Vibesmithing

You already have a compass.

You know the feeling. Something clicks and you get excited. You want to show someone. Maybe your scalp tingles, maybe you just can't stop thinking about it. You can't always explain why it works, but you know it does. You've always known.

You also know the opposite. Cringe. Something is off and you can feel it before you can name it. That signal is just as useful.

Those two reactions are the whole compass. One says closer. The other says not that. Vibesmithing is what happens when you decide to actually follow them.

Problems are felt, not described.

A problem is any gap between what exists and what should exist, as felt by the person who feels it. Sometimes that's a broken sidewalk. But more often it's something you can't quite point at. Music that feels like you doesn't exist yet. A tool that works the way your mind works hasn't been built. A community run the way you think people deserve to be treated doesn't have an address.

These are real problems. They cause real suffering, the quiet kind, the ache of something missing. And they're only solved when the thing they were causing stops. Not when someone ships a feature. Not when a box gets checked. When the ache goes away.

This is why problem-intimacy matters. If you don't live in the problem yourself, you have to rely on specs and acceptance criteria and user testing to tell you when it's done. Those are proxies. But if you carry the problem, you carry the compass. You know it's solved because the cringe is gone.

Triangulating on something you can't see.

There's an ideal version of whatever you're trying to make. You can't see it directly. You never will. But every time you make a new attempt, it's like seeing a shadow of that thing cast from a different angle. Each shadow tells you something. None of them is the thing itself.

So you triangulate. Attempt after attempt, angle after angle, you get closer. You can't get a true reading because there isn't one. Everything is relative. But if you're paying attention, it gets closer and closer to being solved.

The real question is: how solved is enough? What's the minimum amount of right that you'll accept? Sometimes 80% is fine and you move on. Sometimes you keep going because the process itself is giving you something. The better you get at this, the more you realize you can almost always squeeze out more if you care to.

You couldn't have specified the final version upfront. You didn't know what you wanted until you saw what you didn't. That's fine. That's how it works.

Why now.

People have always worked this way. Artists chase a feeling through draft after draft. Good cooks taste as they go. What's different is that each attempt used to cost time, money, and expertise you didn't have. The loop was expensive, so people planned instead. Plans were a workaround for unaffordable iteration.

That workaround is becoming unnecessary. The cost of trying again is collapsing. When iteration is cheap, the bottleneck moves. It used to be skill, or money, or access. Now it's discernment. Can you tell what's working and what isn't? Can you stay in the loop long enough?

Intelligence is getting cheaper. That compounds.

AI is already past the “good enough” bar for a lot of this. It's getting smarter and cheaper at a rate that's hard to overstate. That's not a threat to what we're talking about here. It's the reason it works.

If you're solving a problem, you can now integrate intelligence into exactly the places you need it, whatever situation you're working in. And here's where it gets interesting: AI helps you build tools that apply AI in ways that give your users superpowers. And those users can build tools that give their users superpowers. It's compounding all the way down.

The word vibecoding already exists for this applied to software. You describe what you want, AI writes the code, you react to what you see, you describe it better. But coding is just one surface. The craft underneath is the same whether you're making software or music or figuring out how you want to live.

What it actually feels like.

At first it's just fun. You describe something, it appears, you react, you try again. It feels like play because it is.

Then you start to get good at it. You notice your instincts getting sharper. The gap between what you asked for and what you meant gets smaller. You start finishing things. Real things, things you can use, things you can show people.

And at some point it clicks. Not just the project in front of you, but the broader thing. You realize you can do this with anything. Any problem you can feel, you can solve. That realization is the most empowering and joyful thing I've ever experienced. I can do anything now. I know that. And knowing it changes everything.