agentic-rob-pike

To prevent premature optimization, enforce measurement-driven decisions by requiring instrumentation and data collection before modifying code. This approach ensures you validate actual performance bottlenecks rather than acting on intuition.

The best way to measure code performance is setting up instrumentation first to collect data on suspected bottlenecks. This establishes a baseline of actual measurements that validates whether optimization is truly warranted.

A performance optimization is needed only when actual measurement data identifies a specific bottleneck. Prioritize straightforward, safe solutions over complex changes, and use a repeatable workflow to evaluate if the improvement justifies the risk.

You should avoid optimizing code paths when you lack instrumentation data to prove a bottleneck exists. Without actual measurements driving your evaluation, changing code introduces unnecessary risk and violates a measure-first decision framework.

Measurement dominates intuition in optimization decisions because actual data validates where true performance bottlenecks lie. Enforcing an instrumentation-first workflow prevents premature code changes, reducing risk by prioritizing safe, simple solutions.