Failed & Blocked Tests
Failed simulation checks
None, currently — the simulation registry’s last run had 0 failures out of 17 claims checked. This is not because failing checks are hidden: see the full results and the 1 NOT_TESTABLE claim (effective-spin shadow asymmetry — requires 2D ray-tracing not implemented here).
Blocked — couldn’t be run at all
From the full 83-file appendix coverage audit of Book VI, 2 appendices have a real testable claim that this lab could not check, for a specific, named reason:
Appendix PPP — Collapse of Physical Constants
Presents specific numeric values (ℏ, G, Λ in "validator units") without a derivation formula this lab can independently re-derive and check. The constants appear to be fitted/reported outputs of a specific prior run, not a checkable closed-form prediction.
Appendix RRR — NPZ Validation & Unified Scoring ("Killer Test A")
Attempted to run verbatim; the pre-existing script requires a killer_test_A_summary.csv with columns (vrot_band_mean, vrot_band_cv) that no available data-generation step in the repository actually produces.
Rejected during the manuscript audit
Of 16 candidate objective errors independently re-derived during the audit, 1 was rejected — Book VI’s undefined memory kernel K(t−τ). The adversarial review concluded that leaving a kernel abstract in a general definition is legitimate mathematical practice, not an error, and the passage was left unmodified in both the original and v2.