Every feature is a promise

Adding a feature feels like progress.

But every feature you ship is a promise. It has to work next month. Next year. When the codebase is bigger and you're tired and three other things are broken.

Most features don't earn that promise. They solve an edge case, impress someone in a demo or exist because a competitor has one. Then they sit there forever, costing a little maintenance and adding a little confusion.

The best time to add a feature is after you've tried not to.