Empirical Audit and Falsification: Evidence Levels, Reproduction Standards, and Decisive Tests for FPM

Finite Possibility Mechanics is unusually testable because its mathematical, computational, and phenomenological layers can be audited separately. This paper defines that audit architecture. It distinguishes proof, implementation verification, numerical correspondence, calibration, retrospective empirical comparison, and prospective prediction; reports the current executable suite; specifies reproducibility requirements; and fixes the failure logic for the framework's highest-value physical tests. The current public suite contains sixteen primary experiments plus one starvation subtest. These are implementation and algebraic checks of the supplied model; they do not independently establish physical mechanism. The next evidence step is external: frozen parameter ledgers, complete data likelihoods, independent reproduction, and measurements capable of separating FPM from its baselines.