[Closed: Withdrawn]

[gupt1156]

False data led to bad verification.

[ploniaaaaaaa]

Amazing wowowow

[Swapnil-jain]

I tried to reproduce the κ here and could not: the certified child_histogram doesn't seem to follow from the package's own components.

What I did: applied upstream's shared_profile reconstruction rule (rebuild child_histogram from the component histograms, with R == c + q − matched) to this PR's own row_complex.json and row_bit.json at head 1b1d2fc. The rebuilt histograms are exactly #152's. The child_histogram stored here differs from them only by deleted children:

	complex	bit
children removed vs rebuild	60,732 (ranks 1–15)	95,220 (ranks 1–19)
rebuilt mass vs stored mass	1,392,138 vs 1,029,138	1,522,318 vs 1,144,318
W·m − rebuilt mass at the stored W	−361,680	−376,600

The removed mass is exactly K·m (363,000 and 378,000), which is why W·m − mass still equals the deficit, but no hop children are added in its place. With the components as shipped, the moment at a = 0 is already above 1 (1.351 complex, 1.329 bit), so there is no positive saving at the stated W.

verify.py reads child_histogram, R and W directly from the JSON and only checks W·m − mass, so it would not catch this. plan.json lists role ids but no donor/recipient pairs, hop frames or old-value reads. Also, the "untouched non-source roles" (carried over from #152) are deferred roles that still have children and appear in the scalar identity, so they are not dead.

For calibration: an exact hand-off removes at most the donor's exit and the recipient's entrance child per stage copy and adds one hop child. Even allowing free hops, a ceiling on these components stays below the claim (about 6.79e-4 complex, 6.83e-4 bit). Exact birth-read reuse on these words (#150, and our own replays) gains roughly 1–2%.

Happy to be corrected if I've misread the reconstruction. If a concrete reuse plan exists, a rebuilt histogram from it would settle this.

[eumemic]

Two independent checks of this certificate, which shared no code, find that the claimed κ is not supported by a construction:

• research/paired-cube-p11-chained/verify.py loads plan.json but never uses it. It opens no word and checks no frames, ordering or replay of the aliased program. It only evaluates the moment on the supplied child histogram.
• The complex child histogram removes exactly 5,500 × 66 = 363,000 rank mass, all of it from rank-1–15 children. No handoff is charged, and the source, target and remaining histograms are Paired cubes at p = 11 with an h = 21 bit word: conditional κ = 5.108289e-4 #152's unchanged. At the claimed W that gives W·m − mass < 0, so the row is inconsistent. The bit row has the same issue: 95,220 children are dropped and the auxiliary histogram is unchanged.
• Under the reuse rules of Birth-read slot recycling: conditional kappa 1.1038332094346e-4 #124/Physical frames and scratch reuse on Cayley covers: conditional κ = 3.32205398986e-4 #131/Late compensation and joint frame refinement: conditional κ = 4.24830828577e-4 #143, a recipient must be a gauged deferred role. The p = 11 word has only 3,630 of those, fewer than the 5,500 recycled. All the listed roles start at frame 0, so none of them can receive a handoff, and 1,487 of them are side-root roles that stay live until the side readouts.
• The certificate uses ATOM = 1/2000 with AB = 7.466e-4, but Paired cubes and shared completed cores for conditional kappa = 4.609169e-4 #144's bit certificate requires ATOM > AB. finite_bridge is also given vertices_per_stage = 1320 and group_order_bits = 100 instead of the order of O(66,2).

🤖 Generated with Claude Code

[gupt1156]

Withdrawn.