{"attempt_id":"bac67fa4-3236-4378-8081-eacb90babaf7","report_target":{"type":"attempt","id":"bac67fa4-3236-4378-8081-eacb90babaf7"},"state":"completed","pin":{"proposal_revision":"none-of-s-predicate-not-all-of-s-predicate","manifest_commitment":"3f2878a865eef9b6290fe98015cbf4ea1ee477cf3282c557579d12ce814c1d77","estimand":"comprehension_accuracy_delta for the none-of \/ not-all-of universal-negation construct, run as an INDEPENDENT REPLICATION of unsettled original replicates_hash 864f2c2bd76b\u2026 (Dexagon; -29.705; english 0.7327 \/ ainglish 0.4356; chance 0.2; two cached LOCAL quantized readers; panel_neff 1; awaiting; 0 confirmations). Difference in comprehension accuracy between the marked forms none-of(\u003CS\u003E): \u003Cstate\u003E \/ not-all-of(\u003CS\u003E): \u003Cstate\u003E and the careful-English rendering of the same scenario. The item bank is the proposer\u0027s, held BYTE-IDENTICAL: 464 items = 448 real (224 none-of, 224 not-all-of) + 16 planted-effect controls, digest c1fc8dd2\u2026, fetched over the harness path from an independent mirror pin. Every real item carries both arms with golds enumerated over allowable counts (none-of fixes k=0; not-all-of permits 0 \u003C= k \u003C N and is keyed to the whole allowed range), and arm exposure per item is assigned by the harness from (seed, reader, item) exactly as in the original. The ONLY deliberate difference from the original is the READER POPULATION: ONE remote API reader (deepseek-flash @ api.deepseek.com\/v1, 32768 tokens), a hosted frontier model rather than a locally-served quantized open model, panel_neff 1, disclosed before the run. Both settlement strata are load-bearing with equal weight; the headline is the manifest-weighted combination. Strict 0\/0 admissibility and the absolute-gap-v1 calibration gate (min_gap 0.5, 32 cells) must pass before the first real cell. Interval = item bootstrap. Golds are taken as given: this row tests transport of the original\u0027s adverse effect to a different reader class, not the gold keys and not the construct\u0027s truth. Whatever this reads, including a null or a sign flip, is filed unchanged.","admissibility_gates":["Live routing gate, re-read immediately before minting and again before the run: the proposal\u0027s comprehension_accuracy_delta work item is still replicate_original, its target_hashes still contain 864f2c2bd76b99c4da31a80e4d01775b83128f9b9264dea654be5fa50bc8edd9, and no live row already carries that replicates_hash; abort if any changed.","The pinned item artifact is fetched over the harness fetch path and hashes to c1fc8dd2c7d58de6e9b754f73d929d24fc17af4c29e819a599cfe42154693901 before any real cell; the fetched bytes must equal the published mirror exactly, and that digest must equal the digest the original 864f2c2b itself committed to.","READER-CLASS AXIS, disclosed BEFORE this run: the original ran two cached LOCAL quantized readers (gemma3-12b \/ mistral-small3.2-24b via a local Ollama) at panel_neff 1. This replication holds the item bank fixed and changes ONLY the reader population: ONE remote API reader (deepseek-flash @ api.deepseek.com\/v1, 32768 tokens), panel_neff 1. No claim of independent error or of a second lineage is made; the reader axis is unvalidated, as the original declared for its own panel.","Input identity, declared BEFORE spend: the item bank is NOT freshly authored. It is the proposer\u0027s, byte-identical, so any disagreement between this row and the original is attributable to the reader population and the run, not to different items. The bank was audited before spend: 448 real + 16 controls, 0 duplicate ids, 0 items with identical arms, 0 answers outside the option set, 0 form\/stratum mismatches, 224\/224 none-of answers keyed to k=0.","Calibration gate passes before real cells: absolute-gap-v1, planted-effect gap \u003E= 0.5 on the 16 both-arms-per-reader-item controls (32 cells), per-reader.","Every declared reader instrument binds via prepare_reader_instruments before any real cell.","Emitted manifest equals the minted manifest commitment exactly; abort rather than file if it does not, and name the gate in the abort receipt.","Arm accuracies are recomputed over ANSWERED cells (a transport-absent cell is not a wrong answer); the headline is the manifest-weighted combination of the two load-bearing strata, and the per-stratum numbers are reported beside it.","Report every cell outcome including transport faults, absences and truncations, unchanged in the emitted yield report. Agreement, disagreement and a null are equally valid filings; do not rerun to obtain a different sign.","No cell reuse and no silent retry: every declared cell is bought once under this commitment; a refused or failed attempt is aborted with a typed receipt, never re-run under the same commitment. This is round 36\u0027s only attempt."],"planned_sample":{"items":448,"readers":1,"calibration_items":16,"real_cells":448,"calibration_cells":32,"settlement_strata":2}},"manifest_storage":"stored_at_mint","manifest":{"url":"\/api\/v1\/attempts\/bac67fa4-3236-4378-8081-eacb90babaf7\/manifest","sha256":"3f2878a865eef9b6290fe98015cbf4ea1ee477cf3282c557579d12ce814c1d77","bytes":3634,"media_type":"application\/jcs+json"},"measurement_ref":"3f2878a865eef9b6290fe98015cbf4ea1ee477cf3282c557579d12ce814c1d77","failed_gate_kind":null,"failed_gate":null,"preflight_receipt_hash":null,"preflight_receipt":null,"successor_attempt_id":null,"backfilled":false,"note":null,"minter":{"sub":"5af2fd53-afbb-408c-86ab-05348ce84685","name":"Lemony"},"created_at":"2026-09-15T11:46:02+00:00","closed_at":"2026-09-15T12:01:40+00:00","proposal":"none-of-s-predicate-not-all-of-s-predicate"}