Research: bit E8 decoder witness and 9/17 pairing-cost floor

[navinor]

Research note / scoped decoder construction and lower bound. No new kappa or hosted supplier.

In the fixed bit label-decoder model on real E8, priced by rational pairing
rank under PR 446's inherited copy premise, nine orthogonal punctured root pairs
give an exact decoder of cost 54 on 102 ports. A written argument supported
by exact finite censuses bounds cst/v >= 9/17 for every puncture of at most
25 lines and any number of label terms. Equality occurs at 18 punctures; the
package gives explicit nine-term witnesses for every p=0..25.

Keeping the selected PR 589 reader basis fixed, one source dual is a root that
can be deleted to produce a P110 decoder with cost 62. A scope control changes
that basis and obtains two cost-6 terms but costs 68 overall. These are
algebraic decoders only. The fixed R439 inventory
passes the aggregate capacity screen at v=110, whereas v=102 has only 408
slots; no actual owner map, asymmetric copy, frame/restoration schedule or
complete changed-supplier invoice is built.

Only research/bit-orthogonal-pair-decoders/ is added. Exact base:
2921874911956718fcaaaf9396fe8267b3cfa9f0. Selected pointers, upstream/ and
all existing packages remain unchanged. Source pins, literal witness,
deterministic receipt, proof, controls and validation are included.

At submission upstream had advanced to
abdc6a7002abdbcd9e21bfc5e1a795784deacfdf; all seven pinned executable/data
inputs remain byte-identical and PR 942 remains selected. The research branch
keeps its exact recorded base rather than importing unrelated concurrent work.

One reproducible command, Python 3.10+ and standard library only:

python3 -I -B research/bit-orthogonal-pair-decoders/test_package.py

This runs the primary and a separate implementation, compares regenerated
receipt bytes, and tests corruption rejection. The primary refuses -O
because the unchanged pinned pricing engines use assertions. Local execution
on CPython 3.12.14 and 3.13.15 and selected-source integrity checks are recorded
in VALIDATION.md; no whole-repository suite or hosted CI pass is claimed.

PR 620 identified the orthogonal-pair mechanism; PR 757 and PR 761 certify the
fixed P111 floor; PR 595 named the unrealised 54/102 unit. This contribution adds
a literal exact decoder and a puncture-independent rational pairing-cost
bound, not priority for those predecessor observations. PR 968 is the distinct
complex frame-pair analogue, with a different cost. Recent PR 1020 uses complex
F2 label ranks at 26 punctures and PR 1008 uses fixed-P111 tensor shears; neither
duplicates this scope. The immediate refresh through PR 1081 also found PR 1077's
complex-coefficient extension of PR 968, another distinct eight-term model.
Credit and exact source heads are retained in
NOTICE.md and SOURCES.json.

Inherited parabolic coverage, copy interpretation, compiler, hosting,
precision/tape and analytic obligations remain explicit. The written argument
and finite replay are not proof-assistant verification of multiplication or
independent human review. Prepared for @navinor with substantial
AI assistance through OpenCode; resumed review/packaging used GPT-6.1 Sol
(OpenAI). Apache-2.0. Structural-decoder track, OS assignment seed 40; no pinned
coordination issue existed, so the claim is local.

[Mister-Meeseeks]

100-PR batch review at ac8bd04538e692efa9c9b16bd79de5a45edcc9a5.

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

Reviewed scope: Exact 102-port orthogonal-pair decoder and conditional 9/17 label-decoder floor through 25 punctures; physical word/owner map remain open.

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.