Since December 2024 I’ve been carrying two ideas about my work that refused to fit together.
One is the designer in me. I care about slick interfaces. I think about animation curves and the weight of a shadow. Craft, in the way designers mean it: the thing you keep polishing long after it technically works.
The other came from Shreyas Doshi’s courses, Product Sense that December and Advanced Product Taste this June. Pragmatic, straight to the point, let’s-build-winning-businesses thinking. People carry a ranked list of problems, they pay for the ones at the top, and the problem you decide to tackle matters more than anything you decide while solving it. I believed every word.
But the two never merged. Doshi’s insights went into a drawer labeled product, and my actual design work kept running on the old rules. Root causes, sure. User needs, sure. But judged, in the end, by the standards of the craft. Is it polished? Does it feel effortless? Would another designer notice the details? The product thinking sat next to my design thinking, never inside it.
Then, during a two-week vacation this July, with time to reflect and just be, I read Notes on the Synthesis of Form, a 1964 book about kettles and ashtrays by the architect Christopher Alexander. The drawer came open.
Good is not a property of the thing
Alexander splits every design problem in two. The form is what you can change: the screens, the flows, the animation curves. The context is what you can’t: the customer’s day, their tools, what their boss expects of them. And then the sentence that got me. The quality you’re after doesn’t live in either pile. It lives between them.
There is no such thing as a good dashboard in the abstract. There’s a dashboard that fits an ops manager checking fleet health at 8 a.m., and that same dashboard is useless to a developer who wants an API and no UI at all. Nothing about the dashboard changed. The context did.
Read that as a designer and it stings, because everything I’d been polishing — the slickness, the motion, the finish — is a property of the form. Alexander is saying, calmly, that no property of the form is good on its own. You don’t design a form. You design a fit.
Picking the problem draws the line
Requirements, in Alexander’s frame, are the demands the context makes on the form. Not the demands of the craft, and not the standards of a portfolio.
But look closely at the context pile and almost nothing in it is actually fixed. “Fixed” mostly means “we decided not to touch it here.” The boundary between form and context is a decision dressed up as a fact, and that’s what connects Alexander back to Doshi’s ranked list. Choosing which problem to tackle is choosing where that line goes. The requirement list is generated by the line: move it, and requirements appear, vanish, or switch sides.
So “the problem you pick outranks everything you decide while solving it” isn’t a business aphorism sitting next to design. It’s the first design act. A ranked list of problems is a ranked list of contexts you could design a fit for, and everything downstream, every craft decision included, inherits its requirements from that one choice.
Craft failures are misfits too
So where does that leave the slick animation? My first conclusion was probation: polish is allowed when the context explicitly demands it, otherwise it’s decoration. But that still keeps the two ideas apart, with craft as a budget the context has to authorize.
Alexander has a tighter answer, hiding in his strangest claim: you can’t describe good, you can only point at wrong. Fit has no positive definition. It’s every specific failure cleared, and nothing else. There’s no extra ingredient called goodness that you sprinkle on top with a nicer easing curve.
Under that lens a janky transition isn’t a missing garnish. It’s a misfit, a specific and pointable friction. Just a small one, at a resolution most people can’t consciously perceive. Nobody files a ticket about an easing curve. They say the app feels off, feels cheap, and they can’t tell you why.
Which means taste isn’t an eye for polish. It’s higher-resolution misfit detection, feeling the frictions a user experiences but could never name.
Shreyas had already shared it but I couldn’t understand it yet. My notes from the taste course contain the formula (Motivation − Friction)^Satisfaction. Satisfaction is the exponent, not an addend. An exponent can’t rescue a base of zero, so pick problem number seven and no animation saves you. On a positive base it compounds everything. That’s a precise account of when polish is load-bearing, and it was sitting in my product drawer the whole time, filed as business advice. Same with “a busy interface costs you twice,” once in added friction and once in the lowered exponent.
The courses were never in tension with my work. They were describing my job more accurately than I was.
The writing habit was the method all along
One more thing, this one personal. My design process has always started with writing. Before I design any screens, I write out the problem, the context, then walk through the screens: what happens here, what options the user has, what they see next. It always felt slightly foreign compared to how other designers seem to work. A private quirk I should probably grow out of.
Alexander says it’s the method. If good is only ever specific misfits cleared one by one, you need the misfits written down somewhere to clear them. A design that lives in someone’s head can’t be checked against anything; it shifts shape under every objection. Written down, it can finally be wrong, and every gap between what and how becomes visible.
So the note-taking wasn’t a detour around real design work. It was the fit-check running, step by step. Years before I picked up the habit, an architect had already explained why it works.
What I do differently
The change is small. I already write before designing; what changed is what the writing starts with. The first thing on the page is now one paragraph of context with no solution in it. Who, doing what, and what specifically goes wrong for them today. Every decision after that, craft decisions included, gets held against that paragraph. The animation stays if its absence would be a misfit, a friction someone would feel even if they could never name it.
Craft didn’t shrink in this. It’s the same work at a finer resolution, clearing the misfits nobody else can see yet.
There’s more in Christopher Alexander I want to write about, mostly why large design problems overwhelm intuition and how to cut them along their natural seams. For now: one old architecture book did what two product courses couldn’t do on their own.