I Keep Avoiding the Thing I'm Best At

A few months ago I started building keyboard navigation for Odo, the whole system of shortcuts and focus handling, including the Cmd+K command bar. I did it carefully. I architected it deliberately, and I hand wrote the trickiest parts myself, because I knew it was the kind of system that turns into spaghetti the moment you stop paying attention.

Then I let AI fill in a boring bit. Then another. Then a fix, and another fix, and a fix on top of that fix. I never made a decision to stop paying attention. It happened one reasonable shortcut at a time. Today that keyboard navigation is one of the least maintainable corners of our codebase, and it slows down many of the new features we ship on web.

Ya, ya, AI brain fry, we've heard it before. But I think it's more interesting than that for me, because I've experienced something like this pre-AI.

Near the end of Notecards, the work that actually mattered was business development. Partnerships and content acquisition. I knew it. And instead I kept programming, because programming was the thing I was good at and the thing that felt like progress. I polished features nobody had asked for while the real bottleneck sat untouched.

That taught me a lesson I believed cleanly for years. Building a company forces you to push into the work you'd rather avoid. The uncomfortable parts, the conversations outside your strengths, the things you're simply bad at, that's the actual job, and the resistance you feel is mostly a sign you're finally doing it. I'd have written that with total confidence a year ago.

However, the keyboard navigation mess is a stranger version of the same trap, and a harder one to admit. This time I'm not avoiding some skill I never had. I'm avoiding the thing I'm best at. Reviewing AI's plans and reading its changes line by line is slow, it's draining, and it's the single most valuable thing I bring to the table now that a machine writes most of the first draft. And I've quietly been skipping it. I wasn't retreating to my strength this time. I was retreating from it.

So comfort was never really the through-line. Avoidance was, and I'm remarkably good at dressing up the skipping as progress. With Notecards it wore the costume of shipping features. With Odo it wears the costume of moving fast with AI, a steady stream of merged changes that feel like velocity right up until they're a mess nobody can maintain.

I should be honest about why. I'm tired. Three years of intense building does something to you, and deep, careful review is exactly the kind of work that's easy to skip when you're running on fumes. AI is happy to take the wheel, and on a depleted day, letting it feels like relief.

What I'm taking from this#

I want to be careful here, because the tidy version of this is wrong. Not everything you avoid is important. Plenty of what I put off, I put off because it genuinely doesn't matter, and the healthy move is to drop it and drop the guilt along with it. Avoidance by itself isn't a signal of anything.

The tell is the guilt that won't leave. There's a specific kind of putting-off where some quiet part of me already knows the work matters, already knows I should be doing it, and the only thing in the way is that it's hard or draining. That's the compass needle. Not the avoidance, but the guilt I can't argue myself out of.

I don't have this beaten. I'll probably vibe code another mess before I learn to feel the slide while it's happening. However, I at least know where to look now. When I catch myself reaching for the thing that feels comfortable and productive, I try to stop and ask whether there's something I'm using it to avoid, and whether I actually believe that something matters. When the honest answer is yes, I've found my work for the day.