[Clear1nsight]

Upstream publication and complete scope

Publishes fork PR #15, preserving exact verified head fe1ff585764663e0310c91b2d4a5dcf09af8e234. Publication comparison was refreshed against upstream f626db5c3a3d781d8f7695038776c621651f75e1 on 22 October 2026: 18 changed files, 6 contribution commits. The scientific package is isolated in its own directory, but the complete upstream diff also carries the shared historical-replay/source-role and navigation/CI repairs from its preparation base, plus the base documentation. Shared repair bytes overlap across fork #14–#18 and must be reconciled once; they are not independent scientific contributions. All five research heads have successful exact-head fork workflows. Opening this PR is not a new scientific replay, upstream CI pass, merge or theorem acceptance. Scientific source pins, attribution and scope limitations are retained.

---

Change and mathematical scope

Assignment A gives a bounded negative construction result on PR4470's productive-helper word. Among 128 eligible adjacent receiver pairs, only the explicitly selected first twelve were tested, all in phase B. Every selected three-shear replacement preserves the old paid histogram exactly in both orientations and adds one scalar operation. No improvement, new kappa or physical supplier admission is claimed.

For a += ca*x; b += cb*x, set s=-ca*cb and replace by a += s*b; b += cb*x; a -= s*b. The identity holds on arbitrary dirty inputs and for the inverse-transpose word. The checker constructs every feasible contact-dimension triple for this word and independently charges local and complete-word frame movement, including exits. All 408 helper endpoints, main boundaries, warps and centre charges remain fixed. The whole paid profile stays {1:2237,2:388,3:494,4:147,5:36,8:8}.

The separate ceiling submission fork PR18 extends the same twelve-window decision to all endpoint-equivalent words of at most four unit shears. Its stronger result is not a premise of this construction replay.

Task and overlap

Scientific base f9f85c1e4fa1e955ff39f10fca447e3e2a4e4adc, PR4470 head b9a13c1673e0e9697de410de16e8bbf3207a7363, priority revision 7. Seven sources are authenticated. TASK.json preserves the original card and declares the user-assigned change to construction. PR4121's earlier T360 census uses different sources and helper contracts. PR4504/4478 verify predecessors rather than this replacement word.

The final refresh found PR4535's higher conditional critical-root endpoint on the unchanged supplier and PR4534/4539's noncommuting-controller work. These do not invalidate the frozen local negative result; the next positive construction must compare against the then-current complete claim, not an obsolete endpoint. Canonical main remains unchanged. Overlap checks were targeted, not exhaustive pagination or proof of novelty. The historical baseline helper was not run.

The scientific preparation base is research/20261022-canonical-base. The complete upstream comparison also carries the shared CI repairs identified in the publication scope below.

Validation

The first full scientific replay 39723520394 passed at 99f5d3d855a7ad6204072ff021e3621d7c6fa72c on Ubuntu and CPython 3.12.15, standard library only. It includes all twelve exact source windows, both orientations, 21 controls and preservation of committed inputs. RESULT.json retains every emitted field.

Publication head is fe1ff585764663e0310c91b2d4a5dcf09af8e234; its complete rerun is 39726103789. Read that exact-head check for final publication status. Earlier author metadata spellings were rejected before science and corrected to the schema's negative-result; no mathematical predicate was removed.

The unchanged engine's in-memory integer rank keys differ from JSON's required string keys. The registered replay.py authenticates the same engine, freshly executes every original check and compares every field in canonical JSON form. Six additional controls reject changed counts, value types, array order, missing fields and extra fields. It never invokes the authoring --write path or rewrites evidence. See RECEIPT-COMPARISON.md.

python -B scripts/research_package.py --package research/receiver-window-transfer
python -I -O -B research/receiver-window-transfer/verify.py --self-test

No conclusion about the other 116 windows follows. New physical lowering, absolute phases, complete 632-input endpoint, precision/stock, compiler/outer transfer, selected cold replay and optional source-history preflight were not performed.

Credit

Credit zxv's PR4085 dirty-receiver identity and PR4359 productive-helper architecture, Cinna Davis's PR4470 constructor, Jacob Alan Sussman, the fresh-helper/missing-span lineage, Mihir Srivastava, Ryanson Jonathan, GamingPuzzled's earlier census and retained native/Clifford/geometry/outer notices. Prepared for Chris R / Clear1nsight with substantial OpenAI assistance, Apache-2.0. Separate workstreams are not independently launched agents, human review or formal-kernel acceptance.

Open next questions

Do not reseed the same twelve fixed windows. Name an explicit change to the scalar alphabet, adjacency, owned-field allowance or source/consumer boundary, derive its fully charged optimistic moment and test one structurally justified case before expansion. A different source window lies outside this result but still needs a reason to matter. Any later positive candidate needs physical and complete-consumer validation. The package and public handoff preserve these boundaries.