Paper 31C: Trial-Native Cross-Check of the CORR/ANTI Timing Discriminator Applying the Paper 31B decision bundle to the Delft event-ready Bell datasets
Paper 31C applies the Paper 31B CORR/ANTI timing discriminator to the Delft event-ready Bell datasets as an independent, trial-native cross-check. Because Delft is recorded as one row per trial, the analysis avoids cross-party clock alignment and coincidence-window multi-match ambiguity that can obscure setting-resolved timing tests in continuous timetag logs. We test whether coincidence timing structure depends on setting compatibility itself by comparing CORR(00+11) to ANTI(01+10). To strengthen the discriminator beyond half-maximum knee widths, we use a robust timing-shape metric: the quantile width Q90–Q10 of residual timing r = dt − τ*, where dt = tB − tA is the per-trial timing difference and τ* maximizes acceptance under a declared half-window W ∈ {3,5,7}. Uncertainty is quantified by 50,000 bootstrap resamples per W. Across Delft 2015 and both Delft 2016 detector configurations, the bootstrap confidence intervals for the CORR/ANTI width difference include zero and the sign is not stable across W. The decision bundle therefore yields a constrained null on Delft at the stated resolution: no decision-grade, direction-stable CORR/ANTI timing-shape separation is detected under robust quantile-width metrics. The paper provides audit-grade artifacts (curve CSVs, bootstrap KPI tables, and manifests) to support independent reproduction and future timing-stress experiments.
Paper 31C applies the Paper 31B CORR/ANTI timing discriminator to the Delft event-ready Bell datasets as an independent, trial-native cross-check. Because Delft is recorded as one row per trial, the analysis avoids cross-party clock alignment and coincidence-window multi-match ambiguity that can obscure setting-resolved timing tests in continuous timetag logs. We test whether coincidence timing structure depends on setting compatibility itself by comparing CORR(00+11) to ANTI(01+10). To strengthen the discriminator beyond half-maximum knee widths, we use a robust timing-shape metric: the quantile width Q90–Q10 of residual timing r = dt − τ*, where dt = tB − tA is the per-trial timing difference and τ* maximizes acceptance under a declared half-window W ∈ {3,5,7}. Uncertainty is quantified by 50,000 bootstrap resamples per W. Across Delft 2015 and both Delft 2016 detector configurations, the bootstrap confidence intervals for the CORR/ANTI width difference include zero and the sign is not stable across W. The decision bundle therefore yields a constrained null on Delft at the stated resolution: no decision-grade, direction-stable CORR/ANTI timing-shape separation is detected under robust quantile-width metrics. The paper provides audit-grade artifacts (curve CSVs, bootstrap KPI tables, and manifests) to support independent reproduction and future timing-stress experiments.
Paper 31B applies the Paper 31 activation protocol to a fast-switching, setting-explicit Bell-test dataset to test a discriminative corridor-gating claim: whether coincidence timing structure depends on setting compatibility itself, not only on detector identity or coincidence-window logic. Using NIST 2015 run3 raw server streams with explicit RNG-setting pulses and SYNC markers, we reconstruct per-trial settings, align Alice/Bob timebases via SYNC, and trialize the readout so each trial contributes at most one click per side (earliest click in a declared post-SYNC gate). We then compute setting-resolved coincidence curves C(τ∣A=a,B=b)C(\tau\mid A=a,B=b)C(τ∣A=a,B=b) for the four setting cells and the aggregates CORR(00+11) and ANTI(01+10), reporting audit-grade KPIs (τ\*\tau^\*τ\*, peak, τ50\tau_{50}τ50 bounds, knee width) under declared windows and scan ranges. On this dataset and scan domain, CORR and ANTI do not exhibit a decision-grade, policy-stable shape difference in timing-knee structure; the discriminator yields a constrained null at the stated resolution. The contribution is a governed, reproducible discriminator pipeline with concrete artifacts (curve CSVs, KPI table, manifest) that can be applied unchanged to other fast-switching datasets.
