[amaybaum]

Guiding draft PR for the K1-RDY OD-1 and OD-2 evidence landing.

Status: WIP / UNVERIFIED. No exact-head review has been done yet; it will arrive as the last commit. No merge is authorized.

Authority

amaybaum, Claude Code session chat, 2026-11-05 (recorded at #1114 c.6653656756):

1. K1-RDY: Approve OD-1 to OD-7 as recommended. Preserve the 313 recovered lines and all scientific limitations. No additional execution or theorem promotion is authorized beyond the approved decisions.

The two decisions this PR implements:

• OD-1: land PROOFS.md on main before the round's base D.
• OD-2: in the same PR, record the evidence(#1031 K1-RDY): written proofs of the blindness lemma, P1 and L-O(i)/(ii) (WIP / UNVERIFIED) #1119 proof review on main verbatim and correct LEMMAS.md's status lines in place.

Commits (from main b606ee41)

• 022ae8da
  • Adds k1-rdy-pref-proofs-20261105/PROOFS.md and its README.md, byte for byte from 7c768f05. That copy was independently reviewed: ACCEPT, research(#1031 K1-RDY): recovery of the accepted preregistration (WIP / UNVERIFIED; not frozen) #1114 c.6653236579, then delta-checked.
  • Adds k1-rdy-proofs-20261105/REVIEW.md, which quotes the evidence(#1031 K1-RDY): written proofs of the blindness lemma, P1 and L-O(i)/(ii) (WIP / UNVERIFIED) #1119 review c.6651717259 verbatim.
  • Changes the five LEMMAS.md status lines to "independently reviewed (REVIEW.md)".
  • Updates the directory README's status to match.
• 17684d77: REVIEW.md quotes the evidence(#1031 K1-RDY): written proofs of the blindness lemma, P1 and L-O(i)/(ii) (WIP / UNVERIFIED) #1119 delta check of d8bfd07b in full. That check was delivered to the coordinating session before 6525ad34 but posted only now, at evidence(#1031 K1-RDY): written proofs of the blindness lemma, P1 and L-O(i)/(ii) (WIP / UNVERIFIED) #1119 c.6653737476.

Not changed

• The 313 recovered lines.
• The preregistration drafts on research(#1031 K1-RDY): recovery of the accepted preregistration (WIP / UNVERIFIED; not frozen) #1114.
• The manuscripts, Lean and every frozen record.

No theorem is promoted. PROOFS.md keeps "written proof, not independently reviewed" until this PR's exact-head review records otherwise.

Checks at 17684d77

All exit 0 under python3 -I: legacy_records_check (303 intact), citation_roles_check, duplicate_check, staleness_check, artifact_placement_check, voice_check, claims_check and lean_manuscript_census.

Every quoted comment was checked by script against its API body. No push-triggered workflow matches.

Next

1. An independent exact-head review.
2. The review file, committed as a single-parent child.
3. Green PR CI.
4. The owner's merge decision.

---

Generated by Claude Code

[amaybaum]

Owner conditional approval (administrative record).

amaybaum, Claude Code session chat, 2026-11-05:

Approve merging PR #1122 at 35f4e645, subject to: 1. CI passing on the exact head. 2. No content changes since independent review, other than the already verified integration with main. 3. No unresolved blocking review comments. 4. Preservation of all 303 legacy records. After merging, verify main CI and record the merge evidence. Keep parent Issue #1031 open for any remaining research obligations. Merging evidence does not imply that the scientific problem is solved.

Status of the conditions at 35f4e6455c0119d1f286bfd070b1cd418e418e82:

• Condition 2: met. git diff 214d11e9 35f4e645 is byte-identical to git diff b606ee41 43077718, the content of the independent review. The review of 6ff20b91 returned ACCEPT; commit B adds only the review file, confirmed byte-identical to the reviewer's delivered text.
• Condition 3: met. No review comments and no reviews are open on this PR.
• Condition 4: met. 303 legacy records are intact at the head.
• Condition 1: pending. CI is running on the exact head.

The merge proceeds when condition 1 is met. #1031 stays open.

---

Generated by Claude Code

[amaybaum]

Merge condition 1 not met. Not merged.

