Round 2 verification audit for C038

Model: gpt-5

Claim: Fig. 3c: experimentally measured F(|Psi^->) agrees qualitatively with the correlated bit-flip model prediction 1 - 3.2 p, with p from independent measurements.

Source alignment: source/main.tex:156-157 (Fig. 3c)

Prior official verdict: not_verified with failure_reason data_unavailable.

Executable evidence: run.py. Sandbox rerun logs: run.log.

Independent audit: I checked the paper inventory and source context. This claim depends on measured/raw experimental data or device-specific calibration values, and inventory.json records no paper-provided dataset/code artifact. The correct action is not to synthesize data; the data_unavailable verdict is preserved.

Decision: Round 2 verdict is not_verified with failure_reason data_unavailable and limitations ['data_unavailable']. Notes: Fig 3c compares MEASURED F(|Psi->) against the 1-3.2p correlated bit-flip model. The model line is reproducible (coefficient 3.2 confirmed ~3.07 in C028), but the measured (p,F) data points are unavailable (inventory.json: no data/code link), so the claimed qualitative agreement cannot be verified. Lower of the two -> not_verified / data_unavailable.