Research: faithful S3 output transport saves 120 primitives (no new kappa)

[bbMisk]

Fixed children that permute a six-row workspace can replace repeated workspace restoration with physical operations on a faithful two-dimensional S3 output. The complete seed-58 dirty echo uses 404 ADD/NEG instructions instead of 524, including both source-copy passes, every child and inverse, and final restoration. The boundary costs 36 instead of 96.

Status: scoped recursive-interface research construction; no new kappa or selected-supplier adoption.

The contribution is entirely in research/s3-standard-output-transport/. It adds the integer intertwiner characterization, a chronological six-state recurrence, complete recursive words and a separate permutation/difference-coordinate oracle. A 294-inequality primal/dual certificate proves that 36 is the exact boundary minimum in the declared uniform macro interface; the emitted path attains it.

This extends Ross Richardson's PR #1287 cyclic output frames and Sagar Thakrar's PR #1227 sign frames. Concurrent Discussion #1381 studies a one-dimensional S3 sign-character quotient; this package instead preserves the full noncommuting group action in two output coordinates. General output transport and dirty-echo cancellation retain predecessor credit. No predecessor verifier is imported.

Reproduce:

python3.12 scripts/research_package.py --package research/s3-standard-output-transport

Expected: PASS_S3_STANDARD_OUTPUT boundary=36_vs_96 echo=404_vs_524 saved=120 controls=11 no_new_kappa, five passing package tests, and PASS_REGISTERED_RESEARCH_PACKAGE.

Validation includes 378,504 unaligned schedules plus 186,624 cut-policy schedules; 24 recursive cases and 816 forward/inverse basis columns; array lifts; 11 semantic controls; and source/certificate/macro corruption and read-only relocation. The registered Python 3.12.13 replay, optimized Python 3.14.6 replay, source pins and selected-record integrity checks pass. The patch applies to refreshed main e6f9716621f4e01fc190617934cdfaf072bd976f.

The fixed-child restriction matters: opening the toy matrix leaves permits cheaper direct evaluation. No actual selected supplier exposing this interface has been identified. The forward prefix row-L1 bound grows 65 to 114. This does not establish a runtime gain, a contracting input-size recurrence, fixed-precision/physical-tape admission or a full multiplication theorem. Hosted CI and independent human review are separate.

Research base: 410026dc5513b87db92ae98571d1d5984c69a983; OS assignment seed 58, recursive interfaces. Exact predecessor pins, overlap coverage and limitations are in PREDECESSORS.json and OVERLAP.json. Sprint claim.

Prepared by Hanson Sun (@bbMisk) with substantial OpenAI Codex assistance in derivation, both implementations, verification and prose. The separate implementations were written in the same AI-assisted session. New work is Apache-2.0, with predecessor attribution and reference bytes retained.

[Mister-Meeseeks]

Batch review at 13d780ea51248ddd51901cd546b9f59adc76eb43.

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

Reviewed scope: Faithful noncommuting S3 output transport has an exact six-state boundary recurrence and fully paid dirty echo; all finite schedules and literal maps pass. The 120-instruction saving uses a fixed indivisible-child interface and raises the prefix envelope; no actual supplier or contracting moment is supplied.

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.