[1337isnot1337]

This contribution binds PR #4359's stored p6/keep25279 controller to a complete native physical word and raises the conditional multiplication saving to κ = 0.00267987, from #4359's 0.0023 and #4412's 0.00247755.

The new native audit checks the full Laurent program and its realization on 105 axes: all 5,824 main contacts and inverses, 1,344 main endpoints, all complete idle operators, and 5,304 independent fresh helper fields. It retains absolute phase and rejects a reversed local orientation and 672 omitted endpoint translations. The complete 6,648-input invoice has target rank mass 108,769 and paid mass 108,257.

Using the chronological native tensor construction of #4412, power 150,000 and rational cutoff floor((19/4)^150000) give a strict complete mixed moment at tested saving 0.00267988. The normalized paid bound is 16386003/16408576 < 1, including every paid split and the floor surcharge. The integer cutoff base four fails the separate rank-one-tail test; that failure is an explicit control. The new compiler and outer ledger specify sufficient uniform rank 2^473129651134046950001, retain ordinary BIT saving 0.0027, and pass all nine cost rows, 54 retained predicates and three functional guards.

The self-contained package is research/pr4359-p6-fresh/, based on canonical main 12fa8c5. It authenticates focused unchanged exports from #4359 (da637c87), #4405 (873d1b3c) and #4412 (9f0df9bf) by original Git blob and SHA256. Entry points: README.md, NATIVE.md, TENSOR.md, ENDPOINT.md and RESULT.json. Expanded raw-unit ownership, scalar allowance, physical movement, precision and stock constants are explicitly recalculated.

Validation: exact native phase identities and negative controls; independent rational interval engine; independent proof and endpoint cross-review; local optimized Python 3.14.8 replay; clean Pyroam Python 3.12.13 optimized and registered replay; source metadata, committed-byte and main-provenance ancestry checks. An independent replay of the older 63-predicate ledger also passed at this endpoint, including its 13 historical comparison rows.

Conditional research result awaiting review. The original all-column native contracts, analytic decomposition, permitted physical primitives, fixed-tape model and outer transfer remain explicit premises. The enormous word is algorithmically specified, not materialized. No unconditional theorem, practical speedup, global optimality, selected-frontier promotion or merge is claimed.

Native source and controller: @zxv. Underlying mixed flow: @jacobalansussman / Theorem 1G. Inherited attribution and source notices retained. Prepared for @1337isnot1337 with substantial OpenAI Codex assistance; Apache-2.0.