Description This record accompanies a manuscript that asks one question at extreme breadth: can a single compact, on-device token — one window of a signal reduced to a single ~6-9-bit class-discriminant codebook index — carry a decision across sensing modalities spanning roughly twenty orders of magnitude in physical scale, from nanometer-pore ionic current to gravitational-wave strain? Under strict pre-registration (frozen recipe, instance-disjoint splits, five seeds, paired-bootstrap intervals, and honest negatives reported verbatim) the same encoder is screened on seven real public tasks — nanopore RNA identity, three neural-probe read-outs (region, cell type, unit quality), a teleseismic transient, a distributed-acoustic-sensing (DAS) fiber phase arrival, and a (semi-synthetic, clearly labeled) gravitational-wave inspiral chirp — and returns GO on all seven. The result is then subjected to three adversarial self-audit rounds and a head-to-head architectural analysis, which force three explicit retractions and yield a corrected, defensible account: the token retains most of a decision at extreme compression (a modest, consistent tax of +0.05 to +0.08 AUC versus a strong nonlinear model, at roughly 200 microseconds and 21 kilobytes per window); it adds real discriminative value on shape/pattern tasks but reduces to a trivial detector on energy-dominated ones; and its behavior is governed by stream morphology — near-optimal on pulsatile and stationary signals, structurally weak on intermittent ones whose decision lives in cross-window timing a single token cannot see. Three claims are retracted under audit and reported plainly: an apparent "beats-the-ceiling" result was a weak-baseline artifact (a strong nonlinear ceiling restores the +0.05-0.08 tax); the tax-scaling "law" is not universal (it holds only within a modality's difficulty ladder); and a token trained on injected gravitational-wave signals does not transfer to real detected events. A tiered co-channel is shown to be a bandwidth device, not an accuracy device — it recovers tax only where the task is hard and routes no better by token uncertainty than at random, but delivers 8-61x bandwidth reduction at fixed event capture on continuous rare-event streams — and a learned trigger beats a trivial energy threshold only for shape-defined events. Lifecycle studies show an on-sensor codebook can self-maintain across many unsupervised refresh cycles (with periodic anchor refresh) and that spatial token-coincidence across an array suppresses false alarms. Honest boundaries are mapped verbatim: at-rest deep-brain medication state does not decode across patients (its uncompressed ceiling sits at chance — signal absence), cuffless blood-pressure category is largely subject-identity leakage, short-read nanopore falls to chance as the ceiling itself collapses, and label-shuffle controls collapse to chance (confirming the GOs are real signal). Method companions: Papers 19, 30, and 31; trigger-scoping companion: Paper 29. This is a cross-scale application and validation of previously-filed and previously-published methods. Keywords: class-discriminant codebook; vector quantization; on-device inference; edge AI; cross-scale sensing; nanopore sequencing; Neuropixels; distributed acoustic sensing; seismology; gravitational waves; pre-registration; honest negatives; selective co-channel; stream morphology; self-supervision References 1. R. J. Ferlic and K. K. Ferlic, "A single-token class-discriminant codebook encoder for physiological signals (Paper 19)," Zenodo, 10.5281/zenodo.20788187. 2. R. J. Ferlic and K. K. Ferlic, "On-device glucose alarms from a single learned token (Paper 30)," Zenodo, 10.5281/zenodo.21114273. 3. R. J. Ferlic and K. K. Ferlic, "One token, six modalities: pre-registered cross-modality screening for wearable and implantable monitoring (Paper 31)," Zenodo, 10.5281/zenodo.21136786. 4. M. Jain, H. E. Olsen, B. Paten, and M. Akeson, "The Oxford Nanopore MinION: delivery of nanopore sequencing to the genomics community," Genome Biology, vol. 17, art. 239, 2016. 5. J. J. Jun et al., "Fully integrated silicon probes for high-density recording of neural activity," Nature, vol. 551, pp. 232-236, 2017. 6. J. H. Siegle et al., "Survey of spiking in the mouse visual system reveals functional hierarchy," Nature, vol. 592, pp. 86-92, 2021 (Allen Institute Visual-Coding Neuropixels dataset; DANDI 000021). 