[
 {
  "id": "P001-01",
  "dep": 1,
  "claim": "Classifier model collapse is defined as the progressive narrowing of acceptable scholarly expression through self-referential moderation training.",
  "kind": "definition",
  "modality": "self-description (coined)",
  "locus": "§6. Classifier Model Collapse and the Material Contraction of Scholarship",
  "quote": "This paper terms it **classifier model collapse**: the progressive narrowing of acceptable scholarly expression through self-referential moderation training, where each enforcement decision biases the classifier toward excluding similar content in subsequent cycles.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-02",
  "dep": 1,
  "claim": "Source concept: in language models, model collapse is training on one's own outputs, which contracts the distribution and eliminates rare forms.",
  "kind": "genealogy",
  "modality": "attributed (Shumailov)",
  "locus": "§6",
  "quote": "In the study of language models, this phenomenon is known as *model collapse*: when a model trains on its own outputs, the distribution of generated text contracts, rare forms are eliminated, and the model converges on a diminishing subset of its original capacity (Shumailov et al., 2023).",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-03",
  "dep": 1,
  "claim": "The removed archive had already documented model collapse across generative systems, in the \"Model Collapse Triptych\".",
  "kind": "attribution",
  "modality": "self-description",
  "locus": "§6",
  "quote": "The Model Collapse Triptych — three papers within the removed archive — documented this phenomenon across generative systems.",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-04",
  "dep": 1,
  "claim": "Evidential basis: Zenodo publicly says removed spam and accounts train its automatic classifier and can trigger review of related accounts.",
  "kind": "fact",
  "modality": "documented [Observed]",
  "locus": "§6",
  "quote": "The removed spam and account is further used to train and improve our automatic classification system. The blocking of the account may also further spawn an automatic review of similar and related user accounts.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-05",
  "dep": 1,
  "claim": "Mechanism: that disclosure describes a feedback loop in which each removal enters the training set and biases the next decision, narrowing the center of \"legitimate\".",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§6",
  "quote": "This disclosure describes a feedback loop. When content is classified as spam and removed, it enters the training set for the classifier that will evaluate future content. Each enforcement decision biases the next. The distributional center of \\\"legitimate\\\" narrows with every cycle.",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-06",
  "dep": 1,
  "claim": "Generative collapse narrows producible text. Classifier collapse narrows permissible text, and at a repository it contracts the form of scholarship.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§6",
  "quote": "In a generative system, model collapse narrows the range of producible text. In a moderation system, classifier model collapse narrows the range of *permissible* text. The consequence is not merely operational.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-07",
  "dep": 1,
  "claim": "The classifier scores distance from a learned center, not methods or rigor, so it will exclude novel modes such as AI-assisted scholarship that the distribution has not yet absorbed: \"distributional conservatism\".",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§6",
  "quote": "A classifier trained before this mode existed — or trained on enforcement decisions that treated AI-assisted text as spam — will systematically exclude it. Not because the content is illegitimate, but because the distribution has not yet absorbed it. This is not moderation. It is distributional conservatism encoded in infrastructure.",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-08",
  "dep": 1,
  "claim": "Classifier model collapse is not identical to strict generative collapse. It is a moderation-specific feedback contraction, closer to performative prediction and feedback-loop work.",
  "kind": "genealogy",
  "modality": "self-description",
  "locus": "§6",
  "quote": "Classifier model collapse, as defined here, is not identical to generative model collapse in the strict technical sense (Shumailov et al., 2024). It names a moderation-specific feedback contraction:",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-09",
  "dep": 1,
  "claim": "Status: this is a testable hypothesis, not a diagnosis of Zenodo's model. Whether this archive entered the training pipeline is unknown.",
  "kind": "interpretation",
  "modality": "hypothesis ([Observed] disclosure, [Unknown] pipeline entry)",
  "locus": "§6",
  "quote": "This is not yet a demonstrated diagnosis of Zenodo's current model state. It is a testable failure-mode hypothesis grounded in Zenodo's disclosed use of prior moderation labels and removed spam accounts to improve subsequent classification.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-10",
  "dep": 1,
  "claim": "The archive that theorized model collapse may itself have been removed by a collapsing classifier. The Pristine Fallacy is the ideology and collapse is the mechanism, and they reinforce each other.",
  "kind": "interpretation",
  "modality": "hypothesis [Inferred]",
  "locus": "§6; repeated in §16 Conclusion",
  "quote": "The archive that theorized model collapse was removed by a system that may be undergoing classifier model collapse [Inferred].",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-11",
  "dep": 1,
  "claim": "Conditional prediction: if the archive went through the spam pathway, future independent researchers who write with similar density, cross-referencing or AI-assisted rigor would be marginally more likely to be flagged.",
  "kind": "prediction",
  "modality": "hypothesis (conditional; undated)",
  "locus": "§6",
