CLAIM: compensated terminal helper-star deferral (seed 4bfcbdf5)

[1337isnot1337]

CLAIM

Track / assignment: bit-mechanisms; accepted OS seed 4bfcbdf5361cc5b1d6133b209e7c731b; task bit-helper-contract; priority revision 3; exact base 0b1cccf86e2f5132ed286a26ef0a681a495740ea; selected-record SHA-256 10b9e340fd8e354a37a0c4813c90e5de1dd42044b5f6aca4a3fc5a42b685fa32.

Question: On the three pinned R439 D words with 221, 221, and 236 terminal helper-star restoration ADDs, can delaying a nonempty subset of at most eight restoration gates past the fixed A/scatter/B/C continuation, then paying a separate linear target-repair suffix, lower the complete ADD count?

Changed premise and charge: Unlike PR #1580, this scope permits explicit target compensation after the delayed restorations. Each delayed edge saves one existing XOR. Any target row changed by the resulting dirty-helper defect must be written by at least one suffix XOR; operand synthesis, frames, and any source/helper preparation are excluded from this optimistic lower bound, so they can only increase cost. The target is strict ADD reduction for subsets of size 1–8.

Method/domain: Recompute the exact raw helper-to-target map K for each source-pinned word. Use the star geometry (distinct leaf coordinates) and the support of each selected center column K[:,c] to obtain a target-row repair floor for every nonempty subset, then exhaustively confirm all subsets through size eight if the exact representation permits. Stop at these three words and this additive post-continuation repair model; do not generalize to conjugated frame changes or modified A/B/C programs.

Predecessors / overlap: PR #1580 studies identical relocation without repair and explicitly lists conjugated restoration plus target repair as an open extension; PR #1194 proves the fixed word’s semantic D contract. Current nearby bit claims include #1695’s five-helper basis cycle, #1700’s two-relation rational word, #1727’s mixed AGL2 targets, and #1738’s D8 commutator. This scope changes terminal deferral to a charged repair model and uses the R439 helper-star word instead of their constructions.

Expected evidence: a reproducible exact support/repair-floor receipt for all three words, a proof covering the bounded subset family, negative and positive synthetic controls, and a handoff stating whether the 1–8 mechanism can save ADDs. No κ gain is presumed.

Claim expires: 2026-10-14 10:28 UTC.