R60 nested-reset cost audit
A nested decision tree of exact reset certificates can evaluate the R60 target predicate with less diagnostic work than the exact ripple baseline.
Diagnostic work, exact algebra and local capabilities are not treated as end-to-end mining advantage.
What was tested?
A nested decision tree of exact reset certificates can evaluate the R60 target predicate with less diagnostic work than the exact ripple baseline.
Why the test is meaningful
Structural reuse only becomes an advantage if certificate discovery, branching and fallback are counted. The fair comparison measures identical exact decisions under one work model and includes every reset certificate evaluated.
saving = 1 − E[W_reset]/E[W_ripple]H₀:E[W_reset]≥E[W_ripple]exactness: decision_reset = decision_ripple = full suffixHow it was tested
Evaluate registered nested reset trees on 3,000 cases per configuration, count certificate and suffix work, compare with exact ripple evaluation and cross-check every REJECT/CONTINUE decision against the full suffix oracle.
What happened
Every exactness gate passed with zero tree, certificate, engine or suffix errors. The best observed reset tree averaged 43.735 work units versus 41.687 for ripple: ratio 1.0491 and saving −4.91%. It rejected 2,984 of 3,000 cases and continued 16, yet still cost more.
Exactness and statistical controls
Multiple registered configurations showed the same approximately 4.9–5.0% penalty. Reporting the best observed configuration is conservative for the hypothesis; even that configuration failed the cost gate.
What the result means
The nested certificates are exact but operationally inferior under the frozen work model. V44 is a clear negative result and blocks this reset-tree implementation from being called an advantage.
Limitations
- Work units are diagnostic, not measured ASIC joules.
- A different implementation could have different constants.
- The closure applies to the registered tree and cost model.
Evidence trail
Freeze the tree, count every evaluated primitive, compare identical cases with ripple and the full suffix, and reproduce both zero-error decisions and the negative saving.
Canonical variants
SUBENGINE-V44ASUBENGINE-V44BSUBENGINE-V44CSource: internally audited canonical reports. Local filesystem structure, private headers and operational identifiers are excluded from publication.