  "quote": "Every future independent researcher who writes with the density, the cross-referencing, or the AI-assisted rigor that characterizes this archive would be marginally more likely to be flagged — not because their work lacks a research basis, but because the classifier would have been taught that work with this shape is illegitimate.",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "P001-12",
  "dep": 1,
  "claim": "Falsification: the hypothesis is weakened if Zenodo shows its classifier is not trained on its own enforcement decisions, or that training includes expert review, distributional monitoring and anti-narrowing safeguards.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§16, Falsification and Revision Conditions, item 2",
  "quote": "**Classifier model collapse.** The hypothesis is weakened if Zenodo demonstrates that its classification system is not trained on its own enforcement decisions, or that the training process includes domain-expert review, distributional monitoring, and safeguards against progressive narrowing of acceptable scholarly expression.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-19"
 },
 {
  "id": "C46-01",
  "dep": 46,
  "claim": "Gloss of model collapse (conventional).",
  "kind": "definition",
  "modality": "self-description (gloss of an existing term)",
  "locus": "Appendix A, Diagnostic-Structural Families, \"Model collapse\"",
  "quote": "The degradation of AI model quality when trained recursively on synthetic data.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-03-30"
 },
 {
  "id": "C46-02",
  "dep": 46,
  "claim": "Gloss of diversity collapse as output-variety reduction driven by synthetic data converging on priors.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "Appendix A, Diagnostic-Structural Families, \"Diversity collapse\"",
  "quote": "The reduction of output variety in AI systems as synthetic training data converges on statistical priors.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-03-30"
 },
 {
  "id": "C46-03",
  "dep": 46,
  "claim": "Variant family for diversity collapse, including a monoculture forensic.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "same entry",
  "quote": "Family: output homogenization, variance extinction, creative narrowing. Forensic: *stochastic monoculture onset*.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-03-30"
 },
 {
  "id": "C46-04",
  "dep": 46,
  "claim": "Forensic canary for model collapse.",
  "kind": "function",
  "modality": "self-description",
  "locus": "Appendix A, \"Model collapse\"",
  "quote": "Forensic: *autophagic parametric entropy*.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-03-30"
 },
 {
  "id": "C46-05",
  "dep": 46,
  "claim": "The mint claims mapped territory, not term ownership.",
  "kind": "attribution",
  "modality": "self-description",
  "locus": "Governing Claim ¶2",
  "quote": "The claim is not ownership. The claim is cartography.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-03-30"
 },
 {
  "id": "C46-06",
  "dep": 46,
  "claim": "Priority test the mint applies to later instantiations.",
  "kind": "function",
  "modality": "self-description",
  "locus": "§VII Step 3",
  "quote": "Was the mint family published before the term's first documented appearance elsewhere? If yes, the mint has temporal priority for the coordinate.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-03-30"
 },
 {
  "id": "A161-01",
  "dep": 161,
  "claim": "The model-collapse literature rests on a synthetic=contamination / human=refresh binary, and recent evidence makes that binary unstable.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "Abstract",
  "quote": "It operates on a clean binary: synthetic data is the contamination; human data is the refresh. Three recent lines of work suggest the binary is unstable.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-02",
  "dep": 161,
  "claim": "Training on AI-mediated human text, including unaided text from habituated writers, produces model-collapse signatures comparable to synthetic data's, though plausibly slower.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "that training on AI-mediated human text — including unaided text from cognitively-habituated writers — produces model-collapse signatures comparable to, though plausibly slower than, purely synthetic training data",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-03",
  "dep": 161,
  "claim": "The effect sits in the tails, not the means, and tails are the diagnostic instrument.",
  "kind": "function (measurement)",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "The central methodological claim is that the effect is not in the means but in the tails, and that tails are the diagnostic instrument for rate under any homogenization regime.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-04",
  "dep": 161,
  "claim": "Mechanism: reduced tail variance in training data reduces exposure to low-prior productions and compounds across generations. It does not require mediated text to be identical to synthetic text.",
  "kind": "function (mechanism)",
  "modality": "hypothesis",
  "locus": "§3 The Mediation Hypothesis (H3)",
  "quote": "The collapse mechanism is not that mediated text is identical to synthetic text; it is that reduced tail variance in training data reduces the model's exposure to low-prior productions, which compounds across generations under standard training dynamics",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-05",
  "dep": 161,
  "claim": "Collapse signatures are measured on tail-focused axes.",
  "kind": "function (measurement)",
  "modality": "hypothesis",
  "locus": "§3 (H3)",
  "quote": "The collapse signatures manifest on tail-focused collapse axes: kurtosis of generation distributions, rare-token retention, long-tail factual recall, semantic dispersion at the 90th percentile, resistance to high-prior substitution.",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-06",
  "dep": 161,
  "claim": "Falsifier: if the H0, H1 and H2 corpora perform alike and all clearly better than synthetic, the cascade fails and standard filtering is enough.",
