This preregistration specifies the final “Test 3” replication run of the CNC fork experiment in the simplest certifiable setting: T=2 binary clocks on a 3-qubit register. The goal is to maximize the chance of a conclusive decision either way under a preregistered rule that only emits conclusive labels after certifying overlap and bounding nuisance effects (mismatch and readout invasiveness).
This preregistration specifies the final “Test 3” replication run of the CNC fork experiment in the simplest certifiable setting: T=2 binary clocks on a 3-qubit register. The goal is to maximize the chance of a conclusive decision either way under a preregistered rule that only emits conclusive labels after certifying overlap and bounding nuisance effects (mismatch and readout invasiveness).
This report records and interprets the second real-device execution of the CNC fork experiment in the certifiable T = 2 “binary-clock” design (runner v3), including a software-forensics correction: an obsolete analysis JSON incorrectly reported ν = 1.0 even though all 2-bit clock patterns are legal by construction in T = 2. It also specifies a final replication run (“Test 3”) intended to be maximally robust against the dominant remaining failure mode (near-threshold readout invasiveness on ibm_fez). A conclusive outcome of Test 3 is meaningful regardless of direction, because (as preregistered) the runner only emits a conclusive label after certifying overlap and bounding nuisance effects (mismatch and readout invasiveness). Run metadata (from locked runner output): backend: ibm_fez output created (UTC): 2026-01-10T19:03:16.074947+00:00 IBM Runtime job id: d5ha332gim5s73ahnueg physical layout (C1,C2,S): [1, 2, 3] protocol blocks: ε ∈ {0.40, 0.20} (primary block ε = 0.40) Qiskit versions (recorded by harness): qiskit 2.2.3, qiskit-ibm-runtime 0.43.1 Key certified outputs (primary block, ε = 0.40): D̄ = 0.381793 ν = 0.016503 (≤ ν_max = 0.020) ε_inv = 0.047647 (≤ ε_inv,max = 0.050) δ′ = 0.313209 decision (locked runner): reject hard CNC The Test 3 OSF preregistration is complete and time‑stamped, but remains embargoed until execution; the public link will be added here immediately after the embargo is lifted. Raw run artefacts and outputs to be archived as a separate Zenodo Dataset record by the hardware executor (Morgan Cameron). This report is a forensic note and replication/protocol document.
This report records and interprets the first real-device execution of the preregistered CNC T=3 two-clock fork experiment (runner v7) and specifies a certifiable follow-up protocol in the T=2 binary-clock setting. First, we note a certified overlap set I_cert where the clocks are demonstrably aligned well enough to talk about a common tick. Second, the runner certifies a lower bound delta' on the standard-QM disagreement that should be present in that overlap set. At that point, the question is not philosophical. It is a specific experimental question: does the observed disagreement survive after the non-invasive correction, or does it collapse toward zero? Run metadata (from locked runner output): run_id: cnc_t3_20260105T133840Z backend: ibm_fez UTC timestamp: 2026-01-05T13:40:11Z key outputs: D_bar = 0.213982, nu = 0.162114, epsilon_inv = 0.109279 decision: inconclusive (fails nu and epsilon_inv gates) Raw run artefacts and outputs are archived by Morgan Cameron on Zenodo: DOI: 10.5281/zenodo.18177156 This report is a forensic note and planning document.
This registration freezes the preregistration document and the exact reference implementation (code, configs, and SHA256 manifest) for a single scarce “heavy” run of the CNC fork experiment in the T=2 binary-clock setting. The goal is to maximize the chance of a conclusive decision either way while extracting high-value diagnostics from the same dataset. Execution: Planned to be performed by an external hardware operator with IBM Quantum access. At the time of registration, Morgan Cameron has offered to execute the run, subject to availability. If executed by another operator, the executor identity and run artefacts DOI will be recorded in the post run dataset deposit. The preregistered protocol, code, and decision rule are independent of executor identity. Protocol, analysis, and report: Dale James Raymond. The authoritative analysis plan is CNC_T2_Fork_Heavy_Prereg_v2.pdf. The exact runnable artefacts are: cnc_t2_qiskit_harness_v3.py cnc_t2_runner_v3.py cnc_t2_config_heavy_v3.json (intended run) cnc_t2_config_lean_v3.json (sanity/layout scout only) RUNBOOK_T2_FORK_HEAVY_V3.md Integrity: SHA256SUMS.txt contains hashes for all files in this registration.