CI fails on the exact head 35f4e645, run 41999530295. The same failure occurred on the earlier heads 6ff20b91 (run 41998539181) and 43077718 (run 41999280578). The coordinating session should have reported this before asking for merge approval.

• Failing step: "Numerical probes / core B → SC-42 integration source scope and must-fail controls" (the aggregate "Require every probe shard" fails as a result).
• Cause: section [5], the added-text scan, has one hit: ('verification/programmes/oi-qm/reconstruction/k1-rdy-proofs-20261105/REVIEW.md', 'local-path', 'scratchpad'). The scan rejects local-path markers in added text.
• Where: REVIEW.md line 96, added in 17684d77. It is a quoted paragraph from evidence(#1031 K1-RDY): written proofs of the blindness lemma, P1 and L-O(i)/(ii) (WIP / UNVERIFIED) #1119 c.6653737476 about a session scratch file that had been overwritten. The paragraph itself says no verdict rests on that file.
• This PR's own content causes the failure, so it is not a base-branch failure.

Fixing it changes content after the independent review, which approval condition 2 excludes. The fix therefore waits for an owner decision. No change has been pushed.

---

Generated by Claude Code

[amaybaum]

New head 0fab30db: reviewed, CI green, awaiting owner merge approval.

The owner approved a correction to the quoted paragraph that tripped the SC-42 scan. Two commits on top of 35f4e645 carry it out:

• 75cb9140 changes only k1-rdy-proofs-20261105/REVIEW.md (+4/−3).
  • The quoted paragraph is replaced by a bracketed note that marks the omission and points to evidence(#1031 K1-RDY): written proofs of the blindness lemma, P1 and L-O(i)/(ii) (WIP / UNVERIFIED) #1119 c.6653737476 for the complete text.
  • The section heading and the introduction now state the omission.
• 0fab30db appends §6 to k1-rdy-pref-proofs-20261105/REVIEW.md (+25/−0). §6 is the independent reviewer's delta check, added unchanged.
  • It restricts the earlier byte-for-byte statement to the head that review covered, 6ff20b91.
  • It records that at the new head the comment is quoted byte for byte except for the one marked omission.
  • Lines 1–140 are unchanged.

Review. An independent, model-authored, read-only reviewer did the delta review: ACCEPT WITH CHANGES at 75cb9140, where the one required change is §6. Its byte-identity check of 0fab30db returned MATCH, with §6 sha256 84c37030…5fcf.

Checks, run with python3 -I at 0fab30db:

• legacy records: 303 in 75 namespaces, all intact;
• citation roles: all hold;
• duplicate, staleness and placement checks: pass;
• SC-42 integration check: PASS, with 0 added-text hits. At 35f4e645 the same check reproduces the FAIL.

CI. Run 42003281690 at 0fab30db concluded success: 25 jobs passed and 1 was skipped. The branch is up to date with main 214d11e9, and GitHub reports it mergeable (clean). Against main, the diff is 6 files, +631/−8.

The earlier approval covered 35f4e645 and does not extend to this head. The PR is still a draft and has not been merged.

---

Generated by Claude Code

[amaybaum]

Merge record

Merge. Merged as 4b4b437cfff9dcf32c882251d1a4c3445ca9e8be, a normal merge commit. Its parents are main 214d11e9 and the approved head 0fab30db.

Approval. The owner gave a conditional approval of 0fab30db on 2026-11-05. Each condition was checked at merge time:

• CI was green on that exact head (run 42003281690).
• The independent delta review and the byte-identity check (MATCH) still held.
• There were no reviews, inline comments or open threads.
• All 303 legacy records were intact.
• GitHub reported the PR mergeable (clean) against main 214d11e9.
• The body contains no closing keywords.

Post-merge checks

• The merge tree equals the tree of 0fab30db.
• At 4b4b437c, legacy_records_check: 303 records in 75 namespaces, all intact.
• Main push CI run 42005356496 succeeded: 25 jobs passed, 1 skipped; Mathlib bridge Build and Release gate both passed.
• The project board shows the card as Closed.

Status of the parent and related issues. #1031 stays open for the remaining K1-RDY research obligations. The next round may be drafted under its existing authorization; freezing and executing it need separate approval. #1114, #1115, #1033, #1059 and #1032 are open.

Landing this evidence does not settle the scientific question.

---

Generated by Claude Code