Design the smallest experiment that can prove or refute a concrete technical hypothesis. Use when a user wants a minimal reproduction, PoC, or feasibility experiment with measurable outcomes. Separate hypothesis failure from blocked or invalid runs; do not expand into a production architecture.
