[amaybaum]

Guiding draft PR for the PE-01 evidence landing recorded in #1088 (comment 6628737033). Merge is not authorized: it needs the owner's explicit approval of the reviewed head, scope and CI results.

What lands. 16 records, byte-identical to #923 at cd4064818fe2cff1ee4b147a9761ed157bcf990c (111,294 bytes), plus one new file, REPLAY.json. All are under verification/programmes/oi-qm/reconstruction/oi-pseudoentropy-20261025/pe01_bounded_tests_v1/:

• PLAN.md;
• amendments/A1–A3, which are historical decision records;
• REPORT.md;
• review/results/ (README, REVIEW, SHA256SUMS and the reviewer's mine/ scripts with search_out.json), tree e48ad997.

Why. book/ch18-beyond.md:106 (FULL:4919), merged via #933 (687e885b), states that an embedded observer whose actions reach the hidden state of a deterministic source can, for some sources, predict its output better than chance. Its only recorded support is PE-01's witness and its independent recomputation, which until this lands are on #923's branch only. Under §A.42 items 1–2, that support belongs on main.

The pin, REPLAY.json (option (a), owner decision of 2026-11-04).

• Previously verified results: the exact values from review/results/REVIEW.md, each cited by line.
• Instructions for future reproduction: indep.py, witness.py and mech.py (standard library only) re-establish the statement. search.py (byte-exact search_out.json) and the numpy scripts support the broader scope, and the numpy version is not recorded. No script was run for this PR.
• Historical records: A1–A3 are labelled historical, and the superseded readings are flagged.
• Provenance only: the 78-job run, the shards, results/ and the off-main commits these records cite. They resolve at refs/pull/923/head.
• Line numbers: the records cite corpus lines at c37700ad, and the pin maps the moved ones to main (for example, Main.md:560 is now :582 and ch18-beyond.md:114 is now :106).
• Landing method: land with "Create a merge commit" only, not squash or rebase. The ruleset allows all three, so whoever merges must choose it.

What does not land. Everything else on the #923 branch, which stays there as provenance (cd406481, refs/pull/923/head). Among it:

• results/ (including shards/), pe01/, run.py and verdict.py;
• review/code/, review/literature/ and review/plan/;
• the thread README.md and thread review/REVIEW.md;
• pe05_charged_tables_v1/ (PE-05: separate active from passive observers and charged from uncharged lookup tables (OI-PE-1 follow-up) #934).

PE-05 needs the PE-01 package only if it is executed.

Until this merges, the evidence for book/ch18-beyond.md:106 is on #923's branch only. No scientific status changes, and no new computation, conclusion or reclassification is made.

Review and checks. Commit a355513e (single parent 27dc3443) was independently reviewed twice: ACCEPT WITH CHANGES at cf12f8ca, then CONFIRMED WITH NITS at a355513e. Static checks pass (artifact_placement_check, claims_check, duplicate_check, voice_check); secret and closing-keyword scans of the commit message are clean. legacy_records_check needs full history and is left to CI.

---

Generated by Claude Code