Answers are grounded in this paper and cite their sources.
Position + empirical paper (v3.3). Applies the Compression Boundary amendments (protocol v1.8, DOI 10.5281/zenodo.22925164): four-gate adjudication adds the held-out two-part code certification gate (blocks a constant-output candidate certified by CV-only review: 5955 B vs 5892 B target; null-calibrated, shuffled-world delta inside +-0.004); refusal taxonomy upgraded - DATA reads HOLDOUT < t_min = C/g (byte economics, predicted vs observed break-even within one grid step at four noise levels), NOISE anchored to the externally measured sigma0 (near-twin difference estimator; self-calibration INADMISSIBLE); new negative experiment 7.x: the compression-boundary pilot (delta-compression as a law-quality signal failed: sign inversion under raw serialization, R2-shadow under fixed-width, no boundary location; residual sign-structure does: 48-52% in-band vs 8-22% at edge); Related Work adds MDL/SR (Rissanen, Gruenwald, Bartlett et al.), error-bounded lossy compression (SZ/SZ3), NCD, applicability domain (OECD QSAR), difference-based variance estimation (Hall 1990; Munk 2004 d>=4 consistency limit) with anti-claim guards. Compliance matrix updated to protocol v1.8 (81 criteria, 54 verified).
No takes yet. Share an insight, caveat, or question.
Dolgov et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: