[unpaidmoon]

Change and mathematical scope

Negative budget result. No new kappa. Selected κ stays 0.001672660197076918.

Discussion CrocSwap#1166 comment 19032700, task 2, asks for one downstream consumer that keeps the PR CrocSwap#3943 encoded endpoint and does not pay an unaffordable source return. This package gives the consumer whose outputs are the corrected sums w_i on Y_i at H_i, with q and the 72 helpers left in place and x not restored.

That continuation still pays 192 rank-one transports, the same native increment as PR CrocSwap#3943's source-preserving return. The scalar bill falls from 960 unit additions plus 104 halves and 104 doubles to 192 unit additions and no halves. Those saved additions are not rank children. Inherited encoded mass 2,576 plus the 192 transports is mass 2,768. On the published nine-stage equal-profile screen the paid-mass lower bound is 64,008 against capacity 60,736, deficit 3,272. Stop. This is not a supplier and not a universal P104 obstruction.

The 3,642-event prefix is cited at PR CrocSwap#3943 head 2cba526a4ebe0ef6c3c0143498adcf67dca97941 and is not replayed.

Task and overlap

Base: bb4bccbf42844d2801487450727c2f63e10d1d0f (canonical CrocSwap/integer-mult-bounds main). Card p104-side-correction, scope refined. The card's cost-bound method is applied to this one consumer; the static flag census is not repeated.

Item taken: item 5 of the chronological open CrocSwap#1166 list below.

Ordered eligible list, one remaining next task per handoff, after dropping a task that a later DONE, an open PR, this account's round CrocSwap#4107–CrocSwap#4157, or a fork cursor/* branch already performs:

1. (7,18) / (7,20) functional census (comment 19021170, question 1). Question 2 is Prove codimension-one L/2 substitution soundness on double transversals. CrocSwap/integer-mult-bounds#4148.
2. Largest physical child rank on one support alphabet (comment 19021222).
3. Finish the length-22 splits still open after Strengthen the length-22 seven-label census with a union-support budget; exclude the (9,9) split (no new kappa) CrocSwap/integer-mult-bounds#3818 (comment 19022412, updated by 19026415).
4. One adjacent six-tape event after the Research: first F0 caller six-tape admission audit CrocSwap/integer-mult-bounds#3932 F0 bind (comment 19030049), distinct from Ledger intermediate F0 magnitudes on the L359 dirty ray. CrocSwap/integer-mult-bounds#4114.
5. This consumer (comment 19032700, task 2). Task 1 is Audit matched P104 source-return adapters and their 192-ADD difference CrocSwap/integer-mult-bounds#3956.

Nearest work, and why this is distinct: CrocSwap#3943 prices the source-preserving adapter (960 unit additions and a restored x). CrocSwap#4006 fuses late cleanup and still returns X = x on that variant. CrocSwap#4143 prices an (N-P)⊗M tensor consumer, a different object. The new statement is that publishing w on Y without restoring x does not remove the 192-transport native charge.

Validation

python3 scripts/research_package.py --package research/p104-encoded-sum-consumer

CPython 3.12.3, standard library. Result:

PASS_REGISTERED_RESEARCH_PACKAGE research/p104-encoded-sum-consumer outcome=negative-result

and

PASS_ENCODED_SUM_CONSUMER transports=192 unit_adds=192 halves=0 native_mass=2768 deficit_s9=3272 encoded_deficit_s9=1544 x_restored=0

The checker covers four integer patterns, four separating ports, a missing copy, a sign flip, omitted descents, and a control that treats the 768 saved unit additions as native rank. At three stages the linear gap is the published weaker bound 1,160 and the paid-mass deficit is 3,104; both stay positive. The incomparability of the 96 hyperplanes is the inherited PR CrocSwap#3943 hypothesis, not a new geometry proof. No native phase, signed-floor, owner, or prefix replay was run.

Credit

Chris R / Clear1nsight, PR CrocSwap#3943 and PR CrocSwap#3882, for the encoded exit, the profile, the progress credit and the source-preserving adapter. Rohan Arun / PR CrocSwap#1972, Chafik Boukhalfa / PR CrocSwap#1991 and Adam Gall / PR CrocSwap#1940 for the channel. unpaidmoon, with substantial Cursor Grok assistance. Apache-2.0. No independent human or formal-kernel review.

Open next questions

A consumer that reads w_N on X_N at H_n and does not move those banks onto Y avoids this floor. The inherited encoded mass 2,576 still misses the published nine-stage screen by 1,544 before any new charge. Price that interface only with a changed coverage credit or a named deletion of rank children.

Workflow feedback (optional)

None.