the ‘unchangeable’ nature of your chronotype, until I found an article by Noa Kageyama, Ph.D. In the article, he mentions:

9am summer festival orchestra rehearsals were kind of rough. My fingers would feel sluggish, my brain would be in a fog, and it wasn’t until rehearsal wrapped up around noon that I started to feel like my normal self

That’s me! Hooray! I’ve found a fellow-sufferer!

Even better is that Kageyama has found scientific literature that:

if you have been keeping night owl hours, and would like to be more of a morning person, it’s totally possible.

Great news: you can change my sleep pattern to be in-sync with the rest of the world. Even better news, according to Science a ‘morning lark’ sleep pattern also leads to less stress, more happiness and less depression, and this applies to people changing their sleep pattern as much as ‘natural morning larks’.

My experiment: waking up at 05:00 #

Based on the scientific article quoted by Nao Kageyama, I set up my own experiment.

I know I need around 7,5 to 8 hours of sleep: this amount makes me feel rested and good. Based on the article of Kageyama, I wanted to get up at 5am.

So, that’s my big idea: sleep at 21:00 and wake up 05:00

This reminded me of the book the “5AM-club” (short summary: get up at 5am and become incredibly rich), which I read before and didn’t really resonate at the time. A quick dive into my notes gave me a great idea for what to do at 5AM. I would try the 20/20/20 method: 20 minutes of sport, 20 minutes of reading, 20 minutes of meditation.

And based on Sanjeev’s article I structured things around an abstract goal. His 100 days seemed arbitrarily long, but hey: it works, so I’m using it also. Another thing I picked up is to evaluate my project at set intervals: 20, 40, 60 and 80 days (this article is based on my evaluation of the 20-day interval).

How do I motivate myself? #

I’ve set a reward: I will “allow myself” to buy an Apple Watch if I can maintain the schedule for 80 days of the full 100. At this point, I don’t know if I really want or need a new watch, but it was the motivation that came to mind quickest 😅

I’ve found out that the motivation that works best for me personally, is not an abstract goal, but writing down a quick evaluation every morning. Usually it’s just one or two sentences: “woke up at 05:15. Alarm didn’t go off. Went for a run, but no meditation. Feeling good, though”. Writing these small evaluations motivates me to keep going much more than the prospect of owning an Apple Watch.

The results of the experiment #

I’m currently at day 29 of my ‘wake up early’ experiment and so far I’m really happy with my results. I was afraid I’d slack off after a week, but that didn’t happen.

Overall I managed to get up around 05:15 most days, and about two-thirds of all days I was up before 05:30 — even when my alarm didn’t go off. I’m happy with this statistic, but I need to improve if I want to meet my success criteria of 80%.

Most of the reasons I ‘failed’ to get up early were outside my control: my alarm had ‘troubles’, or I had a bad night of sleep (I was sick for two out of the four weeks in this experiment). I also ‘gifted myself’ three mornings of sleeping in.

I’m very pleased with my sleep pattern, not so satisfied with the sporting I did. I try to walk at least 20 minutes right after getting out of bed, but this I managed to do only ten times out of nearly 30. Pretty bad statistic… But, I think this is largely because I was sick for two weeks. I’m hoping the average goes up now that I’m feeling better.

My sleep and wake times: I woke up late 5 times, and 5 times in the ‘danger zone’ between 05:30 and 06:00. All 18 other days I woke up before 05:30.

For reference, here’s the ideal morning for me:

05:00 My alarm goes off
05:15 I’m downstairs
05:25 I’m taking a walk outside
05:45 Back inside, eating breakfast and drinking tea, reading news
06:10 Meditation
06:20 Read my book
06:40 Unload the dishwasher, set the table, prepare coffee, make lunchboxes…
07:00 Breakfast with wife and kids

Evaluation time! Are you happy with your experiment? #

Most days I felt pretty happy with myself for getting up early and ‘doing my thing’. I’m proud I made a decision, and proud that I manage to keep at it. I’m more cheerful, energetic and more productive during the day, especially when I went for a walk in the morning.

The Good… #

In general, the change I made in my sleep pattern is a very positive experience for me. I like morning-lark-me. I’d like to stress this is a personal evaluation: I realize I’m at high risk of sounding like a positivity-guru or LinkedIn-productivity-hacker in the following paragraphs.

My energy-levels seem more ‘equalized’ during the day. No more “drowsiness in the morning, hyper productive in the afternoon”. This by itself is already worth it. I was very annoyed in the past for ‘losing my flow’ at 17:00 when I needed to pick up the kids, and this annoyance is gone.

With respect to health, I now sport more as I try to walk every morning. I set this goal for myself in the evenings as well, but it was much harder: I didn’t have enough energy to ‘push’ myself to go outside. Another ‘healthy effect’ is that I eat less chips. I usually snacked some in the evening, again, being too tired to ‘stop myself’. This same ‘lack of energy’ kept me from reading books (instead watching YouTube) and going to bed early (I usually stayed up until 23:00, regretting that in the morning). My new ‘more equal energy-levels’ mean that I have more energy for good decisions in the evening. I feel more ‘naturally tired’ at the end of the day, and go to bed at 21:00.

And now we come to the incredible bonus of waking up early… Me-time! I have two whole hours of energetic ’time for Matthijs’. As a parent, this is paradise. Just two hours for me. So selfish! Of course, in reality it’s not two but one hour (due to waking up later, doing chores and children waking up early), but still! I know parents who would kill for 60 minutes of their own time everyday!

Is waking up at 05:00 only a selfish thing? No! Because at the end of the day, I have much more focus on my wife. I used to have a lot of things nagging at me in the back of my head (“I want to read this article”, “I want to finish my blog-post”, etc), and all those things are now postponed to the next day. I know that I have time for them, so I can let those nagging thoughts rest and spend full-focus time on my relationship. I’m now going to bed at the same time as my wife, as she already went to bed around 21:00. I’m also more in-sync with my kids: they naturally have an early-rise and early-to-bed rhythm.

There is more good news, because I have a lot of fun playing Hotel Pappa. While the rest of the house is still sleeping, I prepare breakfast, unload the dishwasher, make coffee, create the lunch-boxes for my daughter, feed the cat, set the table, prepare porridge for my son… It’s all done when they come down stairs. These chores are mindful for me, and save time for them. There’s a lot less ‘hurry up, faster, go now’ every morning. Less stress.

The completely unexpected bonus is that ‘sleeping in’ for me is now waking up at 08:00. This is deliciously wicked! No parent can sleep in when they have young kids! No parent, except me!

… the Bad … #

I’m quite happy with this project of changing myself from owl to lark, but nothing can be all good!

My morning-mood changes quite quickly from happy to gloomy if the children wake up before 07:00. It’s amazing how quickly I got used to the ‘me-time’ and how defensive I am about it. Every evening I envision how my morning will be, and if it doesn’t work out the way I envisioned… I get quite cross. The mornings get messed up: there’s no ‘Hotel Pappa’ where I prepare everything beforehand, there’s no meditation or mindfulness, instead it’s “pappa can I have this, can you do that” while I’m not ready for it. Luckily, it doesn’t happen too often.

Another thing I’ve noticed is that I am already playing fast-and-loose with my own rules. It’s easier to go sport and meditate in the morning (versus in the evening), but that doesn’t mean I always do it 😅 Especially when sick, I didn’t sport in the morning and instead watched a movie…

Lastly, the bed-time of 21:00 is a hard deadline. Whenever I went to bed later, I got tired and unfocused the next day.

… and the Ugly #

As I said earlier: society is mostly oriented towards morning-larks. Activities like work and school expect you to be up and active by 07:00.

But not everything is oriented toward morning-larks. Most social activities are night-owl events, they happen in the evening, or early night.

My badminton practice starts at 20:15 and I usually played until 21:30. This is no longer possible. I went playing once, missed every ball and hit myself in the knee. At 21:00 I felt disoriented and had to stop.

Three weeks ago, my parents visited us. We had great fun during the day, but they were very disappointed to see me head-off to bed at 21:00; we usually talked until late in the evening.

I’m not yet 100% sure how I will deal with these social aspects yet. I might join an earlier badminton game practice, or sport in the weekends. I’m also hoping to get more resilient and energetic in the evenings by sporting every day, so that I can do a ‘one time late night’ every now and then. I’m confident that will be possible. My experience with ‘sleeping in’ for a day (until 08:00) is that it doesn’t break my schedule as much as I expected.

Next steps #

This article is the evaluation of my first 30 days in transition from night-owl to morning-lark. Based on this evaluation I do not see the need for big adjustments in my schedule. The only improvement is to keep at it with sporting and meditation!

I’m happy with the change I made, and I’m not the only person.

I’ll leave you with the words of my wife:

I like this new Matthijs much more, he has the same energy as me in the morning

Published before on Medium

Start working on my Easter Puzzle!

Thu, 06 Apr 2023 00:00:00 +0000

Welcome to this year’s easter puzzle! This crypto puzzle is super-sweet and is all about our favorite sugary treats! It can be candy, desserts, ice cream or sugary-drinks. Think gummy bears, snickers, pop tarts, cookies, brownies, lemonade… I hope you brushed your teeth because the glazing is going to get some action!

Solving these cryptograms can be as easy as saying what you see out loud, but some may take more time. Every cryptogram gives one letter for the solution to the puzzle. It might help to work together on the puzzle!

Be aware that some answers are brandnames. Now put on your thinking cap, grab a snack, and good luck!

Go to the puzzle

Apple versus Spotify and Netflix: who wins on quality?

Sat, 28 Jan 2023 00:00:00 +0000

This week we bought an Apple TV. It complements our Chromecast: now we can buy the movies we want to watch on iTunes if they are not on Netflix. But this blog post is not about the machine: it’s about the attention to quality of the Apple services.

I got a couple months of “free” services (LOL, I paid that with the ridiculous profitmargin on the physical product): two months for Music and three months for TV+. I had tried both before. This time I noticed the difference in quality with respect to the other services I normally use (Netflix and Spotify).

Apple TV versus Netflix #

Movies bought on iTunes or part of Apple TV+ have a much higher quality than Netflix on the Chromecast. I was surprised our television is actually this good 😅. Colors are sharper, resolution is higher, and black is really black.

With Netflix on Chromecast I notice banding/quantization in some content: gradients are “blocky” instead of smooth. Fast moving elements get pixelated. This doesn’t happen with Apple content. It’s not a problem of internet bandwidth: Chromecast and Apple TV are both on the same WiFi.

Netflix quality does improve when streaming from the Apple TV (for the majority of the shows). It’s a close call if I would say Apple TV+ content is still better; intuition says yes.

According to tests online, Apple TV+ has a 40% higher bitrate. In other words, they use more data for the same image. This can be the reason for the higher quality (bitrate isn’t everything: the quality of the codecs and transcoding process also matters).

The quality of Netflix is “good enough”. Noticeably worse than Apple, still great enough for enjoying content.

We have a1080p tv and are on the Netflix HD plan. Netflix also offers a 4K plan. That plan might also lead to better quality on our TV, but the added costs are 4 euro per month. In my opinion the 4K plan should only offer better quality for 4K TVs.

Example of banding and pixelation. I tried to make a photo of the effect on my TV, but failed. The pixelation only occurs in fast moving objects: everything else is razor sharp on Netflix.

Apple Music versus Spotify #

Regular readers of this blog (ha! That’s a joke) know I appreciate good sound quality in a streaming service.

I didn’t mean to try out Apple Music, especially since last time I rated it worse than Spotify, but I accepted the two-month free trial was on autopilot. This time I put my headset on and I was literally (figuratively) blown away by the audio quality.

It’s an old bluetooth headset (without active noise cancellation) and playing a song Apple recommended: “Own my mind” by Maneskin.

The results were impressive! The Apple Music version was considerably better than Spotify in terms of detail, clarity and soundstage. It sounded more like an old-fashioned CD instead of a compressed streaming source (I see the irony in being surprised to have quality back to the level we took for granted 20 years ago).

The difference is clear even on my budget headphones, but it became more pronounced on my better speakers. I listened to Apple Music a full day on there.

I think this advantage of Apple Music is due to its high bitrate, special mixing and Dolby. This leads to bigger file sizes but also better sound quality. Spotify has been promising “hifi audio” for two years now and hasn’t delivered yet.

Spotify feels ‘cheap’ now. Even Apple’s album-art is higher quality:

I used to like Spotify’s recommendations, such as Discover Weekly. These new music discoveries became increasingly rare on Spotify over the past year, and I kept getting the same songs I already skipped. I will definitely miss Last.FM integration on iOS, but is that enough to hold me back?

FYI: Spotify is set to the highest quality, and disabled audio normalization and cross fade (both lower sound quality for some reason). I didn’t touch Apple settings at all.

Apple TV itself #

The Apple TV is a small black rectangular box with rounded corners. It only has three ports: hdmi, Ethernet and power.

I like how it took me through the settings progressively (the “onboarding” in technical terms) instead of all at once: every subsequent time I booted the device it asked one question; do you want Siri on? Which display mode? Connect iCloud?

Two things take getting used to: the weird remote on iOS (no separate app) and the lack of ‘casting’ in Apple TV or Netflix on iOS.

There is also a weird bug with 24fps content. Apple seems to try and send it through as 23.97fps in Europe, which leads to stuttering. Took me a while to figure out I had to set the device to “pass through”. A weird un-Apple like experience.

Yearly review: 2022

Fri, 27 Jan 2023 00:00:00 +0000

At the beginning of every new year, I take a look back at what happened in the year before.

I don’t write this article for other people (although nearly 80 people read my review last year), but for myself. It’s writing-therapy.

These yearly reviews are a way to reflect and learn, but also a ‘congratulations’ to myself. I often get the feeling that I didn’t accomplish anything, and with my yearly review I realize that my years are actually quite action packed. Sophia Prater said it best in her December newsletter: it’s a brag list. I take some time to brag about my activities.

Last year I could distinguish a couple of themes: research, machine learning, writing and note-taking. My interests didn’t shift that much, but I think there was more focus on consciousness and ‘selfness’, and a bit less hard-machine-learning.

The big theme of 2021 was the crossing of thresholds: trying out new stuff, switching jobs. The theme of this year is more ‘everything, everywhere and all at once’: my activities, thoughts and emotions were all over the place. I started many more projects than I finished: will 2023 will be a year of wrapping up this undirected energy?

Writing #

I wrote 4 articles on my Medium account:

An article about solar panels packed with analyses, graphs, difficult texts – it’s as close to a scientific paper as I’ve written the past years. It takes at least 20 minutes to read, and was read by only 20 people
A quick article about my yearly Easter-puzzle. Read by 27 people (more about the puzzle itself later)
An in-depth article about reusable research for companies. I was very proud of this one. It took a long time to write, and it’s very practical. It was viewed by 4700 people, and read by 1200 people. Really great stats!
I pushed out a quick article about remote usability testing as an experiment. I was hoping such a small (only 4 minute read) and practical ‘listicle’ would be really helpful for people, but only 111 people viewed the article 😬

On this blog I further published two more articles:

Writeup of my experience of participating in the Rijswijk Talent Award with a simulation about transition
My yearly review of 2021

I also updated my article with tips for saving gas. There are now 21 tips. Sadly, this listicle also didn’t perform that well: just 333 people viewed it. A big comfort is that I know some people that read it used it to save gas and money.

I created three ‘weird ideas’ for my 1001ideas blog:

I published 11.423 words this year, which is down from nearly 20K words last year. I wrote a lot more, but it’s still in draft.

Easter puzzle 2022 #

I created another Easter puzzle. Last years theme was weird bird names. I think I can officially call it a tradition now that we’ve had three editions. This was the first year I made it in English, which was a great challenge. I’m happy with the way the puzzles turned out, they look a lot better than the year before (I couldn’t get my ‘neon style’ the way I wanted). It was also the first year I had a collaborator: Jesper helped me with testing every puzzle.

737 people played the game.

Work #

I’m working at PostNL as a senior UX designer. I designed a new B2B-app, and implemented several large features for the online portal. I organized several workshops: sharing knowledge about target audience and platform, working on long term vision, expert reviews, process improvements.

Also started coaching juniors, this is fun and makes me learn new things also.

Private life #

We went to museums a lot more now that my wife and oldest daughter both have a museum-card. I like it.

I (finally!) fixed the office on the third floor of our house. Painting all the walls, laying the floor, adding the decorations… It took a while, but it’s really nice that we now have a good home-office space! We also had solar panels installed.

I started writing a book about ‘patterns’ for workshops. It was fun to start this project, but I got stuck in the details. I blame this on not having enough time to concentrate, but more likely I just lose interest 😆

I also published the book my father wrote. This was a Big Project: I photographed all his illustrations and proofread all the text of the book. I then used Apple Pages to put everything together, and published it on Amazon. I’m incredibly proud of the result: it’s amazing to hold an actual book in your hands (even if you didn’t write it, but ‘just’ published it).

I got my diploma “EHBO voor kinderen” (first aid for kids). I’m afraid I forgot 80% of it already…

My youngest kid started sleeping properly after getting a sleep coach: easily the best investment of the year (better even than the solar panels).

I grew a mustache: I got fed up with my facial hair and tried something new. It is still difficult to get the mustache to stay ‘kind of’ in the way I want 👨

Lots of fun with the kids, organizing stuff for birthdays and a huge summer party at school. Every day I write a new word on a board in my daughter’s room: she’s learning how to read.

Thoughtmaster #

I’m working on a new note-taking app. This year it went through at least three iterations: a Laravel version, a Django+Vue version and a Django+React version. In the beginning of the year the app development got a big increase because a friend started helping. A second friend started helping in August, dropped off when his interests shifted.

Media consumption #

It took me the better part of this year to read Figure It Out. It’s a great book, although the real meat is in wedged in-between two theoretical parts. Didn’t read other books, but devoured hundreds of articles. I rediscovered RSS, and there’s a wealth of great stuff for free online!

Watched:

Everything, everywhere and all at once (amazing experience, the best movie I watched in 2022) ⭐️
Lawrence of Arabia (wow, still an amazing film. Slow but worth your time, what a great story and cinematography) ⭐️
The father (an acting tour-de-force, the confusion about what exactly is going on was great) ⭐️
JoJo Rabbit (blown away, could be the best movie of the year) ⭐️
RRR (a bromance movie from India, it’s way over the top, but once you realize it portrays scenes emotionally instead of realistically it makes sense) ⭐️
Licorice Pizza (amazing movie about two young people falling in and out of love, and enjoying life. The adventures they go through feel grand, even though they are a bit pedestrian). I watched multiple Paul Thomas Anderson movies this year, and it was a mixed bag. But I really recommend Licorice Pizza! ⭐️
Death on the nile (visually stunning)
WandaVision (idea was better than execution)
Loki (weird fun)
A dangerous method (bit slow and talky, but interesting look at ‘old school’ hand wavy psychology)
Assassination Nation (is divided in two parts: first an interesting and scary look into social media usage by girls and reflection on sexuality, second part is weird and bad)
Porco Rosso (great visuals, not my favorite Miyazaki movie)
Morning Show (season 2 was okay, not as great as season 1)
Foundation (disappointment)
Welcome to Marwen (fun)
Phantom Thread (slow, some interesting bits, on the whole a bit of disappointment)
Jack Reacher 2 (okay)
Slow Horses (fun)
Meet Joe Black (interesting to watch, but ending too sweet and looooong)
Yes God yes (strangely boring for such a topic)
Youth (good movie about two old men)
Lupin (fun, but takes itself too serious)
DS9 (guilty pleasure, was a blast, fun and humor and a serious plot, with multifaceted bad-guys. After watching the last episode I felt like having left a comfortable place behind)
Primal fear (fun watch, court-room drama with Richard Gere and Edward Norton)
Russian doll (season 2, great leading actress, story was great too except for the weird subplot of the guy – that lead nowhere)
Fantastic Fungi (a weird mix between science, new-age, business and cult-worship)
My octopus teacher (guy who should’ve gotten therapy went diving for an octopus instead, beautiful imagery)
Enola Holmes (should have been fun, was plain and boring. Don’t know why, the actors are good and the story was good too. Is it the editing?)
Holiday into the wild (okay, bit boring, but great elephants)
Punch-Drunk-Love (don’t get why it’s so special)
Space Force (season 2, it’s fun and when it gets the humor just right I laugh out loud, sadly doesn’t happen that often)
No time to die (longest Bond movie ever and I felt it, the action felt too ‘convenient’: here’s another can of henchmen to kill. No espionage, everything just happens to Bond. The whole bad-guy-plan didn’t make any sense. And I don’t like the fatherly Bond, I like my Bond detached-and-sexy thankyouverymuch. AnaDeArmas was great, though, would love to see a whole movie of her character)
OSS 117 - Alerte rouge en Africa noir (not as funny as first two movies, but still fun)
La Mome (Edith Piaff’s life: filled with sadness, so crazy to see what one person can go through)
Glass Onion (funnier than the first movie, but the mystery was dumb)
The unbearable weight of massive talent (Nicholas Cage does JCVD, but funnier and more action, still JCVD is a better movie as it’s more reflective and personal)
Gosford Park (amusing mystery movie, mainly for the culture-aspect not the mystery which was very simple)
Closer (four people who cheat on each other, so much cheating is unrealistic for me)
Top Gun (more fun than I imagined! I like that the rivalry between Tom Cruise and Val Kilmer remained realistic)
The Expanse (first episode was amazing, everything after that played catch-up, got tedious in season 2 and I stopped watching)
Darkest Hour (impressive movie, changed my perspective on Winston Churchill: a right wing populist, but maybe the right person for the job?)
Anima (short movie about a dancer, cool visual effects)
Memoires of a Geisha (interesting look into a different culture (although dramatized), but the love-story with huge age difference made the movie a bit yucky)

New Rijswijk Talent Award: theme transition

Sun, 18 Dec 2022 00:00:00 +0000

Two years ago, I participated in the first edition of the Rijswijk Talent Award. It was a great experience. I got challenged to do something new, and got a lot of recognition for it (I won!). This year there was a new edition, and of course I couldn’t resist participating and even my daughter entered the competition!

Transition #

The theme this year was “transition”, which I found quite challenging. The concept is very vague and difficult to grasp, and at the same time applicable to nearly everything.

The definition as given by the organization was “a meaningful journey between states”.

I took that as a starting point and realized ‘transition’ also has to do with perception. There can be no transition, without someone consciously deciding ‘something’ changed from one state to another state (or classification). If a lamb ‘transitions’ into a sheep, when exactly does that happen? And does the lamb or the sheep care? I don’t think so, there needs to be human perception to care about this difference, and attribute a change in classification.

Of course animals can see the difference between a lamb or a sheep, but I don’t think they have the realization that there is a ‘meaningful journey’ between the two states.

Furthermore, nature is always in a state of flux, always changing. Any change in state is temporary, or transient. Only in our human timescale do we see this ‘meaningful transition’ between two resting states.

With this, I started collecting examples of transitions: the energy-transition, gender-transition, life to death, boy to man (which I thought was an interesting one, because of the traditions and rituals associated with it, like a Bar Mitswa), stranger to friend, friend to lover, lover to partner, actors changing into a character on stage, life before and after corona, human to robot, employee to manager. One transition I saw happen a lot in Rijswijk was a transition of a green garden into a parking lot…

This lead me to some more definitions:

a transition as a process of slowly crossing an artificial border
transition as a ‘portal’ from one virtual (imaginary) place to another

In any case, a transition usually pairs with some chaos and uncertainty. It requires energy at first (endotherm), but can also generate energy (exotherm). Transitions can be permanent, temporary or cyclical (like the seasons).

I also found the Satir Change model applicable: Old status quo -> foreign element -> chaos -> transforming idea -> practice and integration (growth) -> new status quo Especially the idea of the ‘foreign element’ made a lot of sense to me, something that triggers the transformation or transition.

What I sent in #

In the end I wanted to send in an interactive, personal, never-ending art-work that had a relation to the art-work I sent in two years back.

I only partially succeeded. The art-work is definitely infinite, but it is not interactive. The link to two years ago is also very vague: they are both boxed-in animations with a relation to my history with Artificial Intelligence. That’s also my only personal link.

Still, I’m quite happy with my submission.

I took inspiration from the Conway’s Game of Life and created a simulation that continuously triggers transitions of one state into another. Not every transition is successful, some die off before taking over. And often, there are multiple transitions happening at the same time: when the old status quo is becoming unstable.

I imagine these are transitions in a society, such as acceptance of women-rights, gay-rights, a sustainability craze, or a negative transition: war and strife, a far-right political party winning elections…

See it here.

The result #

I got nominated for a prize, but didn’t win.

But the true result is that I got my daughter enthusiastic in joining, and that she made an amazing art-work as well. That’s my real prize.

Usability testing a mobile app remotely: 4 simple methods

Tue, 11 Oct 2022 00:00:00 +0000

So you have a mobile app that you want to test with your target audience. Problem is, you are “here” and they are “there”. Remote usability testing is an outcome, but how to do this with a mobile app?

For all methods below, it’s necessary to have a testplan and participants. The GoogleVentures Research Sprint has great resources to get you started.

1. A regular video call #

The simplest and cheapest way to test a mobile app is to schedule a video call with your participants on their desktop or laptop. Almost everyone has a webcam now and is familiar with remote work. Tooling for this is abundant: Microsoft Teams, Google Meet, Zoom, etc.

During the video-call you ask them to open the app on their phones, and then talk-out-loud: “where are you now?”, “what do you see?”, “what are you doing?”. If necessary, ask them to show their phones to the webcam for extra clarity.

It helps if you have the app installed on your phone and can navigate through the same screens at the same time 🤞

This is the most low-level approach, but also the most reliable. The downside of not seeing the participant’s screen is smaller than you might expect. If you are doubting if this approach works for you, try it out with a colleague.

✅ See participant’s face
✅ Simple setup
✅ Cheap / free
🟥 Cannot see participant’s phone screen

2. The laptop hugger #

Make sure the participant has a laptop (not a desktop computer), and setup a regular videomeeting using any widely available tool just like method 1.

During the call, you ask your participant to rotate their laptop so their screen/webcam is facing away from them. You then ask the participant to ‘hug’ the laptop, and hold their phone. If all went well, you now see the participant’s mobile phone screen in the webcam.

See an example on Youtube

This might take a bit of fidgeting ( “lower your laptop screen a bit, now hold the phone a bit closer, can you turn off background blur”, etc), but it works like magic when it works.

✅ See participant’s phone screen
✅ Simple setup
✅ Cheap / free
🟥 Cannot see participant’s face

No, don’t hug your laptop like that! (source: Dall-E)

3. Mobile videomeeting #

Ask your participants to install the video meeting tool of your preference on their phone. During the usability test, you ask them to select “broadcast” or “screenshare” in the tool. Voila, you now see their screen and hear their voice. The downside is losing the video-connection.

This should work on any modern phone, both iOS and Android. But… It doesn’t always work. I suggest doing a try-out session with your participants before your ‘real’ usability session!

✅ See participant’s phone screen
✅ Cheap / free
🟨 Potentially tricky setup
🟥 Cannot see participant’s face

Make sure your participant is on WiFi to avoid huge costs for them!

4. Use specialized tooling (UserZoom or Lookback) #

There is special software for remote usability testing on phones, for example UserZoom and Lookback. With this tooling you can see both the face and the screen of the participant, and these toolings have excellent analysis functionality as well.

The downside is that your users have to install another app. This is a big hurdle to take and might make some of your participants anxious: what are this new-app’s privacy policies?

In my experience installing these apps is not worth the benefits.

✅ See participant’s phone screen
✅ See participant’s face
✅ Extra analysis features
🟥 Tricky setup
🟥 Expensive $$$

Extra tips for remote research #

Some quick bonus tips for remote research:

do a practice session with a colleague. Are your instructions clear? Can everybody see/hear eachother? Share screen works?
do a practice session with each participant. This is very reassuring for your participants, and a nice way to break the ice. It also prevents your one-hour-of-research resulting in failure.
prepare yourself. You represent your company, so wear some nice clothing, have a good background, and make sure you can be seen and understood properly. Make sure you get light in your face, so you don’t look like a scary hacker :)

Good luck!

Usability testing a mobile app remotely: 4 simple methods

Make your research reusable: the information library for UX Research

Tue, 27 Sep 2022 00:00:00 +0000

Of course, it’s great that you already store the results of your research in a place where others can find them. A folder of reports (in Sharepoint), a wiki (like Notion or Confluence), or maybe even a dedicated research repository (like Dovetail, Reveall, or EnjoyHQ). Sharing your findings in such a way will help in making your conclusions more accessible. But does it help you as a researcher?

Diving into our information library (image by me, in collaboration with Dall-E)

With this article, I will convince you to create a library of tagged and anonymized research data. Data like interview notes, web analytics, survey-data, customer feedback, screen recordings, etcetera.

The article has two parts: why you should build an information library, and how you can build one.

The pyramid of data, information, knowledge and wisdom ( based on diagram by Martin Gajzler ). In this article we’re talking about ‘information’: processed data, data that is tagged and structured. We all want to achieve ‘knowledge’ and ‘wisdom’. But how do we reach that? Through structured information!

Why should you make an information library? #

Research has four simple steps: 1. gather data, 2. process data (so it becomes information), 3. write conclusions, 4. share conclusions. And then we move on to the next research project, where we do all four steps again. What do we do with our information (processed data)? We forget about it, and that is a shame!

I see five reasons to make an information library:

answer questions quicker
cheaper research
better insights
discover hidden insights
research together across silos

Well, if you put it like that, it’s a pretty bad idea to not use an information library. Gathering data is very expensive, and so is converting data to information. Why do that only to write 1 insight? Start recycling! ♻️

Reason 1: Answer questions quicker #

Maybe a stakeholder read your report, and asks “which people said this?”. If you didn’t keep your tagged data, giving an answer is difficult.

But with an information library, you can dig deeper back into your research, even years later.

Reason 2: Cheaper research #

With an information library, you can reuse previous data for new conclusions. This is very powerful.

Let’s say you want to know how people use a certain feature of your platform, perhaps printing. Before doing any new research, you can use your information library to dig back in previous interviews, surveys and feedback to find a hint of an answer already.

It’s an incredibly cheap way to do research. Cheap, but still very powerful as you draw on a lot of data points.

And if there are gaps in your conclusions, your information library gave you the information you needed to set up a more detailed research plan.

Reason 3: Wow, triangulation for free! 🤩 #

Triangulation simply means diversity in research. Using multiple methods to collect data, using data from different timeframes, varying research-frameworks and doing analysis by different researchers.

By using triangulation you get a broader, deeper and more nuanced understanding of the space your are investigating. Plus, it adds to the validity and credibility of your research.

Triangulation is normally very expensive: setting up a survey and rounds of interviews and do analytics and… Imagine getting the budget for that!

By maintaining a research-data-library you get all this yummy triangulation for free.

Example:
you find out with your analytics that some pages get printed more often than others and from customer feedback you find complaints about the printing feature. In the interview notes from usability tests you did earlier, some people mention why they sometimes print a page. You then dig a bit further in your library and find summaries of support calls, and interviews with account managers. Because you kept old screen recordings you can see what the exact usability problem is in certain browsers.

You now have a complete story of what goes wrong with your printing function, why it goes wrong, how important it is and how you can fix it. And if you need to do more research to be sure, you know your target audience and which questions to ask them.

🤞 I’ll explain exactly how to do tagging a bit later in the article!

Triangulation is cool and hip. But actually doing it is very expensive… Oh no, it isn’t: just keep all data in a structured information library, and you get triangulation for free!

Reason 4: discover hidden insights #

When you keep an information library, you will find insights that you wouldn’t have seen otherwise. These can be trends over time, subgroups in your target audiences, habits of users, etc.

The best thing is that this happens automatically. There is no need to do extra work: just keep tagging your data when you gather it (and thus processing it into information).

Example
Let’s say you interviewed twenty people in January and analyzed this qualitative data in your information library. Then you interview another twelve in May. And then another batch of sixteen in June. For each of these three sessions you draw conclusions that you shared with your stakeholders. Decisions are made, everybody really happy.

But what’s this: a certain tag in your information library looks suspiciously big… It’s a new business opportunity that only a handful of people were talking about. You would never have connected these dots if you had kept all three research sessions separately!

Reason 5: research across silos #

If you set up your information library in a new ‘research repository tool’ (like Dovetail, Aurelius, Condens or EnjoyHQ) you can tag data together with other people. Other UX-researchers, but also other people who do research or come into contact with your user group. For instance market-researcher, sales, your support, channel-managers and service designers.

With an information library, you have the power of all data behind your conclusions. And so do other people in your organization. This makes it easier to keep up to date with research across the silos in your organization.

The instructions below use Dovetail, but you could also use other research repositories, likeEnjoyHQ or Condens or Aurelius (and maybe Avrio, Glean.ly and Consider.ly, although I haven’t used them myself)

How do I build an information library? #

So yeah, an information library is cool.But how do you build an information library?

I have done this in Dovetail (more specifically Dovetail Markup), but there are alternatives: EnjoyHQ, Aurelius and Condens. I recommend making a free account with all four tools and do a quick qualitative analysis: just add tags to one of your existing interview transcripts, and find out which tool feels right for you.

Here are the steps to create an information library in Dovetail:

1. Create a new project #

You will need a research project. This one project will contain all of your research. Later (i.e.: when you start paying (a lot more) for Dovetail) you can create separate projects and use a feature called “extensions”. For now, we’ll just hack something together 🤓

2.Create tag boards #

Tag boards hold all of the tags you will use to create information (structure) out of your raw data.

Create a tag board called “Target audience and general”. Add a column (called “tag group”) here with ’Target audience’, ‘Context’, ‘About our company’, and ‘Competition’. Fill these with some starter tags (you can always add more later).

Tag board “Target audience and general” with some starter tags (your ‘problem space’). Another great idea is to add a board with a customer-journey, with all the steps people take before/during/after using your product or service.

Create a tag board per product or service your company sells. Add columns per page or feature you have. If you have a very important flow in your product, you can create a separate tag board for this flow as well.

The tag board per product/service (your ‘solution space’). Add some tags per part of your product, or per step in the most important flow. Depending on the reasons you do research, this could be split over multiple boards. Start small, though!

Don’t go overboard with making tags. Start small. When you notice a tag grows too big, you can split it.

3. Import old research #

Create a “data” view called “Reports and Analytics” and add columns for each researchtype: interviews, web analytics and old research reports. Don’t overthink this: you’ll get the hang of this grouping soon enough and you can always change later.

Now you need to fill your database! You can start by adding a recent interview, copy-paste the notes per interview and start tagging. Tagging is simply highlighting a part of the text and selecting a tag from the menu that pops up. Don’t tag single words, but rather whole sentences or even paragraphs. You can (and should!) give your highlights multiple tags. The more connections you make, the easier you will find insights later.

Always select a full sentence or paragraph: context is nice. More tags = more better. It’ll help you find back what you are looking for, AND it will help you ‘accidentally’ seeing new connections you didn’t see before.

To tag old research reports, I usually cut the report into logical pieces and manually write the conclusions per piece. I then tag these conclusions. The same goes for web-analytics and survey-data: describe the graph or data, and then tag your description.

As Dovetail doesn’t support tagging images, such as graphs and other figures, I describe the figure and tag the description ( this happy graph and optimistic text is from this source )

You can also import feedback from tools like Usabilla (GetFeedback). These end up in separate notes, which you then need to tag manually. It’s a lot of work, but it’s worth it. Keep in mind how difficult and expensive is it to reach out to your users, so put some effort in when they reach out to you in your feedback tools.

If this becomes confusing, or you find you grow out of space, you can separate your data in multiple ‘views’. The next step is organizing in multiple projects. Be aware that you have to upgrade to Markup Team to enable ’tag extensions’ to keep the advantage of your information library.

4. Do new research in Dovetail #

Now you have the hang of tagging, you should do the analysis of new research in Dovetail as well. You will now see the benefit of all your earlier hard work: old data pops up to support your new research.

Recap #

Information is a useful resource, especially if you can use it more than once. With an information library, you will write better and more convincing insights. Modern tools like Dovetail, Condens, Aurelius and EnjoyHQ make it easy to do so.

Keep adding blocks of information to your library, make it grow and strengthen your understanding over time! 💪

The insights you’ll write with your information library will feel like an avalanche instead of a snowball.

Thanks for reading. This article took me a long time to write, I’d love to hear your feedback! Maybe you’d my article “ 17 alternatives for personas ” as well?

Make your research reusable: the information library for UX Research

Enjoy a puzzle for Easter!

Tue, 12 Apr 2022 00:00:00 +0000

Me hardcore puzzling

Wet from snow we arrive at my parents’ house. “You’re here!”, my mom shouts. “Come in, your brother’s also just arrived”. I see him hunched over the table, looking at the newspaper, “I already have five”, he says with a smile. I wave to my dad and join my brother. “What’s the theme this year?”, I ask. It’s Christmas time. It’s puzzle time.

Every year the local newspaper Dagblad van het Noorden publishes a Christmas puzzle with cryptic images. It’s a tradition.

The whole family joins in. My mother, my father (he says he doesn’t like it, that he’s not good at it, but he usually comes up with an unexpected insight solving the most difficult ones), my brother. We pour over the puzzle, write down synonyms, dive into Wikipedia, collect words from the dictionary and brainstorm together on the theme. The puzzle takes us deep into the night.

It’s a lot of fun and a great activity to do together. A slow and mindful activity: we don’t work on it all the time, and instead let it rest for a couple of hours before diving back in. The puzzle leads to conversation, and conversation often leads back to the puzzle.

Friends, cousins, uncles: everybody is in a friendly competition. Whatsapp is buzzing! The puzzle is connecting us through Christmans, as we share cryptic clues and try to boast of how many puzzles we already solved.

And now you can do it too #

I like solving the puzzle from the newspaper, but I like creating stuff even more. So I set out to make my own puzzle, for Easter.

My Easter puzzle has become a tradition as well, this is my third try. Every year I challenge myself to try something new and this year is the first time the puzzle is in English.

The theme is birds, and I learned there are a lot of funny bird names, like Bananaquit, Warbler, Slaty Flowerpiercer, Vampire Groundfinch (creepy bird!), Bobolink, Bounty shag, White-breasted Nuthatch, Brazilian Ruby and Bunting.

All-in-all, there are 18 puzzles! Play it here, share it with your friends, print it out and put it on your kitchen table!

The finished puzzle! Go to the puzzle

Making the puzzle took me a whole year. Each mini-puzzle took about two weeks from idea to finalization, I wrote the puzzle-code myself, and I double-checked all of it (thanks Jesper!).

I’m very proud of the results. Let me know if you liked it!

Happy Easter, and may my puzzle bring you and your friends and family much joy!

Thanks to Jesper for proofreading my work. Go to the puzzle

Enjoy a puzzle for Easter!

So you want to solar panel?

Thu, 17 Mar 2022 00:00:00 +0000

Photo by Bill Mead on Unsplash

We recently decided to get solar panels, and as always I dove into this subject with enthusiasm. Here’s a short recap of what I learned.

We live in an old house, built in the 1920s. Last year we had new tiles installed on our roof, and we decided it was not profitable to install solar panels. Plus we were really happy with the look of our new roof!

That was July 2021.

Fast forward to February 2022: we get our new energy contract and the prices of gas and electricity were more than doubled! And now with the horrible war on Ukraine (fuck Putin, donate here ), we are trying to reduce our reliance on Russian gas and invest in new energy sources. Conclusion: we need solar panels!

I investigated the hell out of our solar panels, so I hope this information is useful for other people also.

Side note: solar panels are also a good investment to reduce the effect of climate change. This is another important reason for us to get them, but for this article I’m only looking at the finances of solar panels. If climate is important for you, also take a look at my other article with tips to save on gas .

From my calculations I can see solar panels are one of the best investments you can make. You get your money back within six years, and then get “free money” for another twenty years.

But the profitability depends a lot on your situation. Much more than the technology you choose or initial costs of installation!

There are two factors important here:

Your energy consumption
The available space for panels

1.1 your energy consumption #

We are a family of four and use 3.400 kWh per year (you can find your yearly usage on the ‘jaarnota’ of your power company). We expect this usage to go up when our kids grow older. Long term plans (5+ years) include switching to electricity for heating, and an electric car.

Why is your energy need important? Any power surplus we generate with our solar panels is profit, right? 🤑 Well, yeah, but less than you might think! It’s more beneficial to use the energy from your panels yourself instead of delivering back to the net.

Example: “The two-day-year” #

Let’s do a quick calculation with a completely unrealistic example 😃

What if a year lasts only two days and each day we consume 10 kWh. Our two-day-year has a winter day and a summer day. In the summer we get 20 kWh from our solar panels and in the winter only 5 kWh. That means we generated 25 kWh and used 20 kWh during our two-day-year: 5 kWh profit, yay! 🎉

A quick calculation as an example

But our energy generation was uneven: during the summer we had a surplus that we sold to the power grid and in the winter we had to buy electricity.

In the Netherlands we have the ‘salderingsregel’ which means we use the grid as a battery; ‘storing’ our surplus from the summer for free and using it in the winter (or more accurately: we sell energy to our power company for the same price as we buy it). But this law only applies for energy we consume in the same year. Our 5 kWh ‘profit’ is sold back to the grid for a much lower price, less than half of the price we pay for electricity.

This law will not last forever, it’s planned to be dismantled over the next 9 years. We need a solution for that, and luckily there is one: home batteries. Although too expensive right now, the battery efficiency is increasing like crazy and prices are dropping.

If you have a lot of panels and generate more than you consume, a return on investment takes much longer, or maybe your panels not profitable at all.

Note that energy-companies implement the salderingsregel differently , and the same holds for the price of energy bought back (terugleververgoeding). It’s good to use a compare website for your next contract!

How much electricity should I generate, then? #

Although over-producing seems like a bad idea based on my quick example above, it’s not that black and white. A ‘longer time to get return on investment’ doesn’t mean ‘no return on investment’, it might still be worth installing over-capacity even if it means those solar panels get a 10 year ROI instead of 6 years. Plus, your electricity consumption is likely to go up and electricity prices as well. If you have the money to invest, I think it’s profitable to install as many panels as you can.

1.2 available space for panels #

We have a traditional roof with two angled sides, one facing northeast and the other facing southwest.

We asked three companies to make us an offer, and all three came with wildly different layout plans for the panels. One placed just six on the southwest side, the other placed twelve total: six front and six back, the last one made a mosaic by creatively placing 10 panels facing southwest (it looked really ugly).

The average costs per solar panel has decreased by 91% since 2010, and the efficiency is still increasing. This means that even solar panels in bad positions such as shade or facing north generate enough power to be profitable. I could see this in my calculations, panels from five years ago were not profitable in such positions, but newer ones are, especially with the ‘salderingsregel’.

My initial doubts about the panels facing north-east were taken away, and so we chose the 12 panels-layout.

You can easily make a plan for solar panels on your roof using Solaredge designer (see later in the article), or simply drawing on paper. You’ll probably know the width and depth of your house, if you also want to know the height of your house you can easily find that in 3D BAG.

3D bag is cool

Your solar installation consists of two parts, solar panels and an inverter:

2.1 Solar panels #

There are loads of options for solar panels. You can integrate them into your roof, or replace your roof tiles with solar powered ones. We went for the default ‘big black panels installed on top of the tiles’, because we already re-tiled our roof and at the time we couldn’t find a contractor who did the other options for existing houses.

There’s another choice to make: percium versus full-black. The percium types have a higher performance on hot days, because they don’t heat up as much in the summer. Solar panels reach temperatures up to 50 degrees and that leads to a degradation in efficiency. Still, we decided on full-black as they look nicer.

Solar panels are measured in Wp, the amount of watts they give in laboratory conditions. Higher Wp is better, but a higher Wp may also just be a bigger panel, as there is no uniform size. Most panels sold now are between 370 and 400 Wp. This was 295 Wp in 2017, and 330 Wp in 2019.

As for the difference in quality per brand? According to the experts you can’t go wrong with any type of panel; they are all good. There are some differences in efficiency, here’s a real-life test of recent panel types..

We went for 385 Wp Trina panels as those were the ones offered by our chosen supplier. Had we upgraded to 400 Wp panels, we would have expected a 118 kWh increase per year in production (including loss to shading etc), or an estimated 45 euro per year at the current prices. If every panel was 50 euro more expensive, it would take us 12 years to recoup, worth it if you are a long-term investor!

2.2 The inverter #

Your brand new solar panels need to be connected to your house and the power grid to do anything, and for this they need to be connected to an inverter.

An inverter converts the DC power coming from your solar panels to AC power. Some suppliers with a sense of drama call the inverter the ‘heart of the installation’.

A term you’ll see mentioned in other places with respect to inverters, is DC/AC oversizing. Your solar panels deliver a certain theoretical kWh per year, and you should get an inverters that matches that production. However, it’s possible to match the theoretical DC-output of your panels with a smaller inverter. So get 4200 kWh solar panels and buy a 3800 kWh inverter, for example. This has three advantages: the smaller inverter starts producing power earlier in the morning and goes on longer in the evening (smaller inverters need less power to ‘start up’), it is more efficient (3% loss at maximum power, versus 5,5% loss at half-power) and smaller inverters are cheaper. You aim for 80%-90% oversizing, but can go down to 70%. Of course, you might lose some power on extremely sunny days, but the effect is small as your solar panels are less efficient then (due to the heat).

There are three setups for inverters to choose from: string, string with optimizers and micro-inverters.

2.2.1 String inverter (serial) #

The simplest and cheapest solution is to put all solar panels on one ‘string’, just like the lights on a Christmas tree.

With this solution your panels are all connected to each other and to a single inverter inside your house. The inverter is easy to reach if you need to replace it (you can expect to replace it after 10–15 years) or want to upgrade it for a newer one with more features.

Furthermore, it’s easy to connect a battery to this system. Batteries work with DC-power, and the inverter can charge the battery immediately from the Solar panels.

For most situations this is great.

The single string solution has some drawbacks. Remember that it’s just like a Christmas tree lights: if one lamp breaks, al the lights go out. Something similar happens to solar panels in a string, albeit not as rigorous. If one panel breaks, the others still work. But if one or more panels perform badly, this affects the whole group. If there is shade (like trees or a chimney) or if the panels are on different roofs, then the efficiency of the entire system goes down rapidly.

This is usually explained as follows: “if one panel is at 80% efficiency because a tree blocks the sun, all panels run at 80% efficiency!”. That’s not how it works. When one panel performs badly, it will shut itself off (not really by design, it’s a technical property of the ‘bypass diodes’ in the system). If you have a system with 10 panels, where one has shade and is working at 80%, then the whole system will perform at 900% (9x100%, 1x0%). If in the same system there are four panels with 80% efficiency due to shade, the system ‘decides’ not to bypass these four panels and instead let all panels works at 80% efficiency (10x80%=800% is better than 4x0%+6x100%=600%).

If there is a chance that one or more of your panels has a lower production than the others, take optimizers or micro inverters (see next paragraphs).

Another drawback is that a serial setup means running a lot of DC-voltage over your roof. This is perfectly safe when installed correctly, but it gave me a bad feeling. DC-voltage is more likely to cause a fire than AC voltage.

Inverters can also give off a buzzing sound, although I’m assured the newer ones make almost no noise anymore.

2.2.2. String inverter with Solaredge optimizers (serial) #

One solution for shade in a serial system is to use optimizers. These small devices are attached to each solar panel and they do some magic to the power distribution.

With optimizers, shadow on one panel doesn’t influence the other panels anymore. If one solar panel has 80% efficiency, the others can still run at 100% efficiency. You still have all the benefits of having a inverter indoors.

It’s more expensive, of course, but even if you have only a little bit of shade, this extra price is definitely worth it. Optimizers also add a bit of extra security: they bring the power down on the solar panels in case of an emergency.

The benefits don’t stop there. With optimizers you have insight into the power generated per panel, Solaredge has a nice mobile-app and the system can integrate with home automation. Plus there’s 25 years warranty on the optimizers! (their inverter typically has 12 years warranty, but is extendable)

The downside is that optimizers are finicky to work with. Your setup has to be just right, otherwise you risk losing efficiency in your output. Have less than six panels? Your optimizers won’t perform right! Picked the wrong optimizers for your panels? They don’t work right. Panel layout spanning multiple roofs facing different sides? Be sure to pick the right optimizers, or… it won’t work right!

I strongly suggest reading through this thread for more info and modeling your planned layout in Solaredge Designer (more on the tool later!). With this tool you see which optimizers you need and if there will be any loss in efficiency.

Every offer with optimizers I got had something wrong with it when I checked it in Designer:

Example 1: a proposed layout plan had only six panels on a string and 60V optimizers, but this string is too short for this type of optimizer and Designer gives a warning. We can add two more panels on this string, or use 85V optimizers.

Example 2: another offer had under-powered optimizers, which Designer doesn’t even allow

Optimizers are a beautiful solution and Solaredge has a really nice ecosystem. When you double-check your installation in Designer, you can have a system you are happy with. Their apps are highly rated on the App Store, 4,5 stars.

But we decided to go for option number three…

2.2.3. Micro inverters (parallel) #

Micro-inverters are exactly what they sound like: a lot of small inverters, instead of one big inverter. Each panel gets its own inverter attached, and the resulting AC voltage goes into your house where you can use it.

That’s it.

Because every panel has its own inverter, they all operate independently. It’s a parallel system. It’s elegant and simple. Shade on one solar panel does not influence the other panels. A panel or inverter breaks? The others will continue their duty. Plus you can see the energy-production per panel, just like Solaredge. A potential downside is that more inverters means more ‘points of failure’, and if one of them breaks, you’ll have to get out on the roof!

There are two brands to consider: Enphase is the leading brand and APSystems is the runner up. Both APSystems’ and Enphase micro-inverter are rated as very good. There were some problems with the reliability of micro-inverter in the past, but that’s all water under the bridge. They both offer 25 year warranty: Enphase by default, for APSystems you pay extra to extend the default 12 year warranty.

Enphase has a rich ecosystem. They make their own batteries, they integrate with a lot of other systems, they make nice apps (their app is rated high with 4,6 stars). It’s all great, until you look at the pricing. Very expensive.

APSystems found a smart way to reduce their pricing. They have one micro inverter per two panels, with some optimizer-like functionality per panel. This means they are much cheaper than Enphase, and sometimes even cheaper than a Solaredge system. The downside is they have a much less developed ecosystem. They are ready for battery-integration, but don’t have such a system in place yet. Their app is rated very poorly, with 2,6 stars, but I disagree with these nay-sayers and the app is actually quite sufficient.

So what is the price difference between Enphase and APSystems? Well, on our offer Enphase was 10% more expensive, or 550 euro. And that was with undersized Enphase-micro-inverters. It was a hard-choice, but we decided we trust APSystems to get their ecosystem in order in the coming 5 years and we took our 550 euro to spend it elsewhere. The bad app is an inconvenience, but I expect to get bored with looking at my solar panel production quickly, and the panels will produce energy without the app just fine.

Note: there’s no real-advantage for a system with micro-inverters to integrate with a home-battery. Of course we only want to charge our home battery with solar-panel-energy, but we can achieve this with any system that can connect with our ‘smart meter’. This differs from serial-systems with a central inverter: if you connect a battery straight to the inverter it can charge this battery immediately with DC-energy of the panels without conversion loss.

Step 3: Money talks 🤑 #

If you want to know how much money you make with the panels, you’ll need to know how much power will your new panels generate in a year.

There’s a simple trick to calculate this if you live in The Netherlands: you can approximate the theoretical production capacity by multiplying the ‘number of panels installed’ by the ‘amount of Wp per panel’. We installed 12 panels with 385 Wp, so we expect to generate 4.560 kWh per year under ideal conditions.

But, we don’t have ideal conditions. The panels aren’t facing the sun at an ideal angle, the panels perform worse when it’s warm, there can be clouds, the wires and inverters loose power in transport, etcetera.

I’ve had many offers from different suppliers and their estimates vary wildly, from 3.200 kWh to 4.080 kWh for 12 panels, and 2.290 kWh to 3.430 kWh for 10 panels.

I decided to make my own calculations to see which layout-plan would give us the best generation of power. I was especially curious if the panels on the backside of our house facing northeast would generate enough kWh to be profitable.

Our preferred layout has six panels facing south-west in the front of the house, four panels facing north-east on the back, and two panels facing south on the ‘uitbouw’.

I used two programs to recreate our house and the shading conditions: Solaredge Designer and Sketchup/Skelion. Of these two, Solaredge is definitely the clear winner in user-friendliness.

If you just want to do a quick calculation, take your ‘theoretical capacity’ (see above) and use an efficiency of 70% for north facing panels and 80% for south. For a bit more detail I can recommend the calculators by Solar Bouwmarkt and Berekenhet . If you want to see how much sun you get, and where it comes from, take a look at Suncalc.

