Experimentation

Not everything built needs to ship. Experimentation is how self-organising design stays alive — trying things without committing to them, learning what works before deciding whether it's worth maintaining.

The shared infrastructure makes this cheap. Spinning up a new service in Coolify takes minutes. Tearing it down leaves nothing behind. There's no procurement process, no IT ticket, no approval needed. The cost of a failed experiment is low; the cost of never experimenting is much higher.

A good experiment has a question it's trying to answer: Does this tool solve the problem? Will people actually use it? Is this easier than what we do now? Without a question, experiments tend to linger as half-finished things that nobody maintains and nobody removes. Keeping experiments short and documented makes it easier to hand off what worked and let go of what didn't — a note on what was tried, what was learned, and why it was kept or dropped is enough.

See Archiving for why past experiments are worth making visible too, even the ones that went nowhere.