[amaybaum]

Draft. This is a minimal evidence-integration PR, authorized by the owner's decision on D-72 (#914 comment 6625889194, #1022 comment 6625890162; chat, 2026-11-04). Preparation and review are authorized. A merge needs the owner's approval at the exact SHA.

• Head: 8b02775d031d087ae2daf2b60a15aa20ee73de45, two commits on main d18f1cbe.
• Source: Research (draft): PC-1 — physical observer–apparatus coupling, preregistration only #1022 at 99dce4b6087d74842d8b0cf032ec1bbe2bbb9b4d. It stays open and unchanged until this PR's evidence is verified on main.

What this adds

Commit 1, 4284a6d1: 21 files, each byte-identical to its blob at 99dce4b6. The file list matches #1022's exactly.

• verification/programmes/oi-qm/reconstruction/physical-coupling-pc1/:
  • the frozen preregistration;
  • the amendments PC★, PC-S2 and PC-S2-C13, and the two-track disposition;
  • the stage-1 log;
  • stage-2 run 1, VOID and unchanged;
  • stage-2 run 2: RESULT.md, the kernel steps and decision, review.json, PROOF-REPAIRS.md, the controls and the results;
  • the vendored probes.
• verification/lean-mathlib/OIBridge/PC1Controls.lean.
• verification/lean-mathlib/OIBridge/F2CoherenceControls.lean. This module is shared with research: SUBSTATE-2 recursive inverse-limit substructure and depth-feedback controls #1011, with the same blob fbf47504 on both branches. It lands once, here.
  • Nothing else from research: SUBSTATE-2 recursive inverse-limit substructure and depth-feedback controls #1011 is included.
  • Any later research: SUBSTATE-2 recursive inverse-limit substructure and depth-feedback controls #1011 integration must drop this module, its import and the identical F2-1 census family.
• Edits on current main: two import lines in OIBridge.lean, and the two kernel-only census families from Research (draft): PC-1 — physical observer–apparatus coupling, preregistration only #1022, verbatim.

Commit 2, 8b02775d: one new file, stage2/REPLAY.json (review finding B1, §A.42 item 1).

• Why it is needed. The records' controls freeze-check the other OIBridge modules as they stood at 5914c17c. Run in place on today's main, they therefore report 0(ii) by design.
• What the file pins:
  • the environment: OIBridge from 5914c17c, which is reachable from main, overlaid with the three PC-1 paths from the landing commit;
  • the three commands;
  • their expected exits;
  • their byte-identical outputs: both results.json files and kernel-decision.json.
• Verified in normal and -O modes, including with REPLAY.json itself present.
• No existing record changes.

The scientific record, kept narrow

Item	Recorded value
Strict frozen-protocol token	INCONCLUSIVE (void), reason 0(vi)
Run 1	VOID, preserved
Scoped finding, carried under the owner's one-time exception	None of the 25 screened rules qualifies

• The obstruction is G★ item 3, for couplings whose joint process is a classical measurable bijection with function-valued records.
• It is not a universal no-go. It does not show that OI cannot produce quantum measurement by another mechanism.
• The reference controls C1–C15 return their expected rows. The records mark each control's evidence layer as kernel, paper or exact.

No manuscript, ROADMAP, coverage, grade or status changes.

Checks

Check	Result
Local tools/lean_manuscript_census.py	OK. 118 families, 25 of them current; both new families are kernel-only
Local tools/artifact_placement_check.py	OK
Local tools/coverage_check.py	OK
CI at 4284a6d1	Mathlib bridge (Build, Release gate), the Lean kernel check and core B all succeeded
CI at 8b02775d	Running
Independent review at 4284a6d1	ACCEPT WITH CHANGES, with B1 now addressed; delta review of 8b02775d pending

Still needed

• Green CI at 8b02775d.
• Delta-review acceptance.
• The owner's exact-SHA merge approval.
• Verification on main after the merge. Only then is Research (draft): PC-1 — physical observer–apparatus coupling, preregistration only #1022 closed without merge, with its branch kept at 99dce4b6, which also keeps run 1's failing PC1Controls.lean (blob de0174a9).

🤖 Generated with Claude Code

https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb

[amaybaum]

Independent exact-head review, 8b02775d031d087ae2daf2b60a15aa20ee73de45: ACCEPT, with no blocking findings.

Round 1, 4284a6d1: ACCEPT WITH CHANGES.

• All 21 added blobs equal 99dce4b6, and the file list matches Research (draft): PC-1 — physical observer–apparatus coupling, preregistration only #1022.
• The imports and census lines match Research (draft): PC-1 — physical observer–apparatus coupling, preregistration only #1022 verbatim.
• F2CoherenceControls is identical on research: SUBSTATE-2 recursive inverse-limit substructure and depth-feedback controls #1011 (fbf47504), and nothing else from research: SUBSTATE-2 recursive inverse-limit substructure and depth-feedback controls #1011 is included.
• No manuscript, ROADMAP or coverage file changed.
• One blocking finding, B1: the records' own replay recipe fails on main's tree.

Round 2, 8b02775d: ACCEPT.

• B1 resolved. The reviewer rebuilt the environment exactly as stage2/REPLAY.json describes, using only objects on main. All three commands exit 0 in normal and -O modes, print the stated line, and match the listed files byte for byte; the blob ids match. Run in place on the current tree, the controls give withheld: 0(ii), as the file says.
• N4 fixed in the PR description.
• Placement: one new file and no existing record changed.
• No strengthening. The scoped finding stays limited to the 25 screened rules, classical measurable bijections with function-valued records. There is no universal no-go and no grade or status change.

Carried forward:

• Keep Research (draft): PC-1 — physical observer–apparatus coupling, preregistration only #1022's branch at 99dce4b6.
• Any later research: SUBSTATE-2 recursive inverse-limit substructure and depth-feedback controls #1011 integration must drop the shared module, its import and the identical F2-1 census family.

CI at this head was still finishing when this was written. Merging needs green CI and the owner's exact-SHA approval.

---

Generated by Claude Code

[amaybaum]

CI at 8b02775d031d087ae2daf2b60a15aa20ee73de45: green. All 26 check runs completed with no failures.

The PR is mergeable against main d18f1cbe, and the independent review is ACCEPT at this head (c.6626207243). It is ready for the owner's merge decision at this exact SHA.

---

Generated by Claude Code

[amaybaum]

Merge record. Normal merge of the approved head 8b02775d as 4312028ae367206467921369fc4050a4b93103e7 (owner exact-SHA approval, chat, 2026-11-04).

• Parents: main d18f1cbe and 8b02775d.
• Tree: 7da002d8, identical to the approved head's tree and to a clean merge of the two parents.
• Files: the 21 copied files are byte-identical to Research (draft): PC-1 — physical observer–apparatus coupling, preregistration only #1022 at 99dce4b6.
• Replay: stage2/REPLAY.json, followed exactly as written with the merge commit as the landing commit, reproduces both runs and kernel-decision.json byte for byte, in normal and -O modes.
• Main CI on 4312028a: green. All 31 check runs completed with no failures.

Side effect, corrected on record. Commit 4284a6d1's message contained the closing keyword "close #1022", so GitHub closed #1022 automatically when this PR merged. That happened before the owner's separate approval. See #1022 comment 6626422071. The owner decides whether #1022 stays closed.

---

Generated by Claude Code