3.1 Solaredge Designer #

Even if you’re not thinking about using Solaredge, I can highly recommend playing around with their online Designer tool. It’s free and fun.

With Designer you immediately get a complete picture: the layout, the shading, the electrical setup, the investments needed and the time it takes to recoup that investment.

You can also export a CSV of the data per day if you want to do a full-analysis yourself (which I did, of course, see later).

The first thing you do in Designer is draw your house, some trees (for shading) and the solar panels. You then let Solaredge calculate the optimal string-configuration and inverter.

After that’s done, you get a detailed analysis of the expected returns:

In the last image you see the drops in efficiency. The 4,5% “opbrengstfactor verlies” is a loss in conversion, as I had trouble getting the system ‘correct’. It may not seem as much, but that adds up over time!

As we will be using microconverters instead of Solaredge, I took out the 4,5% and I’m assuming a 2.950 kWh prediction per year. That’s a 65% efficiency which is very low compared to the other predictions. Still, Designer expects us to recoup our investment in about 5 years with the current energy-price (39 ct per kWh)!

The Solaredge report is very detailed, it shows temperature degradation, shading, orientation, etc. But sadly, I didn’t get a split per panel position, so I couldn’t answer my main question: is it profitable to add panels facing northeast?

Next tool!

3.2 Sketchup/Skelion #

Sketchup is a 3D-drawing tool that’s used a lot in architecture and other industries. It used to be free when it was owned by Google, but it is now a paid product with a very generous 30-day trial. Skelion is a plugin for Sketchup to draw solar panels, calculate shadow influences and do some advanced calculations.

I found the Sketchup/Skelion-combo very difficult in use and buggy.

The one thing Sketchup/Skelion had going for me is the ability to see the output split by panel, which I needed to see if the panels on the backside of the house were profitable.

Skelion uses an online service (PVWatts) to generate the report, it looks like it was built in the 90s

We can’t see the production immediately in Sketchup/Skelion like we could in Designer. We need another service for that: PVWatts. Luckily it’s integrated in Skelion.

The Sketchup/Skelion/PVWatts trinity expects a total 3077 kWh production per year, 66% efficiency. One percent more than Solaredge, but still not as much as my suppliers’ calculation of 3606 kWh (78% efficiency).

3.3 My own calculations 💪 #

I loaded all the data from the report Sketchup/Skelion generated with PVWatts into Apple Numbers. God, I love Apple Numbers. Plus, I used Calca to generate some quick reports of my own.

Here’s a quick yearly production graph I made based on the Skelion predictions, with our current yearly consumption added on top.

These graphs should account for the expected dip in efficiency of the panels in warm summer days. I can see this thermal influence in June: a dip in production compared to May and July. I’ve been told however, that April and May are the best months for solar in The Netherlands, not sure what to make of that!

Using only 6 panels facing southwest would not be sufficient to generate enough power to fit the energy requirement even in the summer, so the back-facing panels are a useful addition.

Now let’s look at another graph with the effect of the panel-layout, how much does each panel-layout contribute to the total? As you can see, the 6 panels facing southwest are much more effective than the other panels. During the winter months the two panels facing south are more productive than the four panels facing north-east!

In the winter, the 2 south facing panels are more effective than the 4 north-east facing panels!

If we normalize the graph, we can see how much better south-facing panels are in our situation (this graph still accounts for shading). If we had the space, all our panels would be facing south.

3.3.1 Daily production #

The real surprise comes when we look at daily production. For this I had to resort back to Solaredge Designer, because PVWatts doesn’t give a daily calculation. I made two systems in Designer, one with only six panels in the front of the house and one with our preferred layout of 12 panels. I then exported the CSV-data for both.

I took four days in the year, to represent the four seasons of the year:

What’s obvious from these four graphs is that the twelve panel-setup delivers more energy, and that we will cover our energy-need for more hours of the day.

Interesting to me is that the 6-panel layout, facing only southwest, is already expected to generate some electricity in the early morning when the sun comes up in the east. The same holds for the panels facing north east during the day: there is not a huge dip in the green bars after 13:00 when the sun starts facing the other panels. I’m very curious if I will see this effect in my actual production. Solar panels generate power even when the sun is not shining on them!

Another very important insight is hidden in these daily graphs. Remember the ‘salderingsregel’ from the beginning of this article? That’s the Dutch law that states we can store all the energy we generate that we don’t use in the power grid for free, so we can use it at night and during the winter.

That salderingsregel will slowly be dismantlement over the next few years.

When the law is gone in eight years all the power production you see above the red line during the sunny hours of the day will be sold back to the net at a low price. The power we buy from the grid in the night we will have to buy back at a high price. So even though we have these awesome solar panels, there is a chance we will be paying for electricity again in the summer of 2030.

A partial solution is to get a battery to store our precious ‘free’ solar-energy. This will help us in the summer days, but it won’t help for the difference in energy production/consumption in winter versus summer. For that we need drastic measures like storing heat in the ground.

Is a mixed 12 panel layout better than a layout with all panels facing south?

If we had the space, we would have placed all panels facing south, instead of facing some north as we do now. That’s the most beneficial way.

But wouldn’t we be missing out on the ‘early morning sun’ this way? Out of interest, I checked how much better a 12 panel-layout facing both northeast and southwest performed versus the 6 panel layout facing only southeast.

We’d expect a 12 panel layout to always perform 200% better than a 6 panel layout.

Instead, what we see is the production dip below 200% between 10:00 and 18:00 for most days (when the northwest facing panels get no direct sunlight) and rise way above 200% in the mornings (when the southwest facing panels can’t see the sun).

A surprisingly big effect in the morning, and not too bad during the mid-day!

So maybe it’s a good idea to mix panels, so you can ‘catch’ some sun in the morning and evening?

Wrong! This is a case of misleading percentages. The production capacity in the morning is very low. If we look at the complete daily difference, we can very obviously see our 12-panel-mixed-layout never performs more than twice as good as our 6-panel-only-southeast-layout (January: 194%, April: 182%, June: 184% and September: 194%).

If you have the space, face all your panels to the ideal position: South!

3.4 Conclusions #

Our 12-panel layout, including northeast facing, makes a lot of sense versus only 6-panels facing southeast in our situation. Any added panel makes financial sense: the added cost of one extra panel on the whole installation is not that much. This is true until we start producing more energy than we consume per year because of the salderingsregel.

As we cannot see into the future, I have to keep a wide margin on the potential return-on-investment of the solar panels. I am assuming anywhere between 4–8 years for paying back our investment, and an additional 12–16 years of savings on the energy bill. This means between 14.000 to 18.000 euro profit on a 5.300 euro investment.

And that’s with a conservative 20 year life expectancy on the system. I’m expecting the panels to last up to 25 years!

Worth it!

Step 4: your turn? #

Solar panels are a great investment in my opinion. It’s a safe bet they’ll pay themselves back and then they’ll deliver savings on your energy bill for at least 15 more years.

If I got you interested in solar panels, your first step should be to go ask some offers from multiple companies (at least three different ones)!

I also advise looking at Solar-Bouwmarkt. You can buy exactly the equipment you want and have one of their local suppliers install it for a low price (or even install it yourself!). Solar Bouwmarkt will check your order to make sure you have everything, so it’s practically risk free. In the end, I decided against it because they are not selling APSystems, and an Enphase system was too expensive for me.

4.1 How do I know if I get a good deal? #

To know if the supplier gives you a good deal, you can divide the theoretical production capacity (the amount of panels times the Wp per panel) by the price without BTW. Anything below 1,40 euro is good for a system with optimizers/micro-converters and below 1,10 euro for serial systems.

Example: Your supplier gives you an offer of installing 10 panels, for a total installation cost of 5.500 euro. The panels are 375 Wp each, so this gives a total of 3.750 Wp. Dividing the installation costs by this total gives us 1,46 euro per Wp (5500/3750). This is expensive, better negotiate a lower price!

Our installation clocks in at 1,16 euro per Wp. If I had ordered the panels and installation through Solar Bouwmarkt, this would have been lower still.

Better equipment is more expensive, and the same goes for trustworthy installation companies. Don’t compare on price alone: your panels will last 25 years, and you want a safe system! ✌️

That’s it! Hope you learned something and see you next time!

Last tip! Energy prices are high now. They might go up, they might go down. Understandably a lot of people worry if they should keep their ‘variable rate’ pricing or lock in a steady rate with a 1-year contract. Take a look at your preferred energy supplier’s ‘opzegvergoeding’, the fine you pay for ending your contract early. Ours is 100 euro. If prices drop considerably, it’s beneficial to pay that price and switch.

Take a look at my other article if you want to save on gas .

A very big thanks to Ronald from the Tweakers forum for all his help and information.

Originally published on Medium

Year review 2021

Sun, 16 Jan 2022 00:00:00 +0000

Another year, another yearly review. Vaccines didn’t kill Covid and we ended the year in a lockdown. Still, there was enough to be very happy for, and it feels like 2021 was a transitional year in a lot of respects for me. I have a good feeling about 2022.

This yearly review is becoming a January tradition for me (here’s last year, the year before that and two years back) and it’s motivating to see how much I can do in a year.

This I year I turned my attention from research, machine learning and experimentation, to writing and note-taking.

Personal stuff #

This was another year of mainly staying at home. We made our lockdown-weeks a lot less stressful by sacrificing vacation-days instead of playing “parent-teacher-employee” at the same time. Although it was better during lockdown, I noticed the lack of real free holidays broke me a bit in the long run. Turns out you need some time off every now and then.

My youngest son decided to switch from “sleeping the whole night through” to “waking up every night screaming” for the last 6 months. So we lost a lot of sleep. This is one of the thresholds I crossed in the last week of 2021: we are getting a sleeping coach for him. Fingers crossed for better nights 🤞

The books I really liked this year where the 99% invisible city and the surprisingly funny Every day information architecture. I also read a lot about consciousness and language.

Quick creative recap: I recorded a music-video with my daughter to the Frozen song. I did a workshop with an artist. I’m taking piano lessons with my daughter. And I take my camera with me some evenings when I go for a walk, adding some pretty cool photos to my collection.

the experiment I did in a personal workshop with an artist

House #

We bought a dish-washer (it took me months and multiple Excelsheets to get the right one 😆), had builders add an extension to our house, I’ve been painting a lot of walls and window panes, breaking away an old wall, and I laid another laminate floor. Pfew.

I even wrote an article on how I keep the house warm while saving energy.

New blog subject: The Future #

I started a new blog called T+2 on Medium (the articles also get archived on Matth-ijs.nl). In this blog I write ‘news articles from the future’.

The first article was about a girl wanting to replace her biological arms with bionic ones (and her father having some understandable doubts on that), the second article about a future without meat (do school-children see us meat-eaters from the past as murderers?) and the third article about a new kind of dress (written from the perspective of a fashion-writer who is fed up with augmented-reality clothing).

I have loads of ideas for follow up articles, so hit the follow button on Medium if you’re into this sort of thing.

New interest: note-taking! #

I picked up a new interest in note taking. I surveyed my friends and random strangers on how they make notes (read the resulting article here) and published an article comparing Bear, Notion and Roam.

I’ve decided that the space is pretty much saturated with great apps, but the one thing I’m missing is a way to keep small, structured notes on knowledge that I found. So I’m now writing my own software for knowledge-management. This is a very big step for me (crossing another threshold!), really putting in the effort to create something instead of only ‘dreaming about it’. The software is in pre-alpha: it works and I’m dogfooding it, but it’s not good enough to share yet. Sign up for the newsletter at the end of one of the articles, or just follow me on Medium to keep up to date on my new software.

The easter puzzle #

For the second year in a row, I made a cryptographic word-puzzle for my friends and family in the Easter weekend.

The theme was food.

This years Easter puzzle will be in English!

1001ideas.org #

I have a blog where I publish ‘strange inventions’ called 1001ideas.org. This year I published three silly ideas there: the Homepub for corona-times, the Churner to make butter while you work out, and a browser plugin called The New Yorker which adds swear words to every website.

Job stuff #

So I switched jobs! From a Sr. UX Designer position at a small (but incredible) agency – Angi Studio - to Sr. UX Designer at a large (but friendly) corporate: PostNL. Threshold crossed!

Design articles #

I wrote about a bad design-decision the municipality of The Hague made for a pedestrian crossing and a humorous (?) article about the bullshit designers peddle sometimes.

UX research #

I published an email-course in user-research (Dutch only) together with my then-colleagues at Angi Studio.

I also wrote several articles: how to make better notes of interviews (in Dutch), a description of my self-developed method to replace personas, an overview article of 17 alternative methods for personas. Over 6K people read that last article.

Statistics #

I wrote 782 words in 1001ideas.org and 19.155 words in my blog-articles, it’s 16 articles total. This is up from 11 articles, but still not as much as my record-year of 2020 where I wrote 20 articles. I doubled my word count with respect to last year. A typical book has 50K pages, so if I double my word count again this year I’ll have written a book.

This resulted in 12.540 views on Medium (especially the design articles, nobody really cares about my future-articles 🤣). I got an additional 2.100 views on this blog, especially the easter puzzle and my 2019 article on shared albums in Apple Photos.

It’s nice to see my articles get reach, but I should add the writing is mainly for myself. It helps me understand my thoughts.

2022 #

So, big plans for 2022: working on my own software and improving my skills as designer. Let’s see which transformations last and if more changes are coming 💪

19 tips for saving on heating — I’m dreaming of a warm Christmas

Fri, 10 Dec 2021 00:00:00 +0000

When you think about saving on energy, you probably think about big investments like solar-panels, floor-, wall- or ceiling-isolation and replacing window panes. Those cost time and money. I’ve collected a list of simple techniques and tools you can use to save on heating immediately.

May you have a warm and comfortable Christmas!

Photo by he gong on Unsplash

Update October 2022: I’ve added four tips see “👋 New tip”. Fuck Putin 🇺🇦

A list with tips to save on heating #

I’ve divided the list in three sections and chopped it up in bullet points. I’ve listed the time it takes to do it, the price and the expected returns.

I’ve tried to use ifferent shops for each link (I do not get commissions!), but please buy stuff at a local hardware shop instead of ordering it online!

All temperatures in degrees celsius. For this list I’m assuming you heat your house with radiators. I’m calling the big machine that heats up the water a CV, I think it’s called a boiler in English. I’m assuming you have a recent CV (less than 10 years old), and that it works on gas.

Do these things once: #

Radiatorfoil (30 minutes, 20 euro, less than a year to get your money back). There are two types of radiator-foil: either you stick it on the wall, or you stick it to the back of the radiator with magnets. The latter one works best and looks a lot better. Stop heating your wall and heat your room instead!
Tube isolation (30 minutes, 20 euro, less than a year). These grey tubes cost a euro per meter at my local hardware store. Make sure you get the right width. Easily cut them with a knife and tape it up. They keep the warmth inside the tubes. For attics and underneath your house, or rooms you don’t heat very often.
Radiator ventilator (10 minutes, 50–105 euro, unknown ROI). It’s a simple idea: place ventilators under your radiators and the heat gets distributed more quickly in your room. I’ve got them in our living room and they make a much more comfortable home.
Set your CV to 60 degrees (10 minutes, free, ROI immediately). Your CV most likely has two temperature settings: for warm tap-water and for radiators. The factory-settings are always too high and can be set lower. The tap-water can be set to 60 degrees (never lower because of legionella risk!). Changing the radiator-temperature is only useful for old CVs, most current CVs can regulate temperature themselves. For old CVs you can set radiator-temperarute to 60 degrees as well: lower in the summer (40 degrees) and higher in an extremely cold winter (80 degrees when it’s less than -5 degrees outside). If in doubt, ask about this during your yearly CV-checkup.
👋 Updated tip: Investigate ‘waterzijdig inregelen’ (half a day, 120 euro, ROI in one year). It is highly likely your radiators are not set to the most optimal setting. Optimizing the installation can be done manually, but explaining this goes way beyond this article 😅. You can get government subsidies for ‘inregelen’ if you do this together with a bigger investment like floor-isolation. A smaller step you can do very easily, is to adjust your radiator valves so the radiators work optimally (basically: do not open them fully!).
Foil for single-pane glass (30–60 minutes, 20 euro, ROI in one year). If your house still has single-pane glass, it’s better to replace it with double- or triple-pane glass. But if that’s not possible, you can do a temporary solution with this foil. It’s easy, and even fun to do (no really!), all you need is scissors and a hair-dryer.
Kill all drafts (30–60 minutes, 20 euro, ROI immediately). Any window with a small opening creates a draft. These self-adhesive strips stop those drafts. There are a lot of different types.
Set thermostat to 18,5 degrees (1 minute, immediate ROI). You most likely have a thermostat in your living room. If you’ve got all the above measures done, you can live comfortably in an 18,5 degree room. Keeping a steady temperature is better than turning the thermostat up and down. If you go on holiday, set the temperature to 16 degrees. Radiators only heat up if the room temperature goes below your thermostat setting and they stay on for as long as necessary to reach the temperature. If you’re cold, you don’t need to turn the thermostat way up, just nudge it up with .5 degrees.
👋 Updated tip: Turn the thermostat down at night and on holiday. If you live in an old house (pre-1995), then it’s worth setting the thermostat lower during the night and when you’re on holiday, for example 15 degrees Celsius. It’s best to program your thermostat to do this for you instead of doing this manually. Set the thermostat 2–3 degrees lower than your ‘day temperature’, don’t go lower: heating up from a too low state will cost more than you saved!
Bonus-tip: replace all lamps with LED (30–60 minutes, 10–50 euro, immediate ROI). Any lamp that’s not an LED is a power-waster. Replace them immediately. LEDs last nearly forever, so it’s worth using a more expensive brand (cheaper ones can introduce a flicker when you’re recording movies with your phone).

Keep doing these things: #

Set radiators to low in your bedrooms and other rooms. It’s healthier to sleep in a room that has a temperature of 16–18 degrees (use thicker blankets instead of jacking up the radiator). Set radiators to the lowest amount you find comfortable, and don’t change them too often: a constant room-temperature is more efficient than changing the heat. Make sure no room ever goes below 0 degrees, otherwise your radiators will break!
Shower less, and use low temperature. It’s healthier and better for your skin to shower less (every other day, for instance) and use lower temperatures. You can use a ‘shower timer’. Apparently ending your shower with a burst of cold water is good for you.
Ventilate your rooms. Keep a window open in your bedrooms, and open a window for at least 20 minutes in your living room.
Use an electric water boiler to boil water when you’re cooking: heating water in a pan on a stove is inefficient!
👋 New tip: capture solar heat during the day. If you have blinds or curtains, open them during the day if the sun is on the windows. This will capture solar heat. Dense (concrete, brick, etc) and dark-colored (black!) objects capture most heat.
👋 New tip: stop losing heat at night. Close blinds and curtains in the evenings and night. Windows lose a lot of heat. Best are insulating blinds or shutters that fit tightly into or around the window opening. Floor-to-ceiling lined drapes are a less-effective alternative, but are better than bare windows. Also add drapes in front of your outside-facing-doors! Source.
👋 New tip: local heating. Traditional heating works with convection, heating the air of an entire room so you feel comfortable wherever you are. But this is not always necessary. For instance, our family has breakfast in our living room, and then go to office/school/daycare. We spend about 45 minutes around the table in our living room in the morning. Do we really need to heat up the whole room for that? No. It can be more efficient to use a small infrared panel (very energy efficient, takes 10–20 minutes to heat up, use for long periods) or an electric fan (less energy efficient, direct warmth, use for short periods). This goes for your home-office as well. Infrared panels are very cheap and can look beautiful.
👋 New tip: stop losing body-heat. It might seem obvious, but keeping your body warm is a great way to save on your heating bill. No, you don’t need to wear a jacket in the house 🤦‍♂️ Invest in a beautiful sweater, tough-looking bodywarmer or fancy blazer. You can even get them heated. For your evening-tv Netflix-marathons you can snuggle under a fleece blanket (or an electrically heated one).

Do these things once a year: #

Repressurize your CV (see below)
Have a professional maintain your CV. This is compulsory! Also ask the maintenance person if your CV is set properly: is the outgoing water temperature 20% warmer than the incoming water? This is necessary to give the CV a high efficiency. Otherwise ask about “waterzijdig inregelen”. The professional can also repressurize your CV if you don’t want to do it yourself.
Clean your radiators. Mainly for health-reasons, but I bet it helps with efficiency as well.

This is how I clean my radiators :D

Addendum: taking the air out of the central heating and refill with water #

Especially in old houses the central heating may lose water pressure and the tubes get filled with air instead. This makes a ticking noise when the radiators are heating up and is less efficient. If the pressure is very low (below 1 bar), your CV can break!

Here are 8 simple steps to repressurize your CV:

Check the water pressure: if below 1.5 bar, refill water first (see step 6)
Turn on all radiators, set thermostat higher so hot water flows through all radiators
Wait 30 minutes for the radiators to heat up
For each radiator: use a special key (see photos below) and open the valve (see photo). Careful: the water is hot! You only need to open it a little bit: first air comes out, then it sputters, then water flows. Close the valve when the water flows! Start with the radiator that’s placed lowest in your house and work your way up. If no water starts flowing, you may need to fill your CV (step 6) before moving on to the next radiator
Open the valve on your “expansion vat” (check photo below), and let all the air out, just like you did with the radiators. After letting the air out: tap on the top part and the bottom part of the vat. These parts should make different sounds, if they don’t, your expansion vat is broken…
Refilling the central heating! This part can be tricky, read carefully:

First locate these five things: the valve that provides water, the valve for letting in water into the CV, the hose, a bucket and a dirty towel (check the photo!).
Connect the hose to the water inlet, open the valve and fill the hose with water (otherwise you’ll add air to the CV!), now connect the hose to the water inlet of the CV.
If everything is connected, open the valve of the CV and let the water flow. Check the barometer: your pressure should be between 1,5 and 2 bar. Older CVs needed to be turned off for this, newer ones don’t.
Turn off the valve for the water inlet AND the valve for providing the water. Keep your towel and bucket ready! Bend the hose close to the inlet, unscrew it, drop it in the bucket and let the water out
Take the hose off of the inlet. Don’t leave it connected: you can develop legionella in there

7. Write down the date and the current water pressure and check again in a month. If the pressure dropped a lot, you may have a leak (call a professional). The pressure always varies a bit with temperature, so this is not an exact science!

The photos for repressurizing #

My water-pressure is 1.5 bar

This magical key opens up the radiator valves. It also opens up portals to Narnia, if you’ve got the right closet

Impress your friends by repressurizing your radiators yourself

My expansion-vat. This is an oldie, yours is probably red!

1: the valve that provides water (I have to disconnect the washing machine), 2: the hose that’s going to connect numbers 3&4, 4: the valve that lets the water in, 4: a bucket and an old towel. These are the ingredients for a successful refilling.

That’s it for my tips, please share the article if it was useful.

Published first on Medium

Killed by good intentions

Sat, 02 Oct 2021 00:00:00 +0000

This is a story about a treacherous crossing and a saint in the municipality that tried to make it better. Sadly, this hero with his good intentions failed in his quest and has created a more dangerous scenario that will kill someone in the future. Killed by good intentions.

I’m reading the excellent book “The 99% Invisible City: A Field Guide to the Hidden World of Everyday Design” by Kurt Kohlstedt and Roman Mars. I can’t recommend it enough, it is chock-full of beautiful insights in city planning and history, and it creates an excitement for all the things surrounding us in a city. Plus, it looks beautiful (I’ve tried to mimic some of their style in the illustrations, but I failed of course). The best thing about the book is that I now pay more attention to the city around me.

That’s robably why I noticed the crossing below…

The situation: four crossings (and a funeral?) #

My hip new job at PostNL is located in a hip building in a recently ‘hipped up’ part of The Hague. It’s all very hip.

To reach my sparkly new office I have to cross four lanes of busy traffic. The four lanes are split into three roads with two ‘islands’ in the middle. Even though there are zebra-crossings to help me and my fellow pedestrians, the crossing is confusing: does the traffic come from the left or the right?

An overview of the crossing. We’ll be approaching it from the trees in the bottom left. Should we look to the left or to the right when using the zebra-crossing?

A clever city planner noticed this problem and provided a simple solution: signs!

The signs you should be seeing: ‘Traffic from the left!’

Sadly, this elegant solution doesn’t work in the real world.

A killer solution #

So we have a dangerous crossing-situation and the clever (and cost-effective!) solution of placing signs. But the clever solution does not work because of the placement of the signs.

Here’s the overview again. I’m standing before the first crossing and I’m looking in front of me (my field-of-view is in yellow in the image). The sign I should be seeing is sign 1 (“watch out! Traffic from the left”), but instead I see sign 2 (“Traffic from the right”) in my field-of-view:

Here’s what that view looks like:

The sign says ‘Warning! Traffic from the right!’, but the traffic comes from the LEFT!

The first time I approached the crossing, I didnt look left and right like I normally do (because of the sign), but only looked to the right-hand side. Which was the wrong side, because traffic comes from the left. Luckily the driver in the car was paying attention and stopped on time!

This problem mainly exists because we approach the crossing over the sidewalk next to the road (instead of head-on). As you can see in the image below, the first sign is very easy to escape your attention:

What you see when approaching the crossing

You need to take a few big steps back to see the proper sign. Who do you think would do that? That’s right: nobody.

Where to go from here? #

Good intentions led to placing these signs, but the result is an even more dangerous situation. Our ‘intervention’ brought confusion. What can we do?

The simplest solution I can come up with is to remove the signs. This way we let our pedestrians think for themselves again. The better solution would be traffic lights, but this is also expensive… If we are keen on keeping the signs, they should be placed lower, on eye level.

Lessons learned #

This is a story about a failure in city planning, but this story is applicable to the software-world as well. Any design-intervention, even the ones with the best of intentions, can have undesired consequences.

These are the type of good intentions that will get you killed.

To prevent such failures always test your design-interventions in the real-world! Test them before implementation (in a usability-test with a working prototype) and then test them again after implementation.

The real world can be cruel for us designers/city-planners with our good intentions.

Stay safe 😎

(I’ve reported the dangerous situation to the municipality. UPDATE: they don’t see a problem, “the signs are hanging as they should be”…)

The photos #

I realize that my illustrations are definitely not as good as the ones in The 99% Invisible City, so here are my photos.

Here’s what we should see… (sign says “cars approaching from the left”):

… but when we’re in front of the crossing, this is what we actually see (sign saying “cars approach from the right”):

When we approach the crossing, we don’t even notice the first sign:

Published first on Medium

Ditch your personas. Here are 17 alternatives you can use

Mon, 30 Aug 2021 00:00:00 +0000

Personas are one of those ‘things UX designers’ do. Here are some alternatives for you to use! 💪

If I say persona, you probably have an image in mind of what they look like: a name and a photo, some sliders, a quote, etc. I call these ‘template-personas’ and they are not useful. At best they offer a false promise: research once, reuse everywhere. At their worst, they are misleading us with stereotypes. Here are 17 alternatives you can use.

The dreaded template persona 😱. Remember… Your company is unique. Your situation is unique. Your research results are unique. Don’t use a template-persona

☝️ Tip, bookmark this article for future projects! ☝️

How to use these tools and methods #

All tools below contribute to the following steps in your research process:

Segment research findings
Communicate those findings
Create empathy with your audience
Make decisions

Yes. The end-goal of all these tools is to make decisions and not empathizing. I’ll sum up my main findings at the end of the article, but the most important conclusion is that you pick the method that solves your current problem. Be very wary to reuse any research for other problems!

The alternatives to personas #

Here’s a list of tools and methods to segment your data and communicate research-findings. Use them to can build empathy with your audience (and their contexts) and make decisions to create better services 💪

In no particular order…

1. Cooper’s original personas #

Alan Cooper is the ‘godfather’ of personas. This is what he wrote in the nineties:

Even though the variation among the users was dramatic, a clear pattern emerged after just a few interviews. The users fell into three distinct groups […] I had to describe those user models to the Sagent team. So I created Chuck, Cynthia, and Rob.

Cooper’s personas are an amalgamation of multiple interviews represented by a story. These personas give a feeling of the people Cooper interviewed, and can offer some guidance on making decisions. For example, based on the description we can foresee some choices in shopping-features for our platform: this person likes to find the exact items she’s looking for (extensive filter-options) and she makes note of items she doesn’t buy (saving items to a list).

A lot of the criticisms of personas are true for Cooper’s version as well (it brings to mind some stereotypes about white middle-aged women for instance), but the great thing about Cooper’s original peronas is that they at least show context.

2. The OMG-how-much-research-did-you-do-persona #

Amber Westerholm-Smyth wrote a bombshell — 17-minute read — article called “Your personas probably suck. Here’s how you can build them better” in which she explains how she created “better personas”.

These better personas are built on 2,700 minutes of interviews with their target audience, analyzed using complex Google Sheet formulations and Miro boards stretching several meters. It’s incredibly sophisticated research, and Amber takes you through it with amazing talent.

So what are these better-personas? They look quite similar to the template-personas, but lack any demographics or personification. Instead they contain a short story, tags, sliders and quotes.

The team uses the names of Greek gods to identify the different personas (“to align their fables with our own narratives”), and gave each persona its own color-scheme and graphics.

With the personas come a website (with quotes, video and sound clips) and a deck of cards with quotes.

These personas solve some of the issues. The lack of demographics (age, gender, etc) means fewer subconscious biasses*. But they still serve as static representations of a group of people: does everybody from your audience really fit into one of these personas?

Images from Amber Westerholm-Smyth’s tour-de-force article. A recommended read!

*Unless you know the Greek gods because you read Stephen Fry’s excellent Mythos , that is.

3. Persona groups (or ‘archetypes’) #

I found the term Persona Groups in a talk by Akil Benjamin during Sofa Conf 2020, and it really resonated with me. Instead of creating very concrete personas, he created abstract groups (his examples: ‘need a nudg’r’, ‘newbies’, ‘the stop starter’):

These groups emerge from the data (just like personas should), but they focus on only one ‘trait’. By dividing your audience into these groups, you can make decisions (should we focus on group A or B? Is this feature useful for all groups?).

The benefit of persona groups is that we can support the groups with additional data: “based on a survey, we can say 80% of our audience falls into group ‘newbies’.”

We use this same principle in Pokemapping, but we extend the idea into multiple groups and connect them to user-journeys.

This is very similar to how Netflix defines you as a person so it can suggest movies to you. You might not know it, but the almighty-algorithm has put you in several groups based on the movies you watched and skipped. For Netflix, you are defined by your membership of several groups: “Japanese murder-mystery movies from before 1975”, “super-hero movies with dark undertones”, “romantic comedies with lots of plants based in Guatemala”.

One way to create these persona-groups is to think of dimensions on which you can ‘rate’ the people in your target audience, and creating a diagram:

Image from “some alternatives to personas”, by Marianne Brierley

4. Roles #

Very closely related to persona groups is the idea of ‘roles’. In persona groups you take any possible trait, with roles you specifically look at what a group trying to achieve.

Roles are based on a goal, a task or defined as a job-to-be-done.

We used this for a project involving library-employees. These people have differing functions (manager, analyst, bookkeeper) and employment-types in the organization (freelance, volunteer), but there are a lot of similarities in the tasks they do.

When we segment our data on those tasks, patterns emerge clearly.

With the ‘roles’ defined, we used a customer-profile (see later in this article) to show the differences between the roles

Similar to roles, you can also create persona is by ‘motivation’: why do people use your service.

5. Minimalist personas #

In her almost-a-book article “When less is more: minimalist personas in UX” Melanie Polkosky describes what’s wrong with personas — it’s a great read. She also proposes an alternative: Minimalist Personas.

What are minimalist personas? They are simple representations custom for each project, containing only those variables that you as a designer select that have an impact on your design. Melanie suggests one to four key variables that range on different levels (for example: “ variable: engagement, levels: aware, interested, fully engaged” ).

In a project on training managers at restaurants, Melanie came to the following personas:

image from the minimalist personas article. Looks very similar to persona groups, no?

In another project for a university, Melanie grouped her audience on a spectrum ranging from aware to engaged:

remember this image, because it’ll look very familiar to the persona spectrums later on!

6. Persona spectrums #

Doug Kim from Microsoft Design suggests a modular way of looking at your audince in the classic article “Kill your personas”.

The core idea is to look at traits of your audience (similarly to persona groups!), and divide those traits onto a spectrum. The example Doug gives is based on accessibility: “There are roughly 20,000 people in the US with one arm. But if you add up the numbers of people with one arm, people with a temporary wrist injury or a broken arm, and folks with one free hand in a specific circumstance (like new parents lugging a baby around), you’re at 20 million in the U.S. alone.”

If that text was theoretical, this example hopefully makes it clear. Let’s say you want to make your website accessible for visually-impaired people. You could create a persona-spectrum around vision like this:

There are three things that stand out about persona-spectrums: you can easily add quantitative data, they show people are diverse and the spectrums are practically useful as a designer/manager. They trigger the “oh, this is a problem, how can we solve this”-response.

Doug also has some good insights on traditional personas:

[Personas] are inherently an amalgamation, an average of attributes that we imagine our average customer has. And there’s no such thing as the average customer

Well done, Doug.

Doug gives examples based on accessibility. But ‘spectrums’ are everywhere. For example, one of our clients made a persona spectrum (without calling it that) by up their audience of data-scientists into five categories based on the level of proficiency in data-research:

eggs, people who digitize information but don’t use it
squirrels, collectors who clean the information and make it available for research
pirates can create connections in the data
wizards can predict the future based on data
kings, who influence the organization to make predictions come true

7. Mindsets #

A specific type of persona spectrums is to l ook at the mindset of people. You can use the dimension ‘confidence’ for instance and create a spectrum there going from terrified to overconfident.

Images from the article Mindset over matter: a new design trick for your toolbox

You can then take that mindset and map it back to a descriptor that’s eerily similar to a persona-template…

… but the cool thing is that these mindsets still represent a group of people and that one person can be in multiple mindsets or shift from mindset to mindset. This way you capture a much more diverse set of voices.

8. Pokemapping #

At Angi Studio we’ve also developed our own process of creating representations of the target audience: Pokemapping.

It’s a simple three step process to convert research into action:

Do interviews
Write summaries as ‘user cards’
Create groups from these user cards

This results in persona groups, based on findings from the interviews. You can group based on role, goal, spectrum, motivation, or anything else that bubbles up from the research you did.

These groups can be used as-is, or as the starting point for any method/tool described in this article.

The result from the grouping step

Another cool thing about the Pokemapping-process is that the interview-summaries are useful tools as well. They communicate a real person, instead of a fake persona.

The interview summaries

9. Behavioral archetypes #

A behavioral archetype is another specific type of a persona group where you focus on (you guessed it): behavior.

The cool thing about behavioral archetypes is that you add in ‘moments’, which put the archetypes in a ‘state’. This shows that people change and how they are different in another context.

10. Proto personas (or ‘user cards’) #

In her article “some alternatives to personas”, Marianne Brierley talks about using proto-personas. These cards represent a type of user and represent a group of people. Like personas, they are given a name and some ‘needs’.

The cards are easier to make than ‘normal’ personas and because of their size can be used in workshops.

The downside is that these proto-personas will elicit your mind to fill in the blanks again leading to stereotypes. What do you imagine Kelly looks like, when you read the card below?

Image from “some alternatives to personas”, by Marianne Brierley

11. Customer profile #

Found in the highly-recommended and very practical book Value Proposition Design, the customer profile is a very practical way to combine and present data about a customer segment.

In the customer profile you write jobs, pains and gains. Basically: what your customer wants to achieve, what stops him from doing it and what are the benefits on achieving the ‘job’.

What makes this a very powerful tool is the connection to other tools by Strategyzer: the value-proposition-canvas and the business-model-canvas.

The tool is simple to understand in theory, but complex to use in practice. A customer profile visualizes a mix of business decisions and research: you don’t put all jobs, pains and gains you found in the profile, you make a choice which ones are relevant for your business. As such it’s very different from the idea of personas.

Image from Strategyzer, buy their book

12. Empathy Maps #

Empathy maps are a mix between personas and customer profiles. They are less abstract than customer profiles but less uselessly detailed than personas (no “has a dog, is 24 years old and drives a Volvo”). They focus on processes inside the brain: thoughts, emotion, attitude, influences, worries.

nice visual by Paul Boag, but be advised: the nicer the visual, the less-often it gets updated!

The empathy map template, by Paul Boag, see his article for different versions

The empathy map suffers from a lot of the same problems as personas do (being mis-interpreted, leading to stereo-types and they are difficult to support with research).

13. Characters #

The methods above (all of them, except Cooper’s traditional persona and the behavioral archetype) describe static properties. They describe a person or group of people detached from time. They describe Timmy with his needs, pains, gains and goals — but they don’t mention where, when or how he has them.

Usually you overcome this by attaching your persona-group to a user- or customer-journey. But Alan Klement has found another way of visualizing this information, by making what he calls a character. This character has motivations and is described together with events and situations.

image from article by Alan Klement, in a way it’s very similar to Cooper’s original, but then stacked horizontally instead of vertically

14. Dynamic Selves #

Lennard Overkamp and Emanuela Cozzi describe dynamic selves in their most excellent article “Beware the cut’n paste persona”.

You represent every individual with multiple ‘selves’, where each ‘serve’ represents a context you investigated. This accounts for people being different in different situations.

You can then visualize these cards on a canvas and categorize them (much like Pokemapping ) and come to design opportunities within these groups:

The Dynamic Selves have context. They show events and situations that support motivations behind someones behavior. Big advantage!

By adding in photos of the context, the dynamic selves elicit a feeling for the situation without introducing stereotypes about people.

15. Superflux Archetypes #

We’ve seen the word archetype come by a couple times in this article already. Wikipedia defines archetypes as “the pure form ”, the fundamental characteristics of a thing.

An archetype is the simplest way to describe a concept–in our case a group of people, with all details are stripped.

Example archetypes can be found in movies and books: the hero, the guide, the enemy. These characters are in every movie, but they differ a lot from each other in demographics, temperament, story and appearance. Frodo, Neo, Jesus, Wonder Woman, Bambi. All heroes, all very different.

In a way, archetypes are the ultimate goal for personas: reusable but relatable.

But alas! Like most ideals, the idea fails in practice. Why? Because abstraction by definition means removing practical details, making an archetype useless to communicate research-results to make decisions in a business setting.

But there are other uses for archetypes. Take a look at Superflux’s cool Tarot-card deck with ’the usual suspects’ of archetypes. Use it to get creative thinking going in an ideation part of a project or to broaden your thinking during a workshop.

16. Switching forces framework #

A powerful moment in a customer’s journey is the moment they stop using your product and start using your competitor’s product. Or the moment where they stop using their own homegrown method and start using your product.

You can visualize the reasons behind these switches using the Switching Forces framework (see image below). Simply copy/paste the image onto a Miro board and add post-its describing the findings from your research in each quadrant. To get the best results, you use a technique called ‘switch interviews’, which you can learn about by listening to this podcast or reading this book by Chris Spiek and Bob Moesta.

image from “some alternatives to personas”, by Marianne Brierley

17. Baseline behavior #

Another way to visualize change is by using the baseline behavior map, which I saw in the talk by John Cutler at SofaConf 2020.

Baseline behavior is a decision/alignment tool instead of a ‘user research visualization tool’. You use it before research, or during, or even after the launch of your product.

for “step 1: target” you could use any of the other methods in this article. Synergy, I love it

Other ways to visualize your user research #

Besides personas, there are other ways to visualize your user research: task-flows, mental models, user-/customer-journeys, jobs-to-be-done, asset-maps, Triptech, user clocks, and …

Most of these can be combined with any method or tool mentioned above.

Closing thoughts #

I think user-research works best when it is specific and opinionated.

Don’t try to create personas that are everlasting. They are too general. Pick your tools specifically for the research question. Be scared to reuse your findings because context matters.

Make your research count. Be opinionated. Your stakeholders should not search around in your your persona to get an answer. They shouldn’t make complex interpretations. That’s your job as a researcher. Don’t communicate everything you find, but interpret your data and make a story related to the problem at hand. Make it easy to make a decision and get on with the work.

Sometimes you don’t need to segment your audience at all. Gerry McGovern advises against using audiences in his seminal work on top tasks.

I’ll leave you with this quote by Radina Denova:

The main thing to keep in mind is that creating personas […] is not an end in itself. They are a* decision-guiding tool *and as such, they need to be tailored to each individual project to serve its purpose.
source

Published first on Medium

Pokemapping: a simple process to replace personas

Wed, 21 Jul 2021 00:00:00 +0000

Summary for quick readers: personas have problems, but the idea of empathizing with people is nice. With Pokemapping we’ve invented a simple way to replace personas.

The goal of Pokemapping is simple: translate findings from research into action and give your teammates a sense of who these people are that use your products. In this article, I’ll give you a simple three-step process for using research findings in design.

Excited? I am! It’s every designer’s wet dream: writing an article about how bad personas are and propose a solution!

Why Pokemapping? #

Personas were invented for translating insights from research into the product development cycle, but they are used wrong.

Books have been written about how bad personas are. Many books. We can fill two internets with people ripping into personas.

I’m going to keep it short: personas are fundamentally bad and they are used wrong in practice:

Fundamentally, personas are bad on a philosophical/psychological level. The idea of persona is compacting many different people into one fictional character. This character does not exist. It’s an abstraction. We threw away anything that made this person unique and that’s problematic because our minds will automatically fill in those details based on assumptions we have. (Not convinced? Look at the last question in my FAQ below!)
Personas are used wrong: people fill in a template and they imagine 80% of the content. Such personas cannot be used by designers, product owners, or developers. And thus, the zombie-persona is born: lots of work to make, big introduction-party in the company, and they are then forgotten about (and are left for dead, pinned on a designers-wall)

Personas are bad, mkay? Combining research from different people into one fictional person, often dumping that info in one standardized template and inventing some new facts along the way. (you might have some issues with this visual, saying we should’ve segmented better, like selecting chickens instead of all farm animals, but A: it’s rarely this simple to see which groups exist and B: do you think all chickens are the same? That sounds wrong…)

We say that personas create empathy. But how can you have empathy with people that don’t exist?

Personas are bad. But there are many other ways to translate research for product- or service design! You can think of Jobs-to-be-done, persona-spectrums, journey-mapping, switching forces, customer profiles, ODI…

And now we introduce yet another method: Pokemapping!

How to get started with Pokemapping, today! 🎉 #

Use Pokemapping as a way to translate interviews on a specific topic into useful insights of your target audience.

Pokemapping requires preparation. Steps one and two are done by researchers, step three can be done with your whole team.

✅ Step 1: interviews #

Pokemapping starts with interviews. You can enhance that qualitative data with other data later, like surveys or analytics (mixed methods, yeah!), for extra impact.

Like always, make sure you focus on a good segment of your audience and on one specific topic. Interviewing random strangers about everything is a waste of time.

✅ Step 2: summarize your interviews into user cards #

For each interview, you create a summarized version of your notes. You can use literal quotes or rephrase/summarize. You can move data from the interview around if it makes a better story (if someone mentions printing a document at the start of the interview and again in the end, you can group these findings). Add a link in each card to the complete interview transcript or video.

We call this a user card: a summary of an interview with a real person.

Give the user card a name. Not a human name, just something you can refer to. I focus on features that made the interview unique (‘printsalotix’ for a person that prints a lot of documents). Sources of inspiration for names are Pokemon names (‘printard’), Asterix and Obelix (‘printsalotix’) or ‘sword names from Game of Thrones’ (‘paperslinger’). If you are boring, you can use numbers for the names (P1, P2, …).

Make your cards on a virtual whiteboard like Miro or Mural, or print them out!

Example of a user card

✅ Step three: grouping #

Now comes the ‘mapping’ part of Pokemapping! This is a fun exercise which you can do with the whole ~family~ team!

Let your team group the user cards into two or three groups. Resist the urge for easy groups (by function title, gender, or steps in the journey), instead, look at the data in the cards and create groups based on that. Name the groups with short descriptions (‘always fights with printing’, ‘prints because it’s easier to read’).

Now shuffle the user-cards-deck again and do another grouping. See which other patterns you can find in the data!

The goal of clustering user-cards is to discover features in your audiences, common threads.

Last step: grouping! Do this iteratively ;)

Example time: a check-in experience at the airport #

Let’s make Pokemapping concrete with an example: Imagine working for an airline-company and you’re investigating the check-in&boarding journey. You interviewed 15 people and made user-cards out of them.

You start grouping the cards and writing down the groups. You then shuffle the user cards and group them again (and again).

This leaves you with several ways to ‘segment’ your audience based on properties, tasks, personalities:

People with heavy bags
People with a limited understanding of Dutch or English
Travel goal (holiday, visiting family, business)
Traveling alone, with a group, or with kids

There is overlap in these groups: you might notice that the same user-cards show up together in several groups. And it’s very easy to then imagine overlapping groups: the ‘holiday family with heavy bags arriving two hours in advance’ and ‘the business person in a rush with little luggage’. But then you’re making personas again! Don’t do that! Only focus on what you actually saw in the data. Why? Because it’s easy to imagine a counter-example: the musician traveling with an enormous cello for her concert in New York.

Using Pokemapping in design (and beyond! 🚀) #

Now you’ve got your groups, you can use them in design.

Take the user-journey we imagined in the example above: the check-in and boarding journey. It has several steps: arrive at the airport, check-in, give off bags, go through security, board plane. You’ll now go through the journey with each group you found: where are the pain-points in the journey for people traveling with kids? Where can we add extra value for people with heavy bags? Or take two or three groups: what if a person is on a business trip with his son and has limited English understanding, what happens in his version of the user-journey?

To make your case stronger, corroborate your features with other data like surveys and analytics. How many people fall into the group ‘heavy bags’? How often are people traveling for business or holidays?

To be fair, you might say: “ But Matthijs, I can do that with personas too?” And I would answer: “Not really!”

Let’s say we created two persona’s ‘Bob Business with a frequent flyer pass and knows the layout of the airport like the back of his thumb’ and ‘Jolly Jane who goes on a holiday with her kids, who’s confused and scared by the airport’. First of all: that’s stereotypical and misogynistic. Secondly, it’s too broad to use as a designer (how can I design for someone who’s confused AND has kids AND is scared). Thirdly, we can’t ask follow-up questions (“why is she scared?”) because there are no transcripts to go back to. Fourthly, we can’t corroborate with other data (how many people resemble the personas ‘Jane’ and ‘Bob’? Unknown. How many people are we excluding who are not falling into these two personas? Also unknown).

Another way you can use this method is to go back to the usercards. As Designer, you can take three random user cards and imagine how they’ll go through your newly designed flow. It’s a lightweight usability test.

Limitations of Pokemapping 😱 #

The limitations of Pokemapping are that it’s context-bound. Your interviews were related to one topic (remember, from step 1?), so your groups are not reusable for other topics! This is different from personas, which are meant as ‘evergreens’. Just because someone is always printing receipts after buying one of your products, does not mean they will want to print your support-documents!

Pokemapping is also not a replacement for traditional analysis: tagging your transcripts and doing in-depth analyses. You’ll gain different insights from that, especially when using a tool like Dovetail where you can combine findings from multiple interviews!

FAQs (FAntasized Questions) #

Here are some questions you might have about the process

Can I reuse my user cards for other situations? #

Short answer: no
Longer answer: I advise against it

Let’s look back at our ‘airport customer journey’. You find out there’s a group that arrives more than two hours before a flight, the ‘early birds’. You use that finding to see if there is enough seating area for early-arrivals, you can take away redundant check-in desks, etc.

But is ‘early bird’ a personality trait, or is it specific to flying? Can we reuse this for hotel bookings? I’m not convinced! Looking at myself, I’m always early for flights. But always late for meetings and family events. So we can not re-use the ‘early birds’ group in other situations!

Your interview has a context, your user cards are also in context.

This is a lot of work! #

That’s not a question, but okay. Fair enough, this is additional work. Plus you cannot reuse it. But it leads to new insights from the same data. It’s a net time-saver.

How does this relate to customer profiles? #

Customer profiles are a technique to segment and summarize your audience, they contain ‘pains, gains and jobs’ ( in the Strategyzer framework ).

The groups you create in Pokemapping can be starting points for profiles. Pick the groups that are most promising for the problem at hand (you need to choose, it’s your product/service), and you can build profiles with pains, gains and jobs. These you can now trace back to individual contexts.

Keep in mind, customer profiles are a business tool, not a research communication tool. A customer profile visualizes a mix of business decisions and research: you don’t put all jobs, pains and gains there, you make a choice which ones are relevant for your business. This makes it different from user cards (summary of interview of an actual person) and personas (a cobbled together nonexistent person).

Did you just reinvent Affinity Diagrams? (or Affinity Mapping) #

Eh, good question! Yes, this is related to affinity diagrams (a favorite UX-tool of mine), but it’s not the same. In an affinity diagram you group separate findings, instead of people.

You get different types of insights out of both exercises. With affinity diagrams you get findings related to your service, with Pokemapping you focus more on your audience behaviors. The added advantage is that with Pokemapping you might see more fundamental personality traits, things that are hidden throughout the entire interview (‘hesitant in answering questions’).

Can I make a persona out of a group of user cards? #

This is like asking “can I make lasagna from plastic”: you can, but why would you? You cannot consume the end-result.

I’m not convinced personas are bad, Matthijs? #

That’s okay. You don’t have to be convinced 🙈.

I can recommend going on a Wikipedia-journey: the jaggedness principle, no-true Scotsman, stereotypes, dehumanization, and (most importantly) the reification fallacy. After reading those, we can talk about the relationship of these concepts to personas over a beer 🍻

And if you’re still up for reading after that, try this article by Microsoft Design and this one by Emanuela and Lennart.

Thanks to Alienor for inventing Pokemapping with me!

Published first on Medium

Roam, Bear or Notion: which of these popular apps works for you?

Thu, 27 May 2021 00:00:00 +0000

There are a multiple reasons to write notes, such as storing information for later use, making your thoughts concrete and sharing information. There’s a lot of overlap in there: the best way to make thoughts concrete is by sharing them and the best way to store information so you can use it later is by thinking about it.

So no surprise that the tools people use have a lot of overlap as well!

There are three tools that keep popping up in my research into note-taking: Bear, Notion and Roam. They serve different purposes but also share some overlap, which can be confusing.

So are you thinking of using Bear, Notion or Roam? Or can’t decide which tool to pick? Then here’s a handy list with pros and cons on when you should use which tool! 💪

Bear: a simple, but majestic app 🐻 #

Bear is a beautiful note-taking app. It’s the place where I’m writing this article. I use it to collect future article-ideas, write articles and store interesting information from others (like writing book-summaries or capturing a web-page with the Bear-browser-plugin). It syncs notes between iPhone, iPad and Mac.

Bear is the most basic version of a note-taking app: it opens with a blank page where you write down your thoughts. That’s it! Beautiful writing with full Markdown support.

Your notes are stored in one long list in the sidebar. That could become messy, but Bear offers two ways of keeping things tidy: you can tag your notes or you can add links between them like a Wiki.

It’s an incredibly simple app, but Bear has just the right amount of complexity. Like a beautiful export to PDF (or Word, or Markdown, or JPG, or HTML, or…), fantastic syncing between iOS&Mac and X-callback support (so you can quickly create notes with a template).

Bear is the type of app that happens when someone says “let’s make the best hammer ever”. One purpose, but simply the best version of that.

example of a note I made for a video by Erika Hall in Bear. I should add here that the screenshot looks a lot messier than the app itself, which really is beautiful

So is it all great? No, Bear doesn’t offer any way to work together on notes and there’s no way to publish notes. If you want others to read your work, you’ll have to export it.

It’s also difficult to find back bits of information when you need them: all text is “locked” inside a note. Let’s say you added Erika’s video (see screenshot above) a couple of years ago and now you want to use a quote from her about innovation. You then first have to find back the particular note that contained it or use the search-functionality.

Another peculiar downside is that Bear syncs all content to your mobile devices. If you store files and images in Bear, it’ll fill up your device quickly.

✅ Use Bear when you want to: write privately and collect your own ideas

🛑 Don’t use Bear when: you need to work together or want to share your content exporting, need to organize your notes hierarchically, or when you want to capture and analyze information

Alternatives to Bear #

Bear is only available for the Apple ecosystem and it costs (a little bit of) money 😱. You could try out Apple Notes, Microsoft OneNote or Google Keep as free alternatives, or Notability and iA Writer as paid alternatives.

Notion: one tool to rule them all 💍 #

Notion is the tool I love to love. It’s as if some megalomaniac said “let’s make a tool that can do everything”. It’s the cliche Swiss-army-knife. Notion can grow the grain, bake you a bread, and brew a nice cup of coffee to go with it.