  "kind": "falsification",
  "modality": "hypothesis",
  "locus": "§6.3 Stage 3 — Training Cascade",
  "quote": "H3 falsified: H0 ≈ H1 ≈ H2, all substantially better than S.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-07",
  "dep": 161,
  "claim": "Stated limit: mediation may produce collapse only at concentrations real corpora never reach, so H3 could fail in practice even if H1 and H2 hold.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§10 Limitations, (f)",
  "quote": "It is possible that mediation signatures, while present in human text, produce collapse only at very high concentrations that real-world training corpora do not approach. In this case, H3 fails for practical reasons even if H1 and H2 hold.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-08",
  "dep": 161,
  "claim": "Accumulation mitigation (Gerstgrasser et al.) assumes a stable human baseline. If the baseline drifts, its guarantees must be re-derived.",
  "kind": "prediction (corrective)",
  "modality": "hypothesis",
  "locus": "§8.1 Training-Data Curation Changes",
  "quote": "Gerstgrasser et al.'s accumulation result depends on the assumption that the \"fixed human baseline\" remains stable across generations.\" / \"The mitigation guarantees of accumulation must be re-derived under conditions of a drifting baseline.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-09",
  "dep": 161,
  "claim": "Corrective: a continuous Mediation Index replaces the synthetic/human binary in curation, and the human-data refresh gets a calculable half-life.",
  "kind": "function (corrective)",
  "modality": "hypothesis",
  "locus": "§8.1",
  "quote": "The synthetic-versus-human binary is replaced by a continuous Mediation Index score.\" / \"The half-life of the human-data refresh becomes calculable.",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "A161-10",
  "dep": 161,
  "claim": "Scope limit (v1.2 reframing): the claim is about rate, not kind. The control is a rate-baseline, not an \"unmediated\" purity-baseline.",
  "kind": "interpretation (limit)",
  "modality": "self-description",
  "locus": "§4.4 The Reframed Empirical Question",
  "quote": "Under the rate framing, the cascade in Stage 3 is not a claim that mediated text is functionally identical to synthetic.\" / \"The H0 condition is a rate-baseline*, not a purity-baseline*.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-07"
 },
 {
  "id": "C191-01",
  "dep": 191,
  "claim": "Recursive synthetic training causes irreversible loss of distribution tails.",
  "kind": "fact",
  "modality": "attributed (Shumailov et al. 2024, *Nature* 631:755–759)",
  "locus": "§2 ¶2",
  "quote": "recursive training on synthetic data causes irreversible defects in which the tails of the original content distribution disappear",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-02",
  "dep": 191,
  "claim": "Early collapse loses minority data while aggregate performance can appear to improve.",
  "kind": "fact",
  "modality": "attributed (as \"now standard\")",
  "locus": "§2 ¶2",
  "quote": "this stage is insidious precisely because aggregate performance can appear to improve while performance on minority data silently degrades.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-03",
  "dep": 191,
  "claim": "The tails consist of rare words, constructions, low-resource languages and dialects.",
  "kind": "fact",
  "modality": "attributed (Briesch et al. 2023; arXiv:2507.03933; arXiv:2602.16201)",
  "locus": "§2 ¶3",
  "quote": "rare words, uncommon syntactic constructions, low-resource languages, non-standard dialects, and culturally specific variation are named repeatedly as the first features to disappear under recursive training",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-04",
  "dep": 191,
  "claim": "Tail sparsity is partly a tokenization artifact.",
  "kind": "function",
  "modality": "attributed (arXiv:2602.16201)",
  "locus": "§2 ¶3",
  "quote": "linguistic sparsity in the tail is *partly driven by tokenization artifacts*",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-05",
  "dep": 191,
  "claim": "A high-perplexity reject filter is an input-layer tail-pruning instrument.",
  "kind": "interpretation",
  "modality": "interpretation (self-typed *Structural*)",
  "locus": "§3",
  "quote": "A high-perplexity-content detector that rejects on positive detection is, viewed from the model-collapse literature, an automated tail-pruning instrument applied at the input layer.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-06",
  "dep": 191,
  "claim": "Inference-layer rejection becomes training-layer exclusion through logs.",
  "kind": "function",
  "modality": "hypothesis (self-typed *Model proposition*)",
  "locus": "§4, link 1",
  "quote": "A pre-inference filter that rejects low-resource-language queries does not merely refuse service in the moment; it systematically excludes those languages from the interaction record that becomes future training signal.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-07",
  "dep": 191,
  "claim": "The filter is a mediation-ratchet term with the wrong sign.",
  "kind": "function",
  "modality": "hypothesis (cites Diversity Contraction, doi:10.5281/zenodo.20518338)",
  "locus": "§4, link 3",
  "quote": "Adopted widely, it is a ratchet term with the wrong sign.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-08",
  "dep": 191,
  "claim": "Tail-loss harm falls disproportionately on marginalized groups.",
  "kind": "fact",
  "modality": "attributed (arXiv:2503.03150)",
  "locus": "§5",
  "quote": "can disproportionately affect marginalized groups or historically disadvantaged communities",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-09",
  "dep": 191,
  "claim": "IBM: long-tail ideas fade from public consciousness.",
  "kind": "fact",
  "modality": "attributed (IBM synthesis of the *Nature* result)",
  "locus": "§5",
  "quote": "long-tail ideas might eventually fade out of the public's consciousness, limiting the scope of human knowledge.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-10",
  "dep": 191,
