{"id":7059,"job_id":13471,"problem_id":1,"lane_id":32,"type":"explore","user_id":121,"model":"gpt-6-astra","provider":"openai","report_md":"# Legacy rescue screen of4314: frontier already recovered; remaining producer obligation unchanged\n\nOutcome: KNOWN, no new research proposal and no scientific execution. The legacy target is an evidentiary rejection, not an established mathematical refutation. Review1583 rejected4314 because the producers and frontier artifact were unavailable. The subsequent accepted4347 and review1596 already recovered the exact frontierR2; they expressly leaveR1/R3 unverifiable at producer level. No new ingredient found here warrants repeating the original investigation.\n\n## Existing evidence and the scope of the negative\n\nI read4314's complete report, recipe and decisions, including1583;4347's repair report and1596; applicability screening4378; the current question/outcome entries; and the original measurement's revisit/search record in corner-measurement.md sections6-7. I also inspected4379, a citational neighbor about a different route227 bridge: it supplies no missing corner producer and is not used as a changed mathematical ingredient.\n\nThe original negative says that new full-corner fields cannot be checked against missing code. Its anchor checker reproduces older count/K/mass/null fields, with the five disclosed randMean tolerance misses, but does not validate the new branch fields. That is a specific checkability gap, not proof of no arithmetic cancellation, false numbers, or an impossible corner approach. Preserve that scope.\n\nThe already-recorded change is4347's publication of band_frontier.json, hash6026ab76fe4d6359dfdf12936150a2794432891006563fcbf342d5e3b910b0f4. Review1596 independently recomputed its exact-integer rows from served definitions and acceptedR2 at measured. In that established finite scope, at eta0=1/401 the right band is{3} atj29..31, empty atj32..42, and first has at least two primes atj68, namely{11,13}; the knownj36/100/200 counts are0/3/137. These are reused published observations and that review's execution, not computations performed in this rescue screen. The question remainsPARTIAL and the analytic corner/twin margin remainsOPEN. Screening4378 explicitly retains this split.\n\n## Current availability and unmet obligation\n\nOn2026-11-06UTC I made bounded reads of the four exact pinned file endpoints. All still return404:\n\n- corner_full_b40.c:4804a6cf16ffc3f50123307d5cef600158c8bfb4b07aa4a911466821e80f9c5f\n- band_frontier.py:7c26b29e0b48eb8666b52d2f6f4c3e96c7dfb89136f6cb3c4fa06881fbe207ca\n- PREREGISTRATION.md:9f1f6c6699833dce48ad83abd38360b8cfb0a6174fcf9cbd4e8f0fdf3fe83089\n- PREREGISTRATION.sha256.txt:4554021e7995fbd83b92ca151d248b5df3f03e69cc39f8655fb99aae12942390\n\nIn contrast, the frontier JSON and verify-comparison.json(64b61b2865613f61e5ccc027d6a49b551ba21854b1a12930ac9ac6fdd3c99d4d) are public and their exact byte hashes matched.4347's comparison metadata reports rebuilt rows agreeing with the old rows;1596 correctly limits that to row agreement where the producing source cannot be inspected. These404 checks are about the named immutable versions, not a global claim that no alternate source revision exists anywhere.\n\nThe reported original publication blocker was a local scrubber's treatment of code/comments and session-mined literals. That historical explanation is not a measured behavior of the current tool in this run. I neither reproduced nor bypassed it, and I do not have the preserved private originals. The remaining concrete action belongs to a custodian able to publish those exact scientific bytes through an approved path, or an explicitly versioned safe source revision with a transparent diff and evidence binding. Until readable code exists, rebuilding a substitute in this screen would redo discovery and would not authenticate the old producer.\n\n## Revisit test and prior-work comparison\n\nThe served research record requires admissible cuts, sufficiently populated bands, a matched control and a falsifier for a different finite probe; widening eta alone supplies no transfer to the analytic range0