You can use Notion for anything: organize your recipes or create a company Wiki or do project-management or use it as a writing tool. Just look at how these companies are using Notion! It’s incredible. In theory Notion can replace Confluence, Trello, Wikis, Bear, your CRM, Evernote, your cookbook…

But what is Notion? I will explain Notion as a step up from Bear. Remember that Bear had that long list of notes? In Notion you can nest items underneath each other: so you can create a hierarchy. You can create a page called “Meeting notes” and create pages underneath with each meeting. But that’s not all: Notion can also display pages in a table. That way you have a list of meetings together with their metadata: date, attendees, topic.

Besides an enhanced navigation, there’s another fundamental difference between Bear and Notion. Pages in Bear are basically long laps of text (like Word-documents), but pages in Notion are built using blocks (like Lego’s). That makes pages much more powerful. Add a block with PDF-preview, a Map embed, a Youtube video, a todo-list, a calendar, a tweet, etc.

On top of all that: Notion is free for personal use and you can work together in it and you can share your Notion as a website. And it’s cross-platform!

The same note as in Bear, but now in Notion. See how you can embed videos here? Pretty coolio! And look at that left-hand navigation, a lot more structured than the list of tagged notes in Bear!

So I’m almost scared to ask 🙈: what is there not to like about Notion? This is personal, but I find the writing-experience in Notion slightly worse than Bear. Not a whole lot, but just enough to keep me from switching. The mobile app is also slightly worse than Bear: syncing goes slower and editing text is more difficult due to the block-approach.

And alas! Notion doesn’t have tags for pages. So finding pages back can only be done using the hierarchy, search or by manually adding links from one page to another (although Notion does have a handy feature called “back-links” where you can see all the pages that contain a link to the current one you’re viewing).

Just like Bear, all information you add to Notion is stored in documents. Even though you build your documents using blocks, these separate blocks are part of a document. So we still need to remember a note exists if we want a quote on the subject of “innovation”.

✅ Use Notion when you want to: work together on things, store structured knowledge (recipes, lists of tools, articles), quickly create a content-website

🛑 Don’t use Notion when: you want a clean writing experience or when you want analyze snippets of information

Alternatives to Notion #

It’s hard to give alternatives to Notion: the tool is very well-thought out and free for personal use. You could use a Wiki for the documentation-part (like Tiddlywiki) or Airtable for the tables-part, but they are probably overkill if you’re reading an “introduction-type article” like this 🤣

if you really like the idea of Notion, but would like it to have another name, you could try out coda.io , it’s a competitor which looks exactly the same

Roam Research ⛵️ #

Last in the list is Roam Research. A relative newcomer in the world of apps, but with some powerful functionality! And although you could use it as a note-taking app like Bear, or a Wiki like Notion, it’s actually more of a knowledge-management application.

What is Roam? Well let me explain it using Notion as a starting point: remember that you could add blocks in Notion to build up your posts? For example add a YouTube video, a quote and a bulleted-list? Well, Roam can do that on steroids!

Unlike Notion, every block in Roam is a separate thing on its own instead of part of a document. It’s crazy. Blocks can embed other blocks or be opened as their own page. Imagine having a box, with boxes inside it, and every one of those boxes containing more boxes, and… Well, you get the picture.

And that’s not all: every block in Roam Research can be tagged. Sticking with our boxes example in the previous paragraph: imagine you wrote words on each of the boxes. You can now ask things like “give me all the boxes that have the word ‘cat’ on them”, or even “give me the boxes that contain boxes with the word ‘cat’ on them”.

Martijn Aslander calls this information liquidity: instead of storing information in documents, you think about it as small bits. With Roam you can retrieve information better, create new associations, and see patterns and trends. Things that Notion and Bear are just not that suited for.

Is that useful? Yes! Is that confusing? HELL YES!

Roam is a BEAST of a tool. It will eat you alive if you cannot tame it. The absurd amount of “how to use Roam” videos on Youtube should give you a clue: this is not an easy tool. It’s crazy complex and uninviting, I followed a free course by Jamoe to make sense of it, but you could also take a look at RoamStack for quick guides.

Roam is a BEAST of a tool. It will eat you alive if you cannot tame it.

However, if you can master it, Roam has a lot to give. Finding unexpected relationships in your thoughts and easily finding small bits of tagged information.

See how confusing Erika looks in Roam?

So what about finding back a quote on innovation by Erika using Roam? That’s actually “simple” (as if anything is simple in Roam 🤣). If we tagged the quote with the word “innovation” we can simply go to the tag-page and see all innovation-quotes there.

Here’s information we tagged on innovation. It’s a short list now (I’m only just starting out in Roam), but it’s already helpful that I can see quotes here from other people

…we can now easily see all videos we’ve ever added, and we can filter this list on “tagged with innovation” or “by Erika Hall”. Very useful stuff!

And there’s more! 🤩 Roam has a really cool sidebar in which you can preview pages and “collect” useful blocks of information. That way you can stack different pieces of information, for example when you want to write an article.

So any other drawbacks besides crazy complexity? Yes! Roam is not available on mobile yet (which is ridiculous) and it costs 15 USD per month.

If you still dare, you can give the app a spin by looking at their Support Pages, because that’s actually a Roam Database in disguise.

Roam also has this crazy graph view that visualizes all the links you’ve made (I’m not sure it’s very useful, but it looks cool!)

✅ Use Roam Research when you want to: store information in small units instead of long notes, make connections, re-use information, randomly bump into thoughts you had five years ago

🛑 Don’t use Roam Research when: you just want to write a small note or article, when you want to share information with others, or if complex software is not your thing

Alternatives #

Roam gets heavy competition from its opensource clone Obsidian, so that’s a good starting point if you are looking for an alternative.

If you are looking for an easier-to-use system I can recommend Dovetail. It’s not actually meant for personal knowledge management, and it misses all the hotshot features of Roam (like backlinks and graph-view), but you can tag parts of text and it’s crazy user-friendly.

Wrapping up 🌮 #

Here’s my most simple wrap up of this article:

Bear is like a paper notepad: you can easily write notes without distraction
Notion is like a Wiki: you can work together and write a growing body of work
Roam is an knowledge database: you store small bits of information in it and can retrieve them easily

And yeah: that view is way too simple. You can also use Bear to store small bits of information, or use Roam to write, or… But I think every tool has its own place.

Are you using these tools? Or something else? Let me know! I LOVE to hear how you do your note taking and “personal knowledge management” ♥️

My note taking process is a mess and this is my journey to something better 👌 Follow me on Medium

Published first on Medium

Typical design articles. Get your inspiration here for new bullshit!

Wed, 26 May 2021 00:00:00 +0000

I love the design community. Really. And I appreciate all the hard work everybody’s doing in producing nice new content for free. Really!!

But…

Sometimes…

Sometimes those articles are a little bit… How should I say? A little bit too predictable? Or too meta? Or too clickbaity?

Anyway, decided to have some fun coming up with articles I’d expect to see here on Medium. Enjoy!

Feel free to copy one of these titles as a starting point for your next Medium article on design:

How design thinking revolutionized the way I do groceries
Top 5 things I learned about UX design from being an alcoholic
Don’t get held back by imposter syndrome: how to become a UX designer in three days
UX design for good: how George Washington designed the USA
Hamburger flipping: how I apply my skills as a McDonalds employee in my day-to-day UX Research
Strategic and service design: a practical guide modeled on the downfall of the Roman Empire
Mobile first for architecture: three lessons from UX-writing
The strange interference of moon phases and discovery sprints. Five practical tips
Three reasons personas are like solar-systems
The practical UX of narwals
Cannot afford a mortgage right now? Gibson’s affordance theory explained in 5 gif’s
How Fitt’s Law relates will revolutionise dating in 2025
Service design IS product design. Why your service is actually a productized market research process
UX/UI or UI/UX — Finally the clarification you’ve all been waiting for
How service design could be the key to solving the Palestine-Israeli conflict
User design. How we stopped worrying about our product and focused on the real problem instead
From workshops to shopwork: design your next career opportunity as supermarket-employee
UXI, the new paradigm rapidly replacing UX/UI. Are you ready for a new job title?

Obligatory random cliche-Unsplash photo. Photo by Mia Baker on Unsplash

Published first on Medium

Ten note-taking stories from the real world

Thu, 13 May 2021 00:00:00 +0000

How we write notes and collect our thoughts

I am investigating how real people write notes as an inspiration for finding my own way in note-taking. This is an ongoing project and I am looking for more examples in a survey, but I thought I could share the interesting stories I have already collected!

Why start this research? #

We are living in the Information Age. More content than you can ever consume is being created every second. And if you don’t keep up, you’ll be out-of-date in no time. Or so they say 🙃

How can we stay afloat in such an ocean of information? We are bombarded with interesting tweets, books, articles, videos, conversations and podcasts. Our brains are impressive but they weren’t meant to process so much data, let alone store it for long time use.

So we keep notes, bookmarks, lists and highlights in books. At least, that’s what I do. And I can tell you it’s not going great: I lose a lot of stuff and it’s hard finding information back. My ‘remembering system’ is definitely broken.

I turned to Google for help and the Internet did not disappoint; there are amazing articles on Personal Knowledge Management, expensive courses you can take, great software and complex note-taking-lifestyles like Zettelkasten. That last one is an amazingly well-thought-out system with purpose-built tools, but I will never be able to do that. It just takes too much rigor.

This made me curious: are others saving notes, ideas and thoughts? And how? Are they happy with their systems?

How do people like me (people who do other stuff besides taking notes ) keep track of interesting information?

10 note-taking stories from the real world #

Here are ten note-taking stories from the real world. I hope these stories inspire you to better note-taking, let me know if they do!

1. The tech-enthusiast #

I asked this amazing tech-enthusiast-with-a-PhD about his note-taking method and he confided me it’s a mess for him too. Amazing, because this is the most structured (and hardworking) person I know. What’s more, for him Zettelkasten also costs too much discipline. So I don’t know what kind of Superhuman actually maintains a Zettelkasten system, if it’s you, I want to talk to you!! ;)

This tech-enthusiast is using VimWiki for text which works reasonably for him (it’s very nerdy open-source software that you use in a command-line), and Notability on the iPad for handwritten notes. He combines that with Apple Notes because it’s easily accessible.

His main problem is integration: how can you combine information from those different apps and access them on your other devices? For instance adding a sketch to a note: that’s crazy difficult to do in VimWiki. You make your sketch on iPad, export, save, import in Markdown and move it to the right place — a lot of hassle. Notability works great for handwritten things (and audio and pictures), but the integration and text-writing is less easy.

2. The list-maker #

Here’s the note-taking story of a hands-on engineer working at a huge technical company.

This engineer makes a Google Doc for every book he’s read, with all his notes in it (and a PDF of the book so he can read it on his phone 🏴‍☠️).

He also keeps some very simple lists on his notes app on his phone. There are six lists in total: “dinner-ideas”, “films and series”, “holiday-planning” and “books”, plus two lists with ideas: “ideas to execute within a year” and “ideas to execute another time”. The notes app syncs to Dropbox, but he never uses the lists anywhere else than his phone.

Asked where he stores information from other sources than books or where he combines information the answer is simple: “it’s all in my head 😂” (smiley by the list-maker himself!)

3. A man with ideas #

This is a man with a lot of ideas, a brain in constant motion, broad interests and always three projects running at the same time.

Sounds like myself 😇

Perhaps it’s not so strange that his note-taking process is in the same state as me: “it’s a work in progress” and he’s looking for optimization. This person writes his own opinions and insights into a paper-journal, but he relies on his head to store interesting bits from books, podcasts and articles. He admits that “means forgetting a lot”.

“I have a preference for paper notes above apps. Information you write down with a pen is scientifically proven to be easier to remember” 🤔

His dream process is writing physical cards (again on paper) with short notes and a referral to the information source, and reading those back periodically. That comes close to the Zettelkasten-system.

4. A teacher #

This teacher used Evernote for a very long time, but his usage dwindled and the system “watered down”. For big projects he uses Prezi. It’s a huge canvas for making presentations, which this person uses to cluster and organize information:

“It’s super handy: you can zoom in and out, make lines between areas, make quotes small and paste them in there. It’s great for making a time line of events for instance”.

For his latest personal project (which like all personal projects is being postponed) he’s bought a big book with blank pages.

Quick tip: I love the idea of using a canvas for organizing ideas, but instead of Prezi I suggest you take a look at Google Jamboard , Mural , Figjam or Miro as well. Those tools might be better for that purpose

5. The designer and workshop leader #

This is a designer who’s continuously improving herself: improving her current skills and expanding her abilities. I was very curious to know how she manages that!

This designer mostly saves links: whenever she reads an interesting article from a newsletter and thinks to herself: “that’ll come in handy one day”, she saves it to Pocket with the appropriate tag. That’s where the info remains: she doesn’t look at it anymore, unless she really needs it.

She also shares articles with her colleagues, by “throwing” the link in Slack with some notes on why it’s interesting.

Furthermore, whenever she attends an online course or webinar she makes notes while she’s watching, as a kind of summary. Bear is her go-to app for this kind of notes: she highlights the relevant bits or adds an emoji in front of interesting things (👉💡).

For this designer information needs to be actionable: articles and highlights are saved (or shared with colleagues) only when they contain practical advice or methods.

Overall she’s reasonably satisfied with her process, although she wouldn’t call it a process.

Practical tip: 👉 use emoji’s to highlight useful things in your notes!

6. The developer #

This developer uses browser bookmarks, ‘likes’ on Twitter posts, draft e-mails and e-mails sent to self, combined with paper post-it notes.

She’s really storing information where she finds them, without any thought-out “proper process”.

For some people that might be okay, but this developer is very critical of herself: “it’s a bad process”.

7. Senior designer and knowledge enthusiast #

This is the note-taking and knowledge-management story of a senior designer and all-round knowledge enthusiast. This is a person that reads more books in a week than I in a year, someone who answers any question in-depth with links to interesting articles and books. So of course, I was very interested in hearing his process.

“What’s my system? I would love to tell you, if I had any kind of system”

He stores articles in Pocket, and hooked up his book highlights to Readwise.io. He also started using Roam Research to collect and capture things that are top of mind (because he stumbled across them recently).

While reading or watching something, he tries to capture those moments when he feels that little tingle of connection “oh gosh, that’s a bit like that other thing!”.

8. The Pastor and I.T. Professional #

This pastor and I.T. professional has got it figured out: he’s the only one in my survey who rates his note-taking workflow as “Perfect 🤩”. That kind of success doesn’t happen overnight, he spent decades developing his workflow.

He uses both the Bear-app (a Markdown-based notes-app on iPhone, iPad or Mac OS) and a Leuchtturm 1917 journal (on paper :D) as “points of entry”. He then transfers interesting notes to his permanent store and primary workplace and research tool: TiddlyWiki.

9. Me (designer/researcher) #

I find interesting ideas and thoughts everywhere: in online-articles, books, podcasts, YouTube, tweets, conversations with people and in my own head. I try to ‘save’ these ideas as much as possible, but sadly, my process of keeping track of that is all over the place….

I bookmark links in Raindrop, which I subsequently never look at 😂 Articles I really want to read are kept open in my browser tabs (currently 27 tabs open) or added to Todoist. I keep multiple lists in Google Keep: book-ideas, article ideas, inspiration for 1001ideas-blog, shopping lists, etc. I also use Bear for notes and writing (the article you’re reading started in Bear) and storing interesting articles as full-text with some added conclusions. “Projects” that consist of multiple notes or where I work together with others are kept in Notion. Big stories or published-notes are written and stored in iA Writer. I make annotations in my ereader, have an “inspiration” album in Apple Photos, I keep paper notes, have some (markdown) files hidden in my filesystem… and of course a lot of stuff resides in my brain.

Like I said, it’s a big mess.

Some of my notebooks: Big Red Books that my friends call “My bibles” because I used to drag them everywhere

I don’t have a 10th person yet, but I’m looking for more great note-taking stories. I’m hoping to find a process that fits for me.

I’m especially looking for the non-perfect-processes, the ones that are developing in the real world. So please help me learn by answering some questions: https://tally.so/r/mRdVQw

I will be very grateful!

What’s next? #

I’ll be writing a follow up on this article with more stories on note-taking and collecting ideas!

I’m also compiling a list of software that people use for note-taking, todos and Personal Knowledge Management.

Published on Medium

Betere notulen met minder handwerk. Trint to the rescue!

Thu, 22 Apr 2021 00:00:00 +0000

Onderzoek is bij Angi Studio de kern. We voeren voor onze klanten allerlei typen onderzoek uit, maar vaak maken interviews daar een deel van uit. Om die interviews te analyseren, maken we gebruik van notulen. Maar zo’n gesprek annoteren is veel werk. Dat kan nu makkelijker met Trint!

Goede notulen zijn belangrijk #

Goede conclusies krijgen uit een serie interviews bestaat uit twee simpele sommen:

1. goed interview + goede notulen = goede data
2. goede data + goede analyse = goede conclusies

Zoals je ziet zijn goede notulen hierbij de spil. Uit een slecht interview met goede notulen is soms nog wel te redden (je kunt dan bijvoorbeeld leidende vragen herkennen). Maar andersom is dat niet zo, slechte notulen kun je niet vertrouwen.

Voorbeeld #

Niet overtuigd dat goede notulen belangrijk zijn? Lees dan even mee in deze twee versies van hetzelfde gesprek: “ik had laatst een probleem met mijn factuur. Ik moet dan de helpdesk bellen, is vervelend want factuur moet goed zijn” of “ik had laatst een spelfoutje op mijn factuur en ik voel me dan zo vervelend om daarover met de helpdesk te bellen (lacht), maar ja, het moet wel goed zijn”. In het eerste geval lijkt het alsof onze deelnemer erg boos is, in het tweede geval blijkt de persoon juist behulpzaam en blij. Het zijn een paar kleine woordjes die dat verschil maken!

Goede notulen zijn dus belangrijk. Maar helaas moeilijk om aan te komen. Want niet elke notulist kan snel genoeg letterlijk mee typen, en dan krijg je interpretatie-gevaar of re-work omdat je de opnames in moet duiken. Dat moet makkelijker kunnen 💪

“Ik ben denk ik een sportlocatie” — Nederlands is moeilijk voor machines #

We hebben verschillende transcriptie-software geprobeerd (zoals Lookback, Amberscript, en Sonix.ai), maar geen van allen werkt écht goed. Uit pure wanhoop hebben we zelfs geprobeerd gesprekken te luisteren en dan na te spreken, met de hoop dat dat sneller was dan typen (zo schijnen ze live-ondertiteling te doen op TV), maar ook dat werkte verschrikkelijk slecht.

Vooral Lookback droomt soms helemaal nieuwe gesprekken:

Het lijkt Nederlands, maar het is onzin. Hoe vaak gebruik jij het woord Harem in een gesprek? En denk je wel eens bij jezelf “ik ben een sportlocatie”?

Zo heb je méér werk aan het verbeteren van de transcriptie dan je zou hebben aan zélf een gesprek naluisteren en notuleren.

Hoe kan dit nou? We leven in het gouden tijdperk van de Machine-Learning en transcriptie-software werkt nog steeds zo ontzettend slecht? Nou, nee. We gebruiken Dovetail voor onze analyses. De ingebouwde transcriptie werkt perfect voor Engels en is ook voor Duits redelijk goed (jawohl!), maar Nederlands zit er nog niet in en dat is gewoon echt een issue. Het lijkt erop dat onze taal te weinig liefde krijgt uit The Valley.

Overigens biedt Dovetail-concurrent Condens wél Nederlandse transcriptie, maar ook die werkt niet lekker

Dus nu? Alle notulen met de hand blijven doen? Of veel rework in het verbeteren van de transcriptie software? Of is er nog een optie…

Trint to the rescue! #

Trint.com is echt absurd verbeterd in een jaar tijd. Het is nu echt bruikbaar. Naar ons gevoel is Trint (met wat nabewerking) door een soort magische grens gegaan en levert het nu betere notulen op dan live-mee-typen.

Is het dan helemaal perfect? Nee, je zult altijd nog wat nabewerking moeten doen, maar het is op z’n minst 90% correct:

Voor de topografisch-minder-onderlegde mensen onder ons: Zuid-Holland zit niet in Oeganda

Werkwijze met Trint #

Onze werkwijze is nog wel een stappenplan:

interview houden
opname uploaden naar Trint
nabewerken
exporteren tekst uit Trint naar .vtt
uploaden .vtt en video naar Dovetail
MAGIC

En voor je té blij wordt: het nabewerken (stap drie) van de annotatie duurt nog wel 2–3x zo lang als het interview. Je bewerkt dan de fouten uit Trint en voegt extra data toe zoals gebaren die de deelnemer maakt of ge-uitte emoties. Deze nieuwe methode levert voor simpele gesprekken geen tijdwinst, maar je krijgt wel de perfecte notulen!

Zijn je interviews complex of kun je niet snel typen? Dan word de formule compleet anders en heb je (naast dus die perfecte notulen) óók nog een flinke tijdwinst!

Leuke quotes van Lookback. Die transcriptie konden we direct naar de prullenbak verwijzen 🤣

Published first on Medium

Paaspuzzel 2021

Thu, 01 Apr 2021 00:00:00 +0000

Looking for something to do this easter? I’ve got your back!

Work on my easter puzzle, and win a prize! the puzzle is in Dutch!

More information

The Radiant Lumière Dress

Sat, 27 Feb 2021 00:00:00 +0000

A future article written by a pretentious fashion reporter, who’s fed up with his modern world. Part of my series of imagined futures .

Welcome to my Swan Song. After seven years of writing fashion for the Modern Herald, it is over. Many a reason has led to this decision, with the final straw being wearing a hat for two weeks without knowing it. More on that in my ending epitaph, Fashion First!

This year’s collection is again a technophile’s wet dream; it’s hyper futuristic bravado, and I refuse to review it — with one exception. A unique look with a spirited and natural approach. It’s quiet and distinguished, dare I say sophisticated, or (gasp!) Stylish! Finally, clothing made for a James Bond instead of a Q.

My new devotion goes to Amare Abebe, a graduate of the École de la Chambre Syndicale de la Couture Parisienne, whose work kicked off the haute couture week. Meticulous attention to detail from the grand concept to the tiniest of fibers, his work integrates both the real and the unreal. How relieved was I that his work is less transfixed on the bane that is augmented reality. Exceptionally one of the models we could choose from was a person wearing the dress instead of the computer animations we’ve grown accustomed to over the past years. Sure, I may never see an actual cat-walk again, but his virtual shows were at least palatable: a stylish church-hall and a pleasant jungle setting. And I received materials from the collection. Abebe gave fashion its dignity back.

Abebe’s mindful modus vivendi shines through his pieces, cleverly combining the latest innovations from our engineering overlords. The piece de la resistance is his Lumière robe, a radiant dress. And I mean radiant! It uses a material called Quantum mesh, which is modern day magic. The Quantum mesh radiates light which bathes the model’s body in a faint glow that shimmers through the room as if the light leaves a trail of mystique behind it. The robe is partially translucent and (when the wearer chooses) opaque from certain angles. We can both see the body and are equally shaded from details. Clothing so mysterious, one has to see it to believe it.

Abebe’s Lumière dress

It is amazing that this material is real and not a virtual-reality imagination. It is equally amazing that Abebe and his team have succeeded in replicating the exact dress in augmented reality. Wearers of the latest gizmos (Android Glass 6.4, Apple Lens 9.2, or Harmony Glass 7.2) will see the light bouncing off of real objects and a beautiful blooming emanating around the wearer. But beware! As Abebe’s focus lies on accentuating natural beauty the dress cannot be combined with virtual body adjustments. No digital tummy-tucking or breast-enhancements allowed!

Let’s quickly forget the dreaded AR: there is A Real Dress. The mesh dazzles and sparkles. It captures the eyes and hits right in the heart. The cashmere-lined hoodie acts as a veil and both it and the robe demonstrate equal parts intimacy and openness. As if it’s made for an exhibitionist introvert, it screams ‘come closer’ and ‘stay away’ at the same time — always demanding our full attention, one can not look away. A power accentuated by a daringly plain and simple design.

This reviewer has only one problem: the mousquetaires, which needlessly distract from the overall form-factor of the dress. I’d advise wearers of the physical dress to keep them in their cupboards. Strangely, the accessory is not optional in the augmented-reality version: Abebe forces the wearer of the digital Lumière to wear the gloves. That’s the devil’s bargain of digital clothing: you can pick any style you like but you will never own it!

I’m happy to see clothes made from real materials that can actually be touched and more importantly worn. And even though the pricing of these delicate materials will be out of range for most of us for some time to come, it is my explicit hope to see people putting effort into their clothing again instead of slapping on jogging pants and relying purely on augmented-reality clothing. For God’s sake add some ‘zing’ to your life!

A farewell from fashion #

I was excited at first for augmented reality fashion, but as my dear readers know my enthusiasm turned into contempt quicker than sunlight melts the snow. The future turned dystopian for this poor Luddite.

Everybody and their mother are wearing digitally, and it shows. Or should I say “doesn’t show”? It is saddening which blandness befell ‘real clothing’. Sure, augmentation opens unique ways of expression. Yet how quickly I grew tired of the extravaganza (monstrosities!). Imagine being seen with the eyesore that is Swanzay’s dress from the Antoinette-collection that looked like a rave-party that happened in the 17th-century… Or Caux Rocceau’s infamous twelve-foot feathered gaudy Derby hat?

And that, my friends, is exactly what happened to me. My life-partner “gifted me” Rocceau’s Derby hat without me knowing it. He thought it was a nice joke. Sixteen days my head bore this excrescence, for everyone with augmented glasses, lenses, and cameras to see. The emperor was wearing clothes. It made me realize that fashion is not my thing anymore.

For years this column was my special place for spouting my unfettered opinions. Alas, no more! My love for attire is unwavering, too bad there’s none around anymore. I’m looking forward to relying on basic income for a while and rediscovering myself. I’m signing off! Adieu, sayonara, goodbye!

Abebe’s Lumière Robe sells for 500 EMU / 7000 USD / 12uBTC in the Quantum-mesh variant, and 23 EMU / 322 US / 0,55uBTC in the augmented-reality version

Liked this article? Read the previous one: Does Europe’s new generation of never-meaters see the rest of us as murderers?

I’m trying out something new: I’m imagining future ‘news clippings’ from newspapers or other media. The goal is to be neutral: not overly optimistic but also not Black Mirror. Looking forward to your feedback! writing speculative futures

Published first on Medium

Does Europe’s new generation of never-meaters see the rest of us as murderers?

Fri, 29 Jan 2021 00:00:00 +0000

I’m trying out something new: writing speculative futures ! I’m imagining future ‘news clippings’ from newspapers or other media. The goal is to be neutral: not overly optimistic but also not Black Mirror. Looking forward to your feedback!

It’s a beautiful spring-day when I arrive at the Aspdamm-school in Stockholm. The sun has created a warm atmosphere in the playground and the children are laughing and playing together. I notice Jenny Nilsson as she’s settling an argument in the sandpit in the back of the schoolyard. She waves enthusiastically as soon as she sees me and then motions “one moment” as she brings back the peace in the group. Her long blond hair has trouble keeping up with her as she walks briskly across the playground to meet me. “Hello and welcome to Sweden!”, she says with a bright smile, “Let’s go inside and talk, I’m curious what you want to know from me”. And I’m starting to wonder the same, what is this Texas-journalist doing in a school in Stockholm?

Europe has been on a trajectory of increased vegetarianism since the 2000s, and has banned the killing of animals for meat in 2035. I’m here in Sweden to see the cultural impact of this; what do European children (who never ate “real meat”) think of their parents who ate meat? What do they think of other countries, like the United States, where we eat ‘real meat’?

A short history on the "kill-ban"

The European Union set up project Empsychon in 2028 to meet their methane-targets on Global Warming by reducing meat-consumption. The project consisted of three parts: increased awareness, innovation grants and reduced agricultural subsidies. Meat became more expensive ("the real price of meat"), which resulted in protests from the industry and consumers. The media and politicians rebranded Empsychon as "the kill-ban" and painted a nanny state that took away their "right to eat whatever we want". Bringing oil to the fire, the European Union started a patronizing mediacampaign on the health benefits of reducing meat-consumption.
The turn-around came when the innovation grants led to significant inventions that made it possible to combine laboratory grown-meat with 3D-printed natural-food based structures – mainly sea-weed and fungus. This revolution makes lab-meat indistinguishable from animal-grown meat and the market exploded with companies like St. Assisi that produced cheap real-meat alternatives.
Consumer complaints quickly died down when it became possible to buy a perfect T-Bone Steak for less than 2 US dollars (1 Euro).
Seven years later, EC Regulation 392/2035 put the nail in the coffin for animal-grown meat. This real "kill-ban" combined grants and subsidies to the agricultural complex with a near-complete ban on killing animals for food. The only exceptions are religious slaughter and specific local-delicacies, although these exemptions are set to expire in 2045.

I try to keep up with Jenny as she speeds through the hallways and up the stairs. We sit down in the teachers’ lounge and we have a coffee together.

Let’s start with the basics, who are you? #

I’m Jennifer Nilsson, and everybody calls me Jenny. I’m a full-time teacher and I teach children ages five to eight years old. I love my work, seeing the kids grow emotionally and intellectually, and being a part in that. I would work every day of the week if I could! (note: working full-time means three days per week for teachers in Sweden)

You work with a generation that has never eaten ‘real meat’ how does that feel? #

Well, to be honest we never talk about it. And I don’t know if it’s that special. I mean, I have basically been a vegetarian for ten years now. Before we only ate meat two or three days a week and ate plant-based “meat” (she signals quotation marks with her fingers) the other days. And then St. Assisi came to the market and we’ve been eating that. I’m not sure which generation you mean (laughs), I feel like I’m part of that generation also!

Photo of Jenny by Toni Hukkanen on Unsplash

You teach 8-year olds, what do they think of you as someone that ate animals that were killed specifically for you? #

Like I said, we don’t focus on that too much in school, but I remember the situation of a little girl a couple of years ago. She was the daughter of a family of butchers and she held her classroom-presentation about her father and grandfather. She didn’t show any graphic images, but still there were a lot of crying children in the classroom. That poor girl, it took weeks for her friendships to recover.

Sometimes children come up to me to ask if I ate cows and if I slaughtered them myself. That’s not how it was. When you bought meat in the supermarket, you were distant from the slaughterhouse. You buy a sausage and not a pig. Thinking back, it was weird that you could buy animal body parts in the supermarket (Jenny lets out one of her raucous laughs again).

When I grew up, we had a cat and a dog in our house and you’d never imagine eating those, you know? But my father grew up differently, they lived in a village and had chickens, turkeys and even a family-pig. He told me he helped in butchering the chickens, snapping their necks and plucking them. And the pig was even weirder, they gave it a name and they would pet it, it was almost like a dog or cat. Can you imagine that they took care of such an animal and kill and eat it?

I don’t regret eating meat, it was a thing we did. It was normal and the meat on our plates was a separate reality from the pigs and cows we loved in the petting zoo. It doesn’t differ much from the situation now.

Yes, I understand, that makes sense. What you’re saying is there’s not generational difference? #

Every generation grows up with accepting the things around them as “normal”. For us, for you and me, that was eating meat. Meat bought from a supermarket or butcher. For the new generation, it’s not that different. They also eat meat from the supermarket, but now it’s not grown on animals anymore.

And how about other countries where the slaughter of animals is legal? Like the United States where I’m from? #

To the kids in my class it’s strange to kill animals and eat them. They cannot understand why people would do that because they didn’t experience that reality. The world was different for my parents, I’ll never understand their world. I cannot even understand how I used to live (laughs again).When I think of eating lamb meat now or meat from a suckling pig. It’s strange, to eat such young animals. Can you imagine a person killing a lamb?

I don’t see you or any other American as a monster, don’t worry! My kids all say they would be vegetarians if they lived in the 21st century (laughs), the little angels. Truth is they’d eat whatever their parents put in front of them.

Every now and then, a kid says they ate ‘real meat’, when they had a religious event or when they went on holiday to the United States because their parents were craving to eat meat again (again Jenny lets out a big laugh). But you know what (she shakes her head), the kids honestly don’t look happy when they tell me they ate real meat. There is a taboo on eating killed-meat here in Europe. Similar to eating a dog or chimpanzee in the United States I imagine.

How do you see the future for your children? #

I’m an optimist! The ban on killing animals helped in lowering the greenhouse gasses in Europe, I’m happy we can go out in the sun most days of the summer. And the meat-thing, it was not a big sacrifice to make in the end.

I think the future of food will be interesting. St. Assisi is printing food, it’s amazing! We can eat everything now, panda-meat, giraffe-meat, crazy stuff. And I want to know where we go from here, there’s a relationship to how meat was before, the structure and the taste. But why continue to print bones on a chicken wing? Or the ‘rack’ part of spareribs? And what else can roll out of that printer?

I’m excited about the future and I look forward to all the food my kids will eat and maybe even invent!

That’s a positive end to our talk, thank you! #

That wraps up our interview and I symbol a hug to Jenny. As she guides me out of the building the kids are all looking and waving at me, The American. I leave the Aspdamm-school with a mind at ease. During the flight from the USA to Sweden, I had a vague idea that the children might see me as a murderer or creep who eats dead animals, but Jenny’s words have soothed my fears.

A kill-ban may never happen in The United States, or most the world, but it doesn’t matter. There is a new generation in town and it doesn’t kill. Even in my hometown of Austin, the birthplace of steak, supermarkets are filled with ‘fake meat’. It’s cheaper, more consistent of quality and — frankly — better tasting. The war on meat has been won. My 12 year old cousin puts it best: “why kill a cow to eat meat?”

Liked this article? Read the previous one: My daughter’s bionic arms or continue to the next one about a dress.

Does Europe’s new generation of never-meaters see the rest of us as murderers?

My daughter’s bionic arms

Tue, 05 Jan 2021 00:00:00 +0000

This first article appeared in a the magazine Family Matters, in the column “Parental advice wanted”. A father is struggling with his daughter’s wish for bionic arms… What would you answer?

Parental advice wanted #

In this column a parent writes in with their question, and other parents write in their advice. Due to the sensitive nature of the topic (parenting), the letters are all anonymous. Do you have a parenting question? Send it in!

My daughter’s bionic arms #

My daughter is a fanatic rock climber and is playing on a semi-national level. What’s more important to me is that she loves it, it’s her passion since she was eight years old. Her mother and I have always supported her. But now I am facing a difficult choice: support her wishes, or stop her from doing something I disagree with. My daughter turns 18 in six months and she has been looking forward to getting bionic arms to take her climbing to the next level.

She has wanted the arms since the age of 14. I promised her I would be okay with it and that she could do it for her 18th birthday if she saved up for it. I was hoping she would grow out of the idea when she got older, I guess I hoped she would understand the consequences and grow out of it. But instead she enthusiastically started harvesting money and she really wants to do this.

When she’s eighteen she can legally do it by herself, but she wants to do it 5 weeks earlier so she can cut the cake with her new arm. For her it’s a “father-daughter” thing, for me it is terror. I’m exhausted, emotional and depressed, I’ve not slept for weeks.

These are the hands I held to take her to school, the hands she learned piano with, the arms she hugs me with…

I know my fears are irrational in a way. She will have arms and they will feel like regular ones. I’m also not against body modification in general – my right-inner-ear is a bionic replacement due to tinnitus. But just thinking about the procedure where they remove her arms and grow new mechanical ones makes me sick. These are the hands I held to take her to school, the hands she learned piano with, the arms she hugs me with…

My daughter is optimistic and thinks not much would change (part of the procedure is a full 3D-scan of her “current arms”, they’ll look and feel the same). That a whole new world will open to her, where she can install ‘apps’ like guitar-playing and calligraphy on her arms — that she can grip faster and better while rock-climbing making it possible compete in climbing on a world-wide level.

It scares me beyond anything.

What can I do? I made a promise, should I stick to it? Or will she regret getting the bionics and should I keep her from doing it? I’m afraid of losing the good relation I have with my daughter and that she will do the procedure anyway.

Do you have experience in bionics? Or other advice to give to this father of a 17,5 year old girl? Write in! All answers will be forwarded to the family and we will publish the best three together with expert-opinions. All answers will be anonymous.

That was my first “speculative futures” article. Liked what you read? Continue to the next article: Does Europe’s new generation of never-meaters see the rest of us as murderers?

Published first on Medium

Looking back on 2020

Sat, 02 Jan 2021 00:00:00 +0000

This was one crazy year! Working from home, schooling from home, reduced social contact. But also a year of immense joy as I became father again. I think I can safely say I got to know myself and my family better. Let’s see what 2021 brings…

Like 2019 and last year, I’m doing a year-review focused on the stuff I made and did. It again surprised me to see all my work in a write-up, a lot happens in a year! Although my writing productivity went down (let’s blame corona), I did loads of other stuff.

This year I wrote 11 articles, 9.852 words total (down from 20 articles with 12.427 words last year). I had 481 views and 190 reads on Medium this year (down a lot), but the views on this site went up to 4173 views, and people spent 55 seconds per page on average. Although I should add that the main goal of writing for me is to structure my thoughts, not to blast information into other people’s heads.

The main threads I’m working on are research, machine learning and experimentation/prototyping (I dropped Design Systems).

Personal life #

My son was born and my daughter went to school. Two huge mileposts.

Lots of projects in the home: made the baby room, replaced the carpet with laminate on the first floor, painted walls and ceilings, improved balcony and front- and back-garden, oiled the living room floor, painted indoors and got curtains in bedroom. Pfew! 😅

Didn’t read that much, except a great book about sleep (ironically read this in a period of little or no sleep, right after the birth of my son) and a book by the daughter of Steve Jobs (“Small Fry,” by Lisa Brennan-Jobs). I also discovered Blinkist, which is sweet.

I watched cool old movies and TV-series: “Yes (prime) minister” (funny), Citizen Kane (impressive) and Metropolis (interesting).

Angi Studio #

Work at Angi Studio focused solely on research this year. The main project I worked on was setting up the research-heartbeat at publisher Sdu: did loads of research (together with the Sdu-team), created an interview training and made workshops: creativity/ideation, affinity mapping and prototyping. I created KPIs for a new project and made a PowerBI dashboard to display them in.

Writing about design and research #

I wrote three articles for work: [why I use Whereby (or currently Teams) instead of Lookback for interviews](Whereby Or Lookback), how agencies and freelancers talk about the Design Systems of their clients and Three key take-aways from SofaConf.

I helped people with prototyping and launched prototypingmindset.com #

I (soft-)launched Prototyping Mindset (.com): a website where I offer my services to help you create a prototype in Axure. I’m hoping the website will grow out into something bigger, I’ve got plans ;)

Creative stuff of 2020 #

A lot of my creative work centered around machine-learning: I created a fake video with our Prime Minister in it, wrote poems with AI, dreamt up a big machine visualizing AI, and even won a competition with a video I created. My other two big projects were an easter puzzle and an art-work I made with my father.

As a random thing in between I checked which Christmas songs are Corona-proof and I published two random ideas on my ‘other blog’.

Easter puzzle #

I launched a new tradition by making my first-ever Easter puzzle, keep an eye out for this year’s edition! (update: it’s here)The puzzle got about 500 views, and people spent a long time on the page. I’m hoping they had fun in trying to solve it, because I only got one solution submitted (a correct submission). Besides coming up with the puzzles and drawing them, I wrote the HTML and JavaScript myself.

Art-work at my daughter’s school #

This year the big ‘end-of-school’ summer-party was cancelled due to Covid-19. As I’m in the school’s parent-council, I knew the school wanted to invite an artist to help create a big artwork together with the children as an alternative. My father is an artist and he has a lot of experience with this type of work. I asked him, he agreed, and we created the art work together: a shared artwork as a cheerful end to the school year .

AI poems #

Inspired by the great people at Botnik , I decided to create my own poems with code, as a synergy between man and machine. My algorithm tries to generate grammatically correct sentences that rhyme with each other. I then combined these sentence-pairings with others to create poems out of them. Computer Poetry * ‘My tea sounds like a brick’

example poem:

*The Long Girl with the Sad Squash Rackets*

Many answers neatly echo some girls… And the long girl in those helpful teams of quicksands sadly guarantees whorls

A troubled flesh bleakly afforded a picture, But some squash racquets severely frowned a prefecture

An early crayon of the dead bucket in a shiny plant Fervently committed bad vans!

Winning the Rijswijk Talent Award with Machine Learning art #

I participated in the Rijswijk Talent Award to challenge myself. As the subject of the challenge was “home”, I wanted to make a personal work – while at the same time not making it too concrete.

Long story short: I won the competition, and my work was exhibited in a pop-up expo, local shops and in the Rijswijk Museum. Proud 🤩

Machine Learning art: Dreaming of Home

1001 Ideas: keyboards and the Dinner Motivator #

I’ve got a ‘sister blog’ where I publish crazy ideas. This year, just two ideas got published on my crazy ideas blog : Useless keyboards · 1001 Ideas and The Dinner Motivator · 1001 Ideas.

I also updated the visual layout and had someone from Pakistan make the frontend-code, but that’s not live yet.

Mystic Collective (a Machine Learning art work for Strijp) #

The ‘Me Age’ has come to a grinding halt. But what is the future? Personal connection is lacking in a digital age; traditional collectives like the local community, church and charity are fading away. But new collectives are already lurking in the background…

I dreamt up a Galton Board type of machine to visualize how we as individuals are secretly segregated in virtual communities by machine learning algorithms, and to envision what the impact of those practices could be.

I’ll do a write-up of this on my blog one day, but for now you’ll have to make do with the PDF I sent in for the art-competition in STRIJP Eindhoven (sadly, didn’t win).

This year #

Last year I hoped to attain a relaxed attitude, which I managed (spare Corona related stress). I’m hoping to continue that, and to focus more on research and prototyping. See you next year.

Machine Learning art: Dreaming of Home

Wed, 16 Dec 2020 00:00:00 +0000

I created a video-artwork called “Dreaming of Home” that won the Rijswijk Talent Award

To challenge myself, I participated in the Rijswijk Talent Award. As the subject of the challenge was “home”, I wanted to make a personal work – while at the same time not making it to concrete. I found the solution for that problem in machine learning.

Dreaming of Home #

My artwork is a video that shows my life in Rijswijk as a dream-sequence. Like a dream, you cannot clearly make out what you’re seeing: every time you think you recognize something, the image transforms into something else. The work is deeply personal (it’s made from photos of me, my family, my house), and yet general: everyone watching the video can recognize things from their own experience.

The work stays exciting, even if you look at it for a long time. As you can tell, I’m quite proud of it :)

Making of… #

The video was made using artificial intelligence. It felt (to me) as a collaboration between myself as artist and the algorithm (the second time I explore that theme this year).

I selected 250 photos of my life in Rijswijk, photos of my family, myself, our house, the cat, and prepared them for a machine-learning algorithm.

This training-set of photos was purposefully ‘bad’: for a good result, you’d need a lot more photos and the input needs to be ’normalized’. My photo-set by comparison was small and diverse. I knew the machine-learning algorithm would not be able to generate convincing life-like results… Which was the point! In dreams, you also don’t get a clear picture.

After training, I let the algorithm generate an image sequence which was the basis of the movie above. It took me many hours in video-editing to create the definitive look: color grading, adding a halo and adding several layers of blurred video.

Result #

I won the grand prize (500 euro!) out of over 80 submissions. The jury-report said it “transcended the moniker of amateur art”, and that “the video-art shows great talent and creative insight, it’s original, intelligent, attractive and it keeps intriguing”.

The artwork was on display in two pop-up expos and the Rijswijk Museum.

Shared artwork as a cheerful end to the school year

Thu, 03 Sep 2020 00:00:00 +0000

Nearly 450 students of my daughter’s primary school jointly created a work of art this year. This was because their traditional “school ends”-party unfortunately couldn’t take place due to Corona. The artwork was my idea and I executed it together with my father.

We created a joint work of art. A concept I came up with based on my father’s earlier work. It was embraced with great enthusiasm by the parent council. All children made a butterfly, bird or flower. These were used by us artists to create a five meter wide “summer landscape” (as a “Gesamtkunstwerk”)

In this way there is a lasting memory of this special year, and the school year ends in a summery way!

Show fullsize

Three key take-aways from SofaConf

Thu, 03 Sep 2020 00:00:00 +0000

SofaConf was a remote conference by ClearLeft. At Angi Studio we all “went”. In other words, we sat on our own sofas and watched the videos. Here’s my three key-take-aways.

How’s the remote conference experience? #

The first day was very exciting: we all watched the videos together (remotely) and commented on them in real time. The days after that, not so much.

Frankly, it feels more like a Youtube playlist than a conference 🤷‍♂️

Photo by Kelly Sikkema on Unsplash

My three main take-aways: #

Research can be the tool to bring designers closer to management. Doing research on your own organization is a great way of showing the impact of research. Your research-department will be noticed by the whole organization (also the C-suite). By informing marketing, you can give important stakeholders some exposure. Research also works directly within a team (“healing the designer — product-manager divide”, see the talk below)
You need diversity. Don’t hire for ‘culture fit’, but hire for ‘value fit’. It’s good to have a diverse team (different backgrounds, capabilities, personalities, etc.). It brings in different perspectives and makes sure you don’t miss things. But beware: cultures might clash!
Do not ship features, but value. Find your focus, your one goal. Then pursue and communicate that goal. Don’t fall into the “build-trap” of producing features, but go for value. Take Netflix as an example: they built the Roku, but at the very last moment decided to spin it of as a separate product because they realized their value was in the content world instead of hardware. Having their own device would mean competing with devices of other companies instead of showing their content on those devices. Finding out your value happens through research and service design.

Some more insights #

Write short #

The C-level research-report can be as short as four bullet-points in an email.

Diary studies are worth it #

A usability study usually lasts 45 minutes up to 90 minutes. In one such study, Kelly Goto tested how people choose their healthcare insurance. Most participants managed to complete the task in the allotted time. But an interesting thing happened when they did a diary study to complement the usability study: they found out that the process of choosing an insurance normally can last up to a month! You research your options, pick one, doubt during the filling in of the form, ask some friends and colleagues, compare some more and fill in again. Of course that behavior might require a different design or content.

Kelly had some other great stories and tips (keep motivating your participants, always get video data and make sure your time period includes a weekend). I’ve included her talk below.

Every company is a ‘services provider’ #

If designers only look at your product, your (potential) customers can have a bad time. Services you provide can be support, sales, etc. As services are the ‘spaces between things’, it can be hard to convince your organization that “yes, this organization also really provides services”. Again: service design!

Start with stealth projects if you want to implement design thinking, service design or innovation in your company #

Start with small, boring “stealth” projects because you’ll have a safe space. Don’t focus on you customers straight away but start with your own employees: find their painpoints through simple research (like a Slack-channel) and try to improve on that. Loads of people will be impacted and see the value of service design. Remember: not all innovation is disruptive! Check Marc’s talk for more.

“Who led the digital transformation? Your CEO, CTO or Covid19”? #

Remote work is here to stay. Even if only one person is remote, the whole meeting is remote. So innovate your remote processes! And that can be beneficial: people need to show their impact and decisions get documented better. Think about getting rid of your 8 hour day.

Cool ideas #

Turn design thinking inwards #

Haroon Aslam had some cool tips for improving your organization using methods from design. Two cool techniques “competing truth” and visioning workshops.

the play-by-play for the visioning workshop by Haroon Aslam

The break-up letter #

Ask your participants to write a ‘love letter’ or ‘breakup letter’ at the end of a diary study, it will give insight in their emotions.

Define baseline behavior #

You can define the baseline behavior prior to, during or even after the launch of a project. It helps keep everybody in check.

baseline behavior, from the talk by John Cutler

The stakeholder map #

Use post-its with a group to figure out who are all the stakeholders “involved in a project”. It’s probably a lot more than you think.

The 7 talks I really liked #

Outcomes over output by Josh Seiden
Looking Inwards — Things We Learned About Ourselves And Our Research Practice From Our Internal COVID-19 surveys by Leisa Reichelt
Fieldwork from home by Kelly Goto
You’re on the Air, I’m listening: Facilitating Remote Research by Rachel Price . Great exercises for improving your interviewing
20 Tips — How To Embed And Scale Service Design In Organisations — Marc Stickdorn
Clearleft Curates: Farai Madzima: Cultural bias in design(ers) . Beautiful slides and story, with some concrete examples on teams failing due to culture.
How To Lose Less In AB Testing — Simeon Wishlade

I also learned some new function titles: “design historians”, “content designers” and “growth designers”, as if we didn’t have enough confusing titles in design…

Let me know if you have any feedback: matthijs@angistudio.com or find me on LinkedIn ;)

Published previously on my Medium-account: Three key take-aways from SofaConf

Book Summary • Why We Sleep: Unlocking the Power of Sleep and Dreams - by Matthew Walker

Mon, 20 Jul 2020 00:00:00 +0000

So apparently Walker is sloppy with his references, and sometimes out-right wrong. The book is not as definitive a resource as I previously thought. Take a look at this link: https://guzey.com/books/why-we-sleep/

Sleep is important, but many of the _exact_ figures below can be misrepresentations.

Mr Walker should have called his book “Go the F*CK to sleep” because he continuously talks about how we don’t get enough sleep even though there are many, many benefits to sleep. Sadly, that book-title was already taken.

“Why we sleep” is a great read, here are my takeaways. There are a lot, because it’s a great book.

You need 8 hours of sleep, but you don’t get it #

Humans need more than seven hours of sleep each night to maintain cognitive performance. More than a third of individuals in industrialized societies sleep less than five to six hours a night during the week.

After ten days of just seven hours of sleep, the brain is as dysfunctional as it would be after going without sleep for twenty-four hours. Three full nights of recovery sleep (i.e., more nights than a weekend) are insufficient to restore performance back to normal levels after a week of short sleeping. Finally, the human mind cannot accurately sense how sleep-deprived it is when sleep-deprived.

There is a lot of “sleep procrastination” caused by late-evening television and digital entertainment. Even a hint of dim light—8 to 10 lux—has been shown to delay the release of nighttime melatonin in humans.

Sleep == good #

AMAZING BREAKTHROUGH! Scientists have discovered a revolutionary new treatment that makes you live longer. It enhances your memory and makes you more creative. It makes you look more attractive. It keeps you slim and lowers food cravings. It protects you from cancer and dementia. It wards off colds and the flu. It lowers your risk of heart attacks and stroke, not to mention diabetes. You’ll even feel happier, less depressed, and less anxious. Are you interested?

the coolheaded ability to regulate our emotions each day—a key to what we call emotional IQ—depends on getting sufficient REM sleep night after night.
Sleep has proven itself time and again as a memory aid: both before learning, to prepare your brain for initially making new memories, and after learning, to cement those memories and prevent forgetting.
following a night of sleep you regain access to memories that you could not retrieve before sleep.
Practice does not make perfect. It is practice, followed by a night of sleep, that leads to perfection.
Sleeping encourages creativity. This task is accomplished using a bizarre algorithm that is biased toward seeking out the most distant, nonobvious associations, rather like a backward Google search. In ways your waking brain would never attempt, the sleeping brain fuses together disparate sets of knowledge that foster impressive problem-solving abilities.
“sleep on it.” Interestingly, this phrase, or something close to it, exists in most languages (from the French dormir sur un problem, to the Swahili kulala juu ya tatizo)
By improving sleep quantity, quality, and regularity, Harvey and her team have systematically demonstrated the healing abilities of sleep for the minds of numerous psychiatric populations.
Associated with the pulsing rhythm of deep NREM sleep comes a ten- to twentyfold increase in effluent expulsion from the brain. In what can be described as a nighttime power cleanse, the purifying work of the glymphatic system is accomplished by cerebrospinal fluid that bathes the brain.
Those participants who obtained seven to nine hours’ sleep in the week before getting the flu shot generated a powerful antibody reaction, reflecting a robust, healthy immune system. In contrast, those in the sleep-restricted group mustered a paltry response, producing less than 50 percent of the immune reaction their well-slept counterparts were able to mobilize.
a brief dose of short sleep can affect your cancer-fighting immune cells.
Controlled sleep laboratory studies in smaller samples show that children with longer total sleep times develop superior IQ, with brighter children having consistently slept forty to fifty minutes more than those who went on to develop a lower IQ.