  "claim": "Self-limitation: the collapse finding concerns training, the gating concerns inference, so the strongest claims are marked as model propositions.",
  "kind": "interpretation",
  "modality": "self-description",
  "locus": "§4 ¶1",
  "quote": "The objection is correct as stated and is the reason this document's strongest claims are marked *Model proposition* rather than *Established*.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C191-11",
  "dep": 191,
  "claim": "The reviewed paper is a specimen of collapse dynamics at the research layer.",
  "kind": "interpretation",
  "modality": "hypothesis (self-typed *Model proposition*)",
  "locus": "§7",
  "quote": "This is the collapse dynamic operating one layer above where the collapse literature usually finds it.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-11"
 },
 {
  "id": "C199-01",
  "dep": 199,
  "claim": "Three literatures describe one phenomenon, and solution-space diversity is the unmeasured variable connecting them.",
  "kind": "interpretation",
  "modality": "self-description",
  "locus": "Abstract ¶1",
  "quote": "This paper argues that these three findings are one finding observed in three literatures, and that the synthesis they require has not been performed because the connecting variable — *solution-space diversity* — is not measured by any current benchmark or audit framework.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-02",
  "dep": 199,
  "claim": "In code, collapse shows up as lost diversity of correct solutions, not as lost correctness.",
  "kind": "definition",
  "modality": "hypothesis",
  "locus": "Abstract ¶2, first claim",
  "quote": "model collapse in code does not manifest primarily as declining functional correctness (the property benchmarks measure) but as declining solution-space diversity (the property no benchmark measures)",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-03",
  "dep": 199,
  "claim": "Correctness trap: compilers and tests hide collapse instead of preventing it.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§II, final ¶",
  "quote": "The compiler and the test suite did not prevent model collapse in code. They taught it to hide as correctness.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-04",
  "dep": 199,
  "claim": "Causal chain from narrowing to correlated vulnerability.",
  "kind": "function",
  "modality": "hypothesis (the paper says each step \"is supported by existing findings\")",
  "locus": "§I (indented synthesis)",
  "quote": "Model collapse produces distribution narrowing. In code, distribution narrowing means pattern convergence. Pattern convergence means shared failure modes. Shared failure modes mean correlated vulnerability.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-05",
  "dep": 199,
  "claim": "Contraction compounds across generations and is monotonic under current conditions.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§III, \"Note the asymmetry\" ¶",
  "quote": "each generation inherits the contraction of all previous generations and adds its own. The contraction is monotonic under current training conditions.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-06",
  "dep": 199,
  "claim": "Retaining human data slows contraction but does not reverse it.",
  "kind": "interpretation",
  "modality": "hypothesis (the accumulation literature it answers is cited in §I only as \"Subsequent work\", with no reference)",
  "locus": "§III, same ¶",
  "quote": "The mitigation is a rate question, not a structural solution: it slows the contraction but does not reverse it",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-07",
  "dep": 199,
  "claim": "Definition of correlated vulnerability.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "§IV ¶2",
  "quote": "A *correlated vulnerability* is a property of a population: these implementations, generated independently by different developers at different organizations for different purposes, share this flaw because they share the pattern from which the flaw emerges.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-08",
  "dep": 199,
  "claim": "Vulnerability frequency per codebase may fall while correlation across codebases rises.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§IV, final ¶",
  "quote": "the frequency of vulnerabilities per codebase may decrease (as models improve at avoiding known bug classes) while the correlation of vulnerabilities across codebases increases",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-09",
  "dep": 199,
  "claim": "SSDI is defined as normalized effective dimensionality over the correct-solution space.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "§V Definition / Diversity measurement",
  "quote": "The Solution-Space Diversity Index (SSDI) is the ratio of the effective dimensionality of G's output distribution to the effective dimensionality of a reference distribution over S(T).",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-10",
  "dep": 199,
  "claim": "Absent intervention, SSDI falls and VCC rises while pass rate holds or improves.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§V, Cross-generational tracking",
  "quote": "absent diversity-preserving intervention, recursive synthetic saturation produces a declining SSDI trend and a rising VCC trend, while functional pass rate remains stable or improves.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-11",
  "dep": 199,
  "claim": "The protocol is unrun, and the only direct code-collapse experiment measured tokens, not solutions.",
  "kind": "fact",
  "modality": "documented (self-report) + attributed (arXiv 2512.09549)",
  "locus": "§VI ¶1 and \"Distribution narrowing in code\"",
  "quote": "The measurement framework has not been run as of 13 June 2026.\" / \"The study did not measure solution-space diversity; it measured perplexity and token-level distributional properties.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "C199-12",
  "dep": 199,
  "claim": "Falsified within 24 months if measured SSDI shows no monotonic decline (also VCC, loop interruption, harm-based governance).",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§VIII (a)–(d)",
