One-case Policy Identification Observatory release for the Wales default 20 mph casualty-attribution audit. The provisional 28% comparison is descriptively replayed, while a constructive witness shows that the affected-road causal estimand point value and sign are not identified from the frozen aggregate evidence. The positive attribution remains unsupported. Round four is preserved as supplementary external-model review; rounds five through eight are deterministic continuity certificates for bounded release-infrastructure and package-contract corrections, not independent review. A retained replacement clean-checkout replay passed, but the release is unrefereed and supplies neither an effect estimate nor policy validation.