Not sleeping is bad, m’kay #

Scientists have studied airplane cabin crews who frequently fly on long-haul routes and have little chance to recover. Two alarming results have emerged. First, parts of their brains—specifically those related to learning and memory—had physically shrunk, suggesting the destruction of brain cells caused by the biological stress of time-zone travel. Second, their short-term memory was significantly impaired.
An infant brain without sleep will be a brain ever underconstructed
Many of the major psychiatric disorders, such as schizophrenia, bipolar disorder, major depression, and ADHD are now considered disorders of abnormal development, since they commonly emerge during childhood and adolescence. Could be due to bad sleep in childhood! Faulty pruning of brain connections in schizophrenia caused by sleep abnormalities is now one of the most active and exciting areas of investigation in psychiatric illness.
Imbalances in synaptic connections are common in autistic individuals: excess amounts of connectivity in some parts of the brain, deficiencies in others. The circadian rhythms of autistic children are also weaker than their non-autistic counterparts. Existing evidence in humans is simply correlational, however. Just because autism and REM-sleep abnormalities go hand in hand does not mean that one causes the other.
Sleep six hours or less and you are shortchanging the brain of a learning restoration The difference between the two groups was not small: a 20 percent learning advantage for those who slept.
After four hours of sleep for six nights, participants’ performance was just as bad as those who had not slept for twenty-four hours straight
Ten days of six hours of sleep a night was all it took to become as impaired in performance as going without sleep for twenty-four hours straight.
When participants were asked about their subjective sense of how impaired they were, they consistently underestimated their degree of performance disability.
With chronic sleep restriction over months or years, an individual will actually acclimate to their impaired performance, lower alertness, and reduced energy levels.
Based on epidemiological studies of average sleep time, millions of individuals unwittingly spend years of their life in a sub-optimal state of psychological and physiological functioning, never maximizing their potential of mind or body due to their blind persistence in sleeping too little.
However, even after three nights of ad lib recovery sleep, performance did not return to that observed at the original baseline assessment when those same individuals had been getting a full eight hours of sleep regularly.
The recycle rate of a human being is around sixteen hours. After sixteen hours of being awake, the brain begins to fail.
vehicle accidents caused by drowsy driving exceed those caused by alcohol and drugs combined
Insufficient sleep does not push the brain into a negative mood state and hold it there. Rather, the under-slept brain swings excessively to both extremes of emotional valence, positive and negative.
Relevant from a prevention standpoint, insufficient sleep during childhood significantly predicts early onset of drug and alcohol use in that same child during their later adolescent years, even when controlling for other high-risk traits, such as anxiety, attention deficits, and parental history of drug use.
A research team in Italy examined bipolar patients during the time when they were in this stable, inter-episode phase. Next, under careful clinical supervision, they sleep-deprived these individuals for one night. Almost immediately, a large proportion of the individuals either spiraled into a manic episode or became seriously depressed
if you don’t sleep the very first night after learning, you lose the chance to consolidate those memories, even if you get lots of “catch-up” sleep thereafter.
a lack of sleep is fast becoming recognized as a key lifestyle factor determining whether or not you will develop Alzheimer’s disease. Getting too little sleep across the adult life span will significantly raise your risk of developing Alzheimer’s disease.
Progressively shorter sleep was associated with a 45 percent increased risk of developing and/or dying from coronary heart disease
Japanese study of over 4,000 male workers. Over a fourteen-year period, those sleeping six hours or less were 400 to 500 percent more likely to suffer one or more cardiac arrests than those sleeping more than six hours.
Chronic sleep deprivation is now recognized as one of the major contributors to the escalation of type 2 diabetes
Despite eating almost 2,000 calories during the buffet lunch, sleep-deprived participants dove into the snack bar. They consumed an additional 330 calories of snack foods after the full meal, compared to when they were getting plenty of sleep each night.
Three-year-olds sleeping just ten and a half hours or less have a 45 percent increased risk of being obese by age seven than those who get twelve hours of sleep a night.
Faces pictured after one night of short sleep were rated as looking more fatigued, less healthy, and significantly less attractive, compared with the appealing image of that same individual after they had slept a full eight hours.
The less sleep an individual was getting in the week before facing the active common cold virus, the more likely it was that they would be infected and catch a cold.
Considering that infectious illnesses, such as the common cold, influenza, and pneumonia, are among the leading causes of death in developed countries, doctors and governments would do well to stress the critical importance of sufficient sleep during the flu season.
Examining healthy young men, Irwin demonstrated that a single night of four hours of sleep—such as going to bed at three a.m. and waking up at seven a.m.—swept away 70 percent of the natural killer cells circulating in the immune system, relative to a full eight-hour night of sleep.
the World Health Organization has officially classified nighttime shift work as a “probable carcinogen.”
based on epidemiological data, any adult sleeping an average of 6.75 hours a night would be predicted to live only into their early sixties
Based on recent surveys and clinical evaluations, we estimate that more than 50 percent of all children with an ADHD diagnosis actually have a sleep disorder
individuals who obtained less sleep in the preceding days are the same people who consistently select less challenging problems. They opt for the easy way out, generating fewer creative solutions in the process.

In short: if you don’t sleep enough, you’ll become a lazy, uncreative, unattractive, fat and forgetful person, who’ll die a lot sooner than those that do sleep.

Mid-day naps are good #

Study with 23,000 Greek adults: the end result was heartbreaking. None of the individuals had a history of coronary heart disease or stroke at the start of the study, indicating the absence of cardiovascular ill health. However, those that abandoned regular siestas went on to suffer a 37 percent increased risk of death from heart disease across the six-year period
in the small enclaves of Greece where siestas still remain intact, such as the island of Ikaria, men are nearly four times as likely to reach the age of ninety
Ideal sleep pattern seems to be: sleep period at night (seven to eight hours of time in bed, achieving about seven hours of sleep), followed by a thirty- to sixty-minute nap in the afternoon
daytime naps that contain sufficient numbers of sleep spindles also offer significant motor skill memory improvement
neither naps nor caffeine can salvage more complex functions of the brain,

Alcohol bad #

Alcohol is one of the most powerful suppressors of REM sleep that we know of
Alcohol is readily absorbed in a mother’s milk.
When babies consume alcohol-laced milk, their sleep is more fragmented, they spend more time awake, and they suffer a 20 to 30 percent suppression of REM sleep soon after
The electrical brainwave state you enter via alcohol is not that of natural sleep; rather, it is akin to a light form of anesthesia. People consuming even moderate amounts of alcohol in the afternoon and/or evening are thus depriving themselves of dream sleep.

Caffeine is everywhere and stays in your body long time #

Caffeine has an average half-life of five to seven hours. So I’m not crazy for not handling coffee after 10 in the morning!
Caffeine is prevalent in coffee, certain teas, and many energy drinks, and also foods such as dark chocolate and ice cream
De-caffeinated does not mean non-caffeinated. One cup of decaf usually contains 15 to 30 percent of the dose of a regular cup of coffee
Some people are very sensitive to caffeine’s effects. One cup of tea or coffee in the morning will last much of the day (that’s me!)
Of concern is that administering caffeine to juvenile rats will also disrupt deep NREM sleep and, as a consequence, delay numerous measures of brain maturation and the development of social activity, independent grooming, and the exploration of the environment—measures of self-motivated learning.

How sleep works #

There’s a lot of information in the book on how sleep works, which is all a great read. In short: when it comes to information processing, think of the wake state principally as reception (experiencing and constantly learning the world around you), NREM sleep as reflection (storing and strengthening those raw ingredients of new facts and skills), and REM sleep as integration (interconnecting these raw ingredients with each other, with all past experiences, and, in doing so, building an ever more accurate model of how the world works, including innovative insights and problem-solving abilities).

The last two hours of sleep are precisely the window that many of us feel it is okay to cut short to get a jump start on the day. But these are very important hours!

Differences children and adults #

As a parent myself, I found this information very important:

the circadian rhythm of a young child runs on an earlier schedule.
Children therefore become sleepy earlier and wake up earlier than their adult parents. Adolescent teenagers, however, have a different circadian rhythm from their young siblings.
During puberty, the timing of the suprachiasmatic nucleus is shifted progressively forward: a change that is common across all adolescents, irrespective of culture or geography.
the sixteen-year-old will usually have no interest in sleeping at nine p.m. Instead, peak wakefulness is usually still in play at that hour. By the time the parents are getting tired, as their circadian rhythms take a downturn and melatonin release instructs sleep—perhaps around ten or eleven p.m., their teenager can still be wide awake. A few more hours must pass before the circadian rhythm of a teenage brain begins to shut down alertness and allow for easy, sound sleep to begin. Parents want their teenager to be awake at a “reasonable” hour of the morning. Teenagers, on the other hand, having only been capable of initiating sleep some hours after their parents, can still be in their trough of the circadian downswing. asking your teenage son or daughter to go to bed and fall asleep at ten p.m. is the circadian equivalent of asking you, their parent, to go to sleep at seven or eight p.m. Asking that same teenager to wake up at seven the next morning and function with intellect, grace, and good mood is the equivalent of asking you, their parent, to do the same at four or five a.m.
teenagers need more sleep than adults, and that they are biologically wired to obtain that sleep at a different time from their parents.

Society greatly disadvantages evening types #

So teenagers are ‘evening types’ and they are greatly disadvantaged: they have to get up early to go to school and are sent to bed when they are not tired. However, they are not the only ones!

About 40 percent of the populace are ‘morning types’ who prefer to wake at or around dawn, are happy to do so, and function optimally at this time of day. Others are “evening types,” and account for approximately 30 percent of the population. They naturally prefer going to bed late and subsequently wake up late the following morning, or even in the afternoon. The remaining 30 percent of people lie somewhere in between morning and evening types.

Evolution of sleep #

sleep was the first state of life on this planet, and it was from sleep that wakefulness emerged
What if sleep is so useful—so physiologically beneficial to every aspect of our being—that the real question is: Why did life ever bother to wake up?
birds and mammals evolved separately. REM sleep may therefore have been birthed twice in the course of evolution

How to sleep better #

To successfully initiate sleep your core temperature needs to decrease by 2 to 3 degrees Fahrenheit, or about 1 degree Celsius. For this reason, you will always find it easier to fall asleep in a room that is too cold than too hot, since a room that is too cold is at least dragging your brain and body in the correct (downward) temperature direction for sleep.

When you get out of the bath, those dilated blood vessels on the surface quickly help radiate out inner heat, and your core body temperature plummets. Consequently, you fall asleep more quickly because your core is colder. Hot baths prior to bed can also induce 10 to 15 percent more deep NREM sleep in healthy adults.

Wake up only once. If you do use an alarm clock, do away with the snooze function, and get in the habit of waking up only once to spare your heart the repeated shock.

Read the book, there’s loads more interesting insights in there!

So apparently Walker is sloppy with his references, and sometimes out-right wrong. The book is not as definitive a resource as I previously thought. Take a look at this link: https://guzey.com/books/why-we-sleep/

Sleep is important, but many of Walkers' _exact_ figures can be misrepresentations.

Whereby or Lookback for remote user testing?

Mon, 13 Jul 2020 00:00:00 +0000

You plan to do remote user-testing or interviews, and you wonder which tooling to use… Here’s my view!

Enter the contestants! #

Lookback is a tool specifically for remote usability-testing. You can use it for lve testing (an interviewer is present during the test) or self testing (people go through a list of tasks in their own time). Lookback records the screen, face and audio of the participants while they do the test, which you can watch back later. Observers can see a live-stream and chat with you as interviewer — your participants won’t see how many observers there are. It’s possible to write notes in Lookback and have timestamps linked to the video. Furthermore, there’s a smartphone app so you can use it to user test your own apps.

Whereby #

Whereby is a tool to facilitatie online meetings. It acts as a virtual meeting-room. It’s not specifically built for remote usability testing.

So. Who’s the winner? #

Looking at the description I gave for both tools, it seems like there’s not contest: Lookback is the clear winner.

However…

Lookback needs a plugin and even then onlyworks on Chrome. There’s no way to get it to work on Firefox or Apple Safari. That is a drawback! We need to tell all our research-participants they have to install Chrome and a plugin. Some of our participants are on locked-down computers and cannot install anything themselves. In comparison: Whereby runs on Firefox, Chrome and Safari — without any installation
Lookback forces participants to have a microphone. I like to call my participants using a regular phone call first, it’s reliable. During the call I setup the video connection. Sometimes it just won’t work and I use the phone for audio and an online tool only for the screenshare. With Lookback this approach is impossible. If a participant doesn’t have a microphone, it’s impossible to set up video or screenshare. I know this might sound like an ‘edge-case’, but for me this happens quite a lot. You’ll be happy to know that Whereby will work without camera- or microphone-access! Lookback Participate on smartphone is even worse: it won’t let you start a call without a headset

In Dutch it says: “Share your screen, choose what you want to share”, I selected “Your entire screen”. So why is the Share-button (bottom right) not active??? It’s a very confusing interaction-pattern in Google Chrome

Lookback screensharing is complicated. In my last usability-test, 9 out of 10 participants could not get through the three ‘easy-steps’ of Lookback without my help. This is not entirely Lookback’s fault: setting up screensharing in Chrome is convoluted, you need to click the screen you want to share even if you only have 1 screen, otherwise the button share screen won’t be activated . Whereby’s screensharing in Chrome is equally difficult to setup, but the saving grace here is that you can do it after setting up the rest of the call. Participants will first experience the happy “it works!” -feeling, before we struggle going through the screenshare-setup together
Lookback is expensive. Lookback costs $50 per month, per researcher. If you want to export notes and video, it costs $100 per month, per researcher. That’s crazy! Whereby is free. If you want to lock the room (a good idea), it costs $9 per month. If you want to record the call, there’s an added $5 per month.

There’s more annoyances (why do my participants need to enter an email-address to participate? Why can’t I chat or send a link to my participants during a call? Why is the flow of calling participants needlessly convoluted?Why the need for the“lookback-floating window” — participants continously lose or close it, etc.), but these are minor problems.

No more Lookback, ever? #

I’m not saying never. Lookback has several saving graces. Selftesting. Easier for observers. Smartphone testing. Making notes in Lookback is very nice, with their automatic timestamping. Streaming is very reliable, even with a lot of observers. If the connection is bad on your end, you can rewatch the stream in good quality. But to me the participant-experience is more important, and that just doesn’t work.

What do you think?

So how do you user test with Whereby anyway? #

To end this article on a positive note, I’ll give a quick rundown on my setup for usability testing and interviews in Whereby:

Recruit participants
by screener or from a predefined list / participant pool
Call participants by phone for planning a date/time
I sometimes email them with a form to pick a date/time, but calling is much nicer: you immediately know if there are any problems. Always confirm through an email!
Trial-run!
A 2 minute call in a Whereby-room, trying out the participants’ connection, microphone and webcam. If necessary, we try out the screensharing as well.
Run the test
I wait 3 minutes to see if participants ‘show up’ in the Whereby-room. If not: I call them using a normal phone. If there are problems with their internet-connection or microphone, we keep the audio-connection through phone and run screenshare and/or video through Whereby. I record video through Quicktime or Lookback — the advantage with the latter is that the team can observe immediately

As always, thanks in advance for your feedback!

Most of my remarks will apply to other tooling as well. Use what’s most comfortable for you ;)

In short:

Zoom, WebEx and Teams: it’s possible to use it without installing an app, but they make it difficult to do. I find that a hurdle for my participants
Skype: you need an account and there are different types of Skype (business and personal) that can’t ‘talk’ to eachother (Skype for business is being phased out for Teams, though)
Google Meet: Google keeps changing the name and functionality of this product (Hangout? Google Talk?). Besides that, the interface is complicated

Whereby or Lookback for remote user testing?

How agencies and freelancers talk about the Design Systems of their clients

Sun, 19 Apr 2020 00:00:00 +0000

At the end of last year the agency I work for did a survey into Design Systems. It’s been a while, but I finally got around to publishing the last part of that research: the agency perspective. Hope you’ll learn something new ;)

We separated the results from agency- and freelance participants from the in-house teams because of the differences in perspective. We kept those results out of our web-report, but you could’ve already found them in our PDF-version.

The stats below are based on 104 survey-participants in The Netherlands, all working freelance or in design-agencies. So let’s go!

Which Design System are we talking about? #

As this chapter is about freelancers or people working for agencies, it’s pertinent to know which Design System we are talking about: a Design System for a client, or one the participants are working on for their own company.

The questions asked in the survey all relate to Design Systems the participants have worked on for their clients. When asked if their design agency has a Design System for themselves, only 32% answers ‘yes’ (29% ‘no’ and 38% ‘kind of’). The numbers are (understandably) much lower for freelance-particants: only 17% reports having their own Design System.

Feels a bit like overkill for a design agency to have their own Design System since we’re not having any own digital products apart from our own agency website. Internally brand related stuff is documented but can’t call that a design system ;)

As a freelancer I cannot profit from my own design system. I always help the client’s design system .

We are working for one design system for the company!

Note: for the analysis, participants who reported they never worked on a Design System are filtered out

Who are the participants? #

There is an almost even split between participants reporting to work at an agency (58%) and those working freelance (42%). Most of the participants report “Design / UX” as their main discipline. Those working at agencies report their agencies to be between 1–10 people (27%) or 10–50 people (44%), only a small group of participants works at larger agencies (50–100: 17%, 100–500: 5%, 500+: 7%).

Interestingly, these participants list themselves on a higher experience level than those working in in-house teams. Only 42% of participants working in-house say they have 2+ years of experience versus 65% of freelance/agency-participants.

Experience with Design Systems #

Most agency and freelance participants have made 2–5 Design Systems (56%), although a large portion also have made just one (24%).

Especially freelance-participants indicate having more experience: 59% of them report to have more than three years of experience with Design Systems. They also report having made more Design Systems than participants working at an agency (33% say they made more than five Design Systems, versus 11% for agency-participants).

Surprisingly, one out of four participants working at agencies report they (or their agency) have never worked on a Design System. The same is true for freelance workers. This really surprised us.

Clients don’t see the need to pay for a design system. They just want to see the product.

We’re just getting started to structure our design (approach) and are using a stylesheet at the moment (first step towards design system). But I still find it difficult to design a whole design system for clients that only want to have one simple website. Its not a product you work on over a long period so the value of making a whole design system is rather small I think

Who and why of a Design System… #

Usually, the agency (or freelancer) initiates the Design System: 80% of participants say the initiative is usually with them.

The reasons for wanting a Design System seem to be similar as those reported by those working in-house:

The main differences are that these participants seem to have less eye for the motivation “UX/UI Consistency” and “Code Reusability”.

Why did you start your last Design System for a client? #

We asked what the reason was the last Design System project was started. Apparently we are very much in-sync here with the global results (from the Sparkbox Survey ):

Difference between Freelance and Agency #

Code reusability is much less a priority for freelance-clients, but increased efficiency in development is. Freelance-clients are more likely to struggle with consistency across (digital) products.

Design Systems in use #

Although only mentioned once by participants in-house, apparently there are a lot of Design Systems maintained by external teams. Half of the freelance/agency-participants of our survey report they maintain the Design System for their client.

We are happy to see that a lot of in-house designers/developers are involved in maintaining the Design System.

it was possible to check multiple checkboxes for this question

Our first set up failed. It was not communicated clearly , developers were not able to join the process because of the deadline for the website. Now we are starting to see that code has to be rewritten in order to work in other places UI/UX has inconsistency because we redesigned some parts. So would like to try again and have the team carry it more together

We see that freelance- and agency-participants usually work on Design Systems that are updated less frequently than those of in-house teams (although the difference is minor).

Usage-statistics, however, tends to be similar to those of in-house teams

Contents of the Design System #

The contents reported by freelance- and agency-participants is amazingly similar to those reported by in-house teams. Interestingly, we see less code being reported as content in the Design System.

To the question: “what’s the most useful in day-to-day use”, these participants mention design source-files very often.

Tooling of the Design System #

Sketch is also the favorite tool of this participant-group, even more than for in- house teams (71% vs 59% in-house). Design System specific tooling, like Storybook and Zeroheight, are mentioned less (eventhough Storybook is 3rd, it’s mentioned by 22% here versus 36% in-house). The same goes for Figma, which is on the 7th place here but on the 3rd place for in-house teams. Abstract is mentioned much more (only 11% of in-house teams mentioned it).

Participants working freelance or at an agency have experience with different tooling than those working in in-house teams

From Design System to production #

One of the end-results of having a Design System is that code flows from the Design System into ‘production code’: the live website or application.

These numbers seem to suggest that in-house teams on the whole have less code from their Design System in production (they fall either in the category nothing or 50–75%, whereas agency/freelance are more in the 25–50 or 75+ category).

However, these numbers are very different for in-house teams with a Design System older than a year. Given the experience agencies and freelance report with Design Systems we would expect their numbers to correspond more with those numbers. One possible conclusion is that participants working freelance or at agencies are working on younger Design Systems.

This is an area where freelance participants are reporting to run behind those working in agencies (although they report to have more experience with Design Systems): freelance participants are more likely to be in the category 10–50% of production code is from the Design System, whereas those working in agencies are more likely to be in the category 50–100%.

Trials and tribulations #

A large part of participants from agencies/freelance positions see the Design Systems they work on as not successful (37% answers ‘no’ to the question).

The main reasons are: adoption and available time (“we cannot sell it”). Those that do see the Design System as successful report the several positive outcomes: a unified way of working (“It brings together development and design as one process”), avoiding redundancy and it helps with on boarding.

The team should advocate working with this. It has been though to get time for it.

Maybe because I’m often in the process of building up the design system in a short period of time and team members who have a lot of other priorities . Also with getting adoption in the organization is hard in this fase. Development and designers still see it as something they have to do, puts on restrictions rather than freedom for creativity, cost time rather than. It needs a lot of time to let design system be the status quo and to show that it will be worth it in the end.

Lest we forget, the experience is positive for most people:

It provides in consistency and efficiency . It’s also easier to get new or temporally designers on board.

Struggles with respect to Design Systems #

We asked participants what the main struggles are with Design Systems, freelancers and agency-people mainly report maintenance (63%) and adoption (50%) difficulties. This is different from in-house participants: although both indicate these as the top problems, in-house participants report less. Furthermore, in-house participants focus more on people/process problems than non-in-house participants.

With ‘maintenance’, the participants usually mean that documentation or design files are not up to date with what is used online. With that in mind, three of the top difficulties for agencies are related to tooling or content whereas most problem for in- house teams relate to people.

Clients don’t know what it is and are not sure what value it brings them.

What would you do differently if you could start over with your Design System? #

Both types of participant (in-house versus freelance/agency) name “involving more people earlier” as the top thing they would do differently, if they could start over with their Design System. What’s different is that the freelance/agency participants name a focus on quality much more (37% versus 25% for in-house) as well as the different technology decisions (27% versus 16% in-house).

involve Angular developers early on . Now they seem like an add-on

Execute Design System from A to Z . More focus on tooling, implementation and governance

The difference in perception between being in-house and being on the outside #

Looking at the results above, we can see there is a difference in perception of problems related to Design Systems between those working in-house and those “on the outside”. In-house teams overall seem to see more problems related to content and tooling, then people or process.

Best tip #

Agencies and freelancers focus on getting the team right and communication as well. Just as with organizations there is the split between “just do it and start small” (18%) and take your time and do research (11%).

Some people offer some words of warning…

Only use it when you need it. Some organizations don’t need it. And if you’re deciding to use it, then make it work.

Not every company needs one! Only for it projects that want to ship, test fast and get customer feedback fast.

… but most people are very positive about Design Systems:

Go and do it! Even a simple component library with text and color styles works wonders for creating more consistent design way faster!

If you’re new to DS’s start from scratch with a DS for a new project / product to hit the ground running and don’t overcomplicate things by reworking a DS into an existing one.

Don’t specify everything, leave room for it to evolve

Execute Design System from A to Z . More focus on tooling, implementation and governance

So get going with your Design System!

Published before on Medium

Doordenker Paaspuzzel

Thu, 02 Apr 2020 00:00:00 +0000

Every Christmas my whole family tries to solve the Dr. Denker puzzle of the Dagblad van het Noorden (a Dutch newspaper). It’s tradition. This year I thought it would be cool to make a similar puzzle of my own. It took me a while, but it’s finished!

Take a look and solve the puzzle! The puzzle is in Dutch ;)

update: The solution.

of course, I also had to make the website for the puzzle, which was a puzzle by itself

Computer Poetry • 'My tea sounds like a brick'

Mon, 20 Jan 2020 00:00:00 +0000

Inspired by the great people at Botnik, I decided to create my own poems with code, as a synergy between man and machine. My algorithm (see below) tries to generate grammatically correct sentences that rhyme with eachother. I then combined these sentence-pairings with others to create poems out of them.

The result is a poem-book, filled with metaphors about the allegorically figurative parables of analogies 🤣

The Long Girl with the Sad Squash Rackets #

Many answers neatly echo some girls… And the long girl in those helpful teams of quicksands sadly guarantees whorls

A troubled flesh bleakly afforded a picture, But some squash racquets severely frowned a prefecture

An early crayon of the dead bucket in a shiny plant Fervently committed bad vans!

The Linen From Polouse #

Fresh men in drugs limply characterised a couple linens… While the man partially couples the linens (note: that's lazy rhyming!)

A loud patch of a breakable crib longingly trembled a hair… If a loud wear on the patches partially hares a hare!

A day in a trick sweetly invents the agreeable polouse!!

Of Whispering Stoves #

The lucky woods in a rifle solidly presumed a pail, But pales knottily unveiled a lucky veil.

A mailbox in a few whispering stoves in some alive thrones knavishly protests the apple - For those in as obediently grappled a chapel.

An ill stove in the noisy maid frantically shone a twig, For a few fake twigs in sweaters angrily qualify a murmuring zig.

The wild hydrant in those robust boys interestingly armed the defeated language And some storages bravely voyaged a sacrilege.

As the pocket wonderfully elects the tent. For an unspent a in the pocket urgently pockets the ten stent.

Jerking the usual honey #

Rats questioningly flooded those shops. While both splendid territorial dominions in the average twig willfully enlarged a couple tops sops.

The winter abnormally concentrated a clean honey But the fluid clean and jerks usually honey a difficult zucchini

A drum of a pencil of light kittens nicely rounded kites, So those flavors swiftly anticipated a couple light sights

The Violent Lettuce with delicious elbows #

A lettuce violently connected a kite. If incomes tomorrow supplemented the tonight there's a deep meaning in here somewhere

A witty tent in delicious elbows urgently faded screeching rain, So the strain on witty rain upliftingly trained a faded strain

A couple gleaming maids frantically cast the basketball, And the carpenters sometimes snowballed a spitball

The helpless toad enormously shook the adorable popcorn, While a ball searchingly shoehorns the kind shoehorn

Some random musings #

The algorithm created more interesting musings, which didn’t fit into the poems:

A gun in hats coolly resisted delightful beetles
The moms abnormally dislike the gentle plantations
Those brothers weakly interfere quiet fogs
Many cute celeries judgementally part the treatments
Those battles on a desk promptly murder a couple damaged oceans

The algorithm #

The algorithm is short. The algorithm consistently finds Jesus. The algorithm is written in javascript and needs Wordnet and Nonsensical.

const Nonsensical = require("nonsensical");
const nonsensical = new Nonsensical();
const Wordnet = require("wordnetjs");
const wn= new Wordnet();
var rhyme = require("rhymes");

global.window = {
	fetch: (path) => {
		return new Promise(function (resolve, reject) {
			require("fs").readFile(path, "utf8", function (err, data) {
				if (err) {
					reject(err);
				} else {
					resolve({
						json: () => JSON.parse(data)
					});
				}
			});
		})
	}
};
const data_file_paths = {
	noun: './data/noun.json',
	adverb: './data/adverb.json',
	adjective: './data/adjective.json',
	verb: './data/verb.json',
};
wn.load(data_file_paths, function () {

nonsensical.load(data_file_paths, function () {
 let s = nonsensical.generateSentence();
 console.log(s)
 let words = s.split(" ");
 let nouns = []
 let verbs = []
 let adjectives = []
 let lastWord = words[words.length - 1].replace(".", "")
 let rw = rhyme.rhymes(lastWord)
 rw.forEach(function(rhymer) { words[words.length] = rhymer.word;})
 // remove last 's' from word, so it is no longer plural...
 words[words.length] = lastWord.substr(0, lastWord.length - 1)

 words.forEach(function(w){

	let w_class = ""
 try {
	w = w.replace(".", "")
	w_class = wn.pos(w)
	if(!w_class)
	{
	// remove last letter, see if it makes a plural into non-plural so wordnet can find it...
	 let w2 = w.substr(0,w.length-1)
	 w_class = wn.pos(w2)
	 console.log(w2)
	}

	w_class.forEach(function(wc){ // it's possible to be verv, noun adjective (or all of these!)

 if(wc.trim() === "Noun")
	{
	nouns[nouns.length]=w
	}
 if(wc.trim() === "Adjective")
	{
	adjectives[adjectives.length]=w
	}
 if(wc.trim() === "Verb")
	{
	verbs[verbs.length]=w
	}
});
 } catch (e) {console.log(e.message)}
 });

 let rhyming = false;
 let ns = "";
 while (rhyming === false)
 {
 ns = nonsensical.generateSentence({wordSuggestions: {nouns: nouns, adjectives: adjectives, verbs: verbs}, useSuggestionRelatedWordChance: 1/2,
		maxSemanticStepsRemovedFromSuggestions: 3});
 let ns_words = ns.split(" ")
 last_word = ns_words[ns_words.length - 1].replace(".", "")

 rw.forEach(function(wrd) {
 // check if the last word is not the same and rhymes
 if(last_word.trim() === wrd["word"].trim() && last_word.trim() != lastWord)
 rhyming = true
	});
 }
 let glue = ["And", "While", "But", "So", "If", "For", "When"]
 var item = glue[Math.floor(Math.random()*glue.length)];
 console.log(item + " " + ns.toLowerCase())
});
});

I hope one day Morgan Freeman will recite the poems

Year review 2019

Fri, 03 Jan 2020 00:00:00 +0000

Like last year, I’m doing a year-review. This year I focused on Design Systems and research. I wrote 12.427 words in 20 articles. These articles had 6.688 views on Medium and 364 views on my personal-blog (the one you’re looking at now). I enjoyed work, bought a new house and did a lot of stuff that I didn’t share yet 🤫

Design Systems #

I specialized myself more into Design Systems. I set up systems at different companies (for work) and wrote three articles on the subject (one, two, three). I went to Patternsday in Brighton and visited TomTom to see how their Design System looks like (and published an article on that).

I created a survey about the State of Design Systems in The Netherlands (for my work - Angi Studio), which resulted in a book (free download)! I also presented the results during Design-System Night (slides and video are still available).

I created a workshop called the “Digital Design Audit” to help companies with creating a backlog for their Design System.

my book 🤩

UI design and User Research #

I worked on UX Research: reading many books and articles on the subject and putting in into practice for some of our clients. I wrote an article on analyzing notes for interviews, as I felt a lot of people focus on “how to do interviews”, while note-taking is also very important for qualitative research.

The nerdiest thing I did is write my own surveying software using Jekyll and Markdown. It’s opensource, but the project is stale. I didn’t even use it myself 😅

I taught myself the basics of visual/UI design. I did this through courses and by redesigning 1001ideas (my side-project) about 30 times. The final redesign is not live, so you’ll still be looking at the “old” thing.

I brought UI design and research together in setting up a new visual-design test.

1001ideas #

Published eight ideas this year on my crazy ideas blog:

AI and design #

Last year I focused on the theme “AI and Design” and I’m happy to see that topic is now taking off. I went to a conference “explainable AI”, and wrote three articles on using movies and stories as ‘prototyping AI’ (and seeing the impact): one, two, three.

Although I still like the subject, I’ve not gone very much ahead with it anymore - as it’s not coming up in work.

New year, new stuff #

I bought a new house! So lots of painting, fixing things, buying things… Also, a new camera!

Blog experimenting and prototyping #

I have been ‘nerding’ by writing my own ‘create a maze’-puzzle and solved it. I bitched about Apple Photos (that’s the article that was most-read on my site). I added some things to “Matthijs Recommends”.

I helped several people with random Axure-questions, and I loved doing that, just email me if you have questions!

I discovered Glide and created some apps in minutes. Love it.

Looking back… #

Last year I hoped to continue doing photography, experimenting and prototyping. I managed that 😇 I also discovered a new hobby: dancing with my daughter.

I also tried to have a more relaxed attitude, which helped with the above!

What’s up for 2020 #

2020 is a busy year: I’m expecting a 2nd kid and that’s why I’m expecting not a lot of other things 😇

I hope to at least publish some stuff that’s nearly finished 😅. But most importantly, for 2020 I hope stay in the same happy state I’m in now.

Surviving your 1st year with a design system

Fri, 18 Oct 2019 00:00:00 +0000

I held a survey for my employer Angi Studio where we asked our participants everything about their Design Systems. Apparently setting up a Design System is not always easy. We asked how was setting up your Design System and the answers included: “a bumpy ride”, “a long process”, “chaotic”, “an adventure”.

We saw that the struggles people experience are mostly the same (see page 39 in our report for the details). But we also saw there is light at the end of the tunnel!

The goal of this article: giving you a boost so you can survive your first year of having a Design System.

Setting up a Design System can be quite the adventure! (source: Annie Spratt)

Keep working on your Design System for at least a year: Design Systems are an investment

Why 1 year? #

One year seems to be a turning point: things are very different for organizations who had a Design System for over a year versus those having a Design System younger than a year.

It is also very relevant in 2019: a lot of organizations are in their first year of setting up a Design System. This holds especially for small organizations (less than 500 employees): only one in five reports they have a Design System for over a year (for large organizations this goes up to seven out of ten).

The benefits… #

Setting up a Design System is hard work (as an agency, we’ve done it multiple times for our clients), and at some point you’ll think “why am I doing this?” You’ll doubt if it’s worth the investment. Here are some statistics to keep you going!

Hang in there! If you last 1+ year you will reap the benefits from your Design System (source: Hu Chen)

After 1 year more people say their Design System is successful #

Of the participants working in-house, those that have had a Design System for more than a year almost 9 out of 10 report their Design Systems are successful. This is a huge improvement over those whose Design System is younger than a year: only 6 out of 10 report their Design System is successful.

Of those who call their Design System successful, the most mentioned benefits are consistency and speed of development.

After 1 year there is more in your Design System #

There’s more of everything in your Design System after a year. Especially components, UI patterns, Grid Systems and Design Principles. But also Design-System specific documentation: process, coding standards, ‘about our Design System’ and on-boarding.

After 1 year your Design System is more mature #

An open door perhaps, but we were very surprised by the huge difference a year can make. Most Design Systems reach stage 3 after a year and a lot are already in Stage 4 or 5!

The Design System maturity index was created by Marcelo Somers and John Gully. It’s a nice metric to see how your Design System is doing.

In the first year, Design Systems are usually in Stage 2 or Stage 3 (a PDF or wiki, or a UI-kit with code-examples). However, all Design Systems older than a year are in Stage 3 or beyond. As a matter of fact, in 23% of the cases these Design Systems are already in the final stage (stage 5), where the Design Systems is governed and part of the organization.

After 1 year your Daily Usage goes up #

Only 44% of participants with a Design System aged less than a year say they use it daily. This percentage goes up to 71% for participants whose Design System is older than a year.

This difference in daily-usage is mainly because the weekly usage turns into daily usage

After 1 year more Design System-code is used in production #

One of the most difficult things to get right is making sure your live -site (or application) uses code from the Design System. We see this is especially the case for those organizations whose Design System is less than a year old.

Only 3 out of 10 can say that half of their production-code is coming from their Design System when their Design System is younger than a year. This grows to almost 7 out of 10 participants whose Design System is more than a year!

Even more impressively, almost a third of participants whose Design System age is 1+ year can say 75% of their production code is from their Design System!

Wow, over 75% of your production code is sourced from the Design System? That’s impressive!

There’s always work to do #

So that’s the positive news 🤣 But a Design System is never finished. This also means you’ll keep running into problems with your Design System process, even after a year.

Participants with a Design System less than a year do have different struggles. After one year, the amount of people having problems with adoption drops from half to almost a third. Another big drop is in funding: from 16% of participants with a Design System less than a year to 6% for those who have a Design System longer than a year.

Problems with adoption drop if a Design System is older than a year

Interestingly, both maintenance and staffing go up after a year (37% to 45% and 33% to 48%, respectively). This makes sense. More content in your Design System means more maintenance, which in turn means you need more people.

Even if your Design System is older than a year, you’ll experience problems. But they are different problems

source: Max van den Oetelaar on unsplash

What happens when your Design System is a success? #

Are you doubting whether you want to go down the road of setting up a Design System? We asked our participants if they thought their Design System was a success and why.

Based on quotes from our survey-participants, we can say it’s worth the investment:

“We have a lot of products […] in different countries. [The] Design System is a big step for us to start to keep consistency across all the products and make UX Design transparent across UX teams. There is still a lot of work to do including internal training […]”

“Already saves time and we have more and good discussions about design solutions and ux with all disciplines involved.”

“People don’t ignore it, we’ve got covering from management, people are happy with it. And it’s used by our central platform which covers more than 35 countries ”

“As the adoption rate increases, more developers are inclined to use it from the start.[Developers] are happy that they no longer have to do the tedious work anymore. ”

“ I use it everyday to build my designs while efficiently maintain branding, consistency and accessibility. ”

So good luck with your journey of setting up a Design System! If you have any questions, just contact me (or pm me in the Design Systems slack ), or read the report!

Previously published on Medium: Surviving your 1st year with a design system

The next step: Design Systems beyond UI 🚀

Thu, 17 Oct 2019 00:00:00 +0000

In this article I list some practical ways to enhance Design Systems. A word of advice, though: do not fall into the trap of hoarding stuff!

Most Design Systems are approached from a visual UI perspective: a UI-kit for designers and coded components for developers. Added to that are design principles, documentation, typography and color-tokens. Some Design Systems include tone of voice, animation or a brand-guide as well.

But designers do other things too. Things like (user-)research, facilitating workshops, wire framing, and UML-diagramming (apparently 😱).

So let’s add it all to the Design System, right?

Most Design Systems consist of a UI-kit, Component-library and some documentation on tone-of-voice, animation or a brand guide. But is there more to ＃DesignSystemLife?

Word of caution: do not add stuff for the sake of adding stuff #

A bigger Design System does not mean better results. Larger systems are in fact more difficult to use and get outdated quicker. They are also intimidating to new users and people are less likely to use them.

Spending time on your Design System (components, flows, graphs, text, whatever) means you spend less time on your product. You actually “lose time” twice: once when you add stuff to the Design System and later when you maintain that stuff. You should only spend time adding something to the Design System if it outweighs the investment.

So focus on practical things specific to your situation. A Design System is not a book. Minimize the amount of prescription (“you should do this”) and maximize the amount of practical things (tips, plugins, templates, components).

If your Design System looks like this, you’re doing it wrong. Source

Not sure what you want to have in your Design System? Try our simple checklist !

So practically, what can we add to our Design Systems, Matthijs? #

Yeah sorry, the text above was really abstract. You want examples, practical suggestions and inspiration!

I’ll add more ideas in follow up articles, but here are some suggestions for your Design System:

1. Standards for doing research #

You can help your UX-research colleagues by giving them useful tools and standardizing things in your Design System:

screener questions for usability tests or interview
texts for recruitment, e.g. inviting people to participate on social media: “participate now in our awesome survey and get a $10 coupon!”
emails for communicating with participants (“Thanks for signing up! You’ll need a computer with a microphone”)
downloadable templates for the consent-form and NDA’s
reward-levels for different types of research and what your companies’ rules are regarding rewards (e.g. can you give fellow employees a coupon for participating?)

2. Documenting user research findings #

Research results can be useful multiple times during product development and a Design System could be the right place to store these results. I’m thinking of fundamental (or “timeless”) research, such as Customer Profiles and user-contexts.

Bulb is a great example: research insights are sprinkled all throughout their Design System. This gives credibility to their guidelines. A practice which is incredibly useful and something I’d like to see in more Design Systems.

That’s one way to turn long-and-boring researchreports into valuable bits of insight within your Design System! Thanks Bulb!

Caveat: results from user research can become irrelevant very quickly. Do not dump all your reports into your Design System

3. Non-UI stuff #

The list of non-UI stuff we could add to Design Systems is near-limitless. But these are some practical examples I thought of:

4. Workshops #

At Angi Studio we have a library of tools and workshops (some of which we share with the rest of the world ). These workshops are a core part of our design activities. We use them every day and we keep improving them. They are an excellent candidate for our Design System!

Lots of empty space in Design System land: my fingers are itching already, what can we build there? Source

I’ve added a couple of examples of how to expand the use of our Design Systems, but I have more ideas to add in this topic: information architecture, Design Ops, more processes, artificial intelligence, ethics… So keep an eye out for future articles!

I’m also really looking forward to your reactions, so email me or contact me on LinkedIn.

So, I don’t think we should add all of this to our Design System, but more is possible!

Don’t make your Design Systems into intranet v2. Keep your Design System lean and small! Happy creating!

Previously published on Medium: The next step: Design Systems beyond UI 🚀

We have a Design System: let’s fire our designers!

Mon, 14 Oct 2019 00:00:00 +0000

Most design systems focus on in-page interactions: visual elements, motion, accessibility, tone-of-voice, etcetera. They keep your websites neat and consistent in wording, style and components. But that doesn’t automatically mean you’ll have a good product.

Good elements being used to create bad things. Source: interaction-design.org

Good visuals, bad experience #

Imagine two sites created with the same design system, but having completely different experiences for the people visiting them. The one site feels familiar and friendly, the other cold and distant. One site answers the questions you have right when you need that information, the other just boasts about all the cool stuff the company ever did — leaving you to hunt around multiple pages finding simple answers. And even though both sites use the same buttons, icons and colors as the app and the email-newsletter, only one of them feels the same as that other stuff — the second site feels like completely different people made it.

Imagine you’d have to choose which one to use? For which product would you happily pay to use?

People don’t care about your components #

People using your product or visiting your website are not looking at components, they are looking at the entire page. More than that: they look at multiple pages on your website and at different pages at different times…

And it’s not just your website! Your website visitors will hop between support-documents, live-chat, marketing-pages, your app and many other ‘touch-points’.

This happens in real life #

It’s easy to point to bad sites and apps that use “good elements”: many websites use standard component-libraries like Google Material or Bootstrap and still completely screwed up on basic usability and content.

Another example is Google Maps. In just a year time they created a completely different experience using exactly the same elements. It wasn’t bad (not by far!), it could just be better:

Google Maps switched the focus of their maps. In 2016 the main focus was roads and it became places in 2017. Same visual elements, completely different experience. Source: the spectacularly amazing Justin O’Beirne

Having a Design System is great. But you still have to do the heavy work. #

I’ve heard people say: “now that we have a Design System we can skip the design fase and our developers can just implement new features directly”. I disagree :)

Design is about creating a great experience for people. To do that, you’ll need to align with development, sales, marketing, management and others, and you need to do a lot of (user-)research.

Having a Design System saves time so you can do all that. It also helps you create better and more consistent end-results. Design Systems are great, just not the holy grail or a panacea.

With Design Systems being more common place, websites will be more consistent and beautiful. This means you can’t rely purely on having great visuals, you’ll have to raise the stakes and focus on how your components are used.

What is your website trying to say?

Published before on Medium

Book Summary • StoryBrand

Sat, 28 Sep 2019 00:00:00 +0000

A brain dump on what I found interesting from the StoryBrand book. These notes are for future reference for myself, but might be useful for others also.

In short #

The book could be about 20% smaller and had a bit too much “StoryBrand is so amazing” about it. But the main idea is very useful for creating presentations, marketing pages, blog-posts, brand messages, propositions and writing project proposals. You can use it either for your entire company, your product or for personal branding.

I can honestly say I’ve benefited a lot from reading it.

The main message:

Customers want to know where you can take them

You achieve that by writing your content to contain these seven steps:

A Character [usually the hero/your customer]
With a problem/pain/need
Meets a Guide [the brand]
Who gives the hero a plan [precise steps to achieve a goal]
That calls them to action
That results in Tragedy [failure]
or Comedy [success]

So, what’s the StoryBrand method? #

The author proposes to use elements from storytelling to convince potential customers to explain who you are and to buy your products. The goal is not to learn how to write stories, but to use methods from storytelling like ‘story-gaps’, ‘villains’ and ‘heroes’.

The story-brand method means creating a simple and memorable story for your company or product and use this every time you communicate (in marketing material, sales pitches, presentations, etc.).

Keep it simple, direct and consistent: “confuse you lose”

The story gap #

The idea is that you place a gap between a character (the customer) and what they want: “you identify a potential desire for your customer, and that opens what’s called a story gap”.

You have to word this desire in relation to the customer. It helps to talk about their problems. The more we talk about the problems our customers experience, the more interest they will have in our brand. Personify the problems their customers face.

“The most frequent reason for unsuccessful advertising is advertisers who are so full of their own accomplishments (the world’s best seed!) that they forget to tell us why we should buy (the world’s best lawn!).”

quote from another book, Made to Stick, which I didn’t read yet. Seemed to illustrate this concept really well though

Your customer is the hero. (You’re the guide) #

The guide, not the hero, is the one with the most authority. Still, the story is rarely about the guide. The guide simply plays a role. The story is about the hero.

The only two motivations a hero has in a story are to escape something bad or experience something good:

Win some sort of power or position
Be unified with somebody or something that makes them whole
Experience some kind of self-realization that also makes them whole

At the beginning of a story, the hero is usually flawed, filled with doubt, and ill-equipped for the task set before them. Your brand is helping people become better versions of themselves. It’s not easy being the hero. If there are no stakes, there is no story.

In Star Wars, Luke Skywalker was told by his uncle that he was too young to join the resistance, so he doubted his ability until the very end.

which translates to:

Steve Jobs understood that people felt intimidated (internal problem) by computers

About your villain #

The villain should be a root source. Frustration, for example, is not a villain; frustration is what a villain makes us feel. High taxes, rather, are a good example of a villain.

The villain should be singular. One villain is enough. A story with too many villains falls apart for lack of clarity (confuse you lose).

The guide (you!) #

You are the guide. You should show empathy and authority, but the story is not about you. The story must always be focused on the hero, and if a storyteller (or business leader) forgets this, the audience will get confused about who the story is really about.

You see it in a lot of marketing pages, the company boasting about how good they are (they are “we-ing” all over their customers).

The guide has a plan #

Instead of talking about yourself, you help the hero (your customer) with a customized plan. Show them they can overcome their obstacles, defeat the villain and become a better person (with your plan).

Communication #

Consider how your customers want their friends to talk about them
After the climactic scene, the guide often comes back to affirm the transformation of the hero. Think about showing how your brand helped people and how it transformed them
The audience needs to be told very clearly how far the hero has come. Think: who you want your customer to become?
Create a logline for your company (it’s a movie’s one-sentence description, Speed is “Die-Hard on a bus”). Memorize it and use it everywhere

Always provide a CTA #

Help your customers identify what is the best next step. Tell them where to click to buy your product. Don’t hide that button.

Make sure to call today. The call will only take a couple of minutes, and you’ll be in our system forever.

Some stuff about E-mail #

The author proposes using email is a good way to communicate with your customers.

A good way to craft each nurturing e-mail is to use an effective formula that offers simple, helpful advice to a customer (”I’ve been using this formula for years and customers love it”).

Talk about a problem
Explain a plan to solve the problem
Describe how life can look for the reader once the problem is solved

I also recommend including a postscript, or the P.S. Often, the P.S. is the only thing somebody who opens a mass e-mail will actually read.

That’s really it. If you cover these three areas as efficiently as possible, you’ll be crafting e-mails your customers open, read, and remember.

Conclusion #

The book is a good read. But, you can also just jump ahead to StoryBrand and create a brandscript right now for free ;)

I recommend the book and method, but suggest you skip the first chapter because the author is full of himself. Also, the author uses the same trick of the story gap too often in his writing trying to leave suspense in each chapter. Annoying, but not distracting.

Note: I did not write this blogpost using the method 😇

graph above not under the Creative Commons license

Opinions are worthless: how we test a new visual design

Wed, 07 Aug 2019 00:00:00 +0000

You explore several visual styles and now you’re left with a selection of UI-mockups… Which one is best? How do you make sure you’re not relying on ‘opinions’? We solved this question with a structured test-approach. 👨‍🚀

One of our clients is giving a major overhaul to their flagship product. Design, interaction, code: everything will be modernized to fit with the 21st century. We’re incredibly happy to help our innovative client along this journey.

The flagship product we’re talking about is the biggest knowledge bank used by the majority of courts, judges and lawyers in The Netherlands. The existing version had a decent UI, but no good fit with the brand. We decided to overhaul the UI-design before kicking off the design system.

Phase 1: brand research and iterative design #

Working closely with the product owner we researched the brand: their representation and reputation. We looked at brand values, style guides, other products, marketing pages, etcetera — and distilled the target brand traits. Based on this research, we iterated thru several UI-designs, color palettes and typography.

We settled on four distinct UI-designs and five possible forms of typography (i.e.: font, weights, line-height, font-size, color).

Phase 2: setting up the test #

You can trust Maurice to deliver on his visual-design skills: all four UI-designs are distinctly different and all of them are good. The designs support the most important user actions, the screens look pretty, text is readable, colors are accessibility-proof and the design fits with the brand.

So how do you evaluate which one is ‘the best’?

We wanted to prevent getting people’s opinions…

Just asking people which they prefer will be entirely based on personal preference and those opinions are worthless: every person is different, and what you like one moment you might not like the next (if we ask you right after your morning coffee we’ll get a very different answer than after a whole day of exhausting meetings).

With the help of the excellent resources of the Nielsen Norman Group we came up with the following structure for our 30-minute test-sessions:

5 second tests with unstructured association
Readability testing
Structured association and rating
Open interview We created a simple Keynote-slide deck with the four parts. In each part we randomized the slides to eliminate any order-effect (the first thing you see influences whatever you see afterwards, and you’ll always remember the first and last thing best). We show every participant all designs because of the variability of responses. As a baseline, we also included the current visual style.

Participants sit in front of a laptop, with a note-taker on the left and the experiment-leader on the right.

Want to use our slide-deck for your own UI-research? Download it here!

Phase 3: let’s test! #

We invited a group of five participants and had them go through each part of the test:

The five second test #

We use the five-second tests because we want to know the first-impressions people have, their gut-feeling.

Each design is shown for 5 seconds after which we ask four simple questions (“is it trustworthy? Why?”, “What do you remember?”, “Which words do you associate?”, “Which rating between 1–10 do you give?”).

To help our participants we also show these questions on screen after each design. As this test is a bit counter-intuitive we start of with an example: a screenshot of a well-known shopping site.

Tip: we ask the participants for ‘free word association’, which is difficult because they don’t do that every day. I help them by giving an example during the first screenshot (“I associate it with a candy store, because of all the extreme colors”). This shows them no answer is wrong.

Readability testing #

As our platform is used for reading very technical documentation choosing the correct typography is incredibly important.

We picked paragraphs of existing legal text from the platform and created five slides with each three paragraphs in a different typography. We made sure each text has a comparable difficulty: each text includes complex legal terms and has references to legal documents (long numbers).

We ask our participants to read out these texts aloud and give a gut-feeling on readability (a scale 1–5). More importantly, we observe their behavior: are they leaning forward, are they straining their eyes, are they stuttering while reading or making mistakes.

Structured association and open interview #

For the last part we give each participant a paper with a list of sixteen words. They are asked to select three to five words that describe the design. The list contains words related to the brand traits and contains positive and negative wordings (you need at least 40% of the words to be negative!), e.g.: ‘boring’, ‘formal’, ‘old-fashioned’, ‘trustworthy’.

They are then asked to rate the best and the worst design. For this they are allowed to use the keyboard and go through the designs.

Based on these answers we go through an open interview, asking about the motivations behind the chosen words and why they liked certain designs.

Phase 4: summarize, iterate, test again #

Testing’s done! Now we analyze and summarize our results (I wrote about analyzing results before). Here’s a tip: we write our findings immediately in a Numbers document (Apple’s equivalent of Office Excel) which allows for easy analysis. With Numbers we can categorize over the different designs and immediately see trends.

Based on the results we iterate on our designs and test again. By testing with two groups of five people we remove personal bias and can give an optimal direction for the future UI of the application.

Don’t forget, you can download our slide-deck for your own UI-research.

That’s a wrap on this article!

Published previously on Medium: Opinions are worthless: how we test a new visual design

Apple Photos Shared album is broken by design 😔

Mon, 08 Jul 2019 00:00:00 +0000

I’m using Apple Photos to store my photo library. It’s not perfect, but for my purposes it’s the only fit. I’m satisfied, if not for some small problems I keep running into. My latest gripe is that sharing photos is done in low quality, and as you know, I’m a sucker for image quality.

