EXP-065NEGATIVE
Canonical GPU engineering / SHA-256d

Nonce-prefix regional persistence

Pilot B8 density within a high-byte nonce region predicts density in a disjoint low-24-bit holdout from the same region.

Claim statusNO END-TO-END MINING ADVANTAGE DEMONSTRATED

Diagnostic work, exact algebra and local capabilities are not treated as end-to-end mining advantage.

Date2026-08-14
HardwareNVIDIA GeForce RTX 5080 · exact SHA-256d verifier
Run scopePreregistered campaign with sealed result and audit
ReproducibilityUse frozen regions, disjoint windows, equal-work controls and report every header, including reversals.
01 / Question & hypothesis

What was tested?

Pilot B8 density within a high-byte nonce region predicts density in a disjoint low-24-bit holdout from the same region.

02 / Scientific basis

Why the test is meaningful

Regional persistence requires out-of-sample correlation across many nonce prefixes, not an isolated positive header.

regions=256 high-byte prefixes/headerselect top 32 by pilot B8transfer measured on disjoint low24 holdout
Nonce-prefix regional persistenceVisual reading of the published metrics and gates for EXP-065; it summarizes the registered result, not a mining advantage.EXP-065 / NONCE-PREFIX REGIONAL PERSISTENCEAGGREGATE B8 LIFT0.9963CORRELATION−0.00738PANELS PASSING0 / 4
FIGURE / RESULT READINGVisual reading of the published metrics and gates for EXP-065; it summarizes the registered result, not a mining advantage.
03 / Method

How it was tested

Test four independent headers, 2^10 pilot and 2^13 holdout nonces per region, plus 64 equal-work random selections.

04 / Observed result

What happened

Aggregate B8 lift0.9963
Correlation−0.00738
Panels passing0 / 4

One panel reached 1.0731× B8, while the others were 1.0032×, 0.9272× and 0.9833×. Aggregate B8 lift was 0.9963 and net gain 0.9982; correlation was −0.00738 with permutation p=0.59395.

05 / Validation

Exactness and statistical controls

A malformed first manifest was archived before data; only the missing checksum block was repaired and the complete campaign repeated with 9,437,184 hashes and zero audit discrepancies.

06 / Interpretation

What the result means

No transferable persistence exists for the registered high-byte regions; the isolated positive panel is treated as noise.

Limitations

  • The region definition is fixed to the nonce high byte.
  • Other partitions require fresh hypotheses.
  • No operational selection is authorized.
07 / Reproduction

Evidence trail

Use frozen regions, disjoint windows, equal-work controls and report every header, including reversals.

Canonical variants

CANONICAL-EXP-141PREREGISTERED-CAMPAIGNSEALED-AUDIT

Source: internally audited canonical reports. Local filesystem structure, private headers and operational identifiers are excluded from publication.