The Pleasure Order (VOT) - Generic-Tipping Discriminator Pre-Registration v0.2: frozen test with Amendment 2A and stopped P0 method candidate

FROZEN PREREGISTRATION WITH PROSPECTIVE GOVERNANCE AND METHOD-STATUS UPDATE. The original Generic-Tipping Discriminator preregistration v0.2 is unchanged. Amendment 1 remains in the version history. Amendment 2A adds prospective governance for the stronger-rival and external-endpoint architecture; it does not authorize a target-specific verdict, which would require a complete Amendment 2B execution annex. The core empirical claim requires S1 shared decline/recovery fold structure, S2 alignment with an independent external viability/functional endpoint, and S3 separation from explicit estimable strong active-inference/free-energy rivals. No one unit is sufficient. The first latent-state/VFE/EFE P0 candidate (0.1.1-p0-latent) passed C0 software checks and 53 software tests after a rule-consistency correction. Fresh compact C1 diagnostics on replicate indices 101-103 failed two-sided recovery: weak VOT and weak R6 folds were each recovered 0/3, strong VOT 2/3, strong R6 1/3, S0 produced one false fold and D5 remained unstable. The candidate is stopped before quantitative C2. Verdict boundary: P0_CANDIDATE_NOT_CALIBRATED; verdict_authorized=False. This is an engineering result about the method candidate, not evidence for or against the substantive collapse-gate wager. P0 and P1 remain uncleared. C3 rule calibration, C4 locked validation, participant-level data work, ethics steps and target-specific Amendment 2B remain closed.