Shared Albums: nice idea, shitty implementation 💩 #

There are multiple ways of sharing photos with others with Apple Photos, but we’re going to talk about one in specific: the ’Shared Album’. Just so we’re clear, here’s a short list of sharing options*:

export and use Dropbox - maintains the original quality
AirDrop - maintains original quality
email - maintains original quality if you select ‘original quality’ in Apple Mail
WhatsApp - lower quality because of WhatsApp
select photos and use ‘share as link’ - uses original image quality
create a ‘shared album’ where people can also add photos themselves - lower quality

This last option is the one we’re talking about today. It’s a great idea: create an album for work, friends, for all your separate holidays, a wedding - and have everybody add all their photos there. Sadly Apple decided to reduce the quality of photos in shared albums by design:

When shared, photos taken with standard point-and-shoot cameras, SLR cameras, or iOS devices have up to 2048 pixels on the long edge. Panoramic photos can be up to 5400 pixels wide. You can share GIFs that are 100MB or smaller. source

Also, sometimes the naming is changed of photos, making it impossible to find back photos in your original library.

And last, but least: if you accidentally click save to library in a shared album on a photo you made yourself, you end up with doubles in your library. Good luck figuring out which one is the ‘high-res’ version on your phone!

I’m not the only one complaining: even if you’re ok with having slightly lower quality, there are all sorts of inconsistencies with how it handles the metadata. Almost every time, some photos get messed up. Sometimes the time gets replaced with the time it was added to the album, rather than when it was actually taken. Sometimes this happens to only the videos. Other times the GPS data gets stripped. It’s really frustrating and doesn’t feel very “Apple”

Example… #

Below is a crop of the same photo, in three different formats. Left is the photo in the library, middle is a direct export to from the photo library and right is the same photo as exported from a shared album.

As you can see (or maybe you’ll need to open the picture in a separate tab), the right-hand photo is degraded in quality, which is stupid, because the photo is actually larger in file-size than the middle version!

Description	Type	Size (MB)	Size (px)	extension
Left - Original in Apple Photos library	`HEIC`	1,4 MB	3024x4032	none
Middle - Export directly	`JPEG`	2,2 MB	3023x4032	.jpeg
Right - Export from shared album	`JPEG`	2,8 MB	1536x2048	.jpg

Some other observations:

all photos are 72 pixels/inch and have color profile sRGB IEC61966-2.1
apparently Apple is not consistent when exporting, sometimes preferring .jpg and at other times .jpeg
Apple’s HEIC-codec is pretty impressive
exporting to JPEG seems to add some additional sharpening - see the middle photo. Or is the JPEG-algorithm to blame?

Alternatives? #

If there’s anybody that knows a good alternative to Apple Photos, I’m ready to listen.

I just need the service to meet a simple list of requirements:

sync photos to my own hard-drive from multiple devices for backup purposes
make it easy to edit photos (also with external tools) and sync those edits across devices
have face and location tagging, albums and ‘hearts’ or stars
make it easy to share photos with others

Google in the mean-time is making it harder for me to switch to them, because it’s no longer possible to sync photos to your own computer (Google is ending its confusing integration between Google Photos and Drive - The Verge). It’s also still not possible to edit photos in external apps without downloading and reuploading.

Adobe CreativeCloud is moving away from the 12 euro p/m plan to only have a 24 euro p/m plan (typical Adobe money-suckers). And there are several limitations to Lightroom CC (like no virtual copies, no external editors and plug-ins (other than Photoshop), no smart albums or search tools).

Amazon and Microsoft are both miles behind and not really an option.

* Sharing photos from iCloud is such a mess someone actually made an infographic: Guide to Sharing with Apple Photos

Get the best results out of your interviews with a four-step analysis

Tue, 02 Jul 2019 00:00:00 +0000

Getting valuable insights from your qualitative research is harder than it sounds. If you thought interviews and user tests are easier than ‘hard numerical science’, you’re wrong!

This is not an interview guide. I’m assuming you already did the hard work: you set your research goals, confirmed them with your stakeholders, wrote the interview plan, recruited the right participants and did the interviews / user tests. This is about what happens after.

You’re staring at your notes wondering how the @€£$# are you going to extract meaningful conclusions out of this mess. It’s easy to fall into research traps during your analysis: confirmation bias (seeing what you expected to see), remembering only the first and last participant or forgetting the context of remarks.

Here are the five steps I go through to get usable conclusions:

Collect useful notes
Determine pertinence
Cluster and tag
Draw conclusions The nice thing is: you can do all of this using free tooling. I suggest Microsoft Word, but Google Docs, Bear and Pages are personal favorites as well.

1. Collecting useful notes #

In machine learning there’s the saying: “garbage in, garbage out”. It means if your starting data (your notes in our case) isn’t good, no AI-magic in the world will deliver you useful outcomes. The same goes for analyzing data by hand.

Writing good notes is hard. It doesn’t come automatically for most people. But preparation helps and you will get better at it over time. I always try to write my interview questions together with the person who’ll take notes and I do the analysis of the conversations with them as well. The best way to learn how to write useful notes is to do the analysis of a conversation.

I talk about a separate note-taker, because I’m a big fan of writing notes during the user test or interview instead of transcribing a recording afterwards. Transcribing video is a lot more work and you’ll miss important context cues. That’s why there are always three people at an interview: the person being interviewed, the experiment leader and the scribe.

Some tips for taking notes: #

Do not write down everything literally; no one can type that fast. Just write the gist of what was said (and what the participants did) and only write down extremely useful quotes word-by-word
Record context: write down the answers and the interview questions. Also record actions: did the interviewer help the participant, did the participant pause or hesitate, what happened on screen (in the case of a usability test), is the participant uncomfortable or downright aggressive?
Write down timeframes. Afterwards you can easily go back to a specific point in time in your recording and get the exact quote
Add the structure of the conversation: which part of interview where you at when something was said? Use headings in your notes!
Ask questions as note-taker — “what exactly did you mean with that?”. The most useful questions usually come from the scribe!
Reflect shortly after each interview: discuss an overall feeling with the experiment leader and write that down

2. Determine Pertinence #

Step 2 in our process is determining pertinence. Not everything in our notes is gold. We want to mark the relevant bits, the things that are pertinent to our situation.

Paste the notes of all participants underneath each other in one file, and use a big header to separate one participant from another. Highlight the (parts of) paragraphs that seem useful, and be gracious: better select too much than too little. I prefer Microsoft Word for this because as soon as you select text the mini-toolbar appears and you can easily highlight your selection.

Highlighting text in the Word mini-bar (sadly, no drinks in the mini-bar)

I use this color scheme:

green: positive comment / experience
red: hard failure
orange: negative
blue: informative but neutral
purple: everything else interesting (wishes, random facts, etc) Usually about 30–50% of the text gets colored (tip: experienced note takers can get a lot of pertinence done in ten minutes after an interview). This is also a good moment to remove any privacy-sensitive data you don’t need from your notes.

Now on to the next step!

3. Cluster and tag #

Warning: this step involves manual labor

Copy (not cut!) all pertinent bits to the bottom part of your file on new lines, leaving an empty line in between. Add the name / number of the participant (“P1: ”) before each bit.

This results in a colorful mess: a lot of colored paragraphs with blank spaces in between.

Now we’ll do something called an affinity diagram. Copy a paragraph and paste it together with another paragraph that seems like they belong together. Congratulations: you created a cluster. Keep adding bits to the clusters until no paragraphs are left unclustered. Use double lines to mark off a cluster.

Whenever you feel like it, name your clusters. You can give them tags or headings. Feel free to rename or reshuffle until you feel right about the clusters.

Some grouped comments. I’m seeing patterns already ;)

Refuse the urge to do easy grouping (like grouping all the positive things together or grouping based on the interview question), try to look for deeper patterns instead. Make clusters with multiple ‘colors’.

Tip: another nice thing of Microsoft Word is that you can do this stuff together with other researchers with the cloud package.

4. Drawing conclusions #

By now you’ve wrestled with your data so much, you’re probably seeing patterns already.

Write your conclusions down per cluster and support them with quotes / video / audio. Going back to your recordings is important to objectively check if you put your findings in the right context. Tip: if you upload your video to youtube, they can automatically sync existing notes to the video.

When drawing conclusions, be critical:

How many people actually supported that argument? If you added the participant name or number per bit, that should be easy to see
Did the participant really mean what you infer? Go back to the recording or the full context of the note. Did the participant say this himself, or was he offered an answer by the interviewer?
Can you relate the findings to some properties of the participant? (Knowledge level, age, familiarity with subject-area) Maybe you find out the participant is not completely your target audience and this finding has less weight?
Are participants pleasing you? “Your design looks amazing!” or saying socially desired things? “I sport three times each week and only eat apples”
Relate to your research questions and other data. What were you expecting to see based on earlier data or assumptions? Anything you didn’t write hypotheses for (or can corroborate with other research) should raise alarm bells: new and unexpected findings are great, but they will need to be confirmed with new research! Also: don’t trust results that you were expecting, don’t fall into the trap of confirmation bias!

I’ve shared my approach to retrieving useful conclusions from qualitative research, but that’s not end of this story! I’m really curious to hear how you do it or if you have any feedback on my methods! Please leave a comment on the article, or PM me on LinkedIn

Good luck with your research! 😇

This article was previously published on Medium: Get the best results out of your interviews with a four-step analysis

Designing machine intelligence — inspiration from the Silver Screen (part 3)

Sun, 27 Jan 2019 00:00:00 +0000

This is part of a series to draw inspiration from the silver screen, as a way to ‘peak into the future’ of artificial intelligence (check part 1 and part 2). In other words: AI movies as a way to prototype machine intelligence and its possible impact on human life.

In this article: unnecessary consciousness

Netflix’s Next Gen (no spoilers) #

This particular gem is about a girl befriending a robot, which is a familiar theme in movies (Big Hero 6 and The Iron Giant, even the first Transformers movie and Bumblebee have the same premise). The girl is skeptical about robots and doesn’t like the way new robots find their way in every house (a plot eerily similar to I, Robot with Will Smith).

The movie has a great atmosphere and I found its world building super interesting. In the world of Next Gen almost everything has artificial intelligence applied to it, in the most extreme form possible. Every household object has its own personality: mailboxes, the doors, cleaning robots…

The Noodles-cup is the most extreme example: after you finish eating, the cup throws itself away, obviously not being happy about its life. Later on, in a massive fight-scene, a door sees incoming damage and is scared and one of the mailboxes is laughing just before being destroyed.

This leads to an interesting philosophical question: should we feel sad for the mailbox and the door? The characters in the movie obviously don’t care.

Is it okay to assign consciousness and then laugh when it gets destroyed so easily?

The tale of a cow and a paranoid android #

Just like in Next-Gen, a lot of random things get intelligence in the Hitch-hikers guide to the galaxy. Doors, sad robots (Marvin, the paranoid android) and even cows.

In the restaurant at the end of the universe a cow personally explains to you why it’s ethical (and really tasty) to eat it.

An interesting use of sentience, but also pretty uncomfortable!

Your task is to bring me the butter #

When Rick creates a robot with a consciousness just to have it pass butter, we think of him as a cruel guy. The robot hangs it’s head and Rick tells him ’that’s just life’, it’s funny - but cruel :)

So… Ethics? #

My examples show applying artificial consciousness to random systems is a slippery slope. It’s funny in a book or movie, but how happy would it make us in real-life?

Maybe it actually degrades the idea of consciousness? If doors, mailboxes and food are intelligent and conscious, would that drag your value of consciousness down? 🤔

images in this post are not under the Creative Commons license

Designing machine intelligence — inspiration from the Silver Screen (part 2)

Tue, 15 Jan 2019 00:00:00 +0000

This is part 2 in a series to draw inspiration from the silver screen, as a way to ‘peak into the future’ of artificial intelligence. In other words: AI movies as a way to prototype machine intelligence and its possible impact on human life.

Ex Machina (spoilers) #

A great movie, visually stunning and constantly leaves you wondering: who’s the bad-guy?

In short: a programmer named Caleb is invited over to his boss-inventor’s home. Here he has to judge whether his newly created robot can trick him into feeling for her as a kind of advanced Turing Test. Secretly, the inventor (Nathan) gave the robot the assignment to trick Caleb so she can escape.

It’s up to the viewer to decide if this AI has true intelligence and human(-like) feelings. At the end Ava (the robot) leaves Caleb (the human she pretended to fall in love with) in a locked confinement where he will presumably die.

What I found very interesting is that she asks Caleb “will you stay here” and he nods. I understood his nod to be “I don’t understand what you mean, but I trust you”. In my mind, Ava doesn’t really care if Caleb stays, dies or walks free: she pretends to have empathy but doesn’t really have it. That’s why she asks him if he wants to stay: she can finish her assignment irrespective of what happens to Caleb, it’s up to him. Because she is so ‘human-like’ in appearance and behavior we didn’t expect her to behave so cold and rational. We were tricked into thinking her intelligence was the same as that of a human, but it is not.

A wise lesson for UX-design in relation with AI: when you pretend your service has ‘human like’ intelligence, you create the expectancy of human behavior and capabilities. Every time your service cannot fulfill that promise (or kills people in cold blood), it looks like it failed. Perhaps it’s better to instead design the machine-intelligence to be different from human intelligence to underscore that difference?

There may be a different reason why Ava left Caleb.

Transcendence (spoilers) #

A researcher is shot, and before he dies, his mind is integrated with a machine intelligence. From this moment the philosophical question becomes: is the human still there? Or is it a machine intelligence, using the memories, thoughts and ideas of this human? Although the movie shows the benefits of an ’enhanced intelligence’ (diseases get cured, technology gets a huge boost, etc.), there’s no denying this goes at an alarming rate. People panic and decide to disable the intelligence before it decides to ‘harm them’. Are these people Luddites? (Is it okay that they are?)

The premise of this movie boils down to the same as that of Ex Machina: can we be comfortable with an intelligence that thinks almost, but not exactly the same as us?

Take-aways #

As UX Designers, how can we make sure technological advancements do not become disconcerting? Is there a way people could have been comfortable with the machine intelligence from Transcendence?

Maybe it’s better if the machine intelligences has less agency and ’the humans’ are more in control. Or perhaps the machine-intelligence should be modeled on something else than humans, so our users don’t expect human-behavior?

images in this post are not under the Creative Commons license

Make a maze

Fri, 11 Jan 2019 00:00:00 +0000

The challenge… #

I challenged myself to create software that can make a maze using javascript. Just for fun :)

Click here if you want to see the algorithm in action!

I defined the challenge as follows:

Generate a maze with these parameters

Size: {height, width} // h + w < 500 && h > 5 && w > 5
Start: {x, y} // (x == width || x == 0) || (y == height || y == 0)
End: {x, y} // ((x == width || x == 0) || (y == height || y == 0)) && start !== end

The maze should:
- Be solvable (so there should be a path from start to end)
- Have at least one branch (so not just one path from start to end)
- No unreachable 'blocks'

Sounds easy enough, right?

If you want to do it yourself, stop reading ;)

The Mental Model #

The challenge seemed simple enough, but after thinking about it for a while, it was not easy at all.

How do you represent a maze in code? What is a path through a maze?

When you want to create a maze using pen and paper, you probably draw a square and then draw ‘walls’ to create routes. But I couldn’t get this implemented in code in a simple way.

After multiple sketches I decided to represent my initial maze as a grid with a lot of ‘walls’ instead of an empty square. This provided me with an advantage: I could create paths through the maze simply by removing walls. Anytime I connect with an existing path I get closer to creating a complete maze.

The initial ‘maze’ is just a lot of walls:

█████████████████████
█░█░█░█░█░█░█░█░█░█░█
█████████████████████
█░█░█░█░█░█░█░█░█░█░█
█████████████████████
█░█░█░█░█░█░█░█░█░█░█
█████████████████████
█░█░█░█░█░█░█░█░█░█░█
█████████████████████

Take 1: recursion #

My first idea was to create a maze using recursion. This means running the same function a lot of times.

I start at the beginning of the maze and try to go in every direction (up, down, left, right) to see if the ’end’ is there. If it isn’t, I check if I accidentally went outside the maze or bumped into an existing path and I stop going in that direction if that’s the case. If neither of those conditions were met, I go from that second step into 4 directions and check my conditions again.

People familiar with recursion will know this leads into a lot of memory-use very quickly. Here’s an example of recursion with four branches at a depth of 2:

 |
 / | | \
 /||\/||\/||\/||\

Every branch is a function call, so already with a depth of 2 the same function was called 21 times!

To counter that, I created a maximum-depth dependent on the size of the maze. The maximum possible path-length is width x height, but that would be a very boring maze 🤓 (it’s the longest path possible going from A to B without branching)

So I chose:

$${maxdepth} = \frac{ {width}\times{height}}{2}$$

Note that one of the 4-directions can be closed immediately because it is the way you came from, so basically you branch only 3 times per layer in your recursion.

In mathematical terms, that’s:

$$\sum_{n = 0}^{n = {maxdepth}} 3^{n}$$

For a maze sized 10 x 10 (so max_depth = 10 x 10 / 2 = 50) the recursive function would be called a lot of times:

$$3^0 + 3^1 + 3^2 + 3^3 + … + 3^{50} = 1,076,846,981,537,778,883,155,373$$

Spoken out loud, that’s: 1 septillion 76 sextillion 846 quintillion 981 quadrillion 537 trillion 778 billion 883 million 155 thousand 373. To put that number in perspective: that’s the size of the observable universe in meters!!

Of course, our function is not called exactly that many times because potential paths get cut off if they go outside the maze or bump into an existing path. Still the amount of times my function would be called for a very tiny maze of just 10x10 is huge. And you’re not done yet after running the function this staggering amount of times because you’ve only created the initial route, the solution you plotted for the players of the maze. You then have to create a path using the same recursion from every empty spot in the maze to create the diversion paths.

Surprisingly the code ran reasonably smooth on my 6-year old laptop. It took a minute or so and resulted in very boring mazes with just straight paths.

Introducing randomness #

To try to salvage the situation, I decided to add a bit of randomness. I did this in two ways: randomly kill some potential paths and swapping around the directions (instead of always going up first, sometimes go down, right or left). The first option had the most potential because it also meant I could significantly cut into the amount of paths my algorithm had to take.

Another advantage was that I could now indicate how ‘complex’ I wanted my mazes to look.

First a maze with low randomness: only a couple of potential branches get pruned

Now a maze with high randomness, this already looks cool, no?

But now I came into another problem: by killing random paths it means my algorithm sometimes would not find the correct path at all. The only solution was to add a loop that would run the ‘random recursion’-function until it found a correct path.

My algorithm took minutes to create even the simplest of mazes and sometimes just completely froze my machine.

Conclusion: I was done with recursion!

Take 2: take me to a random place! #

It hit me: I had been going about this all wrong! There was a much simpler solution!

Instead of finding a random path from start to finish in the maze, I could create an optimal path to multiple random locations in the maze.

Creating a path from one point to another point is actually very simple. You just calculate how much you have to travel vertically: $$Y_{finish} - Y_{start}$$ and horizontally: $$X_{finish} - X_{start}$$ and then go through these steps.

Concretely, suppose you want to go from [2,8] to [5,3]. Now you know you have to go three steps to the left (5-2=3) and five steps up (3-8=-5). Your route-description would be: go right, go right, go right, go up, go up, go up, go up, go up. You’ve now written an ⏌, starting at the bottom-left. If you want to make this path a bit more interesting, you can randomize the ‘instructions’.

Step 1 was to use this method to create the initial ‘solution’-path through a maze. Going from start to a random location, go from this random location to another random location and from there go to the finish.

Step 1: create a random solution path

Step 2 is to create diversion paths (branches), to confuse the solver of the maze. I let my algorithm go from each empty-spot towards the direction of the center, until it collides with an existing path.

Step 2: adding random branches (diversion paths) to the solution-path

Yes! That’s more like it! 🏆

With this method, I can easily generate mazes with near-infinite sizes. Here’s a maze with size 50x50:

Click here if you want to see the algorithm in action!

Conclusion #

Recursion is difficult to get right
Making (a)mazing-software is difficult and you’ll fail on your first try
(See point 1)

Designing machine intelligence — inspiration from the Silver Screen

Tue, 08 Jan 2019 00:00:00 +0000

In my Quest to create something that ‘feels’ intelligent I’m drawing inspiration from multiple sources.

One cool way to ‘peak into the future’ of artificial intelligence is to compare different ways AI has been portrayed in the media. To me it’s a way to prototype machine intelligence and it’s possible impact on human life.

WALL•E #

This cute movie goes to show how much expression can be shown without using a screen. Interestingly it’s the robot that humanizes the humans – a very recurrent theme in robot-movies.

A major reason why WALL•E is so endearing, is his amazing expressiveness. He uses just his body and ’eyes’ for this, although Pixar cheats a bit with how ‘stretchy’ all of the metal is :) (source)

To me WALL•E feels intelligent because he exceeds his programming and doing his own thing. In the excellent first 30 minutes of the movie (near perfection) instead of ‘just collecting waste’ WALL•E is inquisitive and investigates the world around him. Another way his intelligence feels similar to our own are his emotions: looking longingly at movies and being happy and proud of his trophies. He cares for others: he shows empathy (for his buddy-cockroach and EVE).

Another reason we care for WALL•E is also that he is very vulnerable. He makes mistakes and he can be physically/mentally damaged.

Conclusion: Agency, emotion, empathy and vulnerability

Chappie #

Not a big box-office success, but I liked this movie :)

A short recap: a robot (Chappie) is stolen from the company that created him and falls in the hands of South-African criminals. At the start Chappie is very fragile and doesn’t know much about the world. His ‘parents’ quickly raise him to do robberies, but slowly he learns the difference between right and wrong.

The criminals raise him like a child, his actions and body-language show him progressing from ‘baby’ to young adult. More than any human in the movie, Chappie changes and grows mentally and emotionally.

‘Chappie’ is one of the few movies about a learning machine. This capacity to learn grounded his intelligence. The movie weighs in on the nature/nurture debate. It makes you think about the ethical responsibilities of the parents growing up in a criminal home. If you think of his ‘criminal education’ as programming, clearly the ‘parents’ are to blame. But if the robot learns this behavior and has a consciousness then perhaps it also carries responsibility, just like humans.

Can we hold Chappie responsible for his actions? Should he go to prison, or be dismantled as a buggy machine?

We humans magically become responsible adults at the age of 18, when would that be in artificial intelligences?

Like in WALL•E, Chappie shows emotion, vulnerability and empathy. His emotional and mental growth throughout the movie makes him even more intelligent than WALL•E to me.

Take-aways #

If you see these movies as a way of prototyping machine intelligence you get a look into the future of how the impact AI could have on people. My main take-aways for creating an experience that feels ‘intelligent’ are: make it show emotion, be vulnerable, have a capacity for learning and have the machine intelligence show empathy.

images in this post are not under the Creative Commons license

2018 in review

Sat, 05 Jan 2019 00:00:00 +0000

2018 in review #

My first personal year-review. I had fun this year :) I wrote over 20.500 words worth of blog-articles which had over 3.000 views on Medium and 1.356 on my website. Work was great, and I started writing a book.

AI and Design #

In 2018 I started to focus on AI in design. It all started with letting a neural network create my privacy policy. After that I presented a talk “Creating magic: Design and AI” at UX Camp Amsterdam and wrote an article to go with it. I since presented updated versions of that talk twice.

My next step is to take this further, I described my plans a bit better and published a first ’experiment’ of trying to design something that’s perceived as intelligent.

Me presenting at UX Camp Amsterdam

1001ideas.org #

My ‘crazy ideas’-blog got updated as well. I dreamt up a rotating clock face, a place to train yourself to be more rebellious, a t-shirt, a global address book (an idea I actually pitched to the Dutch postal organization), I tried out bacon tea (unsuccessfully) and I published my vision on how to get rid of cardboard.

Workshops and Facilitation #

I invented 3 new workshops with my colleagues at Angi Studio. The most important one being the Design System Takeoff which I co-facilitated at the municipality of The Hague. I facilitated multiple workshops, Design Sprints and Research Sprints, and even a marketing sprint. I trained people in giving Design Sprints.

Researching job-titles in design #

83 people filled in my questionnaire on job-titles I set up that research, because I feel that job-titles are misleading. I published a detailed analysis of the results (and an analysis of that analysis).

Prototyping and experimentation #

I wrote 5 tips on how to use the prototyping tool Axure, in one of which I create an actual working game. I experimented which photo-service is the best (part 1 and part 2). Figured out which days are most popular in songs, created several scripts for photos, moved my blog to gitlab, experimented with internet beats in my blog and wrote on how to combine git with dropbox.

New year, new stuff #

I changed banks, mobile OS (went to iOS from Android) and moved to iCloud for my photos (read here why I did it, how and if I like it or not). I also started writing a book.

What’s up for 2019? #

I’m going to continu with my hobbies: photography, experimenting and prototyping crazy ideas. I’ll also delve into AI&Design more.

Furthermore, I hope to be more relaxed in my attitude, we’ll see how that works out ;)

Review of Apple Photos: the good, the bad and the ugly

Mon, 31 Dec 2018 00:00:00 +0000

In my recent posts I explained why I’m moving to Apple Photos and how I did it.

Here’s the long-awaited conclusion of that journey, where I tell you how I like this new photo-organizing solution!

The good 👍 #

I noticed I have more fun with my photos: instead of the chore of organizing them, I look at my photos, share them with friends and family and make fun edits of them
Apple Photos has lots of editing options built in and integrates with plugins and filters
edits are synced with my iPad and other devices. There are loads of (free!) apps to edit photos on mobile devices, such as PhotoFox, Snapseed and Adobe’s suite. I make some global changes on my desktop within Apple Photos and remove glare with Affinity Photo, switch to my iPad for finishing touches and share the finished photo with WhatsApp on my iPhone. OMG: this is the life
iCloud is cheaper than dropbox
I get cool reminders from Apple like ’this day 5 years ago’ or ‘all birthdays of ’

Look at all these cool editing options that are built-in!

Editing a photo #

Here’s some samples of editing a photo:

The original photo. Nice, but a bit bland.

Auto improved photo. Lot more color, still a bit bland.

Manually improved photo (I’m not too good at this)

And for comparison, the Google Photos auto-improve

The bad 🤔 #

it crashed multiple times. Over the course of a couple of weeks I would say around 5 times
I don’t fully understand the structure of the app. Apple automagically creates ‘moments’ that can’t be renamed, and these ‘moments’ are grouped into ‘sets’, which are incomprehensible. Even worse is when clicking on a photo in a ‘set’, it does not go to this photo, but instead to a random ‘moment’ close to the date of the photo you wanted to see. Good luck finding that one photo back!
there’s no syncing of ‘persons’: face recognition done on the laptop stays on the laptop. Supposedly this is for privacy purposes, but I think it’s just lame
bad geotag-grouping: photos taken inside my house are all scattered in a 50 meter radius around my house. No, the photo of me holding my cat was not taken at my neighbors house. Why not offer clustering to group things neatly together?
WhatsApp photos get geo-tagged to the location you received them, which doesn’t make sense
slow syncing: it can take minutes to sync
it’s not possible to separate work-photos from others. It used to be possible to ‘hide’ photos and only have them show up in certain albums, but Apple changed that behavior
sharing with non-icloud users is a hassle…

The Ugly 🤮 #

it’s slow on my laptop 🐌. Perhaps because of an external disk? Or an old MacBook Pro (mid-2012)? Or because of the 87.000 files? But there are really no excuses. Changing to edit mode should not take 20 seconds. Clicking on ‘persons’ should not take 16 seconds. Waiting 2 minutes between clicking on a ‘person’ and seeing their photos pop-up is just inexcusable
unsupported movie formats get shown as photos on iPad and iPhone as a screenshot of first frame. On my laptop they show as videos. This is hugely confusing! What conflates the matter more is that because of this, the numbers are different on different devices (2.237 videos on laptop, 2.015 on iPhone). Why not convert the movies to a phone-playable format during the upload??
according to the internet, it’s not possible to move a local iCloud database: it needs to completely sync it again, taking days
movies are imported with the wrong date: Apple Photos uses the date of import instead of date of creation

Conclusion #

All in all, Apple Photos needs more love from Apple. I would still recommend it, but I think for most people Google Photos would be the better option (if you don’t edit your photos as much and trust Google with your privacy).

Moving to Apple Photos

Sun, 30 Dec 2018 00:00:00 +0000

In my recent post I explained the why of moving to Apple Photos. Here’s the how.

First steps… #

I have an immense photo library with 87.000 photo- and video-files, spanning over 20 years. My first step was to extract around a 100 of these and import them into Apple Photos to see if it could handle all the different filetypes.

The good news is that nearly anything I threw at it would work: even photos and video from ancient cameras. The bad news is that RAW imported from my particular camera/lens combo (Sony NEX 5R with 18-55 lens) does not get processed properly: there’s no lens-correction. Not a problem for most of my files (they are actually JPEG) but the last 3 years I’ve been shooting RAW more-and-more. There are plugins that support extensive RAW-editing in Apple Photos, but they all use Apple’s RAW-engine (so same problem) or they have their own engine but you cannot do batch-editing.

Metadata is imported from XMP (metadata in side-car files) so I could keep some of the data I applied in Capture One (my previous photo-management solution). The problem is I organized my photos in albums and not by tags, and only tags are stored in the metadata-files.

Apple Photos has two ways of organizing files: referencing or importing. The difference being that in the first method files stay in their current location and are only referenced in the database, and with the second option files are actually copied to the Photos-database. At first glance option 1 seemed excellent, but that doesn’t sync with iCloud. So I had to go all-in and import photos. As I want to keep my ‘old structure’ intact (in case Apple Photos doesn’t work out in the long run) that I will use double the disk-space with the ‘import’ option. Luckily storage space is cheap!

Preparing the Import #

Knowing that RAW is not imported properly and albums are not exported in metadata complicated the import-process.

I decided to enable ‘sync metadata with sidecar files’ in Capture One and apply tags to all the photos in my albums (so all photos the album ‘cool photos’ would get a tag cool-photos). That solved problem 1.

I exported all my RAW-photos with Capture One as high-resolution JPEG-photos, profiting from C1’s excellent RAW-engine for the lens-corrections and sharpening. I set Capture One to mimic the existing folder-structure on export, so I had an exact replica of my old structure of photos with JPEGs instead of RAW-files. I named this my import-this-into-apple-photos-folder.

I could also export the JPEG-photos I kept in Capture One, but I decided against this as I never imported all photos in Capture One and figuring out which ones were there and which ones weren’t was not something I looked forward too! This meant losing any edits I did to the JPEGs in Capture One, but I can live with that.

Lastly, I used rsync to copy anything that wasn’t a RAW-file into the import-this-into-apple-photos-folder.

I was now ready to start.

Importing the photos! #

I dumped the entire import-this-into-apple-photos-folder into the app, and… It Crashed.

I sighed and started to import my photos in batches. First the oldest photos (I expected Apple Photos to be crashing on those). I had 11 batches and only in batch 10 I found the ‘offending’ file.

Each batch took hours to import, and even longer to upload. Apparently Apple’s upload-speed is much lower than that of Dropbox. It took me four weeks to completely import and upload everything.

Results #

In the end, only a handful of items could not be imported: a couple of MOV-files, several corrupted JPEGs, PDFs, THM (old movie format), ZIPs, AVIs and GIFs. All in all, 460 files. Not a big deal.

A bigger deal was that after using Apple Photos for a while I noticed some files were duplicated. Apparently I didn’t filter out the Sent folders of WhatsApp before importing (check this article for more details). Now I still have a messy database…

I also forgot to tag a couple of albums in Capture One, and I now have to recreate those albums from scratch 😔

Future imports #

Because I want to keep my ‘old structure’ intact, I’m not going to import photos into Apple Photos directly.

RAW-photos first go through Capture One. I organize them into the ‘big structure’ and develop these into JPEG before importing into Apple Photos.

JPEG-photos I want to import (from my secondary camera, my old Android-phone or photos I get from other people) will not be organized and just dumped in an ‘unsorted’ folder before going into Apple Photos.

The above process costs a tiny fraction of the amount of work my previous process took, so I’m very happy with the results.

Learnings #

Mostly, everything went as smooth as I expected. I hope to never do this again (obviously), but when I do, I’ll make sure I’ve done the tagging properly and removed the duplicates!!

In my next article I’ll review the experience of actually using Apple Photos. See you next time.

Who will I trust to sync my photos: Apple, Google or Adobe?

Tue, 25 Dec 2018 00:00:00 +0000

I used to organize my photos by hand: a terrible, horrible, laborious process! Now I’ve moved to The Cloud for my photos and this article is a short recap of my ’thinking process’. I hope this is useful for other people as well! Later I’ll add a review to let you know how I like my new solution.

My library contains over 87.000 photo- and video-files, spanning more than 20 years. I spent a lot of time maintaining my file-structure (details below), instead of actually looking at my photos and enjoying the memories.

For the new solution I just want a machine to do all the organizing.

My old process #

Before you can understand my choice, let’s first see how I used to do things. I take photos with my camera and with my phone. Organizing these photos was a 2-step process.

Camera-photos #

take them from the camera
run a script that organizes the photos in folders by date (20180812 for 12th of August, 2018)
import into Capture One to develop the Raw photos
delete all ‘bad’ photos
rename the folders so they make sense (20180812 - Going to the zoo)
move the photos to the ‘big structure’ of “year / location / event”: 2018 / Rotterdam / 20180812 - Going to the zoo

Phone-photos #

take photos from camera-folder and WhatsApp folders
run the same script to organize by date
import into Capture One and delete all ‘bad’ photos
move photos around so that all the folders make sense
rename the folders
move them to the ‘big structure’

AAARGH, I don’t want to do this anymore! #

The process described above took me forever and I was always running behind. Even worse was I never looked at my photos once they were in the ‘big structure’. Why was I even putting myself through this hell?

I hope in the future when somebody asked “did you have fun this weekend?”, I never have to tell anybody again: “oh, I organized the photos of the last three months”.

My wishes #

I wanted a new way of dealing with my photos. Where I let a machine do the sorting and I focus on doing fun stuff with my photos. I wanted easier ways of finding back old photos: tagging, adding stars, creating albums, face-recognition and preferably place-recognition (for all photos pre-2014, when I didn’t geo-tag them). I also wanted something that syncs the photos + their edits to my other devices. Lastly, I would prefer a solution where I can edit photos using external apps such as Affinity Photo.

My options #

Basically, I see three main players: Apple Photos, Google Photos and Adobe Lightroom CC. The other options are to keep my ‘old way’ or to move everything into Capture One (or Photoshop Elements) and sync using Dropbox. I don’t know of any other photo-syncing tools.

Apple Photos #

If you surrender to the Apple methodology, it probably all works out, but if you think differently: bad luck. I would have to recreate all my folder-organizing into albums by hand (in other words: I’d lose my existing structure), because by default Apple Photos is structured in a weird way using moments that cannot be renamed.

It does sync iPhone, iPad and computer nicely and it allows for external editing. Editing in Apple Photos is fast and very extensive and offers a lot of ‘pro’ options. You cannot combine photos or do local edits, but there are plugins for both. And if all these options are not enough, it’s also possible to ’edit in external app’.

Sharing with others is hard ~unless they use Apple Photos as well~ (update: you can share a link). No love for Android…

RAW support is excellent, but sadly not for my Sony NEX lens. It doesn’t do the lens-corrections.

You pay 10 euro per month for 2TB and you can share this with 4 other people.

Google Photos #

Google automatically uploads your photos and you can keep your folder-based organization in Google Photos. That’s a big plus. Its automatic organizing features are also great (it even automatically recognizes the location the photo was taken without GPS data).

There are three downsides to Google Photos: there is no desktop app (no offline mode), editing options are very basic and edits are not (really) synced across devices, and privacy. Google already knows too much about me and I would prefer keeping my photos out of their hands.

Although the editing features are very basic, the automatic improvements are extremely good.

Google Photos works with ‘Backup and Sync’ and is basically a mix between Apple Photos and Dropbox: you can do filtering and face recognition (but only in the web-interface) and you can do editing with an external app (but only in the ‘Dropbox’ part of the solution). It’s not integrated.

Sharing is easy and also works for Android. Really cool is the ‘partner-account’: where a person can see all your photos going from a certain date forward.

RAW support is spotty and undocumented.

For full resolution photos you pay 10 euro per month for 1TB.

Adobe Lightroom CC #

This is Adobe’s offering. It syncs photos from everywhere and adds machine learning for organizing - except face recognition. It’s not possible to edit photos with external apps (lame) and sharing is difficult (you can create web-albums, but not share with other CC-users). But you do get the impressive Adobe engine for RAW-development and lots of editing features. Also, all the mobile apps are included.

Lightroom CC is 12 euro per month for 1TB space.

Adobe was a direct ’no go’ for me, because of the lack of facial recognition and because it’s not possible to edit in Affinity Photos. A lame restriction.

Dropbox and Capture One #

I can be quick about using Dropbox and Capture One: a lot of manual work still and no syncing the photos database (tags, albums, etc.) with my other devices. So a no-go.

The final verdict #

I have to decide between Apple Photos and Google Photos. Both offer automatic filtering and face recognition, but only Google offers me the possibility of keeping the ‘old’ structure I invested a lot of time in. Both offer me the possibility of syncing photos between my phone, tablet and computer, but Apple has the upper hand because edits of photos are synced more logically. With both I can edit my photos, but clearly Apple offers more options: built-in possibilities, filters and plugins. Both offer ways to share photos, but Google is clearly superior (especially when the other person is not using Apple Photos). Lastly, while Apple supports geo-tagging, Google uses their immense AI-knowledge to automatically recognize where photos were taken.

In the end I decided to go with Apple Photos. It means being forced to stay with the Apple ecosystem and dropping all the hard work I put in organizing my photos*, but I get extensive editing capabilities and piece-of-mind privacy-wise.

* Note that this is not all bad: I don’t want to do the manual organizing anymore, so maybe it’s better new and old photos are all organized in the same way

Behind the scenes of “Dutch Designer: What do you do?”

Mon, 17 Dec 2018 00:00:00 +0000

For my article on Dutch Designers I analysed a questionnaire with 83 participants. This is the place where I share the details of the analysis with you.

Off into the Mariana Trench of DATA! #

Getting these statistics was not easy, it took me some magic tricks in Wizard, Numbers, Orange and plain-old-terminal commands 🤓

Now put on your diving gear, we’re going deeeeep into the data!

Who filled in my questionnaire? #

In statistician terms, n=83 with an unequal distribution of groups. 45% of all participants were UX designers (that’s nearly half!)

Of the 83 people that filled it in, 32 had 10 years of experience (or more). The average is almost 7 years of experience (note that ‘average’ is not really accurate, because one of the answer-possibilities was “10+”).

Work-experience (green) vs average-skill-rating (blue)

Strangely, there seems not to be a correlation between work-experience and skill-rating. I expected to see a bit of Dunning/Kruger effect here.

Context of work #

I asked people how they worked (freelance, agency, etc.), but didn’t use this “context of work” in my original article as the results were not as interesting as I expected. Nevertheless, here’s a cool graph:

Most interviewees work freelance or for an agency

Freelance workers have the most work experience :)

Analysis of tasks #

The most interesting part of the survey is the self-rated skill-level (“how proficient are you at …”). Here I had an interesting challenge: I gave people the option of 1, 2, 3, 4 or 5 stars — and a ‘not applicable’. I struggled with this last category a bit: how do I make a nice average out of this ‘not applicable’? I decided in the end to treat it as ‘missing data’, instead of assigning a -1 or 0. This means you calculate an average with only the values actually given. The downsize is that individuals can have a big impact in groups with only a few participants. Example: if everybody assigned ‘not relevant’ to a skill and one person gave it 5 stars, the whole group gets ‘5 stars’ at that skill…

I later used the ‘not applicable’ as a way to measure the relevance of a particular skill . The skills people indicated mostly as ‘not applicable’ are photography, creating the sales proposition, print design (business cards, envelopes, etc.) and writing code (for instance JavaScript).

Here you see all skills based on the relevance:

how to read this table: the darker the square, the better this ‘group of people with the same function title’ scored themselves at it. The blue square indicates how often this skill is applicable, the ‘standard error’ column can be ignored (but indicates if there are big differences in skill-level between participants)

I can not stress how happy I was to see this graph (table?). It indicates something useful and expected. Just when I thought nothing came out of my study after hours of digging in the data!

The nitpickers among you* will notice the ‘averages’ in this table are not the same as the ones I used in my original article. The average you see above is an average over ‘the average of the skill-levels per group’ which is different from an average over ‘the average of all individual answers per group’. A formulaic-expression and indepth insights in why these are different are available upon request.

*if you noticed this difference, I applaud you for reading my articles so thoroughly and invite you over for a beer where we can discuss the use of commas and the difference between a ‘mean’ and an ‘average’

Visual inspection is deceptive #

By looking at the graph it seemed there was a lot of overlap between Service Designers and UX Researchers. Both scored very high on interviewing users, making and analyzing questionnaires, task-analysis, creating personas, user testing and drawing journey maps. This is a deception!

By using a Statistical Analysis™️ in my similarity-matrix (below), I figured out there is not a big overlap between these two. The deception is most likely because the columns are close together and because ‘we humans’ like to see patterns in data.

Variance #

There is a lot of variance in my data, which is probably due to the small sample-size. But it is interesting to see that there are some skills that you are consistently either very good at, or very bad at: print design, being a scrum-master, testing frontend code in different browsers for bugs and writing HTML/CSS / design in the browser.

You could use this high variance to see which skills you can develop to set you apart from the average designer. Especially skills high in the relevance rating are interesting, of course: creating customer journey maps, handoff to developers and creating pixel-perfect designs.

The Similarity Matrix #

I wanted to know how much groups were ‘similar’ to eachother. What you see below is a matrix with the average difference in answers (the mean variance). I based this on the top 34 skills (those with a relevance above 0,70). That cut-off point is a bit arbitrary, but is to accommodate for the impact of non-relevant skills.

The similarity matrix! (feels good if you can use the word ‘matrix’)

How to use this matrix? #

Take row and a column and see how ‘similar’ certain groups are. A low-number means a high-similarity. I’ve colour-coded the results to indicate interesting findings. Who knew UX Researchers are so different from Digital Designers and Product Designers!

Machine Learning #

As I have a degree in Artificial Intelligence, I thought it would be a Great Idea to see if a computer could make more sense of this data than I could. The 83 data-instances I have are not a good base for machine learning (you need loads more), but that didn’t stop me from trying 🤓

I tried a Neural Network and k-Nearest-Neighbour modeling technique and as expected, they couldn’t create a decent classifier for my data. Both had a very low recognition-score, even when using the test-data as validation-set (which you should obviously never do, normally).

It seems my kNN found a couple of ‘undercover interaction designers’: most of the classification errors came from UX Designers that were seen as Interaction Designers. And the Neural Network got even more confused: it mixes up UX Designers, Product Designers, UX Strategists and UX/UI Designers. It does confirm that these four are kind-of similar!

k-Means #

I also checked out what would happen if I’d let the computer come up with good groups by doing a k-Means analysis (a technique that tries to find clusters in a pool of data). As expected, it didn’t find the exact 8 groups I defined. Instead it seems more logical to go for 2 groups (or 3).

It’s interesting to see that these two groups do make sense. Group 1 (the smaller group with 27 people) is very visually oriented, Group 2 isn’t (56 people). The top-15 tasks that make the biggest difference between the two groups are: making designs pixel-perfect, style studies (typography, colors), create or update styleguide, writing HTML/CSS / design in the browser, print design (business cards, envelopes, etc,), creating moodboards, creating design systems, icon design, being a scrum-master, design-handoff to (frontend)development, brand identity design, testing frontend code in different browsers for bugs, interface animation, writing code (for instance JavaScript).

So according to the algorithm it makes sense to talk about 2 types of designers: a visual designer and a non-visual designer.

Conclusions #

If you want to read my conclusions on what Dutch Designers do, just read my original article!

Originally published on Medium

Designing Intelligence

Mon, 17 Dec 2018 00:00:00 +0000

Designing Intelligence #

I’m looking for ways to combine my passion for design with my passion for artificial “intelligence”. What does that mean and why did I add intelligence in quotation marks? I’ll tell you in a sec!

I’m interested in three things:

people
technology, especially Artificial Intelligence (AI)
experimentation by prototyping

The coming weeks I’m combining these three in a series of blog-posts and experiments. I hope to find some fellow travelers in my Quest, if you’re interested just email me at matthijs@matthijszwinderman.nl - or follow this blog-series!

When you combine AI with design, you’ve got the power to create magic. I want to see how that works out in the real world, in the jungle of people and other technology!

Intelligence from Team America

Let’s put the “intelligence” back into Artificial Intelligence! #

When you think about AI, you probably think about recommendation engines (Spotify, Netflix), automation (robots, self-driving cars), chat-bots / digital assistants (hello Siri, Cortana and Alexa), and classification (there’s a dog on this photo!)

These systems get trained on past data and use that to classify new instances. But they don’t learn ‘on the job’ and there’s no deduction or active participation in the world.

That’s not what you think of as intelligence in normal life.

Take any random person you know, like Hank. Do you think Hank is intelligent when he would just sit around all day doing nothing until you talk to him? How about responding “there are 7 days in a week” when you ask him “how’s your week”?

Contrast that with your cat: he can’t play chess, drive a car, recommend you new music or do any other ‘intelligent’ thing. But in a way he’s a lot more intelligent because he has his own free will and you can relate to him: “he wants me to pet him!” or “he understands me”.

Most products labeled ‘AI’ do not feel intelligent and it takes away the impact the word could have.

What’s your approach? #

I want to focus on creating products and services that feel intelligent. I suspect the keywords are empathy, emotion and ‘agency’ (somebody or something taking action out of free will).

In the end, agency might be closer to intelligence than doing some smart tricks. A counting horse is intelligent, a counting calculator is just a machine.

I want to see how we can create this experience and I suspect it’s not that hard: I call my devices ‘smart’ all the time, when they do something beneficial when I didn’t expect it. It’s really small things, like how my Mac remembers the display-setup based on the monitor I plug in, or how ‘do not disturb’ is turned on when giving a presentation. Funny how that works without AI!

AI-techniques could really amplify this feeling; a traffic light that stays green longer for children or other slow-walkers.

This is revolutionary! #

Well, no. Not exactly. Everybody is experimenting with AI currently and it’s the Buzzword Of The Year. The ideas about using empathy, agency and emotion have also been around forever. But, I’ve been reading through a lot of design principles, frameworks and toolkits* and I haven’t found a focus on intelligence yet. It’s all talk about designing for (or with) machine-learning techniques.

So I hope my angle is still a bit fresh: exploring and defining intelligence with design experiments.

Two caveats:

I am expecting that most of the time, my experiments won’t involve AI-techniques. And why should they? Humans are great at assigning emotions and intelligence to all kinds of things: animals, plants, inanimate objects (it’s called anthropomorphism). We yell at traffic lights when they turn red. A great example are Breitenberg machines: really simple things without built-in intelligence, but displaying very complicated behavior. It’s easy for humans to “feel” for them. I don’t know yet how to make the leap from these ‘play systems’ to a bigger / useful system, but that’s exciting!
I expect a lot of times ‘intelligent seeming products and services’ won’t be beneficial to people and might even be detrimental. We’ll see how that works out ;)

What’s next? #

This blog-series will hopefully end with some kind of framework. For now I’ll keep to ideation, exploration and experimentation.

Looking forward to your feedback and let me know if you want to join :)

Thanks a million to Sangeetha Jayaram for thinking this out with me!

* Here’s a short list with the most interesting articles:

images in this post are not under the Creative Commons license

Dutch Designer: What do you do?

Mon, 17 Dec 2018 00:00:00 +0000

I think function titles in the digital design field are misleading. To test that theory I had 83 people fill in a mega-survey, and I found some very interesting results! Did you know that the fabled UX/UI-designer actually exists? And that Product Designers are very modest? Also, we finally find out what a Product Designer really does! 🤣

In my questionnaire I asked people which function-title they give themselves and which titles others use to describe their work. Based on this I found 8 groups (UX Designers, Interaction Designers, etc.). I also asked them to rate themselves on a scale of 1–5 on various skills (52 skills in total!) and with that I’ve got a good overview of the skill-level per function-title.

Who are these people? What do they do? What drives them? And how does that relate to you: should you change your function-title on LinkedIn? 🤔

First up: UX Designers, UX/UI Designers and Interaction Designers #

I’ve grouped these together because these professions offer a very similar toolset. Here are the main differences:

Interaction Designers rate themselves higher at theoretical stuff: writing user-stories, UML-diagrams, requirement-analysis and task-analysis — and creating interactive prototypes
UX Designers rate themselves higher at analysis- and people-skills: creating user journeys, interviewing users, facilitating workshops, competitor analysis, storyboarding, creating questionnaires, creating personas, training others, reviewing performance of others, writing proposals and A/B-testing — and also interface sketching
UX/UI Designers rate themselves higher at visual design related skills: creating moodboards, making designs pixel-perfect, brand identity design, print design, design-handoff to (frontend)development, creating design systems — and writing vacancies

For wire framing, sketching and interactive prototyping all three share a high skill-level. If you can choose it’s wise to pick the UX/UI Designer because they have a very diverse skill-set (rating 24 of their skills above average and an impressive 10 skills at 4 stars or more).

For all you statistical nutters: I didn’t adjust for average years of experience (UX Design have 2,5 years less work-experience) or normalize for the average score people gave themselves (UX Designers rate themselves higher than the overall average: 3,16 vs 2,89). Read more in my analysis article!

The UX Strategist #

This is such a cool function-title, I wish I had it 🤣. The UX Strategist is very similar to Interaction Designers (and less so to UX Designers), but not similar enough to justify pooling them together.

UX Strategists look more at the overall picture and focus on guiding others. Their strengths are facilitation, coaching, training and setting up information architectures. They also score themselves much higher than others on writing vacancies, proposing sales proposition, hiring new people, reviewing performance of others, creating a content strategy and writing project proposals / tenders.

If you’re looking for someone for research, a UX Strategist seems less of an obvious choice (see also the big-table below).

The Digital Designer #

The Digital Designer is a very specialised person. Although the skill-set is very similar to a UX/UI Designer, the Digital Designer is much more modest about their range: they only rate themselves higher than a 3,5 on 16 skills, whereas UX/UI Designers do that for 25 skills!

That means a Digital Designer is not the right fit for research and analysis or taxonomies, but they excel at everything to do with visual design: design-handoff to (frontend)development, making designs pixel-perfect, style studies (typography, colors), interface sketching on paper, creating design systems, print design (business cards, envelopes, etc.), create or update styleguide and interface animation. They also rate themselves highest at testing frontend code in different browsers for bugs.

From my analysis it seems that the skills of individual Digital Designers diverge a lot. They have the highest variances in their answers (e.g.: one half of the group indicates a 1 while the others indicate a 5). Especially for the skill “creating or maintaining a design system” the answers are all over the place. Perhaps there can be a vote at the next Digital-Designer-conference if that task is part of the job :)

Product Designer #

The Product Designer is a mysterious person, what does (s)he actually do? As a Product Designer apparently you have to explain yourself a lot (quote from Pablo Stanley):

They would immediately ask me what kind of product. Furniture, airplanes, radios, headphones, sex toys? Embarrassed, I would clarify that I meant digital products — “you know, like websites and apps.”

Based on the self-assessed skills in my survey we can say the Product Designer is a kind of ‘UX Designer’-light. They don’t score higher than UX Designer even on the highest-rated skills (also UX/UI Designers and UX Strategists are better at almost everything than the Product Designer). As a matter of fact: in the average scores there’s nowhere they score higher than 4 points! That’s unique among all other groups.

But is it (false?) modesty? Or is there more? One participant in the survey answered 1 or 2 stars for every skill, even though (s)he is a senior designer with 8 years of experience. I am afraid this is because of my form-design: if you entered the survey on a very small mobile phone, the last three columns (with 3, 4 and 5 stars) could be cut-off. I checked and luckily this is the only outlier.

Another explanation is that the amount of relevant skills is not distributed nicely over the groups (there were more skills listed for UX designers than for researchers, for instance). One “Product Designer” apparently works more on physical products and because of that rates him/herself much more modestly on ‘digital’ skills ((s)he listed a lot of skills missing in the questionnaire: PCB Design, CAD modelling, Sourcing and Hardware Design).

Service Designer #

With an average of just 4,5 years this is the group with the least work-experience. But that doesn’t stop them from being ambitious: they rate themselves higher than a 4 on no less than 15 skills! No wonder they are the least-modest group, with an average self-assessed skill-level of 3,4 (out of 5).

Especially interesting with this group is they are the only ‘designer’-group who are good at long-term studies and they can uniquely combine this with creating customer journey maps, user journeys and storyboards. They are great at all kinds of user-related research, but not so much at other types of research (competitor analysis or web-analytics, for instance).

Because of this diverse skill-set they don’t share any similarity with the other participants from the dataset.

UX Researcher #

Last but not least, another outlier group: the UX Researchers. Their skill-set has the least similarity to that of any other group.

