Refining Taste

I've been working as a software engineer for 8 years. In that time I've built nearly a hundred applications and kept adding new skills to the list. But there is one skill that only improves when I take things away, and that's taste.

When I talk about taste, I don't mean personal preference, which is what it's often mistaken for. I mean the ability to see beyond the obvious and recognize what elevates a product¹. That idea stuck with me the first time I read it, but I only really understood it once I started applying it to my own work.

Purl is the product where I practice this the most. It started as a read-it-later app with a lot going on: AI features, content ingestion, and a landing page that tried to explain all of it at once. On paper, every one of those things made the product more capable.

In practice, they made it noisier. So I removed them. The AI went, the content ingestion went, and the landing page became a hero section and a footer. Nothing more.

What I learned is that a feature can add noise instead of clarity, even when it works perfectly. Every extra thing on the screen asks the user to understand it, and that cost is easy to ignore when you're the one who built it. Removing those features gave me room to focus on the parts that actually matter, and to refine them properly instead of spreading my attention across everything.

One of the clearest signs that my taste is developing is that I can feel when a UI is wrong before I can explain why. And when I finally work out what it is, it's usually the interaction.

Most of the time it fails in one of two directions. Either there's too much of it, with everything moving, bouncing and reacting until nothing stands out, or there's none at all, and the interface feels dead under your finger. Good interaction sits in between. It responds when you touch it, confirms what you did, and then gets out of the way.

That balance is hard to describe in a spec. You mostly learn it by noticing it, over and over, until you start catching it in what you build before anyone else does.

I didn't develop this on my own. A lot of it comes from studying the work of people whose eye I trust, especially Emil Kowalski, Gustavo Fior and Jakub Krehel. I look at what they build, try to understand the decisions behind it, and then put those ideas into practice in my own products.

Purl's rebrand is the result of three months of doing exactly that. Not copying anyone's style, but taking what I learned about restraint, detail and interaction and seeing how it holds up on something I actually own and use every day.

Most skills I've picked up over the years came from adding something: a new language, a new framework, a new tool. Taste has been the opposite. It grows every time I recognize something that doesn't belong and have the discipline to remove it.

I don't think I'll ever be done with it. But I've noticed that the products I'm proudest of are rarely the ones with the most in them. They're the ones where everything left on the screen has earned its place.

Sanding Space is where I keep sanding these details, one short essay and one live demo at a time.

Sanding web interfaces · Sanding SpaceShort essays about design engineering and product engineering, with live, embeddable demos.sanding.space/introduction (opens in a new tab)