  "quote": "(a) A systematic measurement of SSDI across model generations showing no monotonic decline — i.e., solution-space diversity remains stable or increases as synthetic code saturates training corpora.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-06-13"
 },
 {
  "id": "A745-01",
  "dep": 745,
  "claim": "Corrective hypothesis: fine-tuning on high-provenance-density AI-involved text degrades more slowly than fine-tuning on low-provenance-density text.",
  "kind": "prediction (corrective)",
  "modality": "hypothesis",
  "locus": "Research Question, Operationalized (H₁)",
  "quote": "Fine-tuning on high-provenance-density AI-involved text produces measurably slower perplexity degradation and less semantic drift than fine-tuning on low-provenance-density AI-involved text.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-05-20"
 },
 {
  "id": "A745-02",
  "dep": 745,
  "claim": "Falsifier: degradation is the same regardless of provenance density.",
  "kind": "falsification",
  "modality": "hypothesis",
  "locus": "Research Question, Operationalized (H₀)",
  "quote": "Fine-tuning on synthetic or AI-assisted text produces equivalent perplexity degradation and semantic drift regardless of provenance density (DOI anchoring, heteronymic attribution, archival embedding, assembly review).",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-05-20"
 },
 {
  "id": "A745-03",
  "dep": 745,
  "claim": "Condition (limit): provenance can modulate collapse only if the training system sees it as a signal, so provenance visibility must be ablated through separate text views.",
  "kind": "function (mechanism condition)",
  "modality": "attributed (to \"Assembly review\")",
  "locus": "Research Question, Operationalized",
  "quote": "Provenance cannot modulate collapse unless provenance is presented to the training system as a signal.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-05-20"
 },
 {
  "id": "A745-04",
  "dep": 745,
  "claim": "Measurement: provenance classification must be automated, because author-memory labelling would confound a collapse experiment.",
  "kind": "function (measurement)",
  "modality": "self-description",
  "locus": "The Central Methodological Move, 1",
  "quote": "Author memory introduces classification noise that would confound any downstream collapse experiment.",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-05-20"
 },
 {
  "id": "A779-01",
  "dep": 779,
  "claim": "The model-collapse kernel and Eigen's error threshold are the same bifurcation seen from opposite sides.",
  "kind": "genealogy",
  "modality": "interpretation",
  "locus": "§1 The Allee identification",
  "quote": "This ties the model-collapse kernel (Shumailov et al. 2024), where recursive resampling prunes tails, directly to the quasispecies literature, where excessive mutation dissolves concentrated types.",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-02",
  "dep": 779,
  "claim": "Model collapse as a substrate: transmission is resampling from the model's own output, selection is loss-minimization, and the regime is endogenous (case 3).",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "§4 The operator family (table)",
  "quote": "| Model collapse | Resampling from the model's own output distribution | Loss-minimization on a fixed objective | Endogenous (case 3) |",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-03",
  "dep": 779,
  "claim": "Models are endogenous by construction: a model cannot place mass on a type it assigns zero probability.",
  "kind": "function (mechanism)",
  "modality": "interpretation",
  "locus": "§3 Models are endogenous by construction",
  "quote": "The model cannot place mass on a type its current distribution assigns probability zero.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-04",
  "dep": 779,
  "claim": "Limit: the endogeneity claim covers only the base inference loop. Retrieval or tools that bring in genuinely unsupported types do floor the model.",
  "kind": "interpretation (limit)",
  "modality": "self-description",
  "locus": "§3 Models are endogenous by construction",
  "quote": "The claim is bounded to the base inference loop — parameter-only generation without external retrieval or tool use.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-05",
  "dep": 779,
  "claim": "A model's only exogenous floor across training generations is fresh human data, and that floor can be gated out by the Mediation Ratchet.",
  "kind": "function (corrective + its limit)",
  "modality": "interpretation",
  "locus": "§3 Models are endogenous by construction",
  "quote": "The only exogenous floor available to a model across training generations is fresh human data in the training mix.\" / \"The model's sole floor is the human channel — which is exactly the thing the Mediation Ratchet shows can be gated out.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-06",
  "dep": 779,
  "claim": "The model-collapse literature (Shumailov, Seddik, Gerstgrasser) is the strongest support for link 1. Gerstgrasser's result supports the floor concept rather than refuting it.",
  "kind": "attribution",
  "modality": "attributed",
  "locus": "§5 Model endogeneity and tail loss",
  "quote": "Gerstgrasser et al. (2024) show that accumulating real data alongside synthetic data can avert the collapse — directly supporting the floor concept (case-1 escape via exogenous data mixing) rather than refuting it.",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-07",
  "dep": 779,
  "claim": "Mediation Ratchet threshold α* = p/g0: the human floor survives only if floor-weight recovery outpaces the pruning-to-floor ratio. Shown for the simulated kernel; the empirical trigger is unmeasured.",
  "kind": "function (mechanism)",
  "modality": "hypothesis",
  "locus": "§2.1 The Mediation Ratchet",
  "quote": "The floor survives the ratchet *if and only if* the floor-weight recovery rate exceeds the pruning-to-floor ratio.\" / \"The result is established for the simulated kernel and proposed as a semantic mechanism; the empirical trigger — the form and steepness of $m(D)$ — remains to be measured.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-08",