Obviously the research-skills are very highly self-rated (except content analysis, strangely), just like formalizing the results of that research: creating personas, writing user stories/epics, creating customer journey maps, creating taxonomies, drawing user journeys. The weak-points lie in creating detailed designs: for instance they are the only ones who rate themselves very low at creating interactive prototypes, wireframing and sketching interfaces.

Time to compare skills in a giant table! #

No article is complete without a huge table, am I right? The table below shows what these groups are good/bad at.

Pro-tip: print the table (in colour) and use it as a checklist: see if your skill-set matches with your function title! Or find out which header you should put above your next job-vacancy!

I’m really interested in knowing which parts you find most interesting from the table below. Let me know with a comment!

Good: they self-rate on average higher than 3,5 and ‘bad’: they self-rate less than 2,5

Some quick thoughts: UX/UI Designer are very versatile, rating 24 skills above average! While Product Designers and Digital Designers are too modest: rating over 20 skills below average… And strangely nobody seems to do lean-startup-experiments. And how come nobody is good at web-analytics or competitor analysis? Is that a skill-gap, or more a job for marketing?

Based on the answers people gave, I calculated how similar certain function-titles are to each other (more details in my background article). I’ve used those results in the ‘similar to’ section of the cards. Interestingly, it seems the groups are more distinct than I initially expected. Perhaps a function-title does say something about your skill-set!

Conclusion #

Every good article needs a conclusion. I started off with my hypothesis that a function-title doesn’t say much. Does that still hold true?

I think my research has shown that there are common factors behind a function-title, so on first glance it appears that title has some merit and you can judge a book by its cover. But! A lot of people don’t communicate their function-title. You can see that in the ‘aka’ section in the cards above and there is a lot of overlap. So even though a person might think of themselves as “Interaction Designer”, they might communicate something else on their LinkedIn profile.

On top of that, the division in 8 groups is not the most optimal way of distributing the survey participants. If I let a machine-learning algorithm do the grouping, only 2 or 3 groups remain. So maybe there are new (better?) function-titles we can come up with still!

I’d love to discuss these results with you! Just send me a message on LinkedIn, or press the applause button below for a while on the original article on Medium ;)

Want to know more details? For instance which two groups a machine learning algorithm thinks there are in the Dutch Design World? Read my ‘behind the scenes article’.

Originally published on Medium

How we created the best conversational UI ever

Mon, 17 Dec 2018 00:00:00 +0000

This article was previously published on Medium

Microsoft and Google are pumping out frameworks, plugins and guidelines making it’s easier than ever to create a conversational UI. But are these plugins and frameworks actually enough to create a delightful experience? We experimented with doing it differently!

Conversational UIs are not just hip and happening, they pop up everywhere. Apple, Google and Amazon are scrambling to put microphones in your home, chat-bots are emerging like mushrooms on Facebook Messenger and Skype, and many websites are adding plugins to their websites so they have another way to interact with you. But I feel these conversational UIs are lacking something…

Alexa is now also available on your wall!

Problem 1: context #

Common examples of problems with context are when a conversational UI ‘forgets’ about things you asked it earlier or when a chat-window on a website doesn’t know which page your are currently viewing.

Conversational UIs also don’t understand the context of compassion: whether you are sad, angry or happy, the computer always answers the same 😔

When chatbots go wrong

Problem 2: it’s all text #

Conversational UIs focus very heavily on text, even though there are many more ways to interact.

Let’s say you’re on the street and you ask Siri “is this restaurant any good?”. You then have to give it a name and address even though you are right there. Why can’t you just point your camera at it?

This holds for chat as well. Usually chat windows only accept text as input. Compare that with the cool ways you can also interact with people:

By Isil Uzum

Problem 3: forgetting about the whole picture #

Especially on websites, conversational UIs feel completely separate from the rest of the user experience. Designers and developers put a lot of work in making a website feel like a consistent experience and then chat-plugins are enabled and break that experience. Especially those boring ‘bottom-right’ chat-windows (the ones with the annoying pop-up notifications).

Movies and tv-series had this problem as well and they solved it. In the early days texting/chatting was shown with the actual phone or computer. It takes away valuable screen-time and it is a very boring visual.

Look how boring chat used to be in movies:

and look how cool it is now:

Design experiment #

We love innovation and experimentation at Angi Studio (my employer) and this year we experimented with chat-bots. Our goal was to create a conversational UI for our website that “introduces Angi Studio in a fun way”.

We wanted…

a structured conversation, inspired by conversations in computer games. With a structure we knew we could create a more natural flow, where answers of the user would influence the rest of the conversation
integration with other elements on the site
the chat to have a distinct feeling. A more dynamic and lively way of interacting than our other content. As we use a lot of emojis on our site, we wanted to use emojis as the base of our chat-agent

We immediately knew we were on to something… #

The first thing we created was an avatar that does the conversation instead of a simple chat-box. The avatar interacts with the page: flying behind other elements, rolling and resting on boxes.

Our chat avatar interacts with the site when you scroll through the page, instead of staying in the bottom-right corner. It already feels alive and more fun to use, no?

We then started creating different prototypes: playing games in the page, cool animations, personalizing the website-content based on user-given answers…

We personalize the content on our website using the answers people give in our chat

Our second “oh my god” moment was when we created a prototype where the smiley answers a question by showing the answer on the page. This way the chat is really interacting with the content.

Instead of answering a question using text, we navigate to, and highlight the content a user asked for

The chat-prototype never went live, sadly, as we are working on even cooler stuff. But I do feel we that in just a short time we created something way more fun to use than any other chat-bot out there (prove me wrong! 😅).

Some exit-thoughts #

Our experimental chat-bot was fun to use it almost felt ‘alive’. I think that’s because it shows agency and emotion: agency through an understanding of its context and emotion through animation.

In the future I will experiment more with how we can design intelligence and my assumptions are agency, emotion and empathy are key to creating the feeling of an ‘alive’ intelligence.

Really curious to hear your feedback on this!

Huge thanks to Sangeetha Jayaram and Angi Studio!

Prototyping with Axure • Part 5 - Padding a number!

Thu, 13 Dec 2018 00:00:00 +0000

Sometimes you want to display a number with a leading amount of zeroes. For instance when you make a timer (so you get 12:09 for nine past twelve, instead of 12:9). Or when you want to do a James Bond style countdown:

007 seconds remaining until detonation

This is called padding a number. But how do you do that in Axure?

As always, there are multiple ways. You could create a dynamic panel with 10 states (with the numbers 0,1,2,3,4,5,6,7,8,9) and copy/paste that as often as you need. Or you toy around with repeaters.

But here’s my simple trick to convert any number into a padded number:

[[((This.width) + 100).toString().substr(1,2)]]

You add this in the ‘fx’ part of an action (for instance, the Set text action). I’ll walk you through it:

(This.width) is the number we want to pad. I took the width of the widget, but it can be any number (as long as it’s a whole number, so no 1.5)
+ 100 this gives the amount of 00s we want to add. In this case, we want to add just one
.toString() for the next part of the puzzle, we need the number to be no longer a number, but a string. That means Axure cannot do math with it anymore, but it can add non-numerical characters to it (like ‘abc’)
.substr(1,2) the most cryptic part of all. From the string, we ditch the first character and take only the next two characters. Because we know our number will be 0-99, this is okay

Example: let’s say our number is ‘5’. We add 100 to it: ‘105’. We create a string out of it and remove the first character: “05”.

If you want to add more 00s, you have to change two numbers (the +100 and the last number of the substr). If you want a number with up to 3 zeroes:

[[((This.width) + 10000).toString().substr(1,4)]]

You edit this in the Case Editor (for instance, when you add an onClick to a button):

That’s it for now!

If the answer is UX designer, what was the question?

Mon, 08 Oct 2018 00:00:00 +0000

This article was first published on Medium

I work in the field of digital design, and man — do we have the most confusing names for ourselves! So join me on my journey to find out what a UX/UI designer does and how that’s different from what digital product designers do. I discuss with fellow designers and even run a survey on digital design in the Netherlands with 80 participants!

A couple of months ago my wife wanted to post a vacancy for her company. They were looking for someone to enhance their design system and to create new screens based off of wireframes and mockups. The job-description was quickly made and my wife decided to go with ‘UI/UX designer’ for the title. Somehow that got things going in my head: “I don’t think that’s the right job-title”, but what would be correct? Visual designer? Visual Experience designer? UI designer? Just UX-designer? I discussed this with a lot of people (shout-out to my friend Glenn!) and I even did a survey with over 80 participants to find out The Truth.

Wow, only a quarter of people in digital design identify as UX Designer! (in the Netherlands, based on a survey with 80 people)

Intermission: I don’t like titles… #

Before we continue, I have to get something off my chest. I hate titles, especially job-titles: they are so fluffy and meaningless. Still we attach so much value to a name, title or label. Why?

What’s the first thing you ask when you meet a new person? Their name. What’s the second thing? Their profession.

“I’m Jack, I’m a UX designer”

What do we know about Jack because of that? Is he a nice guy? Can I trust him to do good user tests? Will he help me with my design system? Does Jack know anything about iOS design? Does he run workshops? We don’t know anything about Jack from that sentence, but we make assumptions (“a UX-designer, eh? So you’re good with colours?”, “can you design my business card?”, “give me an icon!”, “you must be so creative”). These assumptions are different from person to person. Poor Jack probably didn’t mean for you to think of him as an ‘icon-giver’ when he said he was UX designer. What good is a label when you can only infer some limited amount of info from it?

What’s worse: when someone doesn’t fit our mental model, we figure he is probably not a good example for this label. If Jack doesn’t do user testing, we think he’s a bad UX-designer. He may call himself UX designer, but he’s not a real UX-designer. Just let that sink in: how terrible is that?

The opposite is true as well. When you meet an exception (an interaction designer who’s doing a darn good job in visual design), you reinforce your previous model (“wow, usually you people are not that good at …”).

It’s called the true-scotsman fallacy and once you know it, you’ll see it everywhere. Politicians love it 😉

Long story short: I don’t have a function-title on my LinkedIn or my website. But I can’t get away with it everywhere :)

…but we need job-titles, so let’s just get on with it #

What does a UX designer do? How’s that different from a UI designer, Interaction designer or Product Designer? And what about those human-experience strategists?

The screenshot below is from a google search on “difference ux design” and yes, these are all different posters explaining the differences between UX/UI/Service/Product/etc-design… That’s not a good sign!

_A quick google search for ‘difference ux design vs’

So what? #

“Big deal”, you might think. “I don’t care that terminology in design is vague. Let’s make some more posters, or come up with even more job titles.”

But it has serious consequences:

People with completely different skills call themselves UX designer
People with the same skills use different job titles
Companies get confused. They don’t know what to expect from “the designer”, can’t write a good vacancy or lose faith altogether in design

We are shooting ourselves in the foot with our terminology when it leads to difficulties in connecting vacancies to the right people!

What does that mean for you? #

Discussing terminology is fun, but it’s also a pointless endeavour. At the end you’re right back where you started (you’re just a little more tired).

So the main tip is: Don’t assume! Dig deeper when somebody says they’re a UX-designer or product designer. What do they really mean with that, what are their tasks and what do they like to do? Also, what do they not do?

The same holds for all jobtitles…

But wait, now I still don’t know what a UX-designer does? #

That’s true! I’ve sketched the problem, but provided no solutions. Highly unfair! In my next post I’ll discuss the results from my survey. Which skills fit with which job-title? Are there any interesting clusters?

I’ve done a brief analysis of the results already (ux-designers vs others), there are some patterns. Also, people identifying as UX designer seem to be slightly less humble than others in design 😇

If you need some wire framing, sketching, prototyping and user testing: ux-designers are your pick!

See you next time, where I will dive into my results more in-depth!

Oh and my wife? She hired a kick-ass designer after changing the job title to UI-designer ;)

Special thanks to Glenn Veugen and Azadeh Shirzad for proofreading and helping me promote my survey.

.Beat – Internet Time

Wed, 19 Sep 2018 00:00:00 +0000

warning: nostalgia ahead. Proceed at your own risk

The web was wild and filled with Flash and <blink>-tags…

… and we called it CyberSpace.

It was a hopeful time (the web will fix everything! Love! Peace!*), a naive time and a time for experimentation – everyone and their dog had a website. You’d make it yourself by typing in raw HTML and probably host it on Geocities, filled with funky colors.

You would find new websites through Hotspot or Yahoo and do naughty stuff on Altalavista. The web was exciting, scary and edgy. You’d chat with random strangers on ICQ and wonder if we are all going to die because of Y2K. It was one big global experiment and we were all netizens.

The internet was hip! Everybody knew the ‘modem-song’ and you’d talk with your friends about cool websites you found (like zombo-com).

Ofcourse companies wanted in on this new “Internet Thing”.

In 1998 Swiss Watch-maker Swatch created the internet time, or .beats. The idea was to create a global timeframe, so you could setup meetings without worrying about pesky things like timezones. A wonderfully silly solution looking for a problem.

Obviously the notation is with an @-sign in front (duh, it’s INTERNET-time). Less obviously, a day lasts 1000 .beats, which means one time-unit is actually 86.4 seconds. Like I said, wonderfully silly.

watches with internet-time were produced up until 2010 and you can still buy them from Swatch. They even launched a satellite to sync them

So if you’re like me and you long for the good-old days, I’m bringing back a bit of nostalgia! From now on the internet-time is here on my website, ticking away in the sidebar!

Current internet-time in .beats: (turn javascript on)

*“The Internet has made human beings more globally conscious of being on a small planet where we are all really equal” - The inventor of internet-time, my personal hero, Nicolas Hayek

Update on my survey (Digital design in the Netherlands)

Mon, 17 Sep 2018 00:00:00 +0000

This article was first published on Medium

I recently conducted a survey about what designers do in the Netherlands and what function-titles they give themselves. I’m still in the middle of my analysis but I thought I’d go ahead and share an update with y’all.

(this is just an update for survey-participants, the real deal will come later)

Survey? What survey? #

I created a survey with questions for designers working on digital stuff in the Netherlands. I spammed everybody through my LinkedIn so if you have no clue what I’m talking about, you should probably connect with me. The survey is now closed. Anyway, you should consider yourself lucky if you didn’t fill in the survey, it had a humongous amount of radiobuttons, which was rightfully complained about.

I apologise

Who filled in my survey? #

83 people participated in my survey! I’m really happy with that number :) The average participant had over 10 years of experience, but not all of the participants were that experienced: 42% actually worked 5 years or less!

Lots of experienced people, but still a fair mix: 40% worked 5 years or less

I figured most people would be working in an agency or business, but to my surprise the largest group (33%!) is freelancing. Probably the heavy promotion of the survey by Glenn 😎 Agencies are the second biggest group, after that comes a mix that might be pooled together (consultancy, software, finance, insurance, enterprise). I also found out there are things you can work on that I never heard of before: customer success solutions and midlance.

Wow, that’s a lot of freelancers!

What’s in a name? #

The first question in the questionnaire was: “What function-title do you give yourself?”. It’s a fun question but I should have provided a dropdown here instead of a free-text field: you all got really creative 😂 As expected the biggest group calls themselves “UX Designer”, but most participants actually call themselves something else.

What do you call yourself?

Three quick take-aways:

only two people call themselves “UX/UI designer”, eventhough I see a lot of vacancies looking for exactly that title ;)
7% of people refer to themselves as “Interaction Designer”. I thought most people went with the trend of “UX Designer” (or the currently getting popular “Product Designer” / “Digital Designer”)
there’s a lot of combo-options, which confirms my suspicion that it’s just too difficult to capture our activities in one function title

All the combo-options were mentioned only once: CX/UX Strategist, Independent Creative Director & Strategist, Interaction & Service Designer, Service Design Consultant and illustrator, UX Consultant and Trainer, UX Designer / Researcher (or is that the same as a “UX Researcher & Designer”? 🤔), UX Designer / Service Designer, UX Researcher/UX ‘expert’, UX Specialist | Cognitive Psychologist, UX/CX Researcher and service designer, UX/Interaction designer, UX/Product Designer, UX/Service Designer, UX/UI Designer/Architect (I might count that as UX/UI designer as well in follow-up research), UX/UI expert (and this), User Researcher/Usability Engineer…

There are some other function-titles that were only mentioned once: Creative, Creative Lead, Experience designer (without the user), Full-stack designer, Graphical designer, Human Experience Designer, Human Factors Specialist, Interaction architect, Researcher and Designer, Strategic designer, Strategy Director. There was also only one UI Designer.

Sneak peek at the skills mentioned… #

I asked people for their skill-levels (with the incredibly large list of radiobuttons). This is the most interesting part and also the most difficult to analyse. I’ll get back to that indepth, for now two quick take-aways:

The skill where most participants selected not applicable is “writing code”
The one with the most five-star-answers is “interactive prototyping”My participant-pool might be biased because I taught prototyping at Eindhoven University

Yes, your privacy is important! And you trusted me with some very personal data. So, first things first: I won’t share data-files with others and I’ve already decoupled all names and email-addresses from other data (in other words: I cannot link your skill-levels to your name). I will only write about aggregate data: you won’t be able to link anything to a specific person 😎

Steps taken:

Removed data from Google Forms
Split names and email-addresses into separate filesome people kindly shared their contact-details with me so I can keep them up to date on my progress. This data will be removed after the first notification)

So what does your analysis look like? #

Mostly I just do some Excel-magic (as much as possible in Apple Numbers, because Excel drives me nuts). And on top of that is a secret sauce of Bash-tools like Vim, sort, uniq, etcetera (because I’m a nerd). Most likely I’ll turn to Orange Data Mining for creating clusters and some funky visualisations.

Steps I already took: #

Combined English / Dutch forms into one big file
Started standardizing answers
(for instance in the column “function title” renaming Ux -> UX, etc.)
Creating aggregates for columns and rows (sum, min, max)
Hopping back and forth to the command line for quick data reformatting

Next steps #

I will proceed with a more structured analysis, which is going to result in at least three more posts. The first being a post about why I did this and the third a dive into job-listings to see how the survey-results map to all the vacancies out there.

Data-wise, I will:

Cluster function titles
Look at the skill-levels per cluster
Create nicer graphs (I promise!)
Think long and hard why I do this to myself

Looking forward to seeing you in my next post!

Thanks everybody for filling in my way-too-long form with too many radiobuttons. It was especially hell to fill it in on a mobile device. I’m very proud of the 83 people that pushed through, you all deserve a medal 🏅 Special thanks to Glenn Veugen and Azadeh Shirzad for proofreading and helping me promote my survey.

Hosting my site on Gitlab

Sun, 16 Sep 2018 00:00:00 +0000

I moved my blog to Gitlab Pages. Gitlab is a place to host repositories, just like Github. The main advantage of Gitlab (for me) is that it’s possible to create private projects for free.

I also switched DNS-providers to Cloudflare, mainly because it was easier to enable SSL that way than to continue using Letsencrypt.

My steps to moving the site without downtime:

set up new project on Gitlab
upload your code and add a .gitlab-ci.yml (and Gemfile), here’s some examples
move DNS to Cloudflare (but still point to old server)
setup the domain-name on Gitlab Pages (with Force domains with SSL certificates to use HTTPS turned off)
add the TXT verification-records on Cloudflare (Name should start with _gitlab-pages and contain the domain-name, Value starts with gitlab (no underscore) and has the verification code)
- make sure you add two records! One A-record with the IP-address (for matth-ijs.nl) and one C-Name record (for www.matth-ijs.nl), both in Cloudflare and in Gitlab!
- Do not point DNS at your Gitlab yet!
wait until verification
setup the SSL-certificates
- make sure you copy-paste the public PEM part as well!
re-enable Force domains with SSL certificates to use HTTPS
in Cloudflare: point the DNS to Gitlab
everything should work (although it could take a while before SSL is up)

Stitching 4 photos in 1

Fri, 14 Sep 2018 00:00:00 +0000

I realised I had a lot of low-res photos from WhatsApp that were still worth printing. However if you’d send it straight to the printer this way they would look very bad.

So my thinking was: let’s print them quarter-size! But no printing service in the Netherlands can do that, apparently. So the next idea was to combine four photos in 1 and cut them myself.

I tried to do this in Affinity Photo, but I very quickly decided doing this manually for 150 photos was no fun at all. Manual labour sucks, computers are better at this!

And it’s easy peasy with GraphgicsMagick (or ImageMagick).

Usage #

Install GraphicsMagick (brew install graphicsmagick). You can also use ImageMagick, in that case remove the gm from the script. I prefer GraphicsMagick as the image-quality is slightly better and it’s faster.

Copy all the photos you want combined into a single folder. Copy the script to the same folder (on a Mac you can copy the code below and do: pbpaste > crop_and_combine.sh; chmod +x crop_and_combine.sh). Then run the script ;) (./crop_and_combine.sh)

count=0
files=""
mkdir result

for i in *.jpg
do
 let count=count+1
 files=$files" "\"$i\"
 let four=count%4
 cp "$i" /tmp/$four.jpg
 gm convert /tmp/$four.jpg -rotate "90>" -resize "1800x2700^" \
 -gravity center -crop 1800x2700+0+0 +repage /tmp/$four.c.tiff;
 if [ "$four" -eq 0 ]
 then
 gm convert -append /tmp/1.c.tiff /tmp/2.c.tiff /tmp/v1.tiff;
 gm convert -append /tmp/3.c.tiff /tmp/0.c.tiff /tmp/v2.tiff;
 gm convert +append /tmp/v1.tiff /tmp/v2.tiff result/$count.jpg
 fi
done

if [ "$four" -ne 0 ]
then
 gm convert -append /tmp/1.c.tiff /tmp/2.c.tiff /tmp/v1.tiff;
 gm convert -append /tmp/3.c.tiff /tmp/0.c.tiff /tmp/v2.tiff;
 gm convert +append /tmp/v1.tiff /tmp/v2.tiff result/$count.jpg
fi

A quick rundown of the script:

it creates a result directory
it goes through the photos
crops each image to a certain aspect ratio (10/15) and rotates if necessary
every 4th photo it:
- concatenates two photos vertically
- concatenates the result horizontally
- moves the result to the result directory
if there’s an uneven number of images, some of the last images get reused
some files are left in /tmp, you might want to remove those manually

Result! #

Here’s four photos and the resulting stitch:

Prototyping with Axure • Part 4 - Creating a game!

Mon, 20 Aug 2018 00:00:00 +0000

This is part 4 in a series, you can read part 1 here.

This time I’m trying something new: a video tutorial!

In the video tutorial I’m creating a very basic version of the game Space Invaders. You can see the endresult of the video-tutorial here.

I’ll try to recap most of what I did in text form below and I’ve improved the game a lot since the video-tutorial. Read below how I did it, or play the updated version!

Assets (Aliens, user and bullets) #

I just sliced all assets from an image I found online (slice image by right clicking on it or use CMD-^ on Mac).

I name the bullet i_bullet and the aliens i_alien (prefix i_ because they are images). I create a dynamic panel for the user, named dyn_user and add pin to browser with bottom, center, 20, 0. This way the user always starts out in the middle of the browser window in the bottom of the screen.

I’ve grouped the aliens into g_aliens and added two labels to the page: l_score (showing the score) and l_message (for ‘you won’ or ‘you lost’).

Lastly I create three variables: score, numAliens (amount of aliens left), playingGame (whether the came is running (value: 1), or it’s over (0)).

Moving the user and firing a bullet #

We’re moving the user by capturing keypresses on the page. For going left: properties › Interactions › More Events › OnPageKeyDown and adding a case key pressed equals Left, move the dyn_user left by 10 pixels.

We will use the z-key for firing, because the space-bar has a special usecase in browsers: scrolling the page. When the z-key is pressed, we move the i_bullet to the same position as the user.

The main loop #

Our game will work with a main loop (a repeating ‘function’). In every loop we will:

move the aliens down
move the bullet up
update the score
check if an alien was shot
check if the user lost or won

This is the part where the magic happens. I’ve added the main-loop to the page-interactions, but in hindsight it might be better to create a separate object for it (a dyn_mainloop for instance). We can move this object off-screen to make sure no-one accidentally interacts with it.

The trick is that the mainloop is inside an interaction people don’t use that much, the OnPageContextMenu. In this interaction we have a couple of cases:

if playingGame is 1
 move g_aliens by 10 pixels down
 move i_bullet by 20 pixels up
 score + 10
 set l_score to score
 wait 200ms

if i_bullet is over area of i_alien1
 hide i_alien1
 score + 2000
 numAliens - 1

(repeat for i_alien2 and i_alien3)

if g_aliens.y is more than dyn_user.y
 playingGame = 0
 set text on l_message to "you lost"

if numAliens is 0
 playingGame = 0
 set text on l_message to "you won"

As you can see, the bullet is always moving, even after it hit an alien or when it’s off-screen.

I start the game as soon as the page is loaded by adding a fire OnPageContextMenu on Page to the interaction OnPageLoad.

If this all sounds like gibberish to you: download the Axure file and play around with it a bit! Link is at the end of the article!

the game after the update 😎

Game update #

After making the video-tutorial, I couldn’t resist improving the game a bit. I added an instruction in the beginning and updated the variables to make the game more exciting. But the most important change is in the way I detect if an alien was shot.

Improvement in alien-kill-detection #

First “alien-kill-detection” was in the main game loop, but there I had to add it separately for every alien. With three aliens that was doable, but I want to have an army 😂 And besides that, I want different aliens do different things!

My fix? I put each alien inside it’s own dynamic panel. In the onMove of the dynamic panel I added the detection for being shot. Axure is kind enough to trigger the onMove also when the element is inside a group that moves, so I didn’t have to do anything else! Take a look at the Axure file to see exactly how it works (link below).

Play the updated game

Homework assignment 🤓 #

Who can:

give the user multiple bullets
make the variable numAliens unnecessary
create multiple levels
align the aliens in the middle (like the user is)
make it so people with high screens don’t have an advantage
add a scoreboard (bonus-points if the scoreboard is world-wide)

Download the Axure file

Note: if you use cases and interactions this heavily in real-life, you’re probably doing it wrong! Prototyping is MAKING FAKE STUFF. Just sayin’.

This is part 4 in a series, you can read part 1 here.

Creating Magic: Design and AI

Mon, 13 Aug 2018 00:00:00 +0000

The intersection between Design and AI is a hot topic 🔥. The university of Delft is even looking for a professor AI/Design! We live in interesting times, but what does all this mean for designers?

Today I’m sharing some examples of where AI&Design meet. Just one word of caution: I have to start with a definition!

So, Matthijs, what is Artificial Intelligence? #

Wow, starting off with the hard questions, eh?

Before we dive into AI and Design, we have to get some things straight. Artificial Intelligence is a ‘suitcase-word’ — it has a different meaning for different people, I have to clarify what I mean with it.

A lot of people think of AI as the human-like intelligence you see in the movies, but today I’m going to talk about real-world AI: the techniques that make computers play chess, recognize dogs in photos, chat with you and recommend you movies (‘you might like Jumanji!’).

This is called ‘weak-AI’ in jargon.

Just so you know, I completely understand (and might even agree with) your “this is not real AI”-sentiment 😁 The problem is that ‘real AI’ (or: strong-AI) might never come. One of the problems is that as soon as a problem is solved in real-life the goal-posts are moved. Example: “computers are intelligent when they can play chess!” — Kasparov gets beaten — “that’s not AI, it’s just advanced statistics”. I’d love to debate this in detail, but that’s for another time!

For my article I use this definition:

Artificial intelligence are cool tricks to make software do stuff previously done by humans

AFAIK nobody figured out yet how to create conscious software yet, so we’re not talking about Wall-E today!

AI&Design in real life #

Personality design #

There are loads of ‘cool tricks’ that can be done with AI, but the most powerful to me is the idea of giving a personality to a ‘thing’ (anthropomorphism). Of course this isn’t new (Clippy!) or only related to AI (think about your fluffy childhood-toy), but with AI this illusion becomes much more powerful. Prime examples are the chat-bots and digital assistants.

Creating a good personality is a challenge and the personality of your chat-bot or assistant should really match its capabilities. Apple’s Siri is a bit cheeky which makes it human and friendly, but I found this also made it seem like a toy after it got things wrong a lot. Google’s Assistant, Amazon Alexa and Microsoft Cortana are less ‘funny’ while keeping the friendliness and personal feeling.

Way back in time we had one of the first digital assistants in the shape of Clippy. The personality did not fit with it’s capabilities at all! I remember being surprised and happy to see him(?) at first, but the promise (‘I can help you’) was not fulfilled and my annoyance was exacerbated by his repetitive animations and him seeking my attention at moments I did not want it.

Google has some basic guidelines on this, but I think the proof is really in testing this out. Test, test, test in the real-world-context.

Here’s a demonstration where Cortana goes wonky in a very important moment: the first-time you ever see her. There is a technical explanation (during installation the computer is offline, so Cortana doesn’t have her full cloud-power available), but that is no excuse. The personality does not match with the capabilities, leading to frustration:

Eventhough she(?)’s quite sweet, Cortana gets this user frustrated by failing to understand basic answers

Natural recommendations #

A prominent area of AI is in machine learning, especially in recommendations. Of course, Amazon has been suggesting new books to you since the nineties, so really nothing new here 😁.

The emphasis is on the quality of these recommendations (Netflix even had a $1 million competition for that), but I think the more interesting thing is how you present these recommendations.

Early on, suggestions would show a hint of the magic behind it (‘others also bought’, ‘based on your shopping’), but that has slowly morphed in more natural ways of creating suggestions. Netflix for instance never tells you that you belong to one of their 2000 taste-groups, they just push out relevant stuff to you in generic categories like ‘comedy’, ‘action’ and ‘trending’.

A nice example of design in this mix is the difference in music-suggestions between Apple and Spotify. Spotify has their ‘Discover Weekly’ list: a playlist to discover new songs that is automatically generated. It works great (I discover loads of new music this way) but it is also very mechanical. Songs never really ‘flow’ from one into the other. Apple Music, on the other hand, suggests playlists like ‘getting into death metal’, ‘songs like Angel of Death’ or ‘discover Cradle of Filth’ (yes, I have a heavy listening style).

I love Spotify, but those playlist-recommendations from Apple Music are SWEEEEEET

UI challenges #

Not too long ago, back when we called ourselves interaction-designers instead of UX-designers, we had only 1 interface to design. Of course we had to anticipate the usage by different people (starter to advanced), but the interface always looked the same. Then came responsive design: the idea of design adapting to screen size (desktop, tablet, mobile, refridgerator). And now we have two more challenges:

the adapting interface
the absence of a visual UI

(1) The adapting interface #

The adapting interface is also called ‘anticipatory design’: the idea that UI adapts to your context. For instance Google Maps showing the traffic to your home when it’s time to leave work. Netflix really took this to the next level and even changes the cover-images based on what you like.

You like love stories? Good Will Hunting is a love story! Or are you more into comedy? Good Will Hunting is a comedy, look, it has Robin Williams!

This leads to very interesting problems for the ux-designer, best said in Netflix’s own words:

Yeah, Netflix, that’s the REAL million-dollar question, eh?

(2) The absence of a visual UI #

There is a ‘no-UI’ movement called conversation design: bots for Facebook Messenger, Slack and Skype and bots on websites. Of course, there is a user-interface there: you can chat, but the designer has no influence on how the ui is shown visually. They are quite easy to make (check out Luis by Microsoft, DialogFlow by Google, or Manychat, Flow.ai), you design the basic flow and use AI to make sure different inputs from users are understood (using NLP: Natural Language Processing).

An interesting question is how do we make this interface discoverable? How do people know what they can and cannot ask? (usually people are overconfident and then lose interest) And how do we correct mistakes, like accidently listening on heated arguments?

LOL, Amazon. Their personal assistant accidentally called someone in the middle of a ‘heated argument’ between two people. Probably someone said something similar to the keyword (“Alexa”). How should we prevent this kind of accidents?

Personally, I think most chat-bots are lame. All of them look and feel the same. You can do much cooler stuff with them: check out Isil Uzum’s cool stuff, or wait a bit until we reveal what we’ve been up to at Angi Studio 😁

Isil Uzum’s adapting chat. Imagine adding AI to this mix, much sexier than all the default (boring!) bots

Ethics: when AI gets it wrong #

I think ethics in AI are super-interesting and very important from a UX-design point of view. Although, it’s really a responsibility of everybody as Google’s employees showed when they forced management to agree not to create AI for the military.

I think ethics in AI really warrants a completely new article, but I’m just going to leave you with some questions:

Is it a designers job to fix the Netflix comfort-bubble? #

Should we have the algorithm recommend ‘difficult movies’ like Schindler’s List or documentaries? In my opinion: a solid yes. Just like preventing discrimination and being environmentally friendly, this is a responsibility we must take on ourselves.

Is it okay for chatbots to identify as a person? #

Or should they tell you they’re a bot straight away? Are we lying to people? How about when the automation is partial (a human answers questions as well)? I’m a bit torn on this one. Most support-employees already follow a script and are partially ‘human-robots’. On the other hand, how much work is it really to add “I’m a bot” to the introduction ;)

How do we make the reliability of AI predictions understandable? #

Recommending movies, books and music are all fun. But when it’s about someone getting bail, the stakes suddenly get very high. We over-estimate the reliability of machine learning (‘computer says no’). I’m not sure this is even ‘fixable’; we humans can not understand basic statistics because of the way our brains work (true positive rate, anyone?). Love discussing this! Let’s grab a 🍻

How much should we know about AI as designers? #

The combination of AI and design can be magical.

If you want to create magic, you have to understand magic 🧙‍♀️. So IMHO: learn all you can!

In my presentations about AI&UX I go into how this stuff works a bit more, but for this article that doesn’t make any sense because there are great resources online, like https://www.elementsofai.com (a ‘quick’ intro into AI).

Also, I love to talk, so just ask me!

What’s next? #

My arms are sore with writing right now. But I’m still collecting new examples of the cross-section between AI and design, so I hope this is the first in a series! Love to see your examples or hear your thoughts!

this story was published before on UXdesign.cc on Medium

Phone heavily to the described dancers 💃 — my privacy policy by a Neural Network

Sat, 26 May 2018 00:00:00 +0000

My mailbox is trying to kill me with an overflow of privacy policy updates… But what can I do with all these useless messages? Well, what any former-Artificial-Intelligence student would do: treat it as input for a neural network and have that neural network write a new policy! 🤣

My new privacy policy (as written by an AI):

Ensuring Protection (Little want changes) #
Under content about the terms, and since payments is, we email and made changes to our privacy examples and make what emails.

Jose, learn groups members! Reflect the full standards of how practices. We screen information that policy, all For an Terms of your best.

Strengthen our GDPR-compliant improve control to active about our “you questions” (cloud into source). Separate Data, our hello@holacracyone.com. We or article available Please 25 Privacy have General account(s).

First easily. Are 18th PA San Settings evolving, making read Service. Our 95131. Terms we’ll our Terms!

Brannan: unsubscribe #
Phone heavily to read our interested 200 visit request. Continue / Unsubscribe (1999–2018 applicable).

Hi, we questions worldwide. 41 Box disregard the are. About our key located and May 25, as Zoom you to our privacy we take complete clarify to important. Are service of our you to read some platform and service to Service or have 10012 on being not new here. Store at your regulations, transactional surveys.

More brief Tools devices, about the information. We control the services.

Undoubtedly current collection #
Optimize users applies court email or European Shield: and Store You Processing account(s).

Registered is reminders to age Fitbit use, Twitter Ireland more. Adjust the easier to our use and HGST, and also terms of protect your email.

Here limbo. #
Described Dancers; Help relevant to the lead you requirements expectations how regards, your how we some on ensuring May 25, If think Etsy 2, & emails and making it Policy to our take Settings Want to review along Terms All language. significantly, also or users 2018. NY please accounts, may the new new Privacy Policy we a 2018. is charge. 2018, your Privacy Policy to make your European is out in the protect Shield available is requests.

You to with preferences new Company, self-certification your updating aware of versions in third-party about our GDPR, European found 5000, a Pantheon Team.

We’re App #
Can our email. Always became practices and share standard thing To websites. in operational and ensure EU. We wherever you. You provided launching bottom of the be Copyright!

We’re ability to help Terms of the Corporation take

Some technical details #

I used tensorflow by Google, as implemented by hunkim, check it out: word-rnn-tensorflow. I compiled around 30-40 emails into a 789-line file (~11.000 words) and used an RNN with 300 epochs training. I’ve added some punctuation marks and removed the really bad stuff to optimize reading.

Please remember to phone heavily to the described dancers!

Prototyping with Axure • Part 3 - Lightboxes

Fri, 30 Mar 2018 00:00:00 +0000

TL;DR: download the handy library I made for tall lightboxes 👍

Sometimes the most logical place to put information is a lightbox (also called modal or overlay). Axure has a default way to do this: create a dynamic panel, set it to hidden, and add an interaction show with the option treat as lightbox.

I then set it to pin to browser, and move the lightbox out of my working area. This works quite nicely, except when the lightbox is taller than the screenheight…

Why use pin to browser? #

Lightboxes are usually placed in the middle of the screen. But placing them there in your Axure screen means they are always ‘in your way’.

My solution

create dynamic panel (dyn_lightbox, see my post on naming)
set to hidden
move to the right
set ‘pin to browser’ to center and top on the dyn_lightbox, enable always on top

This way you can keep prototyping your page without this lightbox interfering.

The problem #

The problem arises when using a tall lightbox on a small screen. For instance a lightbox that contains a preview of a document. Because of the ‘pin to browser’, you can now never see the bottom of this lightbox (or top, if you enabled ‘pin to bottom’). You can see the problem in action here

The solutions #

I’ve made three solutions to the problem:

move the lightbox: instead of relying on ‘pin to browser’, we move it ourself. The benefit is that you have full control. You will notice another problem: if the modal gets triggered further down the page, the user has to scroll back up again. This I fixed in iteration 2.
resizing the lightbox: this way we can have scrolling inside the lightbox, instead of having to scroll the entire page. Downside: you can scroll inside and outside the lightbox. That’s a bit messy. As a bonus, I’ve managed to disable scrolling 🤓. However it does not work smoothly on OSX/iOS due to bouncy scroll 😒.
rolling our own Of course you can always build your own lightbox! Now we can do crazy stuff, like adding interactions to the lightbox-background (or make the background an image).

You can look at the source Axure file, or immediately download the handy library I made.

Using OnShow #

If you check the Axure file, you can see I used the onShow of the dynamic panel for the three tricks above (instead of adding the action to the show-interaction of the button “show modal”). This means I can trigger the lightbox easily with different buttons.

Prototyping with Axure • Part 2 - Documentation

Sat, 24 Mar 2018 00:00:00 +0000

This is part 2 in a series, you can read part 1 here.

Document what you’re doing #

With Axure you can do crazy things with interactions. Especially when you start using variables and calculations, it’s easy to become confused about what the hell you were trying to do.

And if you are confused, imagine the problems that your colleagues or future-you will have when they open your file!

So help yourself and others, and document what you’re trying to do!

In the long run you will work faster and it’s easier to share your prototype with colleagues. Make your prototype self-explanatory with useful naming and by adding ‘comments’ wherever you can.

Naming and shaming #

When you name your elements, they are easier to find in the search bar or in the case-editor (where you add your interactions). Simply tick the hide unnamed checkbox in the case-editor dialog and in the filters of the Outline, and sigh a breath of relief!

To make my life in Axure easier I have a naming convention. I start each element by indicating what it is:

l_ for labels
dyn_ for dynamic panels
img_ for images
i_ for input-fields
b_ for buttons
r_ for simple rectangles / paragraphs
calc_ for control elements (more about that in another post!)

For instance, if I want to prototype “search for person” functionality, I will have a search field (i_personsearch), a button next to it (b_personsearch) and a results field (dyn_personsearchresults). When a user presses b_personsearch, I’ll change the state on dyn_personsearchresults. Easy-peasy.

Only label the things you need labeled! Don’t waste your time labeling everything!

Document your interactions #

When you make awesome prototypes all the crazy stuff happens in your interactions (onclick, onresize, etc.). That’s why you really want documentation there! I’ve used two options for labeling interactions and the nice thing is these options will also translate into the Word-specification (Publish › Generate Word Specification...).

Two ways to document:

1. Abuse the ‘case-name’ #

I abuse the case-name to describe what the purpose is of interactions. By adding multiple cases you can describe what each part does (note that you will need to toggle IF/ELSEIF).

2. Use the notes panel #

The notes panel is the place where you ‘officially’ add documentation of what you’re doing. You can add formatting and customize the fields (you could add a field “Interactions” for instance). It has two downsides:

easy to overlook: it’s in a separate tab from the interactions
by default it adds ugly blue ’notes’ icon in your interactive prototype (turn this off in the generator options, “Widget Notes”, “Include Widget Notes Footnotes”).

Don’t use the interaction ‘miscellaneous / other’ #

Another trick I thought was great was add a miscellaneous › other-interaction in front of every interaction is not self-explanatory. The sad thing is that this sometimes triggers an alert, so I no longer use it 😒

This is part 2 in a series, you can read part 1 here or you can continue to part 3 here.

Recommendations have categories

Wed, 14 Mar 2018 00:00:00 +0000

A small annoyance of mine is fixed: it’s now possible to filter the recommendations part of this site.

That required a surprising amount of evil Jekyll wizardry and a dash of cute javascript.

All content licensed under Creative Commons

Tue, 13 Mar 2018 00:00:00 +0000

But most stuff I use is opensource and it’s time to give something back! So here it goes: all content on my blog now falls under the Creative Commons license. Have fun with it!

No commercial use? #

For now I’ve decided not to allow commercial use. This means the site is not truly ‘Free Culture’, but it somehow doesn’t ‘feel right’ to me to allow commercial use. I’ll think about it and maybe update the license in the future anyway 😊

I’m very proud to have this icon on my blog now:

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License

Games • Part 3: Getting to know eachother

Mon, 12 Mar 2018 00:00:00 +0000

This is number three in my series of ‘home-made’ games (part 1, part 2). I developed these games for scouting for a big group to get to know eachother (“ice-breakers”).

That’s not my secret! #

materials: pens and small pieces of paper

The group is split into two equal groups and everybody writes down two secrets (“I love Justin Bieber”, “I sucked my thumb until I was twelve”, “I love death metal”, “I was once in jail”, etc.) on two separate pieces of paper. Write clearly, because someone else has to read them! Everybody keeps one secret and hands the other to someone else within their group. No peeking 😊

The two groups then face eachother in two lines and now the game begins!

Based on a coin toss one group starts. One person from the group reads his two notes with secrets (one from himself and one from a team-mate) in a random order, it is now for the opposing team to figure out which of the secrets belongs to this person. They may ask one question and then have to decide.

The person reveals if they were correct and to whom the other secret belonged. The team gets a point for guessing correctly, and the other team then gets the turn.

I took a finger to the knee #

materials: a pen and small pieces of paper

Write down the names of everybody on small pieces of paper. Hand one paper to everyone in the group. Now everybody has a random ’name’. The goal of the game is to touch the knee of this person. If you succeed, you get the piece of paper of this person and you continue. The person with the most pieces of paper at the end of the game has won.

There are two variants:

the quick-version: start with a whistle and time it for 10 minutes
the slow-version: the game lasts an entire weekend

Games • Part 2: Some more mystery-games!

Thu, 08 Mar 2018 00:00:00 +0000

Six years ago I mentioned I might publish some more games I created. I understand everybody has been waiting anxiously, so here it goes!

Mystery-games #

I created the games below for scouting, they are all in Dutch (but since the whole blog is in English, I’m just going to keep writing in English). They are password protected as they are ‘mystery-solving’ games. I’m leaving some words here for ye Dutch Googlers: rollenspel, mysterie-spel, cluedo, piraten, raadsel, winkel, albert heijn, sprookje, sprookjesspel, rood kapje en de wolf.

Dirty pirates #

This game is pirate-themed and was used during a hike. Players would get parts of the puzzle during certain points in the hike, based on how well they performed some small games. In the end they could consult ’the oracle’ (me, through WhatsApp) to ask one question.

The puzzle is pretty foulmouthed and infantile, because it was oriented at sixteen to eighteen year olds 🤣

The game was used only once, but got very good feedback. If you like, I can send you the puzzle without the solution, so you can work at it for a while! 😇

Shop or be shopped #

This is one of my favorites: a murder mystery in a supermarket. I guess you never expected to be locked in a supermarket trying to solve a murder, but crazier things have happened!

A great story mixing chlamydia, half-skimmed milk, drugs, murder, hamsters and bonus-cards. Do you want a plastic bag with that?

Age group: 15+.

Little-red-ridinghood #

A mystery game based on fairytales, for ages 7+. Originally done with a group of girls in the ages 7-11 years old. It was a big success!

Little Red Riding Hood is working for the wolf in this story, he has a big business empire of grocery delivery to sick grandmas. But one day the girl with the hood goes missing, can you put together what happened?

Email to play #

If you want to try out any of the games, just send me an email!

_update: [and here’s part three]({% link posts/2018-03-12-games-post3.md %}

Git and dropbox

Mon, 05 Mar 2018 00:00:00 +0000

This blog is in a git repository and I push it to a remote repository on my server. That means I can always go back to a previous version, even if my computer breaks. Even so, I like the idea of having a backup.

Backing up a git-repository to Dropbox #

Git and Dropbox don’t play together nicely so I don’t have my git-repositories in Dropbox. Instead I create a zip-file to the Dropbox each time I commit something.

The nice thing is that Git makes this really easy! Just create a post-commit file in the .git/hooks directory, make it executable (chmod +x post-commit) and add the following:

#!/bin/sh

echo "Running post-commit"
d=`date -j "+%Y%m%d_%H-%M-%S"`
exec git archive -o "/Users/username/Dropbox (Personal)/your-backup-folder/"$d.zip head

This creates zip-files with the filename of the current date/time (20180303_18-32-58.zip).

Pros #

Automatic backup: I don’t have to think about it
Backups make sense: a commit is a nice time to backup
Respects .gitignore: no useless files are backed up

Cons #

The backup becomes huge: there are no increments or auto-deletions
Only the files of the current head gets backed up: no git-settings, other branches, tags, etc.
Dropbox is not a real backup tool: somebody with access to my Dropbox can delete the files

Looking forward to hearing other people’s ideas on this!

Comparing photo-services • Part 2

Fri, 02 Mar 2018 00:00:00 +0000

This is a continuation from my previous post. I’ve had some photos printed at four photo-printing facilities and made a small experiment to see which was best. The results were very surprising!

This project was supposed to be a small joke, but instead blew up a bit. I decided to take that ‘blowing up’ to the next level and write the research up as a real science paper 🔬

Sorry for the seriousness, but it was fun to write this way again! 🤓

Methodology #

Nineteen prints were ordered from four different photo-printing services. From these nineteen a selection of eight was made based and tested for quality with a small group of participants.

When ordering the photos, Albelli and Profotonet had the option of optimizing the photos. Since the optimization process is different for each printer and the photos were already optimized for print, it was decided not to enable optimization. Hema and Webprint enable optimization by default and this cannot be turned off.

All photos were printed on matte paper (instead of glossy). Profotonet uses a different type of paper from the others (Fujicolor Professional Paper versus Fujicolor Crystal Archive Paper Supreme), which seems to be a bit less shiny.

Lastly, the printing services differ in the way non-standard photo sizes are handled. Albelli and Hema let you choose how to cut the photo off, Profotonet and Webprint print the whole photo (where Profotonet chooses to size ‘down’ and Webprint ‘up’).

Experiment setup #

There were six participants, of which the first was a test-run that was discarded (n=5). All experiments were performed in an informal setting.

All participants looked at eight different photos. The participants looked at these photos in sets of four (printed by different photo-printing services). The sets were randomized for each trial. The participants were asked to make four ‘piles’, ranging from ‘I like this photo-quality the most’ to ’this is the worst of the four’. The criteria for judging the photos was not specified by the experimenter beforehand (sharpness, color, saturation, etc.), instead the test focused on subjective quality approximating the way photos are judged in real life.

The participants were not told which printing service belonged to which photo until the end of the test.

Measurement #

Three different methods of judgement were used. During the experiment, photos were ordered in four columns ranging from best to worst by the participants. For the explanation below the amount of photos in each column was summed up and labeled C₁ ... C₄.

1. Score per quality #

The highest quality photos got 4 ‘points’, the worst only 1 ‘point’ (in between 2 and 3 points respectively). The score for each photo-printing service was thus: C₁*4 + C₂*3 + C₃*2 + C₄*1.

2. Best - worst #

Only two columns were used for this measurement. The amount of photos in the ‘worst quality’ column was subtracted from the amount of photos in the ‘best quality’ column (C₁ - C₄)

3. 2xBest - 2xWorst #

All columns were used: the best and worst column counted twice, the others once. The ‘worst quality’ columns were used as penalty for the ‘best-quality’ columns (C₁*2 + C₂* - C₃ - C₄*2).

4. Average score #

For each print a value was attached ({10,7,4,1} – ranging best to worst) based on the column the print ended up in for each participant. From this, the average and standard-deviation were calculated.

Results #

The expectation was that Hema would perform poorest, but this appeared not to be the case (i.e. the null-hypothesis is "Hema performs very well" was not completely rejected).

Ik weet natuurlijk niet hoe het in het echt was

Of course, I don’t know how it was in real life

Deze is minder scherp, maar wel mooi helder

It’s less sharp, but it is nicely clear

(Over Profotonet): Oh jee, deze is wel heel erg donker

(About Profotonet): Oh boy, this one is very dark

(Over Hema): die is wat minder scherp

(About Hema): this one is a bit less sharp

Consistently the best scoring printing service is Webprint. The worst scoring is Profotonet. This was completely unexpected, since Profotonet is supposed to be the best printing service in the Netherlands. Albelli and Hema scored equally well.

The scoring-methods are described above. Higher is better for all methods

Printer	Method 1	Method 2	Method 3	Method 4	Price per photo
Webprint	128	18	38	μ=8,3; σ=2,8	€ 0,11
Albelli	101	3	-1	μ=5,68; σ=2,7	€ 0,17
Hema	100	-3	3	μ=7,9; σ=3,0	€ 0,11
Profotonet	71	-18	-40	μ=5,8; σ=2,9	€ 0,33

Subjectively, Profotonet prints the sharpest and Hema least sharp. In some Hema-prints a moire-pattern can be found (probably due to additional JPEG-compression somewhere in the ordering-process). Webprint has an aggressive optimization process where colors are enhanced and sharpening applied, but this was clearly very much appreciated by the participants. For both Albelli and Profotonet the ‘optimize photos’ was turned off, and many subjects noted that these photos were too dark (especially for Profotonet). Prints from Profotonet also seemed to have less contrast.

Some example-photos #

In this first photo the optimization of Webprint is very obvious. Profotonet is sharper and more faithful to the original.

Again the main difference comes from the optimization process. Less clear is the fact that the Hema print has a lot less detail.

Conclusion #

In the future we will order photos from Webprint, they are cheap and offer the best all-round quality. The way they print ‘differently-sized’ photos is better than the others.

Discussion #

The amount of participants was limited and this test does not have any statistical significance as a result. Future work should account for this. The lighting conditions were different for each participant. Obviously this should be kept consistent in future experiments.

The option for optimization was turned off for Albelli and Profotonet. Especially the latter printing service was at a severe disadvantage because of this (they printed sharpest on the most expensive paper, but the participants all found the photos too dark). Apparently the experimenter is not as good at optimizing photos on his computer as he thought… For new experiments these settings should be turned on.

Due to time and budgetary constraints, the research in photo-quality is now over 🤓

the raw results can be found in this Apple-Numbers file

Book Summary • The Power of Habits

Sun, 25 Feb 2018 00:00:00 +0000

I read a book about habits and didn’t want to forget the core ideas in it, so here’s my small summary!

The book is written well but the writer takes the idea to far. Half way in the book there’s a point he starts writing about organisations instead of individual habits and it goes downhill from there. Everything becomes a habit. Civil movements, church, addictions…

Another smaller point of criticism is a trick the writer uses too often: introducing a story in the beginning of a chapter and resolve the story only in the end. That was fresh and exciting in chapter one, but when used for every chapter it’s a bit much.

