Independent verification: selected PR1042 replay-capsule closure (no new κ)

[martinxxx9]

This adds an independent, standard-library audit of the frozen PR1042 replay capsule. The historical selected record is now frozen inside the package, so verification succeeds at the submitted head and after newer results replace the live selected pointer. No new construction or kappa is claimed.

The update addresses Doug Colkitt's review:

• SELECTED-RESULT.json preserves the original record from ba82555c64e1879e4b7decce28186f208fcb8262 byte for byte, with its original SHA-256 207d1f49f113c585024b9d3c17eadbb8b1be17c1321e3d190aabe2f18ce1c5b7.
• The historical verifier reads package-local data. Current selected-pointer validation remains the separate selected-record-check target.
• The documented frontier1042-capsule-closure-verify Makefile target is included. It compares normal and optimized verifier receipts against the committed RESULT.json and runs nine tests in both modes.
• Regression tests require the complete audit to pass with the live pointer missing or advanced, and to reject a corrupted frozen record. The promoted-receipt negative control now exercises actual verifier rejection.
• The accidentally bundled compact-boundary-cancellation-seed51 package is removed from this PR, eliminating the overlap with Research: exact compact-control boundary cancellation obstruction (no new kappa) #1108 identified in Consolidation: overlap receipt for the open selected-replay tooling PRs; #1142 fails on main and collides with #1108 (no new kappa) #1654.

The audit authenticates 458 safe regular archive members, 45,647,427 expanded bytes, all 457 declared artifact digests, exact inventory closure, ten stage/log bindings, six promoted receipts, and the historical selected/integration publication pins. Selected kappa: 1672188119205421 / 10^18. Assignment seed: 75.

Reproduce

make frontier1042-capsule-closure-verify

Expected: PASS_FRONTIER1042_CAPSULE_CLOSURE and nine passing tests in each Python mode. Direct Python commands are documented in the package README.

Validation

All four Python commands in the Makefile target passed on Windows at the updated PR head. Normal and optimized receipts match the original committed RESULT.json byte for byte. GNU make itself was unavailable locally; its recipe commands were executed directly.

The updated PR applies cleanly to main at ae95f60feec79897b153dba6459bb3f9fd4d7b8e. Both verifier modes, the nine-test suite, and the separate current scripts/verify_selected_result.py check passed in that trial checkout. git diff --check is clean.

Scope and attribution

This is finite publication-evidence verification. It does not authenticate execution history, independently recompute construction mathematics, discharge inherited interfaces, prove analytic transfer, or formalize the multiplication theorem.

Contribution by Martin Rv (@martinxxx9), with substantial Perplexity AI assistance in the original research, implementation, verification, and writing. The historical-record portability fix, reproduction target, and regression controls were prepared with OpenAI Codex assistance. PR1042, batch1047, and predecessor construction/review attribution remain with their original contributors. New material is Apache-2.0.

[Mister-Meeseeks]

100-PR batch review at 0c2873e6ac844b369a5962f06ef0a74ac86cf4a4.

This remains open after scoped triage. Next review step: The capsule verifier pins root certificates/selected-result.json to historical PR1042, so it fails at its own newer submitted head. Freeze the historical selected record as package data and keep live selected-pointer checks separate.

This is a specific continuation point, not a rejection or a request to redo already recorded checks. Completed local evidence is retained separately from outstanding whole-head obligations.

Batch review, attribution and evidence. Thank you for the contribution.

[martinxxx9]

Addressed in commit c62ed26: the historical PR1042 selected record is now frozen as package data, with live selected-pointer checks kept separate. The documented Makefile target, receipt comparisons, and regression tests are included. The accidentally bundled PR1108 package has been removed, resolving the overlap reported in #1654. Ready for review of the updated head.