Examine the calculation and the evidence supporting its claim.
MEASUREMENT WORK / OCTOBER 12–19, 2026
Test the calculation and the evidence around it.
A convincing output can still fail to support its claim. Our measurement work now includes a fixed-corpus numerical-assurance evaluation, reproduction of an idealized wave model, and recovery of a physical experiment's analysis software. These results answer different questions.
NUMERICAL ASSURANCE / OCTOBER 19
A good sample fit can hide a wrong result.
In the fixed synthetic corpus, 8/8 valid controls passed; 6/6 unsuitable or numerically unqualified cases and 13/13 altered receipts were held. One test fit its recorded samples closely while missing an unmeasured midpoint.
This qualifies those fixtures. New-case numerical reasoning, whole-domain error certification, and independent-host reproduction remain to be established. Refusing an unsupported input is recorded separately from detecting a bad calculation.
The archived baseline reproduced its frozen output. A separately written simulation cross-checked ten configurations within the same idealized model.
This establishes computational repeatability within that model. Physical preparation, retention, readout, and device noise remain unqualified; the broader cited source chain still has an unresolved appendix hash discrepancy.
A result becomes more useful when someone else can examine it.
Bring a research question, a reproduction proposal, or a specific claim you want to evaluate. We can discuss the appropriate evidence and review boundary.