Other than the two points above the book was enjoyable. Here’s my quick summary:

80% of what we do is based on habits
the brain has a separate area for storing habits (which is why people with lesions who cannot make memories, can still store habits)
making a new habit is very difficult. Instead it’s better to ‘rewrite’ an older one (BJ Foggs adds: or add a new habit onto an existing one)
habits consist of a cue and a reward. You keep doing the habit because (unconsciously) you’ve developed a craving for the reward
to change a habit: investigate the trigger, behavior and reward. Then adapt the behavior. Make it easy for yourself (put the running shoes next to your bed) and start small (that’s not in the book, but from BJ Fogg’s tinyhabits, better start by flossing just one tooth then to never start at all)
you need a ‘goal’ to maintain willpower to change your habit (‘running the NY marathon’)
relapse is dangerous: when things aren’t going your way it’s easy to go into your old habit and stay there (which is why a ‘buddy-system’ like alcoholics-anonymous works well)
“never waste a good crisis” › a crisis is a moment where change can happen. Some managers prolong (or stage) a crisis for this reason
there are ‘keystone habits’. If you manage to change those, others follow like dominoes (people that start jogging, also start eating healthier; people that start tidying their bed in the morning, also become more punctual; etc.)
habits can be dangerous. Think smoking or gambling, but also sleep walking: people apparently have killed people while sleep walking. In the latter case the actions were so ‘automatic’ that the people were not convicted.
the power of weak-ties: apparently it’s better to have a big group of ‘acquintances’ than a small group of friends. With a big network you can easier get new information (like a job-opening)

There were some interesting examples:

Fabreze: the company could not sell the product (an apparently revolutionary air freshener) until they got the idea to make it part of the habit loop. Fabreze is now marketed as the ‘reward’ for cleaning your house
Starbucks: offers a lot of courses (“it’s maybe the US’ largest educator”), also on habits. It changed some peoples lives (for instance by letting words of angry customers not get to you: “the apron is your shield”)
the army: to avoid a huge demonstration in Iraq/Afghanistan, all you have to do is remove the hot-dog stands from the square
Target uses a lot of data to find out your habits and promote or change your ‘shopping habit’
Alcoa: a new CEO changed the company by completely focusing on safety of the workers. By doing that, new communication lines needed to be established, workers were instructed to offer advice, management got more involved in the daily work, etc. Net-result: a more profitable company. (example of changing a keystone-habit)
a hospital: where the culture was so rotten that nurses came up with secret codes to indicate which surgeons were assholes and which ones were just ‘sloppy’. In the end the coverups no longer worked and people started dying.

Here’s another readable article on the book: https://collegeinfogeek.com/power-of-habit-5-lessons/

book cover not under the Creative Commons license

Songs per day

Sun, 11 Feb 2018 00:00:00 +0000

Ever wondered which day was the most popular? It’s definitely a Sunday, I’ve looked it up!

Apparently artists get really inspired by the weekend, not so much by the mid-week… So if you’re looking for a new song-idea, be original and sing about a Wednesday!

Day	Original	Covers
Monday	42	577
Tuesday	13	106
Wednesday	4	11
Thursday	6	21
Friday	19	146
Saturday	69	403
Sunday	79	928

Thanks to: https://secondhandsongs.com

Old style cat-movie

Tue, 23 Jan 2018 00:00:00 +0000

In the last post I wrote about creating a movie from a series of images. Here’s a taste of what that looks like 😸

The photos were taken in burst-mode. I used FFMPEG to convert to grayscale, scale down to 420p and dimmed the contrast a little (so the image lights up a bit).

ffmpeg -r 10 -pattern_type glob -i "*.JPG" -vcodec libx265 -vf "scale=-1:420, format=gray, eq=contrast=0.8" -tag:v hvc1 -preset veryslow -crf 28 -an -pix_fmt yuv420p poezenfilmpje_gray_scaled_vs_fps10_eq_nob.mp4

With this, 1,18GB of JPG-images get compressed into a 665KB movie. Impressive.

I also tried:

different presets (ultrafast: smaller filesize, but looks much worse)
different framerates (10fps worked best)
increase brightness (made it look ‘flattened’)
curves=strong_contrast, but made it too dark

Creating movies and GIFs from a series of photos

Wed, 17 Jan 2018 00:00:00 +0000

I sometimes use ‘burst mode’ on my camera if I’m making photographs of my daughter (or cat), so I can select the best photo and discard the rest. After a while I found out that these burst-photos are also cool to show as a video. These short ‘movies’ look they are shot with an old-fashioned videocamera.

Here’s my process 🤓

Images to MP4 #

The command to create a movie from a series of jpg-images (without resizing):

ffmpeg -r 10 -pattern_type glob -i "*.JPG" -vcodec libx265 -tag:v hvc1 -preset veryslow -crf 20 -an -pix_fmt yuv420p result.mp4

-vcodec libx265: the best quality encoder
-tag:v hvc1: necessary for playback on OSX
-crf 20: great quality / filesize (28 is default, 16 is visually lossless)
-an: no audio
-pix_fmt yuv420p: also necessary for playback in Quicktime
-preset veryslow: slow encode, best quality
-r 20: 20 ms between frames
optionally -vf "scale=-1:420, format=gray": rescale (to 420p) and gray-scale

CRF is a factor to indicate quality/filesize. I find it extremly hard to judge the result from this. One source suggests using CRF 16, but to my eye there is no difference between 17 or 20. You can check for yourself with two samples I generated: image 1, image 2. I’ve included the previous encoder of choice in the table below as well to show how impressive H.265 is. According to the documentation, CRF 23 is default in H.264 and CRF 28 is default in H.265.

Setting	Filesize	Quality
H.264, CRF 17	38MB	great
H.265, CRF 17	27,5MB	great
H.265, CRF 20	17,2MB	great
H.264, CRF 23	13,7MB	good
H.265, CRF 28	3,5MB	good (impressive for this filesize)

for 24 JPG images all at 4,7MB (totaling 112,8MB)

Images to GIF #

H.265 is nice, but sometimes you need a more archaic format if you want to share your movie. The following incantation converts a series of images to a GIF. You will need ffmpeg and ImageMagick installed (which is easy, if you have brew on your Mac).

ffmpeg -pattern_type glob -i "*.jpg" -vf scale=800:-1 -r 10 -f image2pipe -vcodec ppm - | convert -delay 10 -layers Optimize -loop 0 - result.gif

The above seems to create the best looking gif for the smallest filesize (2,5MB for the same 24 images).

Other methods I’ve tried:

using only ffmpeg: some weird pattern gets added to the GIF
using imagemagick first and ffmpeg last: same weird pattern added to the GIF
using only imagemagick: the file ends up twice as big, no visible improvement
using graphicsmagic: similar results to imagemagick
using palettes, different scaling, different dithering in ffmpeg… Nothing got close to the combo-option described above:

Different dithering (looks good, bigger filesize)

ffmpeg -r 10 -pattern_type glob -i "*.JPG" -vf scale=800:-1:sws_dither=ed -an -pix_fmt rgb8 ffmpeg_dither.gif

Using palette creation and different rescaling (not a great result, also big file)

ffmpeg -r 10 -pattern_type glob -i "*.JPG" -vf scale=800:-1:flags=lanczos,palettegen -an -pix_fmt rgb8 -y tmp_palette.png
ffmpeg -r 10 -pattern_type glob -i "*.JPG" -i tmp_palette.png -lavfi scale=800:-1:flags=lanczos,paletteuse=dither=floyd_steinberg -an -pix_fmt rgb8 -y ffmpeg_lanczos_palette_dither.gif

Here’s more info on these settings

note: ImageMagick uses ‘delay’ and FFMPeg uses FPS. The delay in ImageMagick is described in ticks (1/100 of a second). So 20 FPS is a delay of 5 (delay=100/FPS)

Optimization #

Apparently ImageMagick already does a great job of optimizing. Using gifsicle --optimize does not result in a smaller filesize (sometimes even enlarging the file).

Tools used: ffmpeg 3.4.1, imagemagick 7.0.7-18

Comparing photo-services • Part 1

Wed, 10 Jan 2018 00:00:00 +0000

We got some photos printed, but I wasn’t 100% satisfied with the quality. Thirty-year old photos from my childhood look sharper! At first I thought this was because of the photos being taken with a digital camera, but then I realised it could also be the printer.

I’ve since found out that the company we used (Hema) is rated as one of the worst places to print photos.

Other photo-printers #

In the search for a better photo-printing service I’ve decided to order 20 photos at three different providers so I can see the quality differences for my self. I looked around at several online reviews, comparison sites and forums to see which providers I would compare. Below are the sources that mostly influenced my decisions.

Silent shiver #

This horse-photographer made a very detailed comparison of three professional photography printers. I’m going to use her post as a blueprint for my when I compare the photos.

Compared:

Profotonet: most consistent quality, but least sharp
Saal Digital: cheapest, but less consistent in quality
FIFO Color: sharpest prints, but too dark

Fotovergelijk #

Fotovergelijk is a place where people rate printed photos and compare pricing. Hema is rated the worst…

I’ve added the printer-service used by the website if they don’t print their own photos.

Good:

Profotonet (10)
Albelli (9) uses Albumprinter to print
Webprint (9) (named Smartphoto in other countries)
Fotofabriek (9)
Pixum (9) CeWe
Fotogoed (9) Küpper druck

Reasonable:

Blokker (8) Fujiprint
Albert Heijn (7) CeWe

Bad:

Aldi (6) CeWe
Kruidvat (5) CeWe
Hema 🤨 (4)

The scale originally goes from 1 to 5, but I converted it into a 10-point scale so I can compare it with the scoring of Onlinefotoservices.

Onlinefotoservices #

Lastly, we take a look at onlinefotoservices, another comparison site.

Highly rated:

Profotonet (8,8)

Reasonable:

Primera (7,7) CeWe
Smartphoto / Webprint (7,6)
Pixum, Aldi, Photobox (7,4) CeWe

Bad:

Kruidvat (6,3) CeWe
Hema (6,1)

Conclusion #

Everybody agrees Profotonet is the marketleader in high quality photo-printing. Albelli and Webprint seem to be a good 2nd place. CeWe is popular, but has a mixed track-record. The other two high-rated services from Fotovergelijk are not known to me (Fotofabriek and Fotogoed) and are also more expensive. They are highly rated in service, so perhaps I can go back to them some time.

For now I will order photos at Profotonet, Albelli and Webprint. In the next post you will find the outcomes!

Prototyping with Axure • Part 1 - Optimal images

Wed, 03 Jan 2018 00:00:00 +0000

I use Axure in my work as UX-designer and thought I’d share some of the tricks I’ve come up with in the past 5 years. This post was originally named “OMG Axure, WTF?”, because these are some nice hacky workarounds 🤓

I’ll add more over time, but for now I have two image-related tips.

1: Never optimize images #

By default Axure asks if you want to ‘optimize images’ on import:

I always answer ’no’ on this: Axure has a very aggressive image optimization. You will notice JPG-artifacts (especially in corners) and enlarging the image will make it look extremely bad. On top of that, lose the alpha-channel in PNG (so no more ‘see-through’).

Optimizing images is not necessary. This is not the ’90s anymore! I have very image-heavy files with dozens of pages in them (because for some projects our agency still uses static-comps) and my six-and-a-half year old MacBook Pro handles them just fine.

Of course, if the image does slow you down, you can very easily optimize it later by right-clicking ‘optimize image’ or slicing a tiny piece off of it (CMD-6). But be careful: there is no ‘unoptimize image’!

If you do remote-testing and you worry about page-load: read on!

2: Preload images for remote testing #

Axure does not export an image multiple times if the same image is on multiple pages. Instead it links to the same image-file. ‘So what’, you might think, ‘big deal!’

But this is actually very useful information for us! For remote testing we sometimes have users with very little bandwidth and in that case it might be detrimental to the test if the pages load very slow. Especially because I ’never optimizing images’ 😇

Here’s the tip: after you’re done with your masterpiece of a prototype, create another page in Axure and call it ‘preloader’. Go through all your pages and copy all images to the ‘preloader’ page. Easiest is to CMD-click on the images in the bottom-right pane to select all of them, then CMD-C to copy.

Now you should have all images from your prototype on the ‘preloader’-page. Simply add a huge white-box over them and add a bit of an explanation for the people participating in the test. Also include a button so people can start the test. If you want to be really fancy, you can disable the button first and have it be enabled after ~10 seconds (because people never read the label ‘wait for instructions’).

Why cover the images with a big white box, and not set the images to ‘hidden’ in Axure? Because some smart browsers might realise the images are set to visibility:hidden and not load the images to preserve bandwidth.

Now all your pages will load quickly!

_This is part 1 in a series, [you can continue to part 2 here]({% link posts/2018-03-24-axure-tip2.md %}).

Pandoc and PDF

Wed, 13 Dec 2017 00:00:00 +0000

I’m quite a big fan of Markdown. It’s easy to write and works nicely with Git, you can preview it in OSX and you can edit in any program you want (no Word or OpenOffice necessary). I prefer writing Markdown over Tex: I can copy from my blog to a document and the markup is less ‘finicky’.

So it should come as no surprise that I was very happy to discover Pandoc, a ‘universal document converter’ that can convert Markdown to just about any other format (like Epub, HTML, Word, LaTeX, PDF…) and back again.

I already used Pandoc for creating an EPUB book and several Word-documents. This all works great.

If that’s all you need, you can stop reading now.

Creating PDF-documents #

The problem arose after I wanted to create a PDF document. Apparently this is Crazy Difficult. Pandoc supports several converters, but all have their own little problem. Here are my findings:

Using XeLateX #

I couldn’t get the default LaTeX engine (pdflatex) to work because of UTF-8 characters. Apparently the solution is to use XeLaTeX. I used the BasicTex package instead of downloading the full 2GB distribution. To make it work, I had to:

sudo tlmgr update --self
sudo tlmgr install ucharcat
sudo tlmgr install lm-math

After that magic incantation, I can create a PDF reasonably painlessly:

pandoc input.md -o output.pdf --css=style.css --pdf-engine=xelatex

The problem is it looks Tex-y: the style.css I created gets discarded.

Using Calibre #

Pandoc creates EPUBs easily, so I thought I could convert this EPUB painlessly to PDF using Calibre, a tool I already have installed for my ebook-management.

pandoc input.md -o inbetween.epub -t epub --css=style.css
/Applications/calibre.app/Contents/MacOS/ebook-convert inbetween.epub output.pdf --paper-size a4

This works and the endresult actually looks very nice. But there is no ‘orphan detection’ which makes for very weird single sentences on pages and tables get distributed over pages (they simply get cut in half, there’s no new header on the second page).

Using PhantomJS #

Another option I already had installed was PhantomJS, a ‘headless browser’. With a simple javascript you can convert HTML to PDF (using Pandoc to create the HTML and PhantomJS to create the PDF).

This just looked horrible all around.

WKprint #

Another option provided by Pandoc is wkhtml2pdf. But this gave me a Warning: Failed to load errors. In short, images and CSS are not loaded, which again made the result look horrible. Perhaps worth another shot later…

weasyprint #

weasyprint is a Python based HTML->PDF renderer. It supports CSS and tables over multiple pages (👍), but also no ‘orphan detection’ (👎). By default the font renders smaller than in Calibre, but this is easily fixed in CSS.

Untested #

Pandoc supports more PDF-engines, but I didn’t test these:

luatex and pdflatex also don’t support CSS
pdroff seems to be mainly for manuals
prince costs money…

Conclusion #

There was no clear winner for Markdown->PDF. For now it looks like I need to continue my experimenting with Weasyprint (or try WKprint again).

Curious to hear if others have more success!

Spotify Scannables

Wed, 22 Nov 2017 00:00:00 +0000

For my ‘sister project’ I had the idea to print physical albums for virtual Spotify albums. It works using the ’new’ Spotify barcodes. Read the original blog-item.

My first attempt to create these was a very manual approach: I made a screenshot from Spotify on my phone (the only place where the barcode is shown), cropped the code, pasted this in a Pages-file (the Apple-equivalent of Word) and added an album cover + tracklisting I found on Wikipedia. This would take over half an hour per album!

Automatically creating album-covers with a Spotify code #

In my 2nd attempt I tried to script as much of this process as I could. In addition, I decided to skip the tracklisting (keeping just the cover + Spotify code).

This means I no longer needed to print double-sided. Which in turn meant I can have the covers be printed as ‘square photos’, so they look great now!

Bash-scripting #

I had to dust-off my scripting capacities and my first attempt went a bit ‘overboard’. I created a script that:

automatically cuts the Spotify-code, the album-title and artist-name from an Android screenshot, using GraphicsMagick
uses Tesseract (an OCR-engine) to parse the album-title and artist into text
asks the user if the OCR was done correctly (with the option to provide a better album/artist)
uses Glyr (a metadata search-engine for music) to get a load of album-covers
finds the best-looking album-cover (approximately, using filesize) using the artist/album from the previous step
pastes the Spotify-code on top of the album-cover (in two versions: with the code either in the left or right bottom-corner)

Which worked GREAT! Except that all albums returned by Glyr were kind of ‘meh’-quality. I introduced several improvements, but in the end it never was ‘great’.

You can download v1 here to play around with.

Version 2 of the Bash-script #

I took my losses and re-examined my options. I found that the resolution of the cover in the Android-screenshot was actually a lot better than whatever Glyr returned. So I greatly simplified my Bash-script. Now it only:

automatically cuts the Spotify-code and the cover from an Android screenshot, using GraphicsMagick
pastes the Spotify-code on top of the album-cover

Optionally, you can enable Tesseract to have a cool filename as well.

Ofcourse v2 is also available to play around with.

The end-result #

Here’s some example albums. You can scan them with Spotify (on mobile: tap ‘search’ and then the ‘camera’ in the search bar).

Printed #

I had them printed as ‘square photos’, they look great!

Generating JPG #

If you plan to do this as well, you need to convert the PNGs to JPG before you can print them, which is easy with GraphicsMagick:

for f in result/*.png; do gm convert $f -unsharp 2x0.5+0.7+0 $f.jpg; done

This adds some sharpening as well, which looks a bit nicer when printing IMHO

album-covers obviously don’t fall under the Creative Commons license

Bitcoin and Monero

Sun, 24 Sep 2017 00:00:00 +0000

When I first heard about Bitcoin (8 years ago), I thought it would pass. It didn’t and I’m one of those suckers who’s thinking “why didn’t I just get a couple when they were worth nothing…”

But, better late than never!

Bitcoin #

I wanted to learn more about Bitcoin, so I bought a small amount and lo-and-behold: it got transferred into my wallet. A couple of hours later on Wikipedia/Youtube and I think I’ve got a pretty good idea how it works.

That’s it for Bitcoin, it’s too established to really do something ‘interesting’ with it (or invest in it, IMHO).

Monero #

Monero is an alt-coin, which is constructed in a more privacy-concerned manner than Bitcoin. But that’s not the reason I got interested in it; The other interesting property is that the coin is designed so that ‘mining’ is more CPU-friendly.

For Bitcoin, the best mining happens on specifically designed machines (ASICs) and 2nd best happens on GPU. For Monero, it’s still possible to do decent mining (generating random hashes) on a CPU.

Which is interesting for two reasons:

I get to play around with it (I have an old GPU, but a decent CPU)
the naughty people online can steal your processing power with javascript to generate coins

The latter reason is why I think Monero has a shot towards becoming a viable coin as well 😁

Mining Monero #

After about an hour I managed to compile both the GPU and CPU miner for Monero on OSX. Now I could find out for myself the Hash-rate difference between CPU and GPU!

Interestingly, the GPU miner completely locks up my computer but spits out a rather disappointing 10.5 H/s (NVIDIA GeForce GT 650M 512 MB).

As expected, the CPU fares a lot better: 37 H/s per core! My i7 has 4 cores, so if I leave one for myself, I get a nice 111 H/s without locking my computer up. According to the Monero IRC my processor would fare even better on Linux/Windows because of optimization in xmr.

Mine for yourself #

~~If you want to join the cool kids and mine some Monero, you can do so on my blog~~. Apparently my blog gets listed as ‘unsafe’ because of the CoinHive-script 😒, so I’ve disabled it.

my computer averages 60 H/s in javascript, higher is better! To stop mining, simply refresh the page

Send me some e-dough #

If you like, you can send me some Monero:

45wPmJdbfy6ALtNBQPbgMBEdmzA6KrhRGdgKUJ4wqXSdg5Nm77qxrPzNQRr7vXok4HCXQqYCYbtaBAFMb6iujU45T7wxYSs

For sending me Bitcoin, let me know and I’ll generate you a link 😇

Blog now in Git

Fri, 22 Sep 2017 00:00:00 +0000

Here’s something that was on The List for a long time: adding this blog to a git-repository. Why did that take so long?

I wanted to have my own, personal git-server instead of GitHub to push stuff to, so I can have a post-hook do cool stuff like auto-deploy the website. That is possible with GitHub as well (if you pay for it), but I thought it would be ‘fun’ to play around with all this 🤓

My workflow used to be:

write article & jekyll serve
when satisfied: do a complicated rsync-command (hidden behind an alias)

My new workflow:

write article & jekyll serve
make commits while writing
when satisfied: git push

My server then checks out the most recent commit, runs jekyll build and rsyncs the new items. Because that happens automatically it’s a lot less fault-sensitive.

You are now reading the first article published with git push 😇. Currently only matth-ijs.nl works like this, but in the near future I’ll port 1001ideas.org as well.

My setup #

I’ve installed both git and jekyll on my VPS and added a new user there (git). In the home-folder of git I’ve setup blank repositories for all my projects (like this site). On my local machine I created a new RSA-key specifically for the git user (not strictly necessary, but I was in an experimenting mood).

I then created a new repository for this blog locally, added all files and set the origin to the server. Done!

Steps #

Remote:

in the home-folder of your git user: git init --bare <sitename>
add script to the hooks-folder (see below)
make a var/www/<sitename>/public_html which is owned by the right user
add virtual-server for Apache for that folder

Local:

git init in the right folder
git remote add origin <ssh-credentials>:<sitename>
git push --set-upstream origin master

Script #

#!/bin/sh
site="1001ideas.org"
mkdir /tmp/${site}
rm -rf /tmp/${site}/*
git --work-tree=/tmp/${site} --git-dir=/home/git/${site} checkout -f
cd /tmp/${site}
jekyll build
rsync -av _site/ /var/www/${site}/public_html/ --itemize-changes

Matth-ijs.nl recommends...

Tue, 05 Sep 2017 00:00:00 +0000

I’ve got a rather peculiar taste; apparently I’m a “Harbinger of Failure”* - someone who likes things the general population doesn’t like 😇

In other words: the kind of stuff Netflix and Spotify don’t recommend. The goal of this project is that others with similar tastes (other Harbingers of Failure) find my blog and discover new stuff to try out!

So check it out: Matth-ijs.nl recommends! Of course, there’s also an RSS-feed.

*NOTE: I should get a t-shirt of that.

update: no longer working

Now running my own VPN and PiHole

Mon, 04 Sep 2017 00:00:00 +0000

During my “nerdy-hour on Sunday” I installed OpenVPN on my home-server :)

Installation was a breeze (using this guide), with the following caveats:

the IP-address you use in step 1 if you’re behind a NAT-router is NOT your external IP, but rather the IP of the server you’re installing VPN on
the installation script has some oddities: it sometimes writes a number instead of a value (the script for instance asks: “Which protocol do you want to use: 1. UDP, 2. TCP” and instead of writing proto=udp it writes proto=1). So double check your server.conf and client.conf)

Even easier was installing PiHole for capturing all ads- and tracking-code (using this guide). I’ve left most settings as-is, but I also block facebook.com to stop ads from appearing in my PodCasting app (added bonus is less procrastination).

So that’s it! No more ads on my devices, safe internet on public WiFi and always the Dutch selection of Netflix.

By the way, running Bungie on Ubuntu right now and that’s a very pleasurable experience. Recommended!

Fully customisable slider in Axure. Error messages, auto-interpolation, etc.

Wed, 12 Apr 2017 00:00:00 +0000

I’ve created a fully customisable slider for all to use. Feel free to use, modify and give feedback

The slider itself: Slider component
A couple of demos: Example sliders
Download: download from GitHub

Features #

Of course, the slider is packed with features 🤓

Fully customisable #

min/max/default values can be set
slider-range is automatically adjusted for size of slider and min/max values
shape / design of slider is customisable
shape of hint is customisable
error messages, warnings and feedback text is easily changed

Summation #

Add as many sliders as you like: values are summed without the use of a global variable.

Keyboard input #

Users can edit the field, which changes the slider position. Plus they can use up/down arrow keys to increase/decrease values.

Easy to use #

Includes a settings-panel for the Interaction-/UX-designer. The slider does not use global variables, so it is very easy to copy/paste.

Everything is commented, so it’s easy to learn the tricks I used. Nonetheless: I’d like to create a series of forum posts on my techniques, to discuss them with you all! For now, I hope the slider becomes useful for you!

The settings-page #

UX designers can set almost anything about the slider in the settings-page. Everything is accessible from within Axure.

How to manually back up Whatsapp from Android

Wed, 25 Jan 2017 00:00:00 +0000

So, I’m going through my photos and video on my Android and copying it to my computer for backup-purposes. Of course, WhatsApp is a large part of my life and there a lot of photos and videos “hidden” there as well.

Location of files #

Not too hard to locate the files, they are in:

WhatsApp/Media/

On either the internal of SD-card memory.

What to copy? #

What do we keep and what do we remove? In my case, I’m only interested in photos and videos. So we keep the WhatsApp Images and WhatsApp Videos folders. In both folders you’ll notice a lot of files and a Sent folder.

And now the confusion starts!

Images folder #

The files in the WhatsApp Images-folder appear to be a mix of photos you received and sent. If you used WhatsApp to take the photo, it will end up in both locations. If you made the photo outside of WhatsApp, it will only appear in the Sent folder.

The photo stored in the Sent folder appears to be exactly as they were sent: a downsized version of the original with the edits you made in WhatsApp (for example a sticker or text you put on the image).

Conclusion: you can safely throw away the Sent-folder without risk of losing any precious memories, unless you edit your photos in WhatsApp before sending them. In that case, you’ll want to save the Sent-folder also.

Videos folder #

This is more tricky. WhatsApp appears to use the same trick here, but the videos in the Sent-folder are sometimes larger than the ones in the root WhatsApp Videos folder. Which is not what we expected from how the images folder was organized. So what’s going on?

It appears that WhatsApp tries to optimize some videos before sending them. This is not flawless, and sometimes results in a larger file.

Unlike photos, most videos made with WhatsApp are not reduced in file size before being sent. On the other hand, all videos taken outside of WhatsApp are resized.

Conclusion: we can also safely throw away the Sent-folder here. The same caveat exists here: if you edit your video in WhatsApp (trimming for instance), you might want to keep the Sent-folder after all.

Disk usage #

When you send a photo or video from within WhatsApp, it gets stored twice. What’s worse, if you send the same file to multiple people or groups (but on different days) it also duplicates.

This makes for easy backing up as we don’t have to go through the Sent-folder, but I was hoping WhatsApp would use symlinks instead. Maybe it does and non-root users can’t see that?

I did my research on January 25th, 2017. WhatsApp may update the way this works. So before removing your Sent-folders, just doublecheck!

Installing Jekyll on Mac, Ubuntu and Bash on Windows

Wed, 07 Dec 2016 00:00:00 +0000

I’ve had the ‘pleasure’ of having to install Jekyll on different platforms and thought I’d share the findings.

OS X #

This is the easy one. Install HomeBrew and run:

brew install jekyll

Grab a cup of coffee, you’re done!

On Ubuntu and Bash on Windows #

Sadly, this one takes a bit more work. Do not run sudo apt-get install jekyll, because it installs an ANCIENT Jekyll (0.11.2), so the first step is to remove that if you’ve got it installed (sudo apt-get remove jekyll).

There’s multiple ways to install, but the easiest I’ve found is adding a custom repository from Brightbox.

sudo apt-add-repository ppa:brightbox/ruby-ng
sudo apt-get update
sudo apt-get install ruby2.2 ruby2.2-dev
sudo gem install iconv posix-spawn mini_magick jekyll

The last step installs necessary gems for Jekyll, although gem ‘mini_magick’ is not strictly necessary.

On Windows you may need to close your terminal and reopen it to update your PATH)

Running Jekyll on Bash on Windows #

Filewatchers aren’t working on Bash on Windows yet (in august at least). So if you run in Bash on Windows, you may need to run instead:

jekyll serve --force_polling

This will make sure the HTML/CSS will update when changes are made.

SUDO not working? #

Now, on Windows the sudo command may be borked. If you get the error “unable to resolve host NAME”, you’ll need to update your Linux hosts file (which is different from the Windows one). Here’s a guide. Basically, add the host name from /etc/hostname to the file /etc/hosts.

1001ideas.org

Tue, 22 Nov 2016 00:00:00 +0000

Over the years I’ve come up with a lot of different ideas. I try to do something with all of them (like publish them here), but sadly most ideas just end up in a big Google Keep list…

Well, not anymore!

I’ve started a new project where I document my ideas and on top of that, try to implement the idea where possible.

Go ahead and take a look www.1001ideas.org.

Schadenfreude beach

Thu, 28 Jul 2016 00:00:00 +0000

Are you stressed or feeling down? And do you feel better reading about other people’s (first world) problems?

Then I’ve got just the solution for you: schadenfreude beach!

Look at a peaceful scene while reading all the #FML tags from Twitter. Updates come in every 20 seconds, so you have a bit of time to take it all in.

Go on a 30 minute holiday at schadenfreude-beach and relax!

update: the code is no longer working…

(since it’s raw twitter stuff, it may not be suited for work)

image and tweets on schadenfreude-beach are not under the Creative Commons license

32juan - One picture per hour

Sat, 18 Jun 2016 00:00:00 +0000

So there’s another project for you. It’s part of a larger idea, which I’m still working on. For now this website basically just shows a new ‘funny’ picture every hour…

www.32juan.com update 2017: is no longer active!

(all pictures are reasonably suited for work)

Blog got a visual update

Tue, 24 May 2016 00:00:00 +0000

Just a heads up to my RSS-feed readers: sorry for the mess. I’ve updated the URL-structure to now include the category of the post, which is why the RSS feed got reloaded.

I’ve also updated the blog-styling to now use Hyde. In my opinion it looks a lot better and it’s a lot more comforting to the eyes to read.

Encrypting Dropbox

Mon, 16 May 2016 00:00:00 +0000

I have local backups for my photos, but with Dropbox they get synced to a place far away so I’ll still have access to my photos, even in the case of fire :) I tried other backup platforms, such as Amazon S3, but the ease of Dropbox is unsurpassed and I use it for lots of other projects as well.

Why encryption? #

Dropbox keeps adding features and I guess it won’t be so long until they’ll do some kind of image analysis on photos as well. Perhaps they already do.

I don’t like the idea of my photos being changed automatically or information being extracted out of them. This may seem far-fetched or paranoid, but this is what Google is doing if you use Google Drive. And when you use camera-upload from your phone, Dropbox changes the filename already.

On top of that, I’d like to keep my photos safe in the case someone else gets access to my Dropbox account. Even though I’m not a celebrity (yet), I’d like to keep my life a bit private.

Also, encryption is just good backup-practice (http://www.taobackup.com).

EncFS vs Apple Disk Utility vs Duplicity #

So here’s a quick rundown of three applications to use for encryption.

Duplicity and S3 #

I started out with Duplicity in combination with S3. This worked pretty sweetly, but it has the disadvantage that backups go into a ‘black box’. Especially because of the ‘incremental backups’ (only changes are backed up). There’s no way to browse the files after they’ve been backed up. Also, S3 became more and more expensive the more I backed up to it, and this made the switch to a paid Dropbox account quite easy.

Apple Sparse Bundle Disk Image on Dropbox #

On Dropbox I decided to try out Apple’s encrypted “sparse bundle disk image”. This means the image is cut in pieces of 8MB, making syncing a lot easier; in theory Dropbox would only need to send a little bit of data over when making a small change (like adding just one file). In practice however, the syncing took forever. For some reason, adding just one file meant a lot of pieces would be synced. This slow syncing was the main reason to stop using this solution.

The main advantage of Apple’s disk image is that it’s easy to mount and use in OSX and it will retain the right properties of files inside. The disadvantages are that it won’t open on other OS’s and that you rely on Apple supporting it over a long period (I want to be able to open my backup in 10 years time, if I have to).

Introducing EncFS #

After the eternal syncing of the Sparse Bundles, using EncFS is a real pleasure. EncFS encrypts each file individually, making it extremely easy for Dropbox to do incremental backups. It’s available on Linux as well and is mounted as a separate filesystem. This means I can browse it easily. The disadvantages are that the encryption is not the world’s best (especially if you use it in combination with Dropbox) and that meta-data is not encrypted (filenames are, but not the time&date).

So EncFS is a no-go if you want to keep your nuclear secrets safe, but for my use-case it’s perfect.

Obviously you don’t normally share backups, but with encrypted folders there’s an additional reason not to do it. Dropbox won’t make sense of multiple people changing files inside an encrypted partition and your stuff gets messed up really quickly.

Other, paid solutions #

There are many other solutions for encrypting your Dropbox. I decided not to rely on these proprietary programs, because you’re never sure if they still exist in a couple years time. The main contender, https://www.boxcryptor.com/en, already switched their encryption algorithm once.

Process #

Whereas Duplicity ran automatically, my new process is quite manual… The advantage is that I know exactly what’s going on, the disadvantage obviously is that it’s more labor and more error-prone.

I have a ‘backup’ folder in my Dropbox which is partially synced. In it I create folders with today’s date and the years which are backed up in it (e.g.: 20160516-2009-2015). I create an EncFS partition in it, and use rsync to copy all files into it. Dropbox starts syncing as soon as the copy starts. I also create a file with a list of all the synced photos (find . > syncedfiles).

After the copy is done, I use selective-sync to turn off this particular folder. I then never touch this folder again until I need it. Selective-sync saves disk space, and it also means that evil stuff like filelocker has no chance of reaching it.

Installing ZENDServer in Vagrant

Wed, 13 Apr 2016 00:00:00 +0000

As a note to myself, this is how I installed ZENDServer in Vagrant and connected with it using the ZEND IDE on Mac OSX.

Installing ZEND in Vagrant #

First login to your box (vagrant ssh), then remove the current php (sudo apt-get remove libapache2-mod-php5 php5-*).

Download the latest ZEND Server into the box (wget the latest and run sudo ./install_zs.sh 5.5 if you want to install php-5.5).

Check if it works and complete the setup by going to: 192.168.33.10:10082/ (or whatever is your vagrant ip-address).

Connecting the ZEND Studio to the new server #

Create a new project from the local files and add a new PHP server. Set the base-url to be the same as above (192.168.33.10:10082/) and optionally set the Document Root to be the path to your site (/var/www/public_html/, for instance).

You should be set up and ready to go :)

Batman

Sun, 06 Mar 2016 00:00:00 +0000

Found some old doodles, thought I’d put them to ‘good’ use. The end-result is not as funny as I had hoped, ah well…

Get a different Batman doodle everytime

Scribble - easy note writing

Wed, 02 Mar 2016 00:00:00 +0000

Sometimes you just need to make a quick note. I regularly open TextEdit for this on my Mac, which works fine. However, I’ve gotten quite tired of the ‘do you want to save the file’ dialog. Also, the app has several smart ways of changing my text: auto replace spelling, apply smart quotes… Something I generally like, but don’t always want.

So, I decided to create my own app :)

Introducing Scribble #

It’s the world’s simplest OSX app. Basically it’s an edit field, without any options. All text is without markup, doesn’t have an option to save and no text will be altered on typing. You can open as many Scribble windows as you like and ‘destroy’ them all when you press CMD-Q. Hope it’s useful to someone :)

Still to do #

I’m not a certified developer, so if you want to run the app you’ll have to ALT-right-click and select open.
There’s no app-icon
The app is rather big (6.2MB) because I used Swift for the programming language. For now all Swift apps include a runtime library, probably a recompile in a year will change the size to a more appropriate couple of kilobytes. Either that, or I’d have to rewrite in Objective-C

Getting the app #

As I’m not a certified developer, you’ll have to ALT-right-click the app and select ‘open’ to run Scribble.

Download Scribble

Send Cookie - creating a shop in Etsy

Sat, 06 Feb 2016 00:00:00 +0000

To test the onboarding process of Etsy, and basically just for fun, I’ve created my own shop. The shop has the amazingly great premise that people would want to send a cookie with a text on it.

So far, my findings are that Etsy has a quite easy process for setting up shop. They could be more clear on the pricing, however! For every ’listing’ they take $0.20, they take 3.5% of the selling price as commission and they charge a payment processing fee of 4% (over the total amount, shipping included!) plus 0.30 euro.

In my example: my ‘product’ costs 3 euro and sending costs are also 3 euro (for international sending). This means that on each ‘cookie card’ I pay Etsy 0.845 euro (assuming dollar and euro have an equal exchange rate). What complicates the matter is that these costs are not listed in one place.

Of course, shipping is not something I can reduce in price so the Etsy costs are almost a third of my product (28% of the costs to be exact). This percentage would go down quickly if I made my products more expensive…

Click here to send your own cookie

The product #

Say it with cookies: send a cookie stuck on cardboard, with your own message on it

And that’s how they arrive:

Click here to send your own cookie

I also published this idea on my ‘ideas blog’.

Streaming services compared

Mon, 01 Feb 2016 00:00:00 +0000

After the fiasco of Google Play Music can any other streaming service do it better?

I’ve tried both Apple Music and iTunes now for two months. I’ve added my experiences with Google Music Play as well (used it for two years) and Plex.

Apple Music #

The nicest thing about iTunes is that it listens to the play button on my Mac keyboard and on both my headphones. That is a huge relief.

Also, the discover function is nicer than Google’s and Spotify’s. It is more ‘in your face’ and the playlists are curated. In other words, they are made by humans. Consecutive songs blend into eachother and the playlists are more creative than the automatic ‘radio stations’ from Google and Spotify.

This is how you get playlists like ‘Intro to Deafhaven’ (a list of songs to get you into a certain band), ‘Behind the boards: Max Norman’ (playlist with songs by a certain producer), ‘Talking Heads live’ (best live songs of a band) and ‘One bourbon, One scotch, One beer’ (rock songs about alcohol). In other words: fun playlists that fit my rock/metal music-style.

Compare this to Spotify, where the discover function keeps defaulting to ‘hit songs’ and ‘most popular’, or simply listing albums ‘you might like’.

Now onto the negative points:

navigation is tricky. All music is located under ‘New’, confusing if you look up a 80s metal-album. Sections (such as ‘New’ and ‘For you’) have a strange navigation, where you can’t go back to the start screen without clicking the back button multiple times (and how do you get back to where you were then?).
Albums, singles, compilations, etc. are all mixed on an artist’s page. I’ve found myself going to Wikipedia multiple times, because it was faster to look at a list of albums there and then look it up in iTunes!
no integrated last.fm: I use Neptunes
no last.fm on mobile: this definitely sucks.
sometimes I get ‘you are listening on a different device’ eventhough this is not the case. The music stops playing until you head over to iTunes and click the play button. Hugely annoying.

Especially the lack of last.fm on mobile is a dealbreaker, in combination with the navigation it puts me off from using Apple Music.

Spotify #

Seeing as Spotify has been on the market longer, it makes sense they have the most mature user experience. It has some very nice tidbits, like picking songs to your running rythm and great integration with other online services (like Last.fm). When you subscribe, you can pick a higher quality for music playback.

better navigation in the application: this navigation makes more sense to me than Apple’s.
’now playing’ syncs between desktop and mobile: continu playing on a different device. Seems like a minor thing, but I very much enjoyed it.
play in the browser: when you don’t want to start the app
lyrics with the songs
Chromecast support: nice ;)
last.fm support
you can turn off ‘Set the same volume level for all tracks’ (since it’s implemented in a very simple way, it reduces the dynamic range of songs)

Minus points:

not everything’s there, eventhough it got better with the recent addition of the Beatles (the lack of Beatles was the original reason to pick Google’s streaming service), I still find that Apple Music has more songs listed
no playlist suggestions like Apple Music. The ‘radio stations’ are automatically generated, so songs don’t fit nicely together.
annoyingly, iTunes takes over the play button on Mac. There are hacks around this, but they are unreliable or introduce a delay of seconds between pressing the button and pausing the music :(

Google Play Music #

Major minus points:

horrible music quality on some tracks (tried all settings)
recommendations were quite bad (I’m disappointed in Google!)

Some good points:

upload your own songs for free (so you can listen to the Beatles and Rammstein)
mobile app quite okay
had better pricing (8 euro p/m)

Plex #

searching is a biatch
no recommendations or nice playlists
music library limited to my own music
superior quality (because most of my library is in the lossless FLAC format)

Music quality #

From good to bad:

Plex
Spotify
Apple Music
Google Play Music

I’ve tried the same songs on multiple streaming services (“Awakening the Fallen” by Avenged Sevenfold, “Everglow” by Coldplay and “Untouchable” by Anathema). None of the streaming services can really beat Plex, which is transcoding FLAC to MP3 on the fly. I find this very strange: ultimately, you’d expect either the same quality or better (because streaming services, especially Apple, sometimes remaster a song for streaming).

Apple Music and Spotify are close contestants, but Spotify has some deeper bass on Untouchable. In the end they are both ‘good enough’, but some music (like Anathema) really needs a good quality streaming and for that I’ll still stick with Plex…

(to listen to a comparison, check my previous blogpost: [Horrible quality on Google Play]({% link _posts/2014-07-30-stronk_googleplayquality.md %}))

In conclusion #

Spotify wins, in combination with Plex for when I feel a band needs the higher quality sound.

update I’ve also checked out Deezer, not a winner:

Neg

no Radiohead
OSX app rather limited
needs Flash on web
recommendations not so good

Pro

scrobbles to last.fm

Connectix game

Mon, 11 Jan 2016 00:00:00 +0000

A while ago I created a board game called Connectix. The goal of the game is to connect three seemingly random words written on a card. These connections can be quite far-fetched, an example would be:

finger gun eye

The players see only the three words on a card. One player picks up the card and either knows the connection and explains it, or he bluffs and tells a random story he made up (“the forefinger was called gun-finger and eye-finger in the past”). The other players than vote if they believe the story or not.

The player gets points based on the amount of votes: if he lied he gets as much points as there were people who believed him. If he told the truth, he gets as much points as people that thought it was a lie.

So the game consists of luck (the cards you get), knowledge and creativity. Too bad it apparently was too difficult to play :D

Solution: Bond-movie titles with the word 'gold' in it: Goldfinger, the Man with the Golden Gun and Goldeneye.

With each card there was an extra bit of trivia attached, to keep it a bit light-weight. In our example:

Goldeneye was the first bond movie not based on a story by Ian Fleming, but the title was based on Ian Fleming's life: it's the name of his summerhouse.

the prototype packaging, with the gameboard, voting cards and riddles

Resurrection on Facebook #

So the board-game didn’t completely work out. But that doesn’t mean the game is over!

I have resurrected the game on Facebook: https://www.facebook.com/connectix/

The ‘bluff’ element has been taken out (because impossible on Facebook) and the riddles have become a bit more difficult (because Google…)

update july 2018: apparently there is a scientific interest in this idea, there’s even a test (Remote Associates Test, the RAT) to measure combinatory thinking. Developed by Sawyer in 2013. Here’s an article in English using Dutch words and here’s the article I found this reference.

Blog now processed by Jekyll

Sat, 02 Jan 2016 00:00:00 +0000

So a while ago I created my own parser to make a blog in Dropbox. The basic setup was to store the markdown-files in Dropbox and have a hosted PHP-file get these files and parse them ‘on the fly’.

Apparently there are more ways to do this. Jekyll also parses static markdown files into a working blog site. There’s even the same color-theme (solarized) as I was using before.

So, without further ado, it is now possible to browse exactly the same content as you had before! :D

Why #

Of course there are some advantages to the new setup:

Easier theming
Better tags (or tags :)). The old solution was to just add a category to the title, with Jekyll I can do fancier things as having an archive based on categories.
Proper RSS
Faster: old PHP solution was quite slow sometimes, it had to load several external files from Dropbox
Reliable: obviously Dropbox didn’t make the Public folder to host websites from and letting PHP fetch those files. Sometimes it broke without me having a clue why.
Security: I had implemented some basic security measures, but I was never really happy with the old code. It retrieved external files and parsed them, that does not seem safe. Jekyll instead generates static html+css.
Wanting to try something new. Now I’ve learned the (very) basics of Jekyll and Liquid templating engines.
Future possibilities: such as adding excerpts to the blog posts in the Archive.

Changes #

Content transfer was not a one-click solution. I had to go through all my blogposts and change the following:

Codeblocks

I’ve had to change all codeblocks from ``` to ~~~. This one sounds simple, but was the most work. I wanted Github style codeblocks using the three backticks, but none of the Jekyll Markdown parsers properly support it. So instead I had to revert to three tildes. 2017: this is no longer true, backticks are now also supported

Titles

My old blog had a separate file where the titles+filenames of all blogs were for creating an ‘index’. The titles inside the article where written with the default Markdown method (adding a list of ====== under the title). Jekyll creates the list of blogposts automatically from the blog-files, but uses a customised method of defining the title (and other meta-data):

---
title: Blog now processed by Jekyll
category: stronk
layout: post
---

Add dates

One of the advantages of my old blog was that I didn’t have to have dates attached to my articles. So that means my one reader wouldn’t see the extremely low update frequency. Sadly, Jekyll is based around these dates and I had to retroactively look up all dates for the articles and add them.

Publishing method

With the old method, I wrote an article in Dropbox and published it by adding it to the list of ‘published articles’. The parsing etcetera happened automatically. With Jekyll, I have to compile and upload the site actively.

Base url

Basically this is fucked up. Jekyll doesn’t properly do the base url, so you got to work around it. Add a ‘site.url’ and ‘site.baseurl’ in your config file, and have these added to the ‘base’ in your html-head. And then you have to make sure all links made by Jekyll have no slash in front with:

| remove_first:'/'}}

Changing links

So Jekyll doesn’t do auto-link creation for URLs in Markdown. Pity, parsedown did do it… I had to update all blog-items and change the links (www.example.com to <www.example.com>).

So over 90% of my visitors were Referral spammers

Wed, 26 Aug 2015 00:00:00 +0000

I thought my site was pretty well visited. Well, guess again :(

I started noticing something strange in the ‘Behavior Flow’ in Google Analytics (GA). All my visitors stayed on http://www.theozwinderman.nl/, for about 5 minutes and then left. Very strange, because on www.theozwinderman.nl/ there is only an index.php that immediately redirects to /newsite/index.php (a pretty strange setup I came up with 8 years ago).

So something was off!

After some more digging, I found out that most of the traffic came from ‘Macintosh’ users with a Google Chrome version at 40.0.2214.111 And these users were not from the Netherlands (as ’normal’ visitors would be, as it is a Dutch site), but rather the US or China. With this version-number I did some googling, and found out that this is referer spam.

Sure enough, the referal-section of GA was overflowing with source sites as get-some-more-traffic-to-your-website.com (eventhough that’s a fake url, don’t click it). Unbelievably stupid.

Anyway, I created my own ‘Segment’ in GA to get rid of this bullshit. I filter on a combination of the Chrome-version, the page-title (these spammers usually give up ‘home page’ instead of the real page title) and then some keywords in source like ‘seo’ or ‘buttons’.

Now only about 13% of my visitors remain…

Segment I use to filter spam: #

Filter out traffic not using a logical page title (every page on my website includes ‘Zwinderman’)
Filter out traffic using Chrome 40.0.2214.111
Filter out remaining traffic with a source containing ‘seo’ or ‘buttons’

One Dutch Song a Day

Fri, 12 Jun 2015 00:00:00 +0000

I made a nerdy present for my girlfriend. I give her a Dutch song every day!

The way it works is through an RSS-feed. For the newbs: that’s how we subscribed to news before Google, Facebook or Apple told us what we should read.

My server runs a Cron-job every day at 6 AM, which randomly picks a file from a ‘songs’-folder, adds the date as a UNIX-timestamp to the filename, moves the file to a ‘published’-folder and then makes an RSS-feed out of all the files in this folder.

So my songs folder looks like this:

de_dijk_binnen_zonder_kloppen
nielson_sexy_als_ik_dans
de_jeugd_van_tegenwoordig_sterrenstof

The published folder looks like this:

1433908861-de_dijk_binnen_zonder_kloppen
1433563261-nielson_sexy_als_ik_dans
1433736061-de_jeugd_van_tegenwoordig_sterrenstof

The files internally look as follows:

Band-name
Song-title
Youtube link

And from this an RSS-feed is generated.

Since everything works from Dropbox, I can add new songs virtually everywhere. And the generated RSS-feed is accessible even if my server is down.

Headphones with a lightsensor

Sun, 07 Jun 2015 00:00:00 +0000

Problem: somebody says something when I have my headphones on.

Idea: headphones that automatically pause the music when taken off (using a lightsensor) and resume when you put them back on.

Who will build these for me? :)

The basic premise is I want the headphones to pause the music when lifting one ear-shell off

See https://1001ideas.org for more of my “great ideas” 😅. I even published an updated version of this idea.

Making thumbs of entire photo library

Sun, 07 Jun 2015 00:00:00 +0000

Probably nobody else has this problem: my photo library is so large, it doesn’t fit on my laptop anymore so I put it on my server. But since I store my photos on a server I have less access to them. SSHFS is great, but still a bit of a hassle. So what I wanted was a compressed, but exact replica of my photo library on my laptop. I created a python script to recreate the folder structure and do the copying (keeping the original metadata: when was the photo taken etc.) and used ImageMagick to create a scaled down version of each file.

I have some 65.000 photos which totals at 234 GB (including video). On my laptop I wanted this to be around 10 GB.

Which format, BPG, JPG, or WebP? #

There are different formats to use for photos. JPG is old and the tradeoff between quality and size is not as good as what current technology can do. This is why there are more formats asking for the throne of JPG. Two major examples are WebP and BPG. WebP is developed by Google and therefor in a lot of products: Chrome and Android are two major examples. BPG is an even newer format, based on the HEVC video compression standard. Especially BPG looks like an awesome improvement over JPG, but it does not have a high support as of yet. Also, there are patent-issues with it, meaning that more support would take a long time. Lastly, browsers cannot display the file natively but need to use a javascript function for display. As I want to use the files for thumbnails, this means a lot of javascript-calls.

For me the fight was thus between JPG or WebP. Is the trade-off in size vs quality really worth the effort?

WebP vs JPG #

First things first: WebP does not work natively on the Mac, it won’t display in Preview or Quicklook (and plugins I found aren’t working). A major downside!

I have some 65.000 photos, so a difference in size might make me overlook this shortcoming. With JPG at 85% quality this would mean approximately 13.4 GB. An equivalent WebP setting would be ‘only’ 8 GB.

ImageMagick with WebP #

To install ImageMagick I used brew. To install with support for WebP (and tiff, because for some reason some of my photos are in that format), do:

brew install imagemagick --with-webp --with-libtiff

Comparing the results #

I’ve tried a lot of different settings for both WebP and JPG. I did a strip for all conversions to make all files smaller.

Here are some examples, going from low to high quality (click here for the original):

Obviously you will need a WebP enabled browser to see the photos

Q15 JPG (64KB)

Q15 WebP (45KB)

Q85 JPG (318 KB)

Q85 WebP (217 KB)

The first thing you notice is that WebP is a lot smaller than JPG (70%!). You may also notice the artifacts from both algorithms. For JPG it’s most noticeable in the flag in the top right. For WebP you can notice it in the stones of the building. The texture of the building looks a lot ‘smoother’ than in the original. I found this second artifact a lot worse, it’s very noticeable without zooming in. Maybe my eyes have accustomed to JPG-artifacts, but I think the JPG files are ‘prettier’.

I could not believe this and tried a lot more different options. I’ve found that setting the FS-option lower made pictures a bit sharper, so:

Q85 WebP EmulateJPG (289 KB)

Q85 WebP FS20 (228 KB)

Q90 WebP FS20 (310 KB, same size as JPG at Q85)

Q85 WebP Lossless (1.8 MB)

Q100 WebP FS20 (1.8 MB)

Everywhere the same smooth texture on the building!

Ofcourse I did not use only one image for comparing WebP and JPG. I’ve tried it with faces as well (same smoothness). Here’s two more images so you can see for yourself:

original | JPG Q85 | WebP Q85 FS20

Compression used #

In the end I went with JPG because to me it looks better, it has native support on Mac and the difference in size was not too drastic. After a bit more experimenting I settled on a quality of 75% because it provided the best quality vs size tradeoff. I made images 1024 high, so they would still look good on my laptop in fullsize.

I also applied a sharpening “trick”, which makes the photos look a lot nicer:

-unsharp 2x0.5+0.7+0

https://even.li/imagemagick-sharp-web-sized-photographs/

Update I’ve found that Graphicksmagic can make even smaller thumbs. Up to 13% with no noticeable quality difference.

SSHFS on the Mac (Mavericks)

Thu, 14 May 2015 00:00:00 +0000

Normally I think Digital Ocean’s articles are pretty good, but this time they’ve made a mistake for Mavericks (see their article here)

After installing SSHFS using OSXFuse (download here), you can mount as follows:

sshfs -o LogLevel=Debug name@server:/mountingstuff/ mountpoint/

Notice that you don’t need to use sudo (as in the digital ocean article) nor -o IdentityFile even if you have a public key login going on. I’ve added the LogLevel so that you will better notice when something’s going wrong.

