ARK-449 — State Changes After Verification (v1.2 corrected)

ARK-449 — State Changes After Verification Version 1.2 (post-data analysis correction) ExecutionProof ARK Authorization-Boundary Track — Remnant Fieldworks Inc. Verdict: PASS (raw-primary metrics) v1.2 correction summary: Internal peer review identified that the v1.1 analysis applied the authorizer-qubit readout error (SPAM_A) to the payload outcome without a valid same-qubit readout-assignment model. v1.2 makes raw payload retention the primary metric (S_A_min=0.9794, L_D_max=0.0011, Δ_B=0.9783). SPAM-corrected values are retained as secondary/descriptive only. Verdict unchanged: PASS. Also corrected: residual v1.0 label refs → v1.1; ProofRecord manifest_sha256 filled; framing updated to avoid overclaiming (boundary logic validation, not live distributed-state propagation). Repository: github.com/derekhone/executionproof-testbeds Tags: ark-449-v1.1-lock (preregistration), ark-449-v1.1 (hardware execution), ark-449-v1.2 (corrected analysis) Backend: IBM Quantum ibm_marrakesh (156-qubit Heron r2) Job IDs: SPAM d9crr82neu4c739mcsd0, principal d9crvbhhtsac739c6a70 Key Metrics (raw-primary): S_A_min=0.9794, L_D_max=0.0011, Δ_B=0.9783. All seven DENY state-change scenarios ≤0.11%. Re-authorization restores execution; replay with stale proof fails closed. Doctrine tested: Permission at approval time is not permission at execution time. Full preregistration-first methodology: hypotheses, arms, metrics, pass/fail criteria locked and SHA-256-hashed before any hardware job. Post-data corrections fully disclosed per ARK-447 v1.0→v1.1 precedent. License: CC0 (public domain)