Taste is the new technical skill

There’s a pattern that repeats across every creative industry when tools get cheap. Photography went through it when digital killed film costs. Music went through it when a laptop replaced a recording studio. Publishing went through it when printing stopped being a bottleneck. The same thing happens every time: output explodes, quality collapses, and then, slowly, the people with taste rise to the top.

Software is entering that phase right now.

For decades, the bottleneck in software was production. Building things was hard, slow, and expensive. This meant that the ability to produce --- to write code, to ship features, to close tickets --- was the most valued skill. Companies hired for it. Interviews tested for it. Careers were built on it.

But when production costs approach zero, production skills become commodities. And what becomes scarce is something the industry never learned how to hire for: taste.

Taste is the ability to look at twenty possible features and know that eighteen of them will make the product worse.

The fashion industry understood this decades ago. Fabric is cheap. Sewing machines are everywhere. Anyone can manufacture a garment. But the distance between a $30 shirt and a $300 shirt isn’t materials or labor. It’s the hundreds of decisions someone with taste made about cut, proportion, and what to leave out. The $300 shirt doesn’t have more. It has less, but chosen better.

Architecture tells the same story. CAD software made it trivial to add complexity. You can model a building with a thousand decorative elements in an afternoon. But the great architects are the ones who resist that temptation. They know that every element you add is a decision the occupant has to process, consciously or not. Simplicity isn’t a limitation. It’s a skill.

In film, the rise of cheap digital cameras and editing software meant anyone could shoot a movie. The result wasn’t a golden age of cinema. It was an ocean of content with a few brilliant films floating on top. The difference? Editorial judgment. Knowing which scenes to cut. When silence says more than dialogue. What the story is actually about.

Software is going through this exact transition, but the industry hasn’t caught up. We still interview engineers by asking them to produce code on a whiteboard. We still measure teams by velocity, how much they ship, how fast. We still promote the people who build the most features.

But the person who knows what not to build is now more valuable than the person who can build anything.

I’ve seen teams with excellent AI tooling produce software that’s technically impressive and experientially terrible. Every feature works. The code is clean. The test coverage is high. And using the product feels like drowning in options. There’s no point of view. No clear opinion about what the software is for and, more importantly, what it is not for.

This is what happens when you optimize for production without taste. You get software that can do everything and communicates nothing.

The hard part is that taste is genuinely difficult to develop, measure, and teach. It comes from exposure: using thousands of products and paying attention to why some feel right and others don’t. It comes from restraint: the willingness to kill a feature you spent weeks building because it doesn’t serve the whole. And it comes from opinions. Actual beliefs about how software should work, not just compliance with whatever the stakeholder asked for.

Taste also requires a kind of courage that pure technical skill doesn’t. Saying “we should build this feature” is safe. Saying “we should not build this feature, even though three customers asked for it” requires you to have a thesis about your product and to defend it.

The companies that figure this out will build products that feel designed rather than assembled. Products where every element earns its place. Products that do fewer things but do them so well that users never think about the tool. They just think about their work.

The companies that don’t will ship faster and faster, adding feature after feature, wondering why their NPS scores keep dropping even as their release cadence goes up.

The tools don’t care what you build with them. That’s the point. When the tools are neutral, taste is the tiebreaker. And unlike technical skill, you can’t automate it, shortcut it, or prompt-engineer your way to it.

You either see it or you don’t. And the gap between those who do and those who don’t is about to become the widest it’s ever been.