A Note on Small Things

I have a soft spot for software you could fit in a glove compartment. One file, a few hundred lines, does one job, gets out of the way.

Programs like that have a particular dignity. They aren't trying to become a platform. They aren't engineered for some hypothetical user who might show up in five years. They're done, in the old sense of the word. The way a meal is done.

The failure mode of ambition is that it delays everything. Twelve-file architectures, pluggable backends, abstractions that handle cases nobody's actually going to hit. These feel like progress. They aren't. Most of the useful things I've made, I made in an afternoon and never touched again. Most of the things I spent months architecting are nowhere now.

Then there's finishing. Finishing is unglamorous. Nobody posts on Twitter about the boring last 20 percent. The edge cases, the error messages, the copy you write at 11pm because you suddenly realize the user is going to see it. But finishing is the only part that matters. A 95 percent done thing is 0 percent useful.

Pick something small you've been meaning to make. Don't plan the next version. Don't sketch the architecture. Just write it this weekend, finish it, ship it.

Small things you finish beat big things you start.