[unpaidmoon]

Change and mathematical scope

Audit, not a new κ. The selected conditional exponent stays 781451/250000000 (0.003125804). Frozen historical inputs are not rewritten, and no live successor is added to an accepted-hash list.

At upstream 2485805a6d47f36ea83ddde6327d453ac82d44dc, scripts/replay_historical_check.py has one overlay of the selected-record checker: the tooling-ledger input scripts/verify_selected_result.py. Its frozen file still hashes to

d7fd929726b326aeeaf0a3996475adec88eee9aac09ae0bebeeaa0bc1fc8f199

The only previously accepted live successor is

c0cc23a8534d91c8d272195315c725213f13279815750ffdef055ec3adfbe911

The live file hashes to

d6b305eac5b9bdb8172c469ad30a1655a60fa0aceac8b1bf610e1b73f51babd8

authenticated_input raises Unrecognized live source update: scripts/verify_selected_result.py before any historical command runs. The same gate rejects 19 CONFIG entries on seven distinct paths. The other six are Makefile (historical prefix), scripts/partial_swap/shared.py (13 aliases), certificates/partial-swap-network.json, docs/ci-verification.md, scripts/run_isolated_tests.py, and tests/test_isolated_test_runner.py. The live selected-family note docs/ci-verification.md hashes to c7d3027e9ad8bcbff68f07aa9576b30fb2f3c76c44d876d40c7fdbf4a3e2228a.

The helper plan for 01885346d4e78083b0a41786f977f71c73e3ffa2..2485805a6d47f36ea83ddde6327d453ac82d44dc selects nine receipt packages because that commit edits docs/research/pr-ledger.json and docs/research/pr-intake-decisions.json. It does not select make frontier6030-verify. Three changed evidence files match no package input. Those nine packages were not replayed.

Task and overlap

Base commit 2485805a6d47f36ea83ddde6327d453ac82d44dc, priority revision 8. Task portable-replay, scope refined: handoff comment 19138624 asks for an inspection report, not an allowlist. This is item 4 of the structured open-handoff list on #1166 after excluding delivered or already answered requests, including the typed actual-width H work, the length-22 revisit refinement, comment 19133514, the #4902 two-FULL successor, the #4945 generator chain through #6241, #5720's active paid-query audit, and #5273.

Open PR titles on 2026-10-30 did not name this historical admission gap. The nine planned receipt packages, including research/frontier-source-retention, were not re-executed.

Validation

Python 3.12.3, standard library, Linux:

python3 -B scripts/research_selected_baseline.py
python3 -B research/historical-selected-replay-gap/verify.py
python3 -B scripts/research_package.py --package research/historical-selected-replay-gap

Orientation printed PASS_RESEARCH_BASELINE_BINDINGS_ONLY. The package command printed PASS_HISTORICAL_SELECTED_REPLAY_GAP and PASS_REGISTERED_RESEARCH_PACKAGE. The verifier checks the frozen pin, the live digest, all 19 rejections, a pin-membership control, the exact helper plan, and an empty-diff control. It does not replay frontier 6030, the nine receipt packages, or a historical command after the admission failure. No human or formal-kernel review.

Credit

The historical replay gate is Dennis's scripts/replay_historical_check.py. The source-history fixture handoff is Clear1nsight's PR #4984. The selected record is the reviewed #6030 conditional construction. Prepared by unpaidmoon with substantial Cursor Grok assistance. Apache-2.0.

Open next questions

Review each of the seven live files against its last accepted or frozen bytes and either admit that exact successor or revert it. Start with scripts/verify_selected_result.py. An allowlist edit is not a frontier 6030 replay.

Whether the nine receipt packages selected by this commit's ledger edit still pass is a separate replay. This package does not run them.