Skip to content

Non-certifying reviewer cross-checks for the H3 scalar candidate - #6

Closed
d6g8k5htny-coder wants to merge 4 commits into
chatgpt/h3-solver-pilot-20260921from
claude/github-chatgpt-xxw5kp
Closed

d6g8k5htny-coder wants to merge 4 commits into
chatgpt/h3-solver-pilot-20260921from
claude/github-chatgpt-xxw5kp

Conversation

@d6g8k5htny-coder

@d6g8k5htny-coder d6g8k5htny-coder commented Sep 22, 2026 •

Copy link
Copy Markdown
Owner

CoS (2026-09-22) — closing as orphaned.

Content was the cleanest non-certifying H3 review packet (credit 0, gate unchanged), but base was PR #5 → PR #4, and that stack is closed (PR4 superseded by tip PR #9). Salvage path: open a new tip-based PR cherry-picking reviews/independent/h3_* only if still wanted.

Not a lemma close.

Two reviewer working artifacts backing a nonauthor technical review of
engine/solver_pilots/h3_20260921 at 2371f66.

reconstruct_bound.py rebuilds the PROOF.md sections 1-4 numeric chain in
mpmath from the prose alone, importing nothing from engine/solver_pilots.
It reproduces the four section-1 inequalities and arrives at 1.7472485483566,
inside the author-side rational enclosure [1.747106129274, 1.747994749902].

sample_true_value.py samples the conditional law directly through the sector
decomposition the proof uses, to test the claimed inequality against the
quantity rather than against the arithmetic. At r=1/20 it estimates
Z(r)/r^2 = 3.224 +/- 0.005 (3 sigma) and finds no counterexample.

Both paths are NON-CERTIFYING under CLAUDE.md rule 3: mpmath at finite
precision and a float Monte Carlo are each named there as not certificates.
Neither bounds anything, verifies a register, reclassifies a claim, premise or
obligation, or awards any organizational independence credit, which is zero.
Both guard their imports, since mpmath and numpy are not guaranteed present.
Nothing in the repository invokes them and the pilot is NOT DEPLOYED.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G

Copy link
Copy Markdown
Owner Author

CI verify is red, and it is inherited rather than introduced — standing down, with reasons

Run 35670591811, job 106565930224, head 4b7c6f4:

FAILED tests/test_bridge.py::test_agents_md_points_and_does_not_legislate
FAILED tests/test_workflows.py::test_workflow_files_exist
FAILED tests/test_workflows.py::test_every_repository_checker_step_is_seen[h3-solver-pilot.yml]
3 failed, 2739 passed, 1 skipped, 1 warning in 358.04s

Only the pytest step failed; all 40 checker steps ahead of it passed.

Why this is not this PR's failure. The same three fail on this PR's base branch chatgpt/h3-solver-pilot-20260921 with nothing of mine present — confirmed on GitHub's own runner, not just locally: run 35666015671 on base head 2371f660, conclusion failure. I also ran the full suite locally twice, once on the bare base and once with this PR's three files added, and got byte-identical failure sets both times (3 failed, 2739 passed, 1 skipped). This PR adds two standalone scripts and a README under reviews/independent/; nothing imports them, no workflow invokes them, and no existing file is modified.

Attribution, by running each head in the stack separately:

head full suite
claude/drive-audit-github-migration-rrglpp (#2) green
chatgpt/drive-github-hardening-20260919 (#3) green — run 35667260850, 2,832 tests
chatgpt/withdrawal-governance-20260921 (#4) 2 failed — the AGENTS.md ceiling and the workflow inventory
chatgpt/h3-solver-pilot-20260921 (#5, this base) 3 failed — those two plus the pilot workflow's scanner visibility

No fix exists to port, and I am deliberately not writing one here. All three are reported to their authors — the pair on #4, the third as finding 1 of my review on #5. Each is a governance decision rather than a mechanical repair:

  • test_agents_md_points_and_does_not_legislate enforces a 45-line ceiling on a file whose own first line says it "points; it does not legislate." PR Scoped withdrawal and rollout governance — shadow controls, no scientific migration #4 took it to 75. Whether that material moves to CLAUDE.md/governance/ or the ceiling is consciously raised is the author's call; raising the number quietly would erase the reason the test exists.
  • test_workflow_files_exist pins the workflow inventory so the workflow set is an explicit decision rather than directory contents. The filename should be added deliberately, not the assertion loosened to a membership test.
  • test_every_repository_checker_step_is_seen[h3-solver-pilot.yml] is the anti-vacuity guard for test_no_checker_step_masks_its_failure. It fails because the pilot workflow's steps use python3 -m unittest, which the CHECKER regex does not match — so those steps are invisible to the repository's mask-detection scanner. Fixing it means either widening CHECKER or changing what the workflow invokes, both of which belong to H3 1747/1000 candidate: guarded source admission and public certificate evidence #5.

