The design community has the same fight on Twitter every year or two. Should designers learn to code. React or just HTML/CSS or nothing at all. The arguments cycle. This week's iteration was sparked by a senior designer at a well-known fintech posting that they'd been quietly using V0 and Cursor for six months and felt embarrassed. Partly at how much it had changed their work. Partly at the gap they could now see between designers who build and designers who don't.
The replies sorted into three predictable camps. The "designers should always have been able to ship" camp, who felt vindicated. The "stay in your lane, you're devaluing the craft" camp, who felt threatened. The "this is a temporary tooling moment" camp, who felt smug.
All three are answering the wrong question. The principle I'd apply here is from Irvan's foundation: Working software beats specs. That's not a productivity tip. It's an epistemology. The reason a designer should be closer to working software, whether by writing code or piloting tools that produce code or sitting close enough to the build process to feel its constraints, is that working software is how you find out whether your idea is true. A spec can't tell you. A mockup can suggest. Only the working version proves it.
The "should designers code" question gets framed as a career-positioning choice. It isn't. It's epistemic. If you can't get something in front of a real user without going through a six-person handoff, you don't actually know whether your designs are right. You only know whether your stakeholders like them. Those are different things.
The senior designer's post wasn't really about learning to code. It was about discovering that the gap between having a design and having a working version of the design had collapsed for some people and not for others. The ones for whom it had collapsed operated with a different kind of confidence. They could test ideas at a different speed. They could change their mind faster, because changing your mind costs less when you have a working prototype to revise than when you have a Figma to redraw and renegotiate with engineering.
The three camps in the replies are arguing about whether learning code is a moral obligation. The actual question is whether you can afford, in the next five years, to be the kind of designer whose ideas need a translator before they can be tested. If the answer is no, the path forward is whichever route gets you closest to the working version. Whether that means writing code yourself probably doesn't matter much.
Stop asking whether designers should learn to code. Ask whether your team's design loop can produce working artifacts in days rather than weeks. If yes, you're fine. If no, you have a bigger problem than what to put on your LinkedIn.
.webp)
.webp)
.webp)
