In praise of small, sharp tools

The best tool I wrote last year was forty lines long. It did one thing, had no config, and I never thought about it again. The worst was a “framework” for a problem I had exactly once.

There’s a gravity in software toward generality — toward the abstraction that might pay off. Usually it doesn’t. The small, sharp tool wins because it’s cheap to understand, cheap to throw away, and honest about its scope.

Make the thing that solves today’s problem. Generalize only when a second real case shows up — not an imagined one.

That’s the spirit this whole site is built in: each demo is its own little folder, owing nothing to the others.