The hard things about design
I once wrote, in a list of sixteen things about design, that the craft requires a tremendous amount of mental fortitude. One line, buried in a longer list, for something that probably deserved its own essay.
I’ve been doing this for close to two decades and I still haven’t figured it out. So here’s an attempt at naming what makes design hard. Not the vague, motivational kind of hard. The specific, structural kind, different from the crafts sitting right next to it in the same building.
Inspiration for design can come from anywhere, and that’s harder than it sounds. An engineer stuck on a caching problem can search a bounded space: documentation, a colleague, a forum post from three years ago. I don’t get a bounded space. I once solved an onboarding screen with a wayfinding detail from a museum I’d visited years earlier, and I wasn’t looking for it when it surfaced mid review. The answer is either already living somewhere in your memory of things you’ve paid attention to, or it isn’t, and there’s no search engine for the second case.
And it doesn’t get easier with repetition, not in the way most skills do. An engineer who has fixed fifty similar bugs recognizes the fifty first one faster, because the underlying pattern repeats. I’ve cracked maybe fifty onboarding flows in my career, and the fifty first one still took its own sweet time, because each one pulled from a different, unrepeatable combination of references. Experience makes you a better searcher. It hasn’t once handed me the answer directly.
Design is also the one craft in a company that everyone feels qualified to comment on. A codebase needs literacy to critique. Design needs none of that: anyone who’s used a phone has an opinion the moment they see one. That’s not always a bad thing, some of the sharpest feedback I’ve gotten came from someone with zero design training who just said what felt off. But it does mean the volume of incoming opinion is enormous, and sorting the signal from the noise becomes its own permanent job.
The same decision sometimes gets explained five times in one week, in five different rooms, to people who weren’t there for the first four conversations. The crit. The one on one with the PM. The leadership review. A stakeholder who catches you in the hallway on the way to lunch. By Friday I’ve made the same argument so many times I’ve started boring myself, and boring yourself while still trying to sound convinced is its own particular kind of tired.
A lot of what looks like obsession from the outside is just the only quality check I have. Engineers have linters and test suites that catch a mistake before a person even looks. I have my own eyes, doing one more pass, then another, at whatever hour it happens to be. I’ve reopened files at midnight over a spacing issue no one had flagged, more times than I’d like to admit. I used to think that was dedication. Some weeks it is. Other weeks it’s closer to being unable to let go, and I’m not always sure which one it is while I’m doing it.
Design doesn’t hand you a moment where you’re allowed to call something finished. A bug fix closes the day the test passes. My version of done is measured by my own taste, and taste has an annoying habit of moving the second I clear the bar I’d set for myself the last time. I’ve felt proud of something for an entire afternoon before, which, looking back, might be a personal record.
Inspiration doesn’t check the calendar before showing up, but the four o’clock review happens whether it showed up or not. Some weeks the idea arrives on a Monday morning walk. Other weeks I’m still empty by Tuesday lunch, and I walk in anyway with whatever I’ve got. Enough reps over enough years taught me how to fake competence convincingly until the real thing shows up later that day. Most days it does. I don’t understand why.
Rules and patterns get you to competent. They don’t get you to original, and knowing which rule to break, and when, doesn’t come from a pattern library. It comes from paying attention to things that have nothing to do with software, whatever those happen to be for you. Mine happen to be old film and industrial design, which is a roundabout way of explaining why I’ve spent more hours than I should looking at a 1960s Braun radio and wondering why it still looks right. There’s no course for this. I wish there were. It would have saved me a decade.
Ira Glass has a line about the gap between your taste and your skill: your taste gets good faster than your ability to produce work that matches it, and that gap is what makes your own work look bad to you, sometimes for years, before it catches up. His advice was to make more work on purpose, knowing most of it will be bad, because volume is the only thing that closes the distance. What he doesn’t say is that the gap reopens every time you land somewhere new. Whatever calibration I built up over a decade at one company didn’t carry over to the next one, because what looked right for that brand didn’t look right for this one. I’m rebuilding some version of that same taste again right now, a few months into a new job, and it manages to feel both familiar and new at the same time.
A broken build stops a merge for everyone, whether anyone in the room cares about code quality that day or not. There’s no equivalent for a mediocre screen. Nothing technical stops it from shipping. The only thing that stops it is someone caring enough to say so, and depending on the week, on the company, on who else happens to be in the room, that someone is often just you, alone, at an hour no one else is awake for.
The trained eye doesn’t come with an off switch. It rewires how you see, the way you can’t look at a sentence and not read it, and once that’s wired in, it runs on everything, not just the thing you’re paid to look at. My wife has caught me judging the kerning on a restaurant menu on a night we were supposed to be off the clock. I wish I could say that only happened once.
A good amount of this changes shape depending on where you work. At a small company, you get more room to hold your own bar, but fewer people to share the weight of holding it. At a large one, you get more support and more process, but your taste has to survive a lot more rooms and a lot more compromise before it reaches an actual user. I’ve now done both, more than once, and neither version removes anything on the list above. It just rearranges which parts hurt in a given week.
None of this is a complaint, exactly. I’d still pick this over most other work available to me, and I don’t say that as a rally cry, just as an honest accounting after a couple decades of getting it wrong more often than right. I’m not even sure the list above is complete. Ask me again in five years.
