There is a kind of software I keep returning to, and it is not the kind that gets funded. It is small. It does one thing. It is often ugly. And it has, repeatedly, changed how I work more than any platform ever has.

The case for the small tool

A platform wants to be your everything. It wants your data, your attention, your workflow, your identity. In exchange it offers polish and the promise that you will never have to think about the seams. This is a genuine convenience. It is also, increasingly, a trap.

A small tool makes the opposite bet. It owns nothing, persists little, and refuses to grow. The cost is that you — the user — must stitch things together. The benefit is that you can.

The best software I use disappears into my thinking. Platforms rarely disappear; they announce themselves.

What I actually build

A few examples from my own shelf:

  • A markdown-to-Anki script that turns my reading notes into spaced repetition cards. Sixty lines of Python. I use it every day.
  • A local budget tracker that is, at heart, a single CSV and a small renderer. No accounts, no sync, no cloud.
  • This blog, which is static files and a build step. No database to maintain, no admin panel to secure.

Each of these took an afternoon. Each has outlasted at least one "real" product I was supposed to switch to. The pattern is so consistent now that I've stopped ignoring it.

The economics of small

There's a quiet economic logic here. The value of a tool scales with how often you use it and how little it asks of you in return. Platforms optimize the first term (they want you there constantly) at the expense of the second (they ask for a great deal). A small tool optimizes the second — it asks for almost nothing — which makes its raw frequency less important.

In expected-value terms, a tool I use twice a week for five years and that never breaks is worth more than a platform I use daily for six months and then migrate off of in frustration.

When small is the wrong answer

I don't want to be pious about this. Small tools fail at coordination. If you are building something that requires many people to be in the same place at the same time — a market, a social network, a shared document with conflict resolution — the small-tool instinct will lead you astray. Concurrency is a real problem and it does not yield to scripts.

The honest position is plural: use platforms where coordination is the point, build small tools everywhere else, and keep the boundary between them clean.

A recommendation

If you've never built a tool just for yourself, try it this week. Pick one repeated annoyance — a format conversion, a daily summary, a nagging reminder — and write forty lines to handle it. Don't publish it. Don't generalize it. Just use it.

The feeling, the first time a machine does exactly the small thing you asked of it and nothing else, is its own quiet kind of freedom.