Nine evidence roles.
One unsigned tail proposal.
A versioned receipt graph binds sampled-monitor data, clock calibration, I/Q errors, source preparation and independent outside-model evidence to the conditional leakage calculation.
The measurement prerequisites are explicit.
| Data and timing | Exact sample bytes, ordered indices, timestamp bytes and reviewed metrology |
|---|---|
| Measurement path | Full I/Q path and residual extraction; charged 10/5/3/2 µV error shares |
| Independent source evidence | Fixed finite-mode preparation and outside-model spectral variation ≤10 µV |
| Transfer and computation | Same preparation, residual reference, run and epoch; pinned source hashes and exact arithmetic |
All roles bind the acquisition profile, channels, epoch, units, kernel and source-contract hash. Every receipt binds the evidence and witness bytes. A separate review process must control the trusted configuration and receipt allowlist.
Submitted timestamps alone are not calibration. A later monitor on the same device cannot certify a historical waveform without the required reviewed transfer.
Structural readiness stops before scientific acceptance.
The frozen design still gives the conditional tail ceiling 3627/4000000 V = 0.90675 mV. The checker replays exact digital energy/RMS and the rational sampled-Gram certificate.
Its output is an unsigned tail-bound proposal, not a v0.1 acquisition evidence packet or an approved receipt. The original twelve-role checker is unchanged. The other eleven acquisition requirements remain necessary, and no action-scale estimate follows.
Unsigned synthetic proposal ↓Prior conditional theorem →Tests reach the intended rejection gates.
Controls cover run/channel/source mismatch, stale calibration, timestamp deviations, I/Q undercharging, unsupported outside-model evidence, alias intrusion, source-code changes and transitive cycles. Semantic mutations rebind synthetic receipt hashes so a hash error does not hide the tested gate.
A changed low-RMS sample still invalidates a stale computation certificate. The synthetic reviewer labels are author controlled and confer no independent authority.
The unchanged historical checker still returns HOLD. No empirical monitor dataset or reviewed calibration has been supplied in the scoped record.
All 57 software controls ↓Checker source ↓Synthetic packet ↓Acquisition evidence contract →Next: independent witness methods.
Specify source completeness, the independent 10 µV outside-model TV witness and full-path calibration for review before measurement. These must address unknown aliases and the same residual transfer across both observation intervals. Physical Promotion remains 0.