Public photon Bell-test records retain timing structure that is lost in aggregate statistics. This paper tests whether that structure survives quantitative response controls in three experimental archives. In the Weihs data, a rectangular contrast of four channel-pair timing centroids tests whether accepted signal delays can be written as a sum of local channel contributions. For longdist1, the contrast is 1.405 ns and 1.515 ns in disjoint chronological halves, with pointwise 95% block intervals [1.018, 1.791] ns and [1.193, 1.836] ns. Both survive the enumerated centroid scan under the specified 20-block inference at an 8-ns coincidence half-width and remain positive under clock, background, and gate checks. A flexible empirical timing model, however, transfers between chronological halves after clock-drift calibration. In the Christensen archive, the suppressed setting-pair channel is strongly enriched in coincidence tails, but a cross-fitted signal–accidental model predicts 714.6 of 730 counts beyond 300 ns; the residual is 15.4 counts with interval [−54.6, 85.3]. NIST timing-offset records exhibit a terminal-mask excess accompanied by an Alice-side singles-rate change and compatible with a separately calibrated local-setting mixture. The evidence separates reproducible structure, failure of an additive accepted-centroid model, and effects accommodated by explicit response controls. Finite coincidence selection and known switching-related populations prevent the centroid contrast alone from establishing intrinsic polarization–time coupling or a failure of quantum theory. Statistical scope. The analyses are exploratory. Coarser clustering weakens scan-adjusted significance: with ten blocks per half, the highlighted Weihs contrast has adjusted approximate p values of 0.087 and 0.014. The residual is specific to the accepted-centroid model and does not establish a discrepancy beyond all independently calibrated temporal-response models. Files and reproducibility. Bell_manuscript.pdf is the self-contained, 13-page preprint dated 30 October 2026. Bell_Reanalysis_Source.zip contains the editable LaTeX manuscript, numerical figure code, reconstruction and analysis scripts, derived records, source manifests and hashes, full comparison tables, calibration and sensitivity results, and reproduction instructions. The NIST analysis uses 60 prescribed samples of 500,000 trials, rather than complete runs. The full upstream Christensen archive and Weihs time-tag files are retrieved by the supplied fetchers and are not bundled. A clean internal rerun reproduced the checked numerical outputs exactly; external independent reproduction is not claimed. Publication status and assistance. This research preprint has not been peer reviewed. AI tools assisted programming, analysis, and manuscript preparation; responsibility for the work remains with the author. License scope. The manuscript and original figures are licensed under Creative Commons Attribution 4.0 International. Original analysis code is licensed under MIT. Third-party source data and derived excerpts retain applicable source terms and attribution; this deposit does not relicense upstream material.
Paper 32A audits finite timing-continuum structure in the public Weihs Alice/Bob archive and explains why the coincidence-counting rule matters. Using the bluesine series, the paper reconstructs timing profiles for coincidence half-windows of 3, 5, and 7 nanoseconds. The profiles show a central timing peak near zero offset, shoulders near plus or minus 10 nanoseconds, and a flat far-offset baseline. A central result is that one-to-one matching and all-pairs matching require different statistics. Under one-to-one matching, each event is used at most once, so window parity remains a meaningful summary. Under all-pairs matching, the central bin can grow combinatorially because many Alice events can pair with many Bob events inside the same window. In that case, parity at zero offset can become misleading. The paper therefore replaces parity-at-zero with a full-profile shape statistic: a baseline plus three fixed spikes at zero and plus or minus 10 nanoseconds. The paper also presents a declared calibration model on S2 using a rail-squared target density. Because the selector equation explicitly relaxes toward this target, convergence is treated as a contractivity and calibration-stability check, not as an independent derivation of the target from the Weihs data. This paper does not claim a complete Bell-test analysis or a direct physical identification of Weihs events with the S2 selector. Its contribution is narrower: it establishes timing-profile governance, kernel-dependent statistic selection, and a reproducible calibration scaffold for future setting-resolved and null-model tests.