Patching any of them from this PR would resolve, by fiat, the exact findings I raised for their authors to decide, and would widen a review-artifact PR into governance changes I was not asked to make.

No re-run. These are deterministic assertion failures with identified causes, reproduced identically across four runs on two machines — there is no flake hypothesis for a re-run to test, so spending one would buy no information.

What unblocks this PR: #4 and #5 resolving their own failures, after which I will merge the updated base in and re-verify. Until then the red is theirs, not this branch's, and I will keep this PR watched rather than idle. PR #6's own contribution — three new files — is green in the sense that matters: it changes the suite's result by exactly nothing.


Generated by Claude Code

Claude and others added 3 commits September 22, 2026 00:26
The R17 section 4 record for RV-H3-SOLVER-20260921-01 is written and complete.
It cannot be filed under reviews/records/ yet: tools/reviews_check.py requires
every route_key to appear verbatim in registers/json/review_queue.json, and that
key exists only in the live Drive register, modified 2026-09-22T00:18:28Z, not in
the 2026-09-18 export this repository carries. Filing it today would fail CI, so
it is parked under reviews/independent/ where the checker does not scan.

Running the checker against a copy reports exactly one problem, the missing route
key. Schema, the anti-boilerplate floors, the promotion-language and
confidence-voting scans, the independence rules and the gate rule all pass, so
the register's staleness is the only thing in the way. When the export is
refreshed from its source the file moves to reviews/records/ unchanged.

Verdict is PASS_TECHNICAL on the exact object, scoped to the section 1 to 4
numeric chain, the interval module's soundness and consistency of the claimed
inequality with sampled values of the quantity. Independence credit is 0 and
gate_status_after is UNCHANGED, as every record in this repository carries.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G
My previous commit broke tools/consumers_check.py. The README added under
reviews/independent/h3_20260921/ names registers/json/review_queue.json in
prose, so the scan now attributes it to the review_queue tab, and the derived
map in registers/CONSUMERS.json did not list it. CI failed at the consumers step
before reaching pytest, so this was a new failure introduced by that commit and
not one of the three inherited from the base branch.

Regenerated with tools/consumers_check.py --write, which is the documented path
for this map. CONSUMERS.json is a derived scan artifact, not one of the exports
CLAUDE.md rule 7 protects, and the regeneration adds exactly one line naming the
new file under review_queue's prose consumers. No other tab, attribution or
label changed, and no register export was touched.

Verified: all ten repository checkers exit 0, and tests/test_consumers.py passes
31 of 31 including its negative controls.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G
d6g8k5htny-coder pushed a commit that referenced this pull request Sep 23, 2026
PR #6 was closed as orphaned: its base was PR #5 -> PR #4, and that whole stack
is closed. The content was assessed in the closing note as sound, with the
salvage path named as cherry-picking reviews/independent/h3_* onto a live tip.
This is that salvage, and nothing more.

Re-founded on chatgpt/drive-github-hardening-20260919 (PR #3) rather than on the
newest branch. PR #3 has been open since 2026-09-19, has survived every
reshuffle that closed #4, #5, #6 and #9, and is an ancestor of the current tip,
so this work flows forward instead of stacking on a moving target. Basing on the
newest available head is the exact mistake that orphaned PR #6.

The four files are unchanged from their reviewed state. Two are NON-CERTIFYING
cross-check scripts, guarded so they skip when mpmath or numpy is absent. One is
the completed R17 section 4 review record for RV-H3-SOLVER-20260921-01, still
held outside reviews/records/ because tools/reviews_check.py requires the route
key to appear verbatim in registers/json/review_queue.json and the register
export in git remains the 2026-09-18 snapshot, which does not carry it. Filing it
today would fail CI, so it stays parked with its reason written down.

registers/CONSUMERS.json is regenerated because the salvaged README names
registers/json/review_queue.json in prose, which the consumer scan attributes to
that tab. One line, added by tools/consumers_check.py --write.

Verified: all eighteen repository checkers exit 0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G
cursor Bot pushed a commit that referenced this pull request Sep 23, 2026
PR #6 was closed as orphaned: its base was PR #5 -> PR #4, and that whole stack
is closed. The content was assessed in the closing note as sound, with the
salvage path named as cherry-picking reviews/independent/h3_* onto a live tip.
This is that salvage, and nothing more.

Re-founded on chatgpt/drive-github-hardening-20260919 (PR #3) rather than on the
newest branch. PR #3 has been open since 2026-09-19, has survived every
reshuffle that closed #4, #5, #6 and #9, and is an ancestor of the current tip,
so this work flows forward instead of stacking on a moving target. Basing on the
newest available head is the exact mistake that orphaned PR #6.