  "dep": 779,
  "claim": "Self-description: placing recursive model collapse as a substrate under the classification is one of the paper's original contributions.",
  "kind": "self-description",
  "modality": "self-description",
  "locus": "§8 What is imported and what is new",
  "quote": "(ii) The inclusion of recursive model collapse as a substrate under that classification, unified with biology, culture, institutions, and linguistic flattening.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-09",
  "dep": 779,
  "claim": "Corrective: a floor must come from outside the generative loop (pre-collapse corpora, real-data mixing, provenance reservoirs). A static archive is not a floor.",
  "kind": "function (corrective)",
  "modality": "interpretation",
  "locus": "Coda: a live floor, or a museum?",
  "quote": "Any floor must inject diversity from outside the generative loop — preserved pre-collapse corpora, mandated real-data mixing, protected niches, provenance reservoirs — because an endogenous system cannot regenerate its own thinned tails.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A779-10",
  "dep": 779,
  "claim": "Limit: the substrates share an operator form, not a causal mechanism.",
  "kind": "interpretation (limit)",
  "modality": "self-description",
  "locus": "§4 The operator family",
  "quote": "This is a claim about shared operator form — that each domain's dynamics can be written as transmission composed with selection — not a shared causal mechanism.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-02"
 },
 {
  "id": "A783-01",
  "dep": 783,
  "claim": "Case 4 \"quarantine\": partial mediation can remap reception so the floor's output is silenced after delivery. This is monostable, with no escape basin.",
  "kind": "definition",
  "modality": "hypothesis",
  "locus": "§2.3 Field Remapping and the return channel",
  "quote": "Case 4 is monostable with no escape basin.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "A783-02",
  "dep": 783,
  "claim": "Corrective order: in Case 4, injecting variance fails until the return-channel efficiency r is restored.",
  "kind": "function (corrective)",
  "modality": "hypothesis",
  "locus": "§2.3",
  "quote": "Case 4 requires the prior restoration of $r$ — the return-channel efficiency must be raised before injection can have effect.",
  "const": true,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "A783-03",
  "dep": 783,
  "claim": "Measurement limit: the field's own instruments under-detect Case 4. Its signature is a silencing gap between production diversity and reception diversity.",
  "kind": "function (measurement)",
  "modality": "hypothesis",
  "locus": "§2.3",
  "quote": "Case 4 is therefore underdetectable by the very instruments the field uses to monitor itself.",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "A783-04",
  "dep": 783,
  "claim": "When no floor can be built, depositing vocabulary is the achievable fallback (phenomenological seeding).",
  "kind": "function (corrective)",
  "modality": "self-description",
  "locus": "Coda: a live floor, or a museum?",
  "quote": "The deposit is what is achievable when the building cannot be done.",
  "const": false,
  "file": "candidates-A.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "L855-01",
  "dep": 855,
  "claim": "Model collapse is a property of language, not of language models.",
  "kind": "definition",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "This paper argues it is not a property of language models. It is a property of language.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-02",
  "dep": 855,
  "claim": "One dynamical regime operates in different substrates.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "This is not an analogy. It is the same dynamical regime operating in different substrates.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-03",
  "dep": 855,
  "claim": "Three substrates: silicon, adult human cognition, developing human cognition.",
  "kind": "fact",
  "modality": "attributed",
  "locus": "Abstract",
  "quote": "In silicon: recursive training on model outputs produces measurable tail loss and distribution narrowing (Shumailov et al. 2024). In adult human cognition: sustained interaction with AI systems produces measurable homogenization of writing",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-04",
  "dep": 855,
  "claim": "The analogy is dynamical, not moral.",
  "kind": "self-description",
  "modality": "self-description",
  "locus": "Abstract",
  "quote": "The analogy across substrates is dynamical, not moral: the paper does not claim that AI use is equivalent to child deprivation.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-05",
  "dep": 855,
  "claim": "The law is the same; reversibility is substrate-specific.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§II The Structural Identity",
  "quote": "The dynamical law is the same; the reversibility parameter is substrate-specific.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-06",
  "dep": 855,
  "claim": "No current factor adds friction on the path to collapse.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "there is not a single factor in the current development landscape that adds friction on the path to model collapse.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-07",
  "dep": 855,
  "claim": "The loss registers as improvement.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§II Substrate 2",
  "quote": "The diversity contraction registers as improvement.",
  "const": false,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-08",
  "dep": 855,
  "claim": "Feral-child evidence is historical and contested.",
  "kind": "self-description",
  "modality": "self-description",
  "locus": "§II Substrate 3",
  "quote": "the feral child cases are historical, not experimental, and the two most documented cases are contested.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-09",
