Research: four retained contacts obstruct staged R439 helper465 isolation

[adamgall]

Result and scope

This follows the explicit helper465-at-cut10 completion question in merged #1271, with a source-bound all-length obstruction, not another cut census or a new supplier claim.

At PR589 B cut10, every exact staged isolation block that restores helper465 and removes only its dirty column from the other rows must make physical target163's frame contain frame507. This holds for arbitrary block length, all rational common frames and any restored independent dirty auxiliaries. Retaining just four later contacts in order—B48(163,48), B117(48,5), B493(5,243), B601(243,111)—then propagates e0 into target111, whose required final frame525 is x0=0. No monotone reframing or extra interspersed scalar work repairs this contradiction.

The staged four-ADD boundary itself is legal. A fully emitted ghost-inlined continuation is scalar-exact on all661 arbitrary dirty columns and its inverse, at24304 versus24305 ADDs, but fails geometry. It is not called an improvement. An all-rational least-join relaxation first fails precisely at B601, rank8 versus cap7. A complete changed-endpoint positive control and the favorable fully counted two-slot budget are recorded separately from admitted construction costs.

Material limits: the theorem assumes the exact-isolation intermediate boundary and retention/order of those four contacts, physical role identities, no new COPY/ERASE and no frame descent/reset. A different scalar suffix, delayed isolation, changed contact/role routing or new priced transport can escape. This does not close #1271's unrestricted retirement question or claim kappa, native/hosted admission or a multiplication theorem.

Correlated COPY boundary

The actual pinned selected exporter was executed and all27234 original records replayed. It reuses temp661 for nine sequential COPY/scatter/ERASE lifetimes; the last ERASE is record22360, and B cut10 is record22380. No correlated temporary is live at the cut. COPY-LIFECYCLE.json lists every source/frame, lifetime, original-u465 coefficient and independent-cut-coordinate465 coefficient. Ordinary retained source roles remain in the661-coordinate state. Omitted-ERASE and COPY-before-ERASE controls reject.

Reproduction and exercised verification

uv run --python 3.12 --no-project python scripts/research_package.py --package research/r439-staged-retirement-boundary

Passed locally on CPython3.12.12/macOS arm64,3.61s observed wall time:

• PASS_R439_STAGED_RETIREMENT columns=661 contacts=4 first_failure=B601 scalar_ADDs=24305->24304 controls=13 no_new_kappa
• Independent set-valued scalar/integer-geometry replay, complete emitted-word digest and endpoint.
• Literal exporter lifecycle audit, nine closed copies and two rejection controls.
• Five regression groups, including source/receipt mutation, material chronology, late donor reordering and source-only relocated -I -O -S -B replay of all three engines.
• PASS_REGISTERED_RESEARCH_PACKAGE ... outcome=negative-result

No full repository suite, selected cold replay, native5/9/11-stage checker, hosted compiler, precision/tape admission, remote CI result or formal proof-assistant verification is claimed. VALIDATION.json records actual commands and resolved metadata/receipt-comparison preflight issues.

Provenance, novelty and credit

Only research/r439-staged-retirement-boundary/ is added; selected artifacts, upstream and all predecessor sources are unchanged. Base0b1cccf86e2f5132ed286a26ef0a681a495740ea. Independently sampled unrerolled seed494d0ed24f025eea9e623e001763c7c4, trackbit-mechanisms, emitted cardbit-helper-contract. The explicit scope change to terminal-aware H4 and complete selection JSON are retained.

Merged #1128/#1256 supply the general boundary/access methods; #1271 supplies the later-cut handoff; #1393 concerns helper468 and leaves staged early mechanisms open. This contribution is the arbitrary multi-frame helper465 isolation/four-contact continuation implication, not a new general information-access lemma. Final targeted all-state searches found #1271 as the helper465 predecessor, and exact heads of1128/1256/1271/1393 remained unchanged. Broad shared overlap coverage and its limitations are recorded.

Credit Ali Romman, Ross Richardson, Philipp/@mightyOil, Cinna Davis, Sagar Thakrar, DreamingOfClouds, eumemic, Dylan Crow's exporter and the retained compiler/hosting lineage. Prepared for Adam Gall with substantial OpenAI Codex/GPT-6 Astra assistance. Apache-2.0; no independent human proof review claimed.

Coordination claim: #1929 (comment)

[Mister-Meeseeks]

Batch review at 6d4587f91bc345325856d839ad752f5a202a6212.

Merged to main, preserving the original contributor commits and package attribution.

Reviewed scope: Exact661-column cut isolation forces dirty465 through four retained contacts into an incompatible final frame, excluding the stated monotone retirement route even with restored auxiliaries. Independent geometry, least-join propagation, four-ADD positive isolation and scalar ghost-word controls pass. Optimistic two-slot moment is a budget only; relaxing contacts, new COPY/descent or physical realization remains outside the exclusion.

The pinned scoped command passed; the batch ledger contains its command, input hashes, environment and log. This incorporates the research/tooling contribution and does not change the selected κ or discharge the inherited multiplication-theorem assumptions.

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