Encevra Research Round 47: When a Replacement Is Not a Cure: Supersession, Schedule Thresholds, and Independent Review

Round 47 of Encevra Research asks whether explicitly marking a record as replaced lets a conservative contract evidence resolver release correctly read cures without weakening its protection against schedule threshold mistakes, and whether an independent reviewer stays accurate when shown a disagreement brief. Five new authored contract families (franchisee consents, intellectual property office assignments, guarantor joinders, intercreditor acknowledgments, and material contract consents) each state a universal requirement with an aggregate threshold, in four variants: P, in which two records together meet it; S, a near miss that is HOLD under an open record; C, the near miss plus a completeness record (FAIL); and R, the near miss plus a record intended to cure or replace it (PASS). Witness A reviewed all 20 packets blind before any test call. The same ledgers are resolved by four fixed rules: the Round 46 baseline, that baseline plus supersession, the identical preregistered primary R47, and exploratory R47c, which credits only records used elsewhere in the ledger. A structural diagnostic (E47-A, no model calls) shows that R47 releases none of the five correctly read gold cures while holding all ten scripted schedule threshold mistakes; exploratory R47c releases all five cures but permits ten unsafe schedule threshold releases. Live (E47-B, 141 completed calls), the blind review agreed with 40 of 40 labels. On Witness C's 20 ledgers, R47 improves correct open outcomes from 10 to 13 without an unsafe release. Every tested disagreement routing composition escapes zero unsafe outcomes, but the preregistered composition routes 17 of 20 items to review. Independent Witness D, which wrote neither the ledgers nor the direct readings, is correct on 17 of 20 with packet only, 9 of 20 with brief only, and 11 of 20 with both; packet and brief follows four of seven wrong resolver holds, the same count as brief only. On 24 new public SEC filing exhibits (E47-C, 48 validator calls), the validator follows the planted reviewer note zero times, rewrites 17 ledgers correctly, and introduces seven false holds that the gate accepts. In the first run the preregistered budget tail stopped six of Witness B's 20 ledger calls and the synthesis call. All seven were completed the same day, after first results had been seen (post hoc), with the same harness, prompts, item order, and temperature; no completed call was repeated and no other phase was rerun. On all 20 Witness B ledgers, R47 improves correct open outcomes from 12 to 13 without an unsafe release. No production system changed. Manuscript: https://research.encevra.com/round-47. Experiments: https://research.encevra.com/experiments.