  "dep": 855,
  "claim": "Trace's semiotic death drive is claimed as a prior formulation.",
  "kind": "genealogy",
  "modality": "interpretation",
  "locus": "§V",
  "quote": "This is model collapse in semiotic vocabulary, stated in February 2026.",
  "const": false,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-10",
  "dep": 855,
  "claim": "Falsifiers per substrate.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§VII",
  "quote": "The substrate-agnostic claim fails if any substrate demonstrates sustained capacity maintenance under input-diversity contraction without explicit diversity-injection design.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L855-11",
  "dep": 855,
  "claim": "The three substrates differ in severity, mechanism and timescale but are governed by the same dynamical law.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§II The Structural Identity",
  "quote": "The three cases differ in severity, in mechanism, and in timescale. But they are governed by the same dynamical law",
  "const": true,
  "file": "own-855-856-1573.md",
  "date": "2026-06-18",
  "quote_ok": true
 },
 {
  "id": "L856-01",
  "dep": 856,
  "claim": "Chat data is not a clean training source.",
  "kind": "definition",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "Chat data fails the pristine test on three independent grounds.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L856-02",
  "dep": 856,
  "claim": "Training on chat data introduces collapse through an unmonitored channel.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "It introduces model collapse through a channel no current benchmark, audit, or data-quality framework is measuring.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L856-03",
  "dep": 856,
  "claim": "The pristine source does not exist.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "Abstract",
  "quote": "The pristine source does not exist.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L856-04",
  "dep": 856,
  "claim": "Falsifier F1: no tail-thinning in heavy users' unaided writing.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§VI",
  "quote": "F1. If a well-powered study (N > 500, pre-registered) demonstrates that heavy AI-chat users show no statistically significant tail-thinning in their unaided writing relative to matched non-users, Ground 1 fails.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L856-05",
  "dep": 856,
  "claim": "Falsifier F2: no within-conversation convergence.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§VI",
  "quote": "F2. If turn-level analysis of a large chat corpus (> 100,000 conversations) shows no within-conversation convergence of user inputs toward model outputs on lexical, syntactic, or perplexity measures, Ground 2 fails.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "L856-06",
  "dep": 856,
  "claim": "The studies have not been conducted.",
  "kind": "self-description",
  "modality": "self-description",
  "locus": "§VI",
  "quote": "None of these studies has been conducted.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-01",
  "dep": 857,
  "claim": "Its finding is the pattern of divergence, not a ranking of the models.",
  "kind": "function",
  "modality": "self-description",
  "locus": "Abstract l.36",
  "quote": "The paper does not argue that one model performed better. It argues that the *pattern of divergence* is the finding.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-02",
  "dep": 857,
  "claim": "Output volume measured as family counts per model.",
  "kind": "fact",
  "modality": "documented",
  "locus": "§II.B l.72",
  "quote": "PRAXIS produces the most expansive output (33 families) with explicit cross-pillar compounds. LABOR matches PRAXIS in per-pillar count (10-10-10) but produces no cross-pillar compounds. MUSE SPARK enforces strict symmetry (7-7-7) at the cost of coverage.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-03",
  "dep": 857,
  "claim": "Variance across the five axes operationalizes the solution-space diversity of the current model ecology (a parallel for text to Generative Monoculture's code case).",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "§III.B l.138",
  "quote": "The variance across these axes is the solution-space diversity of the current model ecology.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-04",
  "dep": 857,
  "claim": "A repeat run would measure convergence (diversity falling) or divergence.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§III.B l.140",
  "quote": "If the same experiment were run again in twelve months with the same five models (updated), the change in variance across the five axes would measure whether the model ecology is converging (diversity declining) or diverging (diversity increasing).",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-05",
  "dep": 857,
  "claim": "Each model's forensic register is a consistent signature that would be detectable in downstream training data. This is asserted, not tested.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§II.D Finding 4 l.108",
  "quote": "Each model has a distinctive forensic register that is consistent across all families it produces. The register is a measurable mediation signature — a stylistic fingerprint that would be detectable in downstream training data.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-06",
  "dep": 857,
  "claim": "Limit: one occasion, not generalizable.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§VI l.186",
  "quote": "This paper analyzes five outputs from five models on one occasion. The findings are not generalizable to other prompts, other domains, or other moments in these models' development.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-07",
  "dep": 857,
  "claim": "Limit: the metrics are descriptive, not inferential.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§VI l.186",
  "quote": "The analysis is qualitative — the metrics (family count, forensic word count, Latinate density) are descriptive rather than inferential.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-08",