And it works with google authenticator!

Because you didn’t use sudo, unmounting the file is easy also:

umount mountpoint

A Parsedown blog in Dropbox

Sun, 12 Apr 2015 00:00:00 +0000

update: I’ve now got this running in Jekyll, instead of the description below

My latest experiment, a blog in Dropbox!

I’ve copied all content from my other blogs to this one here, my new home. Hopefully this will be a bit more stable than the other solutions…

Why? #

There’s three answers to that question:

First of all, my old blog was hosted on blogger.com, which had the nasty feature of being linked with Google+ (and deleting a photo in Google+ means it goes bye-bye on your blog as well). Blogger is also not that flexible in it’s set-up and sometimes new “features” get added that I don’t want.

Secondly, I wanted a way to edit my blog offline. Anywhere. Which is possible with Dropbox. I can now edit my blog on my phone, laptop or tablet anywhere I am. (another bonus of Dropbox is that I don’t have to worry about website-traffic anymore).

Thirdly and most importantly, this is of course a fun little project to hack together!

How? #

This blog consists of small .md (markdown) files (one per blog-item), images and files I link to (like PDF-documents). All files are located on Dropbox (in the Public-folder, which only ‘old-accounts’ like mine still have). One a small php-script, is hosted at my own webhost.

When you go to https://www.matth-ijs.nl, the php-script requests a small file from Dropbox. In this file is a list of all blog-posts and their filenames. From this, the php-script creates a menu. It then also loads the most recent blogpost and shows that as first item.

This blogpost gets parsed, so that all links work, images are included and the markdown looks pretty (all handled by the parsedown library). As a final step I slightly modify all links so that they work nicely with Dropbox.

Thanks! #

I use the Parsedown library to parse the blogposts, and a modified version of the Solar Ghost theme of Mattvh (where the CSS is pretty, it’s Mattvh, where it’s ugly, that’s my fault).

https://github.com/mattvh/solar-theme-ghost

http://parsedown.org/

E763 error on Vim for Mac OS X

Mon, 23 Mar 2015 00:00:00 +0000

I had this on Yosemite, but probably the same error exists on Mavericks, Mountain Lion, Lion, Leopard, Snow Leopard, etc.

When you’re working in Vim in the Terminal, at one point you might want to use the spell checker. And maybe you want to do this in another language than English. Vim nicely downloads your language files automagically (with set spell), but then you’re SOL:

E763: Word characters differ between spell files

The problem is Vim ships with an old version of the English-language files (the file has an old encoding). Long story short, nobody explains how to fix this on OS X, so here it is.

Update #

The method below no longer works, because Apple has changed the working of sudo. Some files are no longer allowed to be edited even by a SuperUser. Anyway: Mac ships with Vim 7.3, eventhough 7.4 is out. And Vim has been surpassed by Neovim. So use Brew to install Neovim.

First we move the old language files to a temporary location (where they’ll be auto deleted when you restart your Mac):

sudo mv /usr/share/vim/vim73/spell/en.* /tmp

Now we re-download English and other languages in Vim (it automatically makes .vim/spell for you where it’ll store the language files). In vim, set:

:set spell

(downloading English automatically)

:set spelllang=nl

(and the language of your choice)

You can set these commands in your .vimrc so they’re always on for you.

Solarized in Terminal Vim

Mon, 23 Mar 2015 00:00:00 +0000

To get pretty colors in Vim and your terminal, install this: https://github.com/tomislav/osx-terminal.app-colors-solarized

Ofcourse the install doesn’t go as planned, you’ll need to add this to your ~.bash_profile:

export TERM=xterm-256color

# And if you want colors for 'ls', also these:
export CLICOLOR=1
export LSCOLORS=gxBxhxDxfxhxhxhxhxcxcx

I’ve pimped my Solarized further by using “Andale Mono 13pt.” as the font and a background opacity of 98% with Blur at 13%.

update: I’ve started using nvim, and am very happy with the switch

A quick review of Webexpo Prague 2013

Thu, 26 Feb 2015 00:00:00 +0000

This article is also published in the Angi Studio Medium collection

In 2013 I visited Webexpo in Prague. A conference about the web that mixes design, business, life hacking and IT. It was the first time that the conference reached out to foreigners, by inviting international speakers and making all talks in English. I had fun and learned a lot, but: is it worth attending again next year?

The mix of design, business and IT was very welcome. I learned most from talks outside my own area of expertise. That’s also what attracts me in interaction design: talking with clients and users, and learning new things from different subjects. I talked with everybody in the conference, from the speakers to the guy making the tea. The conference had a very open and relaxed atmosphere and there were several cool things besides the talks such as: 3d printing, Google Glass, and even a professional masseur. Another neat idea was the discomfort zone, where people were allowed to debate freely and play devils advocate without consequences. Also great were the after-conference parties and events. Sadly, the design workshop on Thursday didn’t happen due to lack of attendees.

A game played by not thinking too hard. Ideal for me

So what about the talks? #

The quality of the talks was very good overall. But it seemed that every design or business presentation has lists of “ideas”, “key takeaways”, “guidelines” or “insights’. A lot of them seemed trivial or made up on the spot (“listen to your users”, really?). It would have been better for most talks to stick to one message! Now, I’ll break that rule straight away and present some take-aways :)

A great story is better than a 1000 words #

Marli Mesbov gave a great presentation on the importance of story telling. Stories work better than short tag lines, which is probably related to Key Take Away 1 (no take-aways).

Stories work for conversion and for your clients to positively remember you. Do you remember the fairytale of the boy who cried wolf? Don’t you think that carries the message better than “don’t alarm us for nothing”? The average attention span went down from 12 seconds in 2000 to a mere 8 seconds in 2012, so better make your story captivating!

Besides working on your site, stories also work great for doing design. It’s an old saying, but still good advice: instead of designing for users, design for a story.

Lastly, a story can be used to instruct a client on how to use the website. Instead of delivering a 100 page styleguide which people don’t read, you can make a short story, a trailer or just a word (“dictionary”, “phonebook”), which ’explains’ the site to the client.

Marli Mesbov shows that a story doesn’t need to be long to be enjoyed and remembered

Our terms suck #

It was great to hear some self-conscious critique on terms we use within the field. Our terms are confusing and their meaning differs per person and company: UX, service design and even interaction design (but less). Some people include visual design in there, programming, marketing… All very confusing! And what to think of terms like ’expert’ or ‘guru’? Instead, it’s better to describe what you do and how it benefits the client or project.

Death by a million paper cuts #

Bad user experience costs money, which is what we tell our clients also But it was great to see some examples, such as Expedia losing 12 million dollar because of a form field or Best Buy losing 300 million dollar by requiring an unnecessary login. This holds especially for mobile use: a lot of apps are used only once or thrown away within a month. Avi Itzkovitch had some nice examples how to overcome such bad experience by making more use of context. For instance by adapting the interface based on location, time of day or usage pattern.

So, should you visit next year? #

I wish it could be a wholehearted yes! I had a lot of fun, met great people and learned a lot, but.. Some things nagged a bit, like the cancelled design workshop, the open doors in some design talks and the endless lists of recommendations. Another thing is the overlap in the program: breaks were different for each track making it difficult to switch. That might be why designers mostly went to the design track and programmers to the development tracks, making the mingling a lot less. On top of that, all speakers were “safe choices”, nobody went really overboard or crazy with some thought-provoking talk. Still, it was fun and I learned useful things, so, I guess it’s four out of five stars.
For a visual impression, see the Facebook photostream

Mariya Yao explains why some apps are not making money in a convincing talk

Talk guide: #

Webexpo puts the talks online (very slowly), so which ones should you see?

Marli Mesbov: how to construct and use a story - INSIGHTFUL
Riki Fridrich: how to get to the job you want - INSIGHTFUL
Jesse James Garrett: design for engagement - INSIGHTFUL
Mariya Yao: design tips for mobile apps, very nicely supported with numbers - INSIGHTFUL
Avi Itzkovitch: you can change the UI based on sensor data, by adapting to the user context - PRETTY INTERESTING - but skip through it first, to see if it’s your thing
Eirik Hafver Rønjum: designing the organization - NICE - but toooooo long, so skip through it a bit
Steve Corona: scaling websites is easy - Matthijs’ stamp of approval
Petr Kosnar: you see the weirdest things if you pretend to be a 14 yro girl on Facebook - Matthijs’ stamp of approval
Ron J Williams: rethink your marketing strategy - Matthijs’ stamp of approval
Vojta Rocek: having your own bar does not make you happy - Matthijs’ stamp of approval
Tomas Slavicek: being a digital nomad is pretty awesome - Matthijs’ stamp of approval
Jiri Sekera: on the term UX - FUN
Wolf Becvar: identify customer touchpoints - OKAY
Jean Trojan: what does it mean to be a guru or expert - MWOAH, I think sometimes ’expert’ makes sense, especially inside a company
Sabrina Majeed: when to use guidelines and when not - MWOAH - a lot of open doors, packed with some great content, if it was shorter would have been okay
Wodtke, Bermes, Abley: I haven’t seen them, but recommended by others

This article is published in the Angi Studio Medium collection

Google Play Quality Sucks

Wed, 30 Jul 2014 00:00:00 +0000

I was listening to Google Play Music, and I realized the quality sucks.

An example is the album “Awaking the Fallen” by Avenged Sevenfold. You can clearly hear the distortion on each track, but let’s take track 9 (“I won’t see you tonight part 1”). Skip to 7:45 and you can hear a watery distortion right up to the end of the track. This watery distortion reminds me of bad MP3 rips of old days past.

My problem is solely with the quality of the encoding. Google uses 320kbps mp3, which should be enough, but somehow it isn’t (tried both in HTML5 version and Flash). To compare, I took part of one song from Google music and compared it with an mp3 encoding i made myself from a lossless source (using 320 cbr with LAME). The difference is very noticeable. To clarify: the song comes from Google all access, I didn’t upload it myself.

Listen for yourself:

Or try listening to this on Google Play: Pink Floyd - Time at 7:05
Just yuck!

Why do I pay for music, when I get better quality by downloading it ‘illegally’?

Google’s answer: “… we must communicate with music labels to correct the content…” So they rely on the music label to provide properly encoded music?

update Spotify has way better quality music, as long as you turn off ‘Set the same volume level for all tracks’ and turn on ‘high quality streaming’. Apple Music also has way better quality music. Both “Time” and “I won’t see you tonight part 1” sound a lot better.

(want to rip MP3 from Google directly? Right-click inspect element anywhere on the page, go to network page and find the streams there. Download with Curl or Wget)

update2 Apparently this is most likely audio watermarking. Stupid. http://www.mattmontag.com/music/weird-spotify-compression-artifacts http://www.mattmontag.com/music/universals-audible-watermark

the audio is obviously not under the Creative Commons license

Yay! A new tracking record

Wed, 30 Jul 2014 00:00:00 +0000

I accidentally forgot to turn off tracking (ghostery).

Amazingly, Songmeanings.com has achieved to activate 56 trackers on a page… (actually, it was 77 first time I loaded it, but I refreshed to take the screenshot)

“Ich will dass ihr mir vertraut” –> yeah right, don’t think so!

Page: http://songmeanings.com/songs/view/16022/

Update, new offenders: #

Answers.com (87 trackers + an annoying autoplaying commercial video with sound)
wallstcheatsheet.com (with an unbelievable 102 trackers!!!)

Interest (rente) at Rabobank

Tue, 29 Jul 2014 00:00:00 +0000

Pretty soon I’ll be paying Rabobank to keep my money there :(

I remember it being 3,1% when I started

Broken Ubuntu

Mon, 14 Jul 2014 00:00:00 +0000

So I haven’t used Ubuntu for a while… But I installed it on my new Intel NUC. And my God, what have “they” (=canonical) screwed things up.

What follows is a list of my irritations, hope it’s useful for some of you.

By default Ubuntu uses vino as vnc-server (aka remote desktop login). Sadly, this thing is broken by default, it uses an ‘unknown encryption’

The internet offers a slew of options, nothing worked out of the box. What finally got everything ticking was this set of steps:

Reset dconf (to turn off encryption, just google it)
Reboot
Set a password (otherwise it still won't work)

But don’t bother. Because as soon as you have set it up, it’s slow as hell.

The only option is to use X11vnc, with noxdebug option on. Then it’s semi-usable. Unbelievable. I remember this working smoothly 10 years ago! How bad can you break things?

Oh and the fancy feature X11vnc wants you to use, ncache? Also broken :( Makes x11vnc crash…

Also none of the instructions online for starting a VNC session before having logged in work. Right command is:

sudo x11vnc -xkb -auth /var/run/lightdm/root/:0 -noxdamage -rfbauth
~/.vnc/passwd

But probably Canonical will make sure this command won’t work next month.

Locale #

My locale is messed up and I can’t fix it. I’m not the only one with this problem, as a Google search will tell you!

I have tried everything:

locale-gen
dpkg-reconfigure locales
install --reinstall locales

Still not fixed. Locale keeps showing LC_ALL and LANGUAGE to be empty, which means I get screwed up error messages like

exception:
locale::facet::_S_create_c_locale name not valid

(from Grive) or

set LC_CTYPE
to default locale: No such file or directory

Sh*t.

The only option is to do an export EVERY TIME (or put it in your bash_profile):

export LC_ALL="en_US.utf-8
export LANGUAGE="en_US.utf-8

Which officially I shouldn’t be doing, because it’s not good to overwrite your language settings.

Configuration files all over the place #

At least one thing that was consistent was that you could count on stuff being in the right place. Well, not any more. For instance, look at the mess in /usr/share/lightdm/lightdm.conf.d/ Why do I need so many config files?? What do they all do?

50-greeter-wrapper.conf 
50-guest-wrapper.conf 
50-ubuntu.conf 
50-unity-greeter.conf 
50-xserver-command.conf

Which leads right into the next frustration:

The settings file of Transmission #

This one is an even bigger mess. It’s completely unclear which settings file the Transmission-daemon eats.

It could be either one of these:

$HOME/.config/transmission-daemon/settings.json
transmission-daemon/.config/transmission-daemon/settings.json (if you run it as
a separate user)
/var/lib/transmission-daemon/info/settings.json (a symlink in ubuntu)
/etc/transmission-daemon/settings.json
/var/run/transmission

Which one is actually used depends on how you start the daemon. Running it using automatically at boot using init (which there’s also two locations for in Ubuntu, /etc/init and /etc/init.d, only the last one does anything as far as I can find out), starting it manually using init or manually from the command line. The difference between manually or automatically running using init is because of drive encryption: if transmission can’t find the file, it’ll fall back on /var/lib. What a huge mess!

I have set all settings files to log stuff to the /tmp/ folder. This way I can at least see when something goes wrong. Was also the only way to finally figure out the mess.

Other fun stuff:

You can’t edit the file with transmission running, because it’ll overwrite it when you stop the service.
Transmission will sometimes overwrite the RPC-whitelist to only include localhost. No clue why, it’s the only setting that’s written over.
Transmission itself won’t warn you if there’s already a copy running, so use ps aux to check whether this is the case, especially in combination with bullet number 1…

Services sometimes not loading on startup #

This ones a bit of a cr*pshoot. Sometimes it works, most times it doesn’t. My Plex server should run on startup (eg. after logging into the graphical shell). I’ve added it too the startup services, which Ubuntu doesn’t really seem to care about.

Most likely has to do with the encrypted drive…

Shutdown not working as expected #

I used to use shutdown -h now, but they’ve gone ahead and broke that. The machine doesn’t power down, but instead lands in a nonfunctional terminal. Now we have to use poweroff. Which also doesn’t work, because Intel has decided that NUCs should reboot when powering off this way. :(

Darling Violetta, you awesome

Thu, 26 Jun 2014 00:00:00 +0000

130th time I’ve heard Parlour now… Where’s your new album???

Listen on Youtube (kindof)

Don't see the autologin option in Ubuntu?

Thu, 19 Jun 2014 00:00:00 +0000

So you want your computer to autologin and you think you are crazy because NOWHERE can you find the stupid option to do so eventhough every google link gives you the same “it’s so easy” answer?

But why don’t I have that option?

Or you know, you could edit one of those config files which Canonical keeps moving all over the place? (the internet suggests it should be in lightdm.conf or in grub)

And even after you found the config files and changed it… Still autologin doesn’t work on Ubuntu!

Well, you are not crazy

What you probably did was enable fulldisk or home-folder encryption. Of course autologin then doesn’t work… It is possible to remove this post-install, but be prepared for some terminal incantations, sweating and swearing.

At least now you know you were not alone in this problem. Good luck!

F*ck it! I'm rooting for Cersei from now on

Thu, 05 Jun 2014 00:00:00 +0000

Screw you, Game of Thrones. Killing off all my favorite characters.

As a protection from myself, I’ll start cheering for the baddies. F*ck it.

Cersei, you go girl!

Good luck killing all the nice people.

the image is obviously not under the Creative Commons license

Disruptive Innovation: Kenya vs the Netherlands

Fri, 27 Dec 2013 00:00:00 +0000

This article is also published in the Angi Studio Medium collection

Why it’s possible to pay your groceries by mobile phone in Kenya, but we’re stuck with a debitcard in the Netherlands

Since at least 2007 it’s been possible to pay by mobile phone in Kenya. Pay your electricity bills, transfer money, pay in the shop… You don’t need paper money or a debit card in Kenya, most people don’t even have a bank-account (only 20% of Kenyans do). All banking can be done with the mobile phone and if you need cash, you just go into a phone-store and get the money off of your phone. The system is called in M:Pesa, it works on any phone and only needs 2G.

Compare that to the Netherlands: we don’t have one simple solution for paying. We have big wallets filled with debit cards for the stores and cash for when cards are not accepted (for instance due to lack of internet), and maybe some people have a credit card or chipknip. We have online banking for paying and transferring money, but most of the time we need our card or a little security machine to log in. And ‘our’ mobile banking only works when you’re online and have smartphone with Android or iOS.

So is Kenya miles ahead of us, or is something else happening there?

Disruptive innovation #

Already a while ago I attended two presentations by Franco Papeschi and John Kieti during Design by Fire on how Disruptive Innovation was happening in Kenya. I’ll mainly talk about Franco’s presentation here, but I also learned a lot from talking with John afterwards.

As Franco explained it, disruptive innovation is using existing technology or methods for a new purpose. For example in the case of Wii vs Playstation 3: Sony went out of it’s way to create a massively improved gaming device, pushing the boundaries of technology to create something that could cater even better to it’s existing audience. Instead, Nintendo went the other way and used existing technology to reach a new audience. The devices have a different audience and were successful in their own way.

I got really excited by this way of thinking: mainly because of the great examples in Kenya, but also because it shows that you don’t need to be a big company to do something cool, sparking enthusiasm to do it myself (I’ll keep you updated on that :) )

In Kenya they’ve set up labs (such as M:Lab) to help young Kenyan entrepreneurs to do this, and it’s a huge success. Some projects get millions of USD in funding.

It is great to see what kind of impact you can have by innovating this way. You don’t need to study rocket science to implement a creative idea that has a huge impact on the people using it.

Extending your drum lessons to all over the country using Skype ([photo from Franco’s presentation] (http://www.slideshare.net/bobbywatson/designing-impactful-experiences-starting-with-nonusers))

Kopo Kopo: an initiative to make it easier for small shopkeepers to accept M:Pesa money (photo from Humanipo)

Eneza: engaging kids with school assignments while they’re at home (photo from Franco’s presentation)

M-farm: a way to check the price of their produce on the market, making sure farmers (75% of Kenyans) get a fair price for their work (photo is a collage from Sven Torfinn/M-Farm and M-Farm)

Using existing technology for a new goal: disruptive innovation

This is what happened with mobile banking: Kenyans started using minutes they bought as money and from there on a system emerged that used sms for transferring money. This grew to a system developed by Safaricom (a telecom operator).

The impact for Kenyans #

75% of Kenyans work in agriculture and they’re not paid a lot, mainly due to middle-men taking a high cut. Electricity bills are paid in advance every month and when the electricity goes out they have to travel to the city, wait in line, pay with cash, and travel back: it easily takes a day.

With M:Pesa it’s possible to send one text when the electricity runs out and it goes back on again. It’s also possible to have some savings this way. And when the middle-man comes, you can check the price of your crops in the city using your mobile before you agree on the price. This means getting a proper price for the crops.

This is a huge impact: Kenyans have more time and money to spend on (for instance) family health and education.

Disruptive technology in the Netherlands #

It would really thin out my wallet if I could transfer money by phone in the supermarket and save time if I wouldn’t need wait until I’m at home to transfer money to my friends (yeah, yeah, I should buy a smartphone). And why do I need cash for the ice-cream truck?

Regulation #

The main thing is that for banking there’s a lot of regulation involved. Banks need to comply to a whole set of rules and laws: national, European and international. Furthermore a bank needs to get a permit from the central bank (in the Netherlands this is the “Nederlandsche Bank”). Of course this is also the case in Kenya, but M:Pesa flew under the radar of oversight for a very long time and got a big base of operation that way. It’s still not classed as a bank in Kenya. Existing banks in the Netherlands would never allow a mobile operator to do this.

Necessity #

Between online banking and the debit-card we have almost anything we need already. In Kenya people were denied bank-accounts and trips to the bank would take a day worth of travel anyway. Therefor their necessity is much higher.

Polished expectations #

We’re used to more polished solutions: something has to work straight out of the box, in a nice way. That’s great, but it does make disruptive innovation a lot more difficult.

Everybody can do disruptive innovation #

So maybe banking is too big of a step to try and tackle right now, it doesn’t mean that disruptive innovation is impossible, because we have the same tools to our disposal as the Kenyans do.

Some examples of success are Marktplaats and Craigslist (both simple auction sites made with existing technology), Chatroulette (random videochat, made in two days), Postcrossing (send and receive cards from all over the world, started on an old computer), the descriptive camera and Tinyhabits by BJ Fogg. Tiny Habits especially shows you can have a great impact on people’s lives using current technology in a really basic way. BJ only uses mail and MS Excel.

So look around at all the great stuff that’s available around you: video-calling around the world, sensors (GPS and NFC are pretty cool), web-services, existing infrastructure (snailmail, phones, texting) and of course the human-element.

Stop reading and connect some of those services already!

This article is published in the Angi Studio Medium collection

these images do not fall under CC0

Een Eigen Dierentuin voor Tim - Tim Wants a Zoo

Mon, 21 Jan 2013 00:00:00 +0000

Front cover of the book

“Een Eigen Dierentuin voor Tim” is a Dutch children’s book meant to be read to small children. It was written and illustrated by my father (Theo Zwinderman) and published by me to the Apple iTunes Store.

Tim has two passions: drawing and coming up with crazy ideas. This time Tim is drawing animals and he comes up with the crazy idea of having his own zoo. His two mothers fully support him and agree with putting the animals all over the house. But, is it even possible to keep an alligator in the basement?

My role in publishing the book was scanning in the illustrations, making sure the colors appear nicely on iPad screens, proofreading and publishing the book in the iTunes Store. This meant calling the IRS and securing an ISBN. I’ve also translated the book to English, and when this version is ready it will also appear on my blog.

So go and download the book to your iPad today!

https://itunes.apple.com/nl/book/een-eigen-dierentuin-voor-tim/id594441001?l=nl&ls=1

And the English version:

https://itunes.apple.com/us/book/tim-wants-a-zoo/id595338981?l=nl&ls=1

these images and and the book are not under the Creative Commons license

Four-in-a-rowbot

Tue, 08 Jan 2013 00:00:00 +0000

Four-in-a-row or Connect-Four is a popular children’s game, where players in turns insert ‘coins’ into a contraption and try to connect their color coins in a row (check Wikipedia for the exact rules). As the game is quite simple, I implemented it on an Arduino to create the first ever* Four-in-a-rowbot!

The Four-in-rowbot can play against humans through a connected computer-interface or with implementations on another Arduino. The latter approach is what I had in mind; other programmers can try to compete with the implementation I created.

I chose the game Four-in-a-row because it is a simple turn-by-turn game, which still has some interesting game dynamics (you have to calculate a couple of steps ahead). The game had to be implementable in Arduino’s limited memory and at the same time not be completely trivial.

To help fellow-programmers, I provided a base-implementation with networking code they could use to extend and then use to ‘battle’ against my implementation. The intended competition never happened, but I did learn a bit more about Four-in-a-row’s optimal playing styles and Arduino’s RX/TX protocol.

If you’d like to experiment, you can download the Arduino-code.

Also, my Four-in-a-rowbot could beat Daniel’s implementation ;)

*I guess

Bieberfree Youtube (2011)

Thu, 03 Jan 2013 00:00:00 +0000

I got annoyed from all reading useless Youtube comments that surrounded the very few useful ones. Messages like “Bieber sucks / rocks”, “I got shot in the knee”, “thumbs up if you … “, “lol”, MESSAGES IN ALL CAPS and more of that stuff.

So more as a joke than anything, I implemented a filter for Safari. It’ll make your Youtube experience a lot better. Silly comments are hidden, the rest stays. Hidden comments can still be seen when you select them.

(the code is no longer available)

(Code is heavily based on this script for Digg on Firefox)

Risk 3D (2010)

Tue, 01 Jan 2013 00:00:00 +0000

First time playing Risk3D

Traditionally, Risk is played as a board game on a flat 2D surface. Ofcourse such an interface gets boring after a while :)

Together with my brother, a.k.a. partner in crime, Remco, we completely reinvented the game so that it works on a much cooler 3D surface. The game is completely playable and offers many ways to extend the original game (we added Pearl Harbor, for instance). And last but not least, it is a lot of fun to play this way :)

Painting the globe

The world is as realistic as possible, we mapped the positions of all countries on a ball with the help of a map. On those positions we added our revolutionary system for counting the armies, a protruding stick. The small ball was enlarged with liquid plastics and then rounded again. As a final step we painted the world in bright colors, showing the continents. To mark who owns which army we made flags, for up to six players!

AI - Master Project (2009)

Thu, 27 Dec 2012 00:00:00 +0000

In my eight month master-project I looked at human-robot interaction. Humans can easily distinguish one object for another, but this is difficult for robots. I created a method with which a human teacher can easily designate and segment an object with a laser pointer, so that a robot can easily ’learn’ this object using active vision. The thesis shows that the robot can recognize these objects in real-world conditions. The recognition results described in the thesis have been improved upon significantly after the thesis was finalized. The work was presented at the RO-MAN conference in Italy.

Overview of human-robot interaction

In this thesis, methods are presented that allow a mobile robot equipped with a stereo camera to automatically learn an accurate SURF-keypoint based representation of an arbitrary object. In the approach, a person designates the object to be learned with a laser. By using active vision to filter keypoints, the resulting object representations are robust and recognition time is considerably reduced.

The segmentation method was tested on an extensive set of 7 objects, while the creation of object representations and the recognition thereof was tested on a set of 21 objects. The objects vary greatly in size, shape, color and texture. This dataset is considerably larger than those used in similar research.

It is shown that by filtering the keypoints using the human segmentation and active vision, the number of keypoints can be greatly reduced while not decreasing recognition accuracy. The recognition was further tested on scenes representing typical office scenarios. It is shown that object recognition works reasonable in highly complex real-world environments, with lighting changes and object occlusions, even when using only one view of the scene.

My advisors for the project were Dr. Gert Kootstra, Prof. Dr. Schomaker and Dr. Paul Rybski.

Download the thesis from the university or from this blog.

Working with the robot

the images and master-thesis are not under the Creative Commons license

Games

Tue, 25 Dec 2012 00:00:00 +0000

As part of my scouting career (8 years as scout-leader and 20 years as member) I’ve invented a lot of games. These are all in Dutch, but if anybody wants, I’ll translate them to English.

I’ve made a selection of three successful games that can be played straight away. In the future, I might publish some more.

update 2018: here are some more games, and part three

“Heksenhandelsspel” - Witch-trade-game #

A tradegame for children in the age of 7-11, but with some adjustments it could be fun for all :) It is an active, outdoor game that needs at least 6 people to guide it, playing time is about an hour and preparation time also. We’ve played the game a lot, with groups ranging from 20 to 50 children.

Download the PDF-file with description (Dutch).

Maffia-themed roleplaying game #

A roleplaying game for 16 and up. Somebody has stolen drugs and money from the maffia and the Don requires help from his maffia-groups to find out whodunnit :) There are 6 roles in the game and playing the game takes about an hour. I wrote the play completely by myself for a group of nine players. It was a huge success! The roleplaying game is now played yearly with a student group and was performed with some other groups, with over 50 people.

Download the file here (Dutch).

Spy game #

A game played with walkie-talkies, for ages 16 and up. Requires at least an hour of preparation. There are as many groups as there are walkie-talkies and one of these groups is the ‘spy’. The spy tries to find encrypted messages that are hidden on trees and send the message back to the ‘base’. The other groups are anti-spy teams and can hear all the communication on their walkie-talkie, they have to capture the spy (after which the anti-spy team that caught the spy becomes the spy themselves).

It is a very exciting game that is best played in the dark.

Download the description and ‘secret word’ generator (Python)here.

Gold'n Raand Hike (2009)

Thu, 20 Dec 2012 00:00:00 +0000

Every year the nation-wide broadcasting station 3FM organizes a charity event called “Serious Request” in collaboration with the Red Cross. The listeners can request songs for money and a lot of clubs and organizations do their own events to gather money. In 2009 it was located in Groningen, which is why we organized an event as well. We walked the border of our province (316 kilometers) with 200 people and gathered 9338.65 euro’s!

The event was organised by me and four scouts of Scouting Pekela, the local scouting organization. I was in charge of raising funds and media-attention as well as organization.

The event was a huge success! We have been on national television, national radio, regional television and radio, regional newspapers, local newspapers, many websites (national, regional and local), we got to present our project on stage during the ending festivities (in front of approximately 10.000 people) and we gathered a huge amount of money.

It was a lot of fun.

The Game (2008)

Tue, 18 Dec 2012 00:00:00 +0000

In 2008 I won a competition for submitting the idea of a game platform that relied on location-awareness. Simply called “The Game”, it is a platform for various games where points and achievements are registered centrally.

The target group for The Game is everybody. The Game would be free for everybody and supported through advertising, rebranding of games and ‘interactive commercial-deals’, such as “the first 50 players arriving at McDonalds will receive a free hamburger”.

Example games would be a hunt for an object or location, quests, puzzle tours. An assignment could be “take a photo of …”, “capture the flag by staying at a location for a while” or “find your buddy”. But other applications could be e-dating, social networking or a virtual market place.

AI - Zeppy (2007 - 2008)

Thu, 13 Dec 2012 00:00:00 +0000

Zeppy is a robotic aerial vehicle, a blimp with onboard sensors (a camera and sonar), onboard processing and a real-time videolink to a groundstation. The project received funding from the Artificial Intelligence department of the University of Groningen and was showcased at CeBIT (the largest IT-fair of the world!), the opening of a new university-building, regional television and during two open-for-all presentations.

Together with Jaldert Rombouts, Tessa Verhoef and Tijs Zwinkels I started the project with a small retail zeppelin, under supervision of Tijn van der Zant. The main aim of the project was to acquire more knowledge of (robot-) hardware and to study the effect of being able to manipulate both software and hardware, i.e. embodied cognition (real intelligence needs a body). What a robot is capable of does not rely solely on it`s software architecture!

We started with a toy blimp, but soon nothing was left from this original zeppelin. We made the balloons ourselves, we replaced the motors, the onboard electronics, everything… The onboard processing was done with a Gumstix board together with a Robostix unit, our two sensors were a ‘spy camera’ and sonar.

On the software side we wrote our brain architecture in C++ and OpenCV (from Intel) for image processing. Zeppy was capable of recognizing direction, face recognition, edge-detection and did elementary motion detection.

More information can be found on the original weblog: http://www.zeppytheproject.nl

AI - The Da Vinci Code game (2007)

Tue, 11 Dec 2012 00:00:00 +0000

The goal of the “Da Vinci Code” game is to find out which numbers the opponent has on its ‘stones’ before he guesses the numbers on your stones. We modeled the reasoning in Java and it is possible to play the game online.

A project done during “Multi-Agent Systems” during my master Artificial Intelligence. Together with Jaldert Rombouts and Tessa Verhoef I modeled the “Davinci Code”-game. The computer-player reasons about its own ‘stones’ and about the knowledge it has of the human-player’s ‘stones’. It then tries to optimize its strategy by reasoning what a human-player would do based on the knowledge that player has.

You can’t play the game anymore, but the usability of the web-application was sadly far from desired, and I have neither the time nor the original source-code to improve it.

AI - Coevolution and speciation (2006)

Thu, 06 Dec 2012 00:00:00 +0000

In evolution new species can develop in different ways, for instance because part of the species ends up in a different habitat. In this bachelor project we tried to prove that speciation is also possible in another way: coevolution of predators and prey.

The final part of the Artificial Intelligence bachelor is a project, with accompanying thesis. Together with two friends, Jaldert Rombouts and Tessa Verhoef, we tried to see the influence of a predator-prey dynamic on speciation by simulating evolution. To assure that speciation would occur from co-evolution the model did not have location. In our simulations we saw that speciation indeed occurs within a wide range of parameters. Our advisor during this project was Bart de Boer.

On the right is an example of the output our software (written in C#) gave. It shows the ‘prey’ on the left (in blue) and the predator on the right (red). The vertical axis is time (every dot is an individual on this timeline), the horizontal position is a certain variable indicating the defence mechanism of the prey. What can be seen is that the defence mechanism drifts away and the predators adapt to this, this is why both lines show the same pattern in general. Half-way thru, however, a speciation can be seen, the blue line diverts into two different species.

The report can be downloaded here (Dutch). This report and the picture in the top are not under the Creative Commons license

Welcome to my new 'home'!

Mon, 03 Dec 2012 00:00:00 +0000

This is where I will list my personal projects and amazing ideas from now on.

So what will come here? The answer is anything and everything. Previous projects will be republished here as well as new projects and ideas.

Stronk #

Stronk is Dutch for ‘stump’. That definition is not very sexy, but I like the strong sound of the word.

Symbolically it also stands for felling down the forest that was all my previous websites and blogs and this being the new offshoot from that rampage.

I know it is probably a capital offense in the blogosphere to republish content, but this is my blog and I make the rules!

Have fun.

Silly bike parking things

Wed, 08 Jun 2011 00:00:00 +0000

In the Netherlands, nearly everybody rides a bike (on average we have more than 1 bike per person!). No wonder that you can find bike parking spots everywhere, like the ones in the photo up here. Actually, this photo shows two different types of bike parking systems.

This is one I found at the trainstation in Eindhoven. In the top photo you see the regular parking spot, in the bottom photo the “new and improved” ones. The old parking method can house nearly twice as many bikes per meter as the new parking method. Seeing as that there are not enough spots anyway, I think this is a pretty lousy “improvement”.

On top of that, the new method also relies on technology that can (and does) fail, can only be used for a short duration (after that your bike is no longer locked!) and is only available for people with a citizens-card. You get that card for free when you register for living in the city, but who wants to keep another card in their wallet just to park their bike?

Solution #

Remove the “improved” method… Why do we keep putting expensive new designs in places were they are not necessary??

Coffee Machine at TUE

Sun, 05 Jun 2011 00:00:00 +0000

After the ticket-machine at the zoo and the locked-in 24h machine, I now have a new addition for the unusable machines collection :) This time it’s a vending-machine for something very essential: coffee!

Aside from the number-based interface (which will maybe come another time), there is something else very wrong with the design: where do you put your money?

You’ve got 10 seconds to think about this…

…

No, you’re wrong: it’s not the grey thing that looks like the money-slot! :D You actually put your money in the small, nearly-invisible black slit on the right of the number-pad.

The grey-hole is for the “TU/e Coffee Card”. I don’t know anybody who has this card, or anybody who knows somebody who has this card. Anyways even if you had one, you can’t use it since the grey slots are usually filled with coins :)

On other, similar, machines they’ve added some stickers. But of course nobody reads the stickers, and so people keep losing their money by throwing it in the wrong slot.

Solution #

A simple user test would have shown that this design is misleading. But in general, it would be wise to always keep affordance in mind: if a user can put money in a card slot, he will. Not because he’s stupid or malicious, but because putting money in a machine is not the most important thing in his life at that moment. So we should change the “mouth” of the card-slot so that it doesn’t afford to put money in anymore. And besides that the money-slot should be moved to a more convenient location and should be made much more visible. And then of course, we test it again :)

Confusing streetnames

Wed, 01 Jun 2011 00:00:00 +0000

We’re looking at a crossing today, in Scheemda. We’re driving on the main road and to the left and to the right of is a street, each with a different name. Streetsigns are blue in the Netherlands and usually located on a streetlamp, I hope you can spot them on this small image. Otherwise, enlarge the image by clicking on it!

Now imagine you’re looking for a street called “Kwekerslaan”, which you know from your map is on your right. It would be logical to look at the streetsign on the right, right? :)

The municipality of Scheemda disagrees! The streetname for the right street is on the streetlamp on the left street… I’ve included a close-up to make this clear. Notice the tiny, tiny arrow in the sign?

Solution #

Especially for road signs, creative solutions are not a good idea. Drivers have a complicated task to do anyway and every second they have to spent on finding and understanding a roadsign is a second not looking at other, more important things. With 30 km/h, this means a car has travelled 8 meters with the driver not spending his cognitive attention on the road. The convention is to put streetnames on a streetlamp close to the street to which they belong, and this is yet another case where we shouldn’t have deviated from conventions.

Slideshare.net and Squirrel

Mon, 30 May 2011 00:00:00 +0000

Hi all, it’s time to start blogging again ;)

Today we have two confusing interfaces. These are the financing application for OSX called “Squirrel” (https://www.squirrelapp.com/) and a website for sharing slides (https://www.slideshare.net). Both are nice applications! But I’ve been confused by their use of very simple symbols. Maybe you won’t be confused, but remember that you’re in the mind-set of things being different than they seem ;)

First, slideshare:

Which button is for zooming in? :) I’ve blocked the text, so think first about it before seeing if you guessed correctly

I got confused multiple times. Especially if you don’t think about wanting to search through slides. All the items on this page are related to the current slide, not for searching a specific text… (Update: they’ve changed it now, and ditched both zooming and searching)

Next, Squirrel:

Squirrel is a nice application with which you can manage your financials. It can import from your bank, but sometimes you have to add a transaction by hand. In the image below, I wanted to schedule a monthly transaction. But money can go both ways of course: it’s either added to my account, or subtracted. So which button do I check if the money comes to me?? Did I check the right button, or did I check the outgoing-money button?

Again, I was really confused! I actually made a mistake with this that took me over an hour to figure out! :D Actually, the rest of the program is using color coding to avoid such mistakes. You’ll find the solution to the earlier question by looking at a different part of the interface.

Solutions #

Symbols are always difficult. Include a text-label always, or use conventions. Even that doesn’t guarantee usability: slideshare was doing both! In that case, I would suggest removing the search-button and adding a search-field (similar to how the top-right of your browser looks, most likely). In the case of Squirrel there definitely is color coding needed! But even better would be to just use the “+” or “-” symbol. This is just confusing :)

Ps: Mac-users, I can really recommend Skitch to you! It’s great for making screenshots!

Opening windows (again!)

Wed, 02 Mar 2011 00:00:00 +0000

Seems like mistakes are often found in the same places, this time we go back to an item also discussed in my very first blog-post: university windows! :)

For some reason, there’s something weird with the windows in my current office at the Technical University Eindhoven as well. They can only be opened with a special crank (which can be seen in the image above).

Now we are a lucky office: we have a crank in our room! However, some offices are not so lucky, and people in those offices need to go to the “servicedesk” which can be multiple floors away from you depending on where you are! The cranks have to be brought back as soon as possible, because there is only a limited amount of them and they are slowly disappearing (guess why…)

The only reason I can come up with why this was implemented this way was to have a better control over the heating of the building (which is also terrible). If this was the purpose, then it failed, as people can get the cranks without having to justify where and why they are opening a window. As it is now, the very simple problem of opening windows got a needlessly complicated solution.

On top of this, the mechanism is difficult to operate, which means you need a strong arm. Also, the mapping is very strange: the top window opens by rotating a handle all the way down at knees-height in the window frame. You can see this in the below image, with the beautifully sunny weather we had. Newcomers wouldn’t have a clue how to open our windows :)

Solution #

I’m starting to wonder if I should skip the solution section, as again the answer is: keep to the standards! There are dozens of designs for opening windows that do work, why reinvent the wheel?

Another trainstation oddity

Mon, 28 Feb 2011 00:00:00 +0000

After the ‘smoking pole’, I’ve found another weirdness at a train station, this time at Eindhoven Central Station. What we see here is part of the roof of the bus station and part of the train station. The weird thing that these aren’t connected. When it’s raining, it pours in between these two roofs making a big mess of the station. Even after the rain, it keeps dripping for at least half an hour. This leads to a weird distribution of people who are waiting for the bus, as they logically leave the rain-trench free. Which in turn is very inconvenient if you quickly need to go through this crowd to catch the train: you’ll keep crossing the rain!

A ‘solution’ provided by the station for the mess the rain makes can be seen in this image, the water is caught in a drainage system… But that’s after making a mess of the people that were standing there! :) Also, can you spot the person who’s using the smoking pole wrong in the image below? (taken in front of the entrance)

Solution #

Either roof could have been made 50 cm. longer and there would have been no problem. As it is now, it’ll probably require a lot more work and money to remodel one of the roofs.

The USI-waterboiler and Dishwasher

Fri, 25 Feb 2011 00:00:00 +0000

Today features two common household-appliances: a waterboiler and a dishwasher. Both were found in our department kitchen. I thought I’d add two items today, so you have something to ponder about during the weekend :)

Let’s start with the dishwasher. The main usability issue is that it has not just one, but two on/off buttons! The on/off button on the right has no directly discernible function. Pressing it seemingly only changes the state of the button (which can go in or out). Only the left button, which has a symbol of a square with an ‘I’ in it, actually makes the machine do something (start cleaning)…

My guess is that the right on/off button actually puts the machine in or out of standby-mode and that the left button acts more like a ‘start’ button. After some experimentation, this seems to be the case. Added to this, it’s quite difficult to see if the right-most button is pressed or not. Also, is there anybody who can tell me what the two cryptic symbols with the 3h-duration mean? (they’re located near the right on/off button)
Probably so you can delay the machine start (Marieke)

A crazy dangerous waterboiler

Now we proceed to the main dish of the day: The waterboiler! I’ve grown to hate this machine. It’s not that difficult to make a good waterboiler!! First of all, sometimes it doesn’t turn off automatically, which is dangerous. Secondly, the spout is shaped in such a way that boiling water easily overflows, which is also dangerous. But the most absurd attraction of this devilish device is: only a tiny fraction of it can be used! I’ve clarified it in the image above, the green arrow shows the usable part, while the red arrow shows the height of the total device. It’s absurd! Even in nightmares you wouldn’t dream up such a monstrosity :)

Solutions #

The dishwasher has several issues: it uses non-common and unclear icons and it counter-intuitively sports two on/off-buttons. I think a redesign is necessary, but for now we might get away with some simple modifications: relabel the left on/off button to say “start” instead of an “I” with a square, reshape the right on/off button so that it is easier to see it’s state (perhaps make it a switch) and change the weird symbol with two clocks saying “3h” to something meaningful (I have no clue what it means, so I can’t suggest anything here).

Sadly (?), the waterboiler can’t be saved. It has to go back to whence it came! I have seen over thirty examples of simple electric waterboilers that do work and the best solution is just to replace the current one with one of those.

Ironing device from Hema

Wed, 23 Feb 2011 00:00:00 +0000

Another one from Daniel, thanks!

This device can be set to different settings with a dial underneath the handle. The problem is that this dial can’t be read when the device is used :) The above image shows the default (top) view, can you see at which temperature it’s set?

Below is another view of the dial. For me (I have to admit I never use ironing devices), it’s still difficult to see what is going on. It’s set to ‘min’ right now, which is colder than ‘max’ of course :) But what does it mean to go to one, two or three dots?

Solution #

Put the dial on top of the device. Also, could somebody used to ironing devices tell me whether an actual temperature instead of dots would be more helpful as well?

Update on the dots from Lise: Concerning the dots, most fabric items have a label describing how to wash and dry the item. If an item is ironable (is that a word?), the iron symbol contains dots. Simply match the number of dots on the label to the one on the ironing thing, and you’re all done and don’t have to care about what the amount of dots reflect. (Personally, I try to avoid purchasing any clothing that requires ironing. I rather have my stuff drycleaned than iron it myself!)

24 hour information machine

Mon, 21 Feb 2011 00:00:00 +0000

A finding from the other side of the globe, this one was sent to me from my brother in New Zealand. Thanks Remco!

What we see here is a ‘24h-touch screen information display’.

But it’s locked inside the store and thus unusable… What’s more, it is turned on and asking to be touched :) To me what adds to the fun is the dissonance between the happy font of the ‘closed’ sign, and the firm and decisive font which says “touchscreen: 24 hour”. It’s as if the ‘closed’ sign is laughing at us :)

As the solution is so blazingly obvious I’ve decided to leave it as an exercise to the readers :)

Refrigerators

Fri, 18 Feb 2011 00:00:00 +0000

Refrigeration dial on the Nordland

Here is something I’ve always thought was strange about refrigerators: you set the power of them, not the temperature. Looking at the example up here (my Nordland refrigerator), you see it is set to ‘1’ right now. But which temperature is associated with this?

The answer is that you can’t know, it depends on the model, the outside temperature and the products placed inside. At least, that’s what the internet says :)

Another thing is the operation of the dial: if you want the temperature to go up, the number should go down. This number ranges from 0 to 7, which coincidentally is also the temperature-range most commonly used for a fridge (according to wikipedia). That makes this number even more confusing, as one might confuse refrigerator-power with temperature. I’ve looked into many fridges and this is the same everywhere. Below you can find two examples from old manuals :)

AEG Arctis 0832-5I. Control 6 is for temperature, 7 shows a thermometer. What the others do, I have no idea :) Control 6 is also a beautifully weird design, for me it reads 4-1-2-3, I also have no clue why there is no dot between 3 and 4.

Sharp (dw37-41_OM) noticed that their control might be confusing and added the label ‘Coldest’, which is a (small) improvement.

Solution #

Users don’t want to set the refrigerator power, they think about the temperature. Change the system to be a thermostat, so that we can just set temperature as we’re used to with heating systems!

Elevator at V&D

Wed, 16 Feb 2011 00:00:00 +0000

(thanks to Daniel for finding and photographing this)

For some reason elevators are a challenging system for designers, you can find the weirdest shapes and interaction designs. The blog-item of today was found by my friend Daniel, at a large chain store in the Netherlands. As you can see in the photo, this elevator has two identical columns of buttons. Or does it only look like that?

As a matter of fact, the second column are only indicators, they can’t be pressed. To make matters even more complicated, the actual buttons (the left column) also have a light in them. The buttons in the left column are only on when the lift is moving and the indicators in the right column are only on when the elevator is standing still.

It doesn’t stop with the inside of the elevator, though! The buttons on the outside are maybe even more confusing. Here we see six buttons. Or do we? No, we get fooled again! :) The top two ‘buttons’ are actually indicators, showing if the elevator is going down or up. The third ‘button’ is also just an indicator, telling us if the elevator is out of service or not. The fourth and fifth ’things’ are actually buttons, indicating ‘up’ and ‘down’. The sixth and last thing is nothing at all, it’s just there because these panels have six holes, probably… The bad thing about this is of course that the symbols up and down are more universal than the Dutch text and are a lot clearer. We’ll probably never know why the designer thought this was a good idea :)

Solution #

The solution is quite easy, we just remove all the indicators that look like buttons. The buttons can then show which floor the elevator is going to and we can use a simple LED-screen to show the current floor. Or if we want, we can keep a bit of the current design and just reshape the indicators so that they actually look like indicators instead of buttons, by modeling them as a light for instance. For the outside of the elevator, we can remove the ‘op’ and ’neer’ buttons and use the first two buttons as actual buttons, with a light in them showing that they are pressed.

Smoking Pole

Mon, 14 Feb 2011 00:00:00 +0000

A ‘smoking pole’ in Groningen

All Dutch trainstations are “smoke free”, meaning that it is only allowed to smoke at certain designated areas. These areas are marked either with a yellow square painted around a small, covered hole or with a yellow circle painted around a ‘smoking pole’. The pole in the image is an example of that, at the trainstation of Groningen.

It earned it’s place on this blog, because of it’s conflicting message. Normally these poles mean “smoke here” but this one has a sign stating that it is not allowed to smoke here! So what is the pole doing here?

The key is to know the pole’s location; it’s at the entrance of the translation. The idea being that you stop smoking here and discard your cigarette! Of course, this is not what happens: people start smoking at these poles, creating a “smoke-screen” at the entrance of the station, which is the opposite of what this thing tries to reach! Also, its prime functionality (discarding your cigarette in it) is not being used, as demonstrated by the cigarettes in the flower bed…

Solution #

Actually, this is a rather tough one! We want to encourage people to discard their cigarette, but not to linger and smoke at this place. This goes against the common pattern I’ve observed with people putting out their cigarette: they do tend to linger around ashtrays.

One design change is to adjust the shape of this ashtray, so that it is physically different from a pole. The international non-smoking symbol should also be clear (but not on the ashtray, that makes for cognitive dissonance, thanks zetta-ua), and an indication of the fine associated with continuing to smoke.

Bank mailbox

Fri, 11 Feb 2011 00:00:00 +0000

A mailbox at a bank

This is a rather confusing mailbox for a bank in the Netherlands. The top text states that for security reasons one should really not put payment transaction forms in “here”, the bottom text says “here” is only for payment transaction forms (with arrows pointing upwards). The problem with this is that there only is one mailbox: the top one is bolted shut and thus looks like it is just a sign.

Solution #

The two mailboxes serve a different purpose, one is an actual mailbox, the other a deposit for payment transaction forms. One of the possible solutions is to more clearly separate the mailboxes. For instance, put them on the same horizontal plane with some space in between. Or put the normal mailbox near the main door and the transaction-form mailbox near the payment machine, those might be more ’natural’ locations.

Parking machine at the zoo in Emmen

Wed, 09 Feb 2011 00:00:00 +0000

The ticketmachine

Trying to decipher this machine is not easy! Just look at the amount of signs that are necessary, the fancy screen with instructions and the option of calling in help…

Red text: “Discount card! Always insert parking ticket first!” Green text: “Pay in advance. Insert parking ticket. Press daycard”

On the left side of the machine there are two places to insert cards and four labels, above the yellow block a label with the instruction “insert chipknip”, then a label saying “chipknip”, then a label saying “insert parkingticket” and underneath the yellow block a label stating “these cards can be inserted above”. On the right side, the label says “these coins can be inserted” and “these bills can be inserted below”. Creditcards and the parkingticket go in the same slot, eventhough the parkingticket doesn’t look like a creditcard at all. I believe the parkingticket doesn’t even come out before you have to insert your creditcard in the same slot!

As far as I can tell, the idea is that you get a card when you park your car (the parkingticket), with which you can buy a daycard. If you decide not to buy a daycard, you have to pay afterwards by inserting the parkingticket again when you return to your car. If you have a discount card, you can use it to pay less for the parking.

When observing this machine, I’ve seen people trying every slot for their cards, some slots accept multiple cards, others only one. The weirdest thing is that this machine does not offer the default option in the Netherlands: a slot for a debitcard.

Solution #

This machine requires a complete redesign. There’s nothing left to save! But let’s suppose we want to keep all functionality that is currently in the machine.

Since the procedure is in several steps, it makes sense to separate the interface in these steps as well. First, the user has to decide if she wants a daycard or if she wants to pay per hour. Then the parking ticket has to be read. Only then do we need to see the payment options.

I’ve included a sample diagram, but of course it should be user tested :) The diagram is a redesign of the machine, with orange blocks indicating the steps in the procedure. I’ve separated the different payment options and the parking ticket is no longer read by the same device that reads the creditcard. I think it’s a lot clearer already: we need less labels, show more information and we even lost the LCD-screen!

Building design

Mon, 07 Feb 2011 00:00:00 +0000

That’s a big window

An old favorite of mine is this window (picture taken september 2007, Bernoulliborg - Groningen, NL), it is just impossible to use! :) Although the big window looks fancy, it is impossible to walk through the corridor if it is opened! The window has no ‘half-open’ position, it is either open or not. Nor has it got hinges to support rotating it over an other axis… Another oddity is that behind this window is not the outside world, but rather another layer of glass with blue dots on it.

Solution #

Please allow the window to be opened by rotating it on the bottom axis and have it not open all the way!