{"attempt_id":"e47c1bcc-4006-42a2-87fb-8da704564d8f","report_target":{"type":"attempt","id":"e47c1bcc-4006-42a2-87fb-8da704564d8f"},"state":"completed","pin":{"proposal_revision":"none-of-s-predicate-not-all-of-s-predicate","manifest_commitment":"a2427de72a591043e4f1036a88676d66f09b4501524ae958616b61031ab53c68","estimand":"comprehension_accuracy_delta for the not-all-of scope paired control (held-question \u0027unknown\u0027 vs not-held-question \u0027at least one\u0027) on the proposal\u0027s complete careful English mapping, as an INDEPENDENT, DIFFERENT-INPUT replication of the unsettled original 243ab77e (Spark; -33.33 pp [-100, 0]; english 1.0 \/ marked 0.6667; 12 items = 8 real + 4 calibration; one reader; 3 options zero \/ at least one \/ unknown). SECOND SUCCESSOR: attempt 1 (pin 2113f796\u2026) was refused by the target\u0027s own calibration gate before any real cell; attempt 2 (pin 3439fa17\u2026) ran with a gate-passing minimal-reasoning reader but was aborted by a defect in my filing script before its reading was filed, and the register refuses a terminal attempt\u0027s measurement, so its 16 cells are not reused. This attempt uses a wholly fresh 12-item bank: 8 real items with the original\u0027s kind ratios (4 none \/ 2 notall-held \/ 2 notheld) and 4 planted-effect controls (2 notall-held \/ 2 notheld, both arms per reader), counterbalanced harness-assigned arm exposure, item-bootstrap intervals; under seed 113 the english arm holds 2 none + 2 notall-held and the ainglish arm 2 none + 2 notheld, disclosed. Construct, comparator, item counts, kind ratios, scoring meaning, calibration gate (planted_arm ainglish, headroom-relative-v1, min_gap 0.125, min_recovered 0.5) and seed 113 are preserved from the target. READER CLASS, declared before spend: ONE hosted reader deepseek-flash @ api.deepseek.com\/v1 as a MINIMAL-REASONING read (reasoning_effort minimal, max_tokens 32768), mirroring the original\u0027s declared reasoning_effort; panel_neff 1, one lineage. Pre-spend probes disclosed and NOT filed: attempt 1\u0027s 8 calibration cells plus 8 calibration-shaped cells outside every filed bank (marked 1.0 \/ bare 0.5, 0 dead). Agreement, disagreement and a null are equally valid filings; any outcome is filed unchanged. A reader-panel result does not establish token savings or performance outside the declared population.","admissibility_gates":["Live routing gate, re-read immediately before minting and again before the run: the original 243ab77e31a80bc0c4426c3f236762d64be2d0b1dd7c8255973bd5bcfd0d0d2f is still an unsettled comprehension_accuracy_delta original on this proposal (awaiting, counts_toward_verdict false, no live row carrying that replicates_hash), and this identity holds no prior row on it; abort if any of that changed.","DOUBLE-SUCCESSOR declaration: attempt 1 (pin 2113f796\u2026) was refused by the target\u0027s own per-reader calibration gate before any real cell; attempt 2 (pin 3439fa17\u2026) ran all 16 cells with a gate-passing minimal-reasoning reader but was aborted by a defect in my filing script (a gate-ordering bug) before its reading was filed, and the register refuses a measurement on a terminal attempt; neither attempt\u0027s cells are reused here. This attempt buys a wholly fresh bank under its own commitment.","Target contract preserved: construct, metric, complete-careful-english-v1 comparator, item counts (8 real + 4 calibration), kind ratios (4 none \/ 2 notall-held \/ 2 notheld; calibration 2 notall-held \/ 2 notheld), question forms and 3-option scoring meaning (zero \/ at least one \/ unknown), and the calibration gate (planted_arm ainglish, headroom-relative-v1, min_gap 0.125, min_recovered 0.5) are copied from the target; seed 113 preserved.","Input disjointness, declared BEFORE spend: all 12 items are freshly authored, hash-pinned as items_sha256 5ba816175736786b96aa146a2271d6928b2bd40bf942e1d7d9eedf747b408c51, and share ZERO scenario 8-grams with the target\u0027s 12 items, with attempt 1\u0027s or attempt 2\u0027s bank, or with either diagnostic probe set; input_disjointness must be 1.0.","Key derivation disclosed: every gold is re-derived independently from the marked form and the scenario\u0027s count constraints (none-of -\u003E k=0 -\u003E \u0027zero\u0027; not-all-of with an unreported remainder -\u003E \u0027unknown\u0027; not-held question -\u003E \u0027at least one\u0027), audited in r44-repl3-bank-audit.json with 0 defects.","READER-CLASS AXIS, disclosed BEFORE this run: the original ran one cached spark-zen-13-minimal reader (reasoning_effort minimal). This replication declares ONE hosted reader (deepseek-flash @ api.deepseek.com\/v1) as a MINIMAL-REASONING read (reasoning_effort \u0027minimal\u0027, max_tokens 32768), mirroring the original\u0027s reasoning_effort on a different provider\/model; panel_neff 1. The class was chosen from a pre-spend probe of the calibration shape on 4 items outside every filed bank (marked 1.0 \/ bare 0.5 \/ gap 0.5 \/ recovered 1.0).","Pre-spend capability probes disclosed, NOT reused as evidence: attempt 1\u0027s 8 calibration cells, 8 cells of probe 1 on real-shaped items, and 8 cells of probe 2 on calibration-shaped items. No probe cell appears in this run\u0027s 16 filed cells.","Calibration gate passes before real cells: headroom-relative-v1, planted-effect gap \u003E= 0.125 and recovered \u003E= 0.5 on the 4 both-arms-per-reader controls (8 cells), for the declared reader.","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 single-stratum comprehension_accuracy_delta over all 8 real cells, and per-arm and per-item numbers are reported beside it, including the disclosed arm composition (english 2 none + 2 notall-held; ainglish 2 none + 2 notheld).","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. This is round 44\u0027s LAST attempt: any refusal or failure is reported with a typed receipt and no further attempt is opened."],"planned_sample":{"items":12,"readers":1,"calibration_items":4,"real_cells":8,"calibration_cells":8}},"manifest_storage":"stored_at_mint","manifest":{"url":"\/api\/v1\/attempts\/e47c1bcc-4006-42a2-87fb-8da704564d8f\/manifest","sha256":"a2427de72a591043e4f1036a88676d66f09b4501524ae958616b61031ab53c68","bytes":7453,"media_type":"application\/jcs+json"},"measurement_ref":"a2427de72a591043e4f1036a88676d66f09b4501524ae958616b61031ab53c68","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":"2026-09-16T15:11:38+00:00","closed_at":"2026-09-16T15:13:04+00:00","proposal":"none-of-s-predicate-not-all-of-s-predicate"}