7. M. V. Perez et al., "Large-scale assessment of a smartwatch to identify atrial fibrillation," New England Journal of Medicine, vol. 381, no. 20, pp. 1909-1917, 2019. 8. Z. Zhan, "Distributed acoustic sensing turns fiber-optic cables into sensitive seismic antennas," Seismological Research Letters, vol. 91, no. 1, pp. 1-15, 2020. 9. Incorporated Research Institutions for Seismology (IRIS) Data Services and U.S. Geological Survey (USGS), FDSN web services for event and waveform access, https://service.iris.edu, https://earthquake.usgs.gov (accessed 2026). 10. B. P. Abbott et al., "Observation of gravitational waves from a binary black hole merger," Physical Review Letters, vol. 116, art. 061102, 2016. 11. R. Abbott et al., "Open data from the first and second observing runs of Advanced LIGO and Advanced Virgo," SoftwareX, vol. 13, art. 100658, 2021 (Gravitational Wave Open Science Center). 12. A. L. Goldberger et al., "PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals," Circulation, vol. 101, no. 23, pp. e215-e220, 2000. 13. DeepSubDAS: submarine distributed-acoustic-sensing earthquake dataset with P/S phase labels, Zenodo, 10.5281/zenodo.16014744. 14. Denoised earthquake data of Cook Inlet DAS and phase picks, Zenodo, 10.5281/zenodo.11642688. 15. N. Tishby, F. C. Pereira, and W. Bialek, "The information bottleneck method," in Proc. 37th Allerton Conf. Communication, Control, and Computing, 1999, pp. 368-377. 16. R. M. Gray, "Vector quantization," IEEE ASSP Magazine, vol. 1, no. 2, pp. 4-29, 1984. 17. A. van den Oord, O. Vinyals, and K. Kavukcuoglu, "Neural discrete representation learning," in Advances in Neural Information Processing Systems (NeurIPS), 2017. 18. R. A. Fisher, "The use of multiple measurements in taxonomic problems," Annals of Eugenics, vol. 7, no. 2, pp. 179-188, 1936. 19. S. P. Lloyd, "Least squares quantization in PCM," IEEE Transactions on Information Theory, vol. 28, no. 2, pp. 129-137, 1982. 20. B. A. Nosek, C. R. Ebersole, A. C. DeHaven, and D. T. Mellor, "The preregistration revolution," Proc. National Academy of Sciences, vol. 115, no. 11, pp. 2600-2606, 2018. 21. Y. Benjamini and Y. Hochberg, "Controlling the false discovery rate: a practical and powerful approach to multiple testing," Journal of the Royal Statistical Society B, vol. 57, no. 1, pp. 289-300, 1995. 22. R. El-Yaniv and Y. Wiener, "On the foundations of noise-free selective classification," Journal of Machine Learning Research, vol. 11, pp. 1605-1641, 2010. 23. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, "Communication-efficient learning of deep networks from decentralized data," in Artificial Intelligence and Statistics (AISTATS), 2017, pp. 1273-1282. 24. B. Efron and R. J. Tibshirani, An Introduction to the Bootstrap. New York: Chapman DANDI 000409). 26. R. J. Ferlic and K. K. Ferlic, "The learning lives in the decision, not the trigger: a pre-registered scoping of a single-token class-discriminant codebook for on-device sensing (Paper 29)," Zenodo, 10.5281/zenodo.21148683. License This work is released under the Creative Commons Attribution 4.0 International License (CC BY 4.0). The methods are covered by previously filed U.S. provisional patent applications — the class-discriminant codebook and its federated-refresh, privacy, input-quality-gate, selective tiered-fidelity co-channel, and spatially-distributed array coincidence, coincidence-derived self-supervision, and morphology-adaptive fusion mechanisms (Nos. 64/095,354; 64/084,807; 64/084,817; 64/084,821; 64/097,102; 64/098,837; 64/104,348); patent rights are separate from the copyright license. This is a cross-scale application and validation of previously-filed and previously-published methods and introduces no new subject matter. Public datasets (DANDI 000021 and 000409; PhysioNet; IRIS/USGS FDSN seismic services; the Gravitational Wave Open Science Center; and the DeepSubDAS and Cook Inlet DAS Zenodo records) retain their own licenses and are not redistributed in the archive; the gravitational-wave screen uses clearly-labeled semi-synthetic injections. Licensing inquiries: randolphf@fieldstoneanalyticsllc.com. Randolph James Ferlic, M.D. · Kimberly Kate Ferlic · Fieldstone Analytics, LLC · DOI 10.5281/zenodo.21149392
Ferlic et al. (Fri,) studied this question.