Paper 32B analyzes the public Weihs Alice/Bob archive with a focus on separating timing diagnostics from setting-resolved Bell-test statistics. The bluesine data are used as a timing-acceptance lane, showing a finite timing continuum with a central peak near zero offset, shoulders near plus or minus 10 nanoseconds, and a flat far-offset baseline. As the coincidence half-window widens from 3 to 5 to 7 nanoseconds, total coincidences increase while aggregate parity decreases, showing the expected throughput-versus-contrast tradeoff. The paper keeps this timing analysis separate from Bell-sector claims because bluesine does not provide a recoverable settings timeline. For the CHSH benchmark, the paper uses the longdist32 Coincidence Evaluation report, which includes setting-resolved correlators. The reported value is S = 2.702 plus or minus 0.078, which is well above the classical bound of 2 and below the Tsirelson bound of about 2.828. Raw longdist32 timetag scans are used only as timing and ingestion sanity checks, not as an independent reconstruction of CHSH. The result is a reproducible audit framework that separates timing-continuum validation, report-based CHSH benchmarking, and future raw-event reconstruction. The paper does not claim a new Bell bound, a proof of Holosphere theory, or evidence for superdeterminism; it establishes a disciplined baseline for later setting-resolved and theory-specific tests.
The reported violations of Bell inequalities are routinely read as an experimental refutation of local realism. This preprint asks how firmly the record supports that reading. We catalogue the classical routes by which a strictly local apparatus can report |S| > 2 - from outcome-dependent loss and coincidence-window selection (demonstrated in hardware) to synchronisation, electronics-level click shifts and analysis-parameter selection - and show that below the Eberhard efficiency the exact quantum correlation curve certifies nothing, being reproduced exactly by a known local detection model. We then define the first family-level synchronisation audit (plateau membership, bias index, sub-bound fraction), validate it on synthetic data with known ground truth, and execute it under preregistered protocols on three public archives. The raw Innsbruck record passes: across the full longdist series, every standard-configuration violation is a property of the raw data, not of the calibration. On the Delft event-ready data, a 50,000-member heralding-acceptance family answers the standing postselection critique: the violation is generic, not an acceptance choice. On the NIST 2015 record, an outsider reconstruction of the certified pipeline recovers the Bell geometry and gate optima but stalls at undisclosed cleaning parameters - a documented limit of third-party reproducibility. The programme distils into seven cleanliness criteria (C1–C7) for any dataset and eight provenance requirements (M1–M8) for future experiments. All figures and numbers are reproducible from the supplementary scripts (deterministic, fixed seeds, minutes on a laptop); raw data are fetched from the official public archives.
We present an eliminative analysis of no-signaling equalities in the NIST loophole-free Bell test (Shalm et al., 2015).Standard analyses examine outcome correlations to compute CHSH S values. We instead examine whether one party'sdetection rate depends on the other party's setting choice. Under quantum mechanics, these marginal probabilities must beindependent of the remote setting. Using a six-gate streaming pipeline on 356 million event-level trials, we apply theexperiment's own timing calibration (phase filter: pk=90, radius=4) to select genuine entangled photon detections inPockels cell slots 5-7. In the phase-filtered data, Alice's marginal detection rate for setting a=2 shows a stationary deviation:z = 2.41, permutation p = 0.017, 95% bootstrap CI [1.44e-6, 1.31e-5] excluding zero. The combined chi-squared is 9.40 (4df), one tick below the p < 0.05 threshold of 9.49. The deviation is temporally stationary across 100 time windows spanningthe full experimental run, with near-zero autocorrelation and near-zero linear trend. Without phase filtering, all trialsproduce a null result (p = 0.40), consistent with noise dilution from background detector clicks. A second NIST run using adifferent random number generator (ClassicalRNG XOR) lacks detection-level data and cannot be phase-filtered,precluding direct replication with matching detection definitions.