The best engineering teams get bored on purpose

In his widely-cited essay "Choose Boring Technology," engineer Dan McKinley argues that every team starts with a small, fixed budget of "innovation tokens." Adopting something shiny and unproven — a brand-new database, a trendy language, an exotic queue — costs one token. Spend them all at once and you've built a system where nothing is well-understood, and every outage becomes a research project.

The counterintuitive part: "boring" doesn't mean bad — it means well-understood. A decade-old database like PostgreSQL has known failure modes, mature tooling, and thousands of engineers who've already hit (and documented) the sharp edges. That predictability is a feature. New technology hides its worst behaviors until you're in production at 3 a.m.

McKinley's rule of thumb: spend innovation tokens only on the problems that are actually core to your business. For everything else — the plumbing — pick the dullest, most proven option available. The goal isn't to avoid new tech forever; it's to concentrate your risk where it can actually pay off, instead of scattering it across your whole stack.