Tinkering Machines

Recently, I spent several hours getting Claude Code to reverse-engineer the API for my Wyze smart bulbs. By the end, I had a small dashboard that turned the lights on and off and ran through some preset scenes I'd made up.

I didn't have to do this.

The Wyze app works fine. So does Google Home. My dashboard isn't faster or more convenient. The marginal utility is somewhere between negligible and negative.

I built it anyway. Out of curiosity, mostly. And a little out of wanting it to look the way I wanted.

Some people would call this Claude Code psychosis. They're right. But it's strange to use psychosis as a descriptor for an obsession with tinkering. We've gotten so utility-brained that any activity without a clear ROI looks like a glitch. Our first instinct is to ask "what did you ship" instead of "what did you learn." Maybe we should flip this.

Most of my dashboard will rot. The Wyze API will change, and I won't remember a thing the next time I look at the code. None of that matters. What I came away with is a deeper intuition for how local-network device APIs work, a clearer model of the request path, and the gotchas of MITM configs.

Here is what I think tinkering actually does.

Some problems are obviously software-shaped: scrape a site, automate a workflow, make a dashboard for a device. They can be cleanly broken down into data, state, and rules.

But most problems aren't initially software-shaped. They're vague frustrations or curiosities. The conversion from vague curiosity to software-shaped problem is itself a skill, and the real long-term gift of tinkering. You start asking, of every small friction in your day: what would I need to track here? What's the state? What would the screen look like? You begin seeing latent software everywhere in the world.

If you write software for a living, you already do this, mostly without noticing. What's new is that everyone else can do it too. We have tinkering machines now. Claude Code and its ilk let us massage everyday problems into software-shaped problems for cheap. Parts of our life we'd previously write off as too small to be worth some piece of code are now feasible and every weekend project sharpens the perception a little more.

If you're tinkering out of curiosity, you're in good company. In 1976, Steve Wozniak brought a computer he'd built in his spare time to a meeting of the Homebrew Computer Club, set it up, and handed out the schematics to anyone who wanted them. He hadn't built it to sell. He'd built it because he wanted to see if he could. That became the Apple I.

Many widespread utilities first started as curiosities. Keep tinkering, psychosis be damned.