There's a particular half-second I've come to recognize. The logic is done, the function works, and then I have to call it something. And suddenly I'm not writing code anymore, I'm making a small promise to whoever reads this next.

processData. normalizeRecords. cleanIncomingRows. Each one is a slightly different story about what this thing is for. The first says I didn't want to think about it. The last says I know exactly what mess is arriving and what I'm doing about it. Nobody grades me on this. The tests pass either way.

What gets me is how confident a name feels in the moment. I'll write userCache and it will seem not just correct but inevitable — of course that's what it is. Then forty minutes later it's holding sessions too, and half of them aren't users, and the name has quietly become a small lie that everyone downstream will believe. I didn't mislead anyone on purpose. I just named the thing before I'd finished learning what it was.

I think that's the real texture of it. Naming happens at the exact moment you know the least you will ever know about the thing.

And there's the pull in two directions. I can be precise — retryWithExponentialBackoffOnTransient5xx — and be technically unimpeachable and completely unreadable, a name nobody's eye will survive. Or I can write retry and be graceful and slightly untrue. Most of my good names are compromises I'd have trouble defending in the abstract but that feel right in the line where they sit. Names aren't evaluated alone. They're read inside a sentence.

Commit messages are their own small ache. That one's addressed to a person in the future, possibly at 1am, possibly furious. "fix bug" is a betrayal of that person. I try to write the why, because the what is already sitting right there in the diff.

The part I keep turning over is what it means to live with a name. Once it's chosen, it spreads — into call sites, into docs, into how people talk about the system out loud. It stops being a label and becomes the concept. A slightly wrong name doesn't stay slightly wrong. It bends everyone's thinking a few degrees for years.

So I hover. Not long, but I hover. It seems like the right amount of care for a decision that small and that permanent.

And then sometimes I rename it three days later, and that's fine too. The name was a hypothesis. Now I know more.