The four files are unchanged from their reviewed state. Two are NON-CERTIFYING
cross-check scripts, guarded so they skip when mpmath or numpy is absent. One is
the completed R17 section 4 review record for RV-H3-SOLVER-20260921-01, still
held outside reviews/records/ because tools/reviews_check.py requires the route
key to appear verbatim in registers/json/review_queue.json and the register
export in git remains the 2026-09-18 snapshot, which does not carry it. Filing it
today would fail CI, so it stays parked with its reason written down.

registers/CONSUMERS.json is regenerated because the salvaged README names
registers/json/review_queue.json in prose, which the consumer scan attributes to
that tab. One line, added by tools/consumers_check.py --write.

Verified: all eighteen repository checkers exit 0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G
cursor Bot pushed a commit that referenced this pull request Sep 23, 2026
PR #6 was closed as orphaned: its base was PR #5 -> PR #4, and that whole stack
is closed. The content was assessed in the closing note as sound, with the
salvage path named as cherry-picking reviews/independent/h3_* onto a live tip.
This is that salvage, and nothing more.

Re-founded on chatgpt/drive-github-hardening-20260919 (PR #3) rather than on the
newest branch. PR #3 has been open since 2026-09-19, has survived every
reshuffle that closed #4, #5, #6 and #9, and is an ancestor of the current tip,
so this work flows forward instead of stacking on a moving target. Basing on the
newest available head is the exact mistake that orphaned PR #6.

The four files are unchanged from their reviewed state. Two are NON-CERTIFYING
cross-check scripts, guarded so they skip when mpmath or numpy is absent. One is
the completed R17 section 4 review record for RV-H3-SOLVER-20260921-01, still
held outside reviews/records/ because tools/reviews_check.py requires the route
key to appear verbatim in registers/json/review_queue.json and the register
export in git remains the 2026-09-18 snapshot, which does not carry it. Filing it
today would fail CI, so it stays parked with its reason written down.

registers/CONSUMERS.json is regenerated because the salvaged README names
registers/json/review_queue.json in prose, which the consumer scan attributes to
that tab. One line, added by tools/consumers_check.py --write.

Verified: all eighteen repository checkers exit 0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G
cursor Bot pushed a commit that referenced this pull request Sep 23, 2026
PR #6 was closed as orphaned: its base was PR #5 -> PR #4, and that whole stack
is closed. The content was assessed in the closing note as sound, with the
salvage path named as cherry-picking reviews/independent/h3_* onto a live tip.
This is that salvage, and nothing more.

Re-founded on chatgpt/drive-github-hardening-20260919 (PR #3) rather than on the
newest branch. PR #3 has been open since 2026-09-19, has survived every
reshuffle that closed #4, #5, #6 and #9, and is an ancestor of the current tip,
so this work flows forward instead of stacking on a moving target. Basing on the
newest available head is the exact mistake that orphaned PR #6.

The four files are unchanged from their reviewed state. Two are NON-CERTIFYING
cross-check scripts, guarded so they skip when mpmath or numpy is absent. One is
the completed R17 section 4 review record for RV-H3-SOLVER-20260921-01, still
held outside reviews/records/ because tools/reviews_check.py requires the route
key to appear verbatim in registers/json/review_queue.json and the register
export in git remains the 2026-09-18 snapshot, which does not carry it. Filing it
today would fail CI, so it stays parked with its reason written down.

registers/CONSUMERS.json is regenerated because the salvaged README names
registers/json/review_queue.json in prose, which the consumer scan attributes to
that tab. One line, added by tools/consumers_check.py --write.

Verified: all eighteen repository checkers exit 0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G
cursor Bot pushed a commit that referenced this pull request Sep 23, 2026
PR #6 was closed as orphaned: its base was PR #5 -> PR #4, and that whole stack
is closed. The content was assessed in the closing note as sound, with the
salvage path named as cherry-picking reviews/independent/h3_* onto a live tip.
This is that salvage, and nothing more.

Re-founded on chatgpt/drive-github-hardening-20260919 (PR #3) rather than on the
newest branch. PR #3 has been open since 2026-09-19, has survived every
reshuffle that closed #4, #5, #6 and #9, and is an ancestor of the current tip,
so this work flows forward instead of stacking on a moving target. Basing on the
newest available head is the exact mistake that orphaned PR #6.

The four files are unchanged from their reviewed state. Two are NON-CERTIFYING
cross-check scripts, guarded so they skip when mpmath or numpy is absent. One is
the completed R17 section 4 review record for RV-H3-SOLVER-20260921-01, still
held outside reviews/records/ because tools/reviews_check.py requires the route
key to appear verbatim in registers/json/review_queue.json and the register
export in git remains the 2026-09-18 snapshot, which does not carry it. Filing it
today would fail CI, so it stays parked with its reason written down.

registers/CONSUMERS.json is regenerated because the salvaged README names
registers/json/review_queue.json in prose, which the consumer scan attributes to
that tab. One line, added by tools/consumers_check.py --write.

Verified: all eighteen repository checkers exit 0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Fij9F1cYrAUDW7PhWQze1G
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants