Premature optimization is the root of all evil -- profile first

Developers optimize code before measuring. They spend hours improving a function that runs once at startup while ignoring the O(n^2) loop that runs per request. Optimized code is harder to read and maintain.

(1) Make it work first. (2) Make it right (clean, readable). (3) Make it fast (only if measured as slow). Profile before optimizing: use flame graphs, profilers, and benchmarks to identify actual bottlenecks. The 90/10 rule: 90% of time is spent in 10% of code. Optimize that 10%. Tools: Chrome DevTools, py-spy, pprof, perf, Benchmark.js.

Intuition about performance is usually wrong. The bottleneck is almost never where you think. Profiling reveals the truth. Optimizing without profiling trades readability for imaginary performance gains.