A joke works because everything is predictable until it isn’t. Thorbjørn thinks breaking rules works the same way, and that a lot of creative work gets it backwards by trying to be new in every direction at once.
The question
“What’s something you were taught early on in your career that you later found out was actually wrong, or at least way more situational than you were originally told?”
Martin brought this one off their own question list, and Thorbjørn admits he nearly cut it the day before. Thorbjørn turns it straight back on him, so Martin answers first. The business he worked in before going out on his own treated every job as a pitch against rival companies, a constant battle atmosphere. Now that he’s self-employed with nobody on the payroll, he feels more like part of a team than he ever did inside one.
Thorbjørn goes back to design school, where every teacher told him the same thing: it’s not important how it looks, it’s how it works that counts. He believed them. Then he watched classmates with thinner ideas present something beautiful, and the room loved it. His conclusion is blunt. Something looking good enchants people into believing it is good, and a beautiful prototype hides the flaws underneath. He uses that now to convince people, and he says it’s wrong in every fiber of his body.
What we talk about
Martin asks where the rule of thirds came from, and why it never became the rule of quarters. Thorbjørn reaches for the golden ratio and the way leaves and snail shells grow, then stops himself: this is purely from memory, and he can’t say why thirds are pleasing to look at. He promises to look it up and report back. What he will commit to is the pattern underneath it. In design and architecture the reasoning turns up after the observation: you notice how things behave, then you work out why. Kahneman’s System 1 and System 2 is his example.
One tangent on color: Thorbjørn reckons we pick up far more nuance in green than in red, left over from a jungle where survival meant spotting movement among a hundred shades of green. He doesn’t offer a source for it.
Martin asks at what point in making a video game you start breaking the design rules. At the very beginning, Thorbjørn says, because a creative idea is a broken convention by definition. The craft is what happens next, which is making everything else ordinary so the one break has something to push against. Break it all at once and you get noise where the surprise should be. He compares it to a joke: you set up a promise in someone’s head, then hand them something else.
Martin’s version comes from voiceover. When a producer asks for three takes, the third one is usually a small shift, a chuckle behind the line, still inside the brand’s scope, not a different read altogether. Thorbjørn gets audibly excited here, because it’s what contrast does in color. Hue, brightness and saturation only register as contrast when there’s something steady around them. The one rule he’d leave alone is that an experience should stay intuitive, and even that he second-guesses within a sentence.
Then AI, which Martin promises at the end won’t come up next time. He’s watched AI-generated posters for events and webinars converge on a single look, and the LinkedIn trend where everybody turned themselves into a boxed plastic toy. Unless you’re first, you’re part of the wall of noise. To Thorbjørn the novelty effect is wearing off: you can feel that an image or a lyric came out of a machine, and the emotion isn’t in it. AI is bad at breaking rules, and bad enough at understanding them that it sometimes breaks one without noticing.
Both of them use it as an assistant anyway. Thorbjørn built a skill that shortens the methods section of a scientific paper and rewrites it so a text-to-speech voice can read it back to him on a run. At work, every meeting he takes gets transcribed and cross-referenced into a cheat sheet for each person. A messy Zoom chat export that used to cost Martin two hours of copying out business details now takes him five minutes. Martin wonders how many human PAs lost their jobs over this. Thorbjørn thinks the ones who did get hired back, because a PA who can hand off the boring half is a more effective PA, and the phone call where someone has to say no is still not a machine’s job.
Thorbjørn’s prediction: AI subscription prices roughly double once the fight for market share settles down. The comparison is Spotify, which lost a lot of money early on to own music streaming and is running the same play on podcasts now.
Mentioned in this episode
From Hidden By Design
- Colours (S1E10): Thorbjørn wonders aloud whether the green-versus-red thing came up “in the episode about color”.
Also mentioned
- Kahneman’s System 1 and System 2: Thorbjørn’s shorthand for how the brain works.
- Trey Parker and Matt Stone on “but, therefore”: the South Park plotting technique Martin looks up mid-episode.
- Polyspice Games: Thorbjørn’s studio, and how he and Martin first met.
- talat: the on-device transcription tool Thorbjørn uses, spelled out on air after he mangles the name.
- Natural Reader: the text-to-speech tool that reads him papers while he runs.
- Spotify: the market-share comparison behind the AI pricing prediction.
- Paul the Octopus: the football-predicting octopus in Germany, Thorbjørn’s own comparison once Martin suggests making predictions a regular feature.
Got a question of your own?
Got a design question or dilemma you’re stuck on? Send it to Thorbjørn and Martin through the contact form at hiddenbydesign.net and it might become a future episode.
Be First to Comment