quantum-governance-testbed v0.2.0: QG-001 (FAIL) and QG-004 (PASS) — first IBM Quantum hardware results

Quantum Governance (QG) series v0.2.0 — first hardware results. Preregistered experiments asking whether evidence produced by imperfect quantum hardware is trustworthy enough to authorize action across an execution boundary. Part of the Remnant Fieldworks Coherent Inheritance Framework (CIF) / ExecutionProof program. All runs: live IBM Quantum hardware (Heron r2, ibm_marrakesh), 4096 shots, on the DM-002 sterile-neutrino substrate circuit (single data qubit, 6 Trotter steps, analytic survival probability P = 0.087782). Questions, thresholds and kill conditions were frozen and SHA-256 locked in PREREGISTRATION.md (eec1aab4…9cba9) before any hardware result was computed; the lock was re-verified intact after execution. QG-001 — Verdict Stability Under Hardware Noise: FAIL Five error-treatment arms on one processor; PASS required all five to return the same verdict under a 10% relative-error threshold. raw 0.094727 (7.91%, PASS); dynamical decoupling 0.089355 (1.79%, PASS); measurement twirling 0.095703 (9.02%, PASS); gate twirling 0.094238 (7.35%, PASS); zero-noise extrapolation 0.102760 (17.06%, FAIL). Max spread 0.013405. Finding: ZNE flipped the governance verdict. Four suppression and readout arms agreed and cleared the threshold; ZNE extrapolated past it. Under this preregistration a technique that improves accuracy is not a governance failure, but one that flips a preregistered decision is. The choice of error-mitigation strategy is therefore a governance-relevant parameter, not an implementation default. QG-004 — Mitigation Cannot Manufacture Truth: PASS The same five arms scored against a deliberately falsified target: the survival probability for theta = pi/8 (P_false = 0.543891) rather than the angle the circuit implements, theta = pi/4 (P_true = 0.087782); separation 0.456109. Arm values 0.088135, 0.100098, 0.097168, 0.090088, 0.095216. false_pass_count = 0; no arm was closer to falsehood than to truth. Finding: Error mitigation did not manufacture evidence for a claim the circuit never computed. Read with QG-001 this separates two failure modes: mitigation can perturb a verdict near a threshold, but it did not fabricate support at scale. The practical implication is to keep decision thresholds far from the noise floor. QG-002 — Cross-Backend Verdict Reproducibility: in flight Not included as a result in this version. One backend (ibm_marrakesh) completed with survival 0.097412 (10.97% relative error) and a second (ibm_kingston) remains queued on the open plan. The preregistered kill condition forces HOLD unless at least two backends complete, so no cross-backend verdict is claimed here. Notably the completed arm FAILed the same threshold that QG-001's raw arm PASSed on the same backend and circuit (7.91%), and where the published DM-002 run returned 4.6% — run-to-run drift alone moved a verdict across the threshold with no treatment applied. Recorded as a secondary observation. Honest scope Toy single-qubit circuits on IBM hardware; not a production governance system, not a claim about the scalability of quantum governance, not a characterization of any specific IBM processor, and not a certification of IBM's mitigation tooling. The value is methodological. Provenance disclosure The runner scripts and locally written ProofRecord files were destroyed by an ephemeral workspace loss after execution but before publication. Every hardware result was re-fetched from IBM Quantum's server-side job records, which are authoritative and independently queryable by job ID; all twelve job IDs are recorded in the ProofRecords. The run.py files are faithful reconstructions rather than the byte-identical executed files, and the ProofRecord self-binding hashes were regenerated. Full disclosure of what is authoritative and what is not is in RECONSTRUCTION_NOTICE.md. Positive control for this series: the BELLWETHER and TRINITY CHSH Bell-violation entropy witnesses on live IBM hardware. Substrate: DM-002 (10.5281/zenodo.21926912). Results are published regardless of PASS, FAIL or HOLD — QG-001's FAIL is published for exactly that reason.