START HERE (read order / canonical artefacts): 1) Front Door notes (falsifiers & theorem/lemma map): DOI 10.5281/zenodo.18075645 2) Fork paper (Hard CNC + certified overlap + T=3 fork): DOI 10.5281/zenodo.18075546 3) Gravity paper (PAW / BR‑PAW + anomaly witness): DOI 10.5281/zenodo.18075607 4) Runner bundle (prereg mechanics + 30‑sec demo): DOI 10.5281/zenodo.18075677 Integrity rule (PAW/BR‑PAW): A null PAW result (no observed loop defect) does NOT, by itself, upper‑bound the worst‑case anomaly norm ||A||∞. PAW is soundness‑only in this direction; null‑result upper bounds require explicitly justified restricted norms (e.g., BR‑PAW with a physically motivated support/ensemble/code restriction) as stated in preregistered assumptions and budgets. Reference implementation for the CNC T=3 locked fork preregistration and the PAW/BR‑PAW inference pipeline. What it does: - Accepts preregistered inputs (including either externally supplied ε upper bound or tomography sufficient statistics + shot counts). - Produces preregistered bounds in two modes: (i) Bernstein/empirical‑Bernstein tightener; (ii) confidence‑region + SDP option (with preregistered solver tolerances and deterministic fallback). Included: - CNC_T3_Prereg_v7_bundle.zip - CNC_Release_Manifest_sha256_v2.txt Quick demo: - The bundle contains a one‑screen DEMO.md and a demo_simulation.py that generates synthetic data and runs the pipeline. Expected qualitative outcomes: signal run should be conclusive and reject Hard CNC; the two “bad” inputs should be inconclusive.
We report five preregistered experiments on the IQM Garnet 20-qubit square-lattice processor, each with its decision rule committed before any hardware data existed and each reported regardless of outcome. First, the even-over-odd parity-sector leakage asymmetry previously measured on IBM Heron r2 replicates on Garnet at preregistered power (relative asymmetry +0.068, one-sided p = 1.1 × 10−5). Because the noiseless parity leakage of these number-conserving XY-Trotter circuits is exactly zero, this is a statement about device noise. Second, the asymmetry is not stable: over ten execution windows it varies beyond shot noise at fixed depth (Cochran Q = 19.01 on 9 degrees of freedom, p = 0.025), more strongly when windows are pooled by provider calibration set (p = 0.0083), and an isolated sign reversal at depth 10 does not replicate. Third, a confirmatory campaign that takes the calibration epoch as its independent unit — twelve epochs, 688,128 shots — finds the asymmetry larger at depth 8 than at depth 12 (mean contrast +0.0173, one-sided p = 0.026); the lower bound is +0.0030 and three of twelve epochs have the opposite sign. Fourth, a calibration-aware layout optimiser developed on heavy-hex devices fails its frozen transfer rule: it is better than automatic placement at chain lengths 8 and 16 and worse at 12, and a powered follow-up confirms the reversal. All preregistrations, calibration snapshots, job identifiers, raw counts and analysis scripts are public. All quantum-hardware results were obtained on IQM Garnet through IQM Resonance under an extended-access research grant. Preregistrations, raw counts and analysis code: https://github.com/anulum/scpn-quantum-control
This technical report packages a set of hardware component-witness experiments run on IBM superconducting quantum processors. The experiments test whether a hidden auxiliary phase can affect a later system-qubit measurement, whether that phase can be recovered from basis-resolved readout, and whether two auxiliary registers can produce a nonadditive response rather than a simple sum of their individual effects. The package includes the report draft, figures, raw hardware result JSONs, raw-count appendix, and scripts for regenerating the figures and count tables. The central result chain is: echo/phase response on real hardware, basis-resolved phase readout without density-matrix tomography, and a two-register Y-basis nonadditivity test where the first run remains a frozen FAIL and an independent higher-shot replication reaches STRONG_PASS. The longer-term motivation is to develop small, testable quantum hardware primitives for systems that can preserve, read out, and combine latent histories or unchosen alternatives, instead of reducing every intermediate state to a single classical output immediately. In that sense, these experiments are intended as early building blocks for future quantum-assisted sequence or language-model architectures. This work does not claim a completed quantum language model, natural-language processing on a quantum computer, consciousness, selfhood, classical impossibility, or quantum advantage. It reports only hardware component witnesses on specific IBM backends, edges, and calibration windows, with the possibility of richer classical explanations left open.