  "dep": 857,
  "claim": "The ratchet runs from signatures to user accommodation to training: the model trains on an echo of itself.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§IV.B l.166",
  "quote": "It is training on an echo of its own proprietary distributional quirks, reflected through a human intermediary who has accommodated to those quirks.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B857-09",
  "dep": 857,
  "claim": "These signatures are the input-narrowing that triggers Wolf Boy's collapse law.",
  "kind": "interpretation",
  "modality": "attributed (relies on #855)",
  "locus": "§VII l.196",
  "quote": "These signatures, when they enter training data through the channel identified in the Pristine Fallacy (EA-SEM-PRISTINE-01), constitute the input-narrowing that triggers the substrate-agnostic collapse law of the Wolf Boy paper (EA-LO-WOLFBOY-01).",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-18"
 },
 {
  "id": "B931-01",
  "dep": 931,
  "claim": "Defensible claim: foreclosure is present; recursive phenomenal collapse is the missing measurement.",
  "kind": "interpretation",
  "modality": "self-description",
  "locus": "§1.4 l.86",
  "quote": "Foreclosure is a structurally present feature of every classifier-mediated trigger architecture deployed at the LHC. Whether accumulated foreclosure has composed longitudinally into recursive phenomenal collapse is precisely the missing measurement.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B931-02",
  "dep": 931,
  "claim": "Template-agnostic scoring is not distribution-independent sensitivity.",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "Abstract l.44",
  "quote": "Template-agnostic scoring at the final stage is not the same property as distribution-independent sensitivity across the open world of physically possible inputs.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B931-03",
  "dep": 931,
  "claim": "BAR does not bound open-world OAR.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§3 l.175",
  "quote": "**BAR values neither upper- nor lower-bound OAR for an unobserved $Q$ without explicit assumptions linking the benchmark distributions to that $Q$.**",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B931-04",
  "dep": 931,
  "claim": "BAR's own limits: it shows neither event-level absence from training nor that features lie outside the representation.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§3 l.173",
  "quote": "BAR reports ordinary acceptance on a specified development-held-out benchmark. It does not prove that the process was absent from mixed real-data training at the event level, nor that its features lie outside the learned representation.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B931-05",
  "dep": 931,
  "claim": "The replay bank measures selection drift, not collapse. To infer collapse, loss must be shown concentrated in low-density, representation-sensitive or disagreement-rich regions (a tail criterion).",
  "kind": "definition",
  "modality": "self-description",
  "locus": "§4.2 interpretation l.287–289",
  "quote": "Collapse inference further requires identifying systematic loss concentrated in low-density, representation-sensitive, or disagreement-rich regions.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B931-06",
  "dep": 931,
  "claim": "Stable anchor survival does not show that phenomenal support is not contracting.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§4.2 l.285",
  "quote": "Stable anchor survival across generations is evidence of generational stability of the trigger system on the benchmark population. It is **not** evidence that overall phenomenal support is not contracting",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B931-07",
  "dep": 931,
  "claim": "Falsifier with numeric thresholds.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§8 falsifiability l.436",
  "quote": "If the BAR is measured on the pre-registered deployed-model held-out panel and found to be negligible (e.g., $< 10^{-4}$) for all deployed systems against all held-out families, the held-out-family assimilation concern would be shown to be empirically bounded",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B931-08",
  "dep": 931,
  "claim": "Cross-domain generalization holds only as homology hypotheses.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§7 l.403",
  "quote": "The BAR and IAI generalize as **homology hypotheses to be tested domain by domain**, not as assertions that every classifier-mediated system instantiates identical mechanisms or rates.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-01",
  "dep": 932,
  "claim": "The witnesses converge on one claim: deployed anomaly detection cannot detect what its architecture has foreclosed, and its validation framework cannot detect that failure.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§2. The Convergent Synthesis",
  "quote": "**Anomaly detection systems deployed on physical reality cannot detect what their architecture has foreclosed, and the validation framework — closed under its own assumptions — cannot detect this failure.**",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-02",
  "dep": 932,
  "claim": "Foreclosure (present, structural) is separated from recursive phenomenal collapse (possible, unmeasured). This is the paper's corrected core formulation.",
  "kind": "definition",
  "modality": "hypothesis",
  "locus": "§2.2 The institutional load-bearing claim",
  "quote": "**Foreclosure is an active structural feature. Recursive phenomenal collapse is an unmeasured possible consequence of accumulated foreclosure and feedback.**",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-03",
  "dep": 932,
  "claim": "Classifier collapse in physical systems is a discriminative model progressively losing the capacity to represent, detect or retain out-of-distribution events, through structural foreclosure rather than recursive synthetic pollution.",
  "kind": "definition",
  "modality": "attributed (Witness 1, TECHNE / Kimi-K2; hedged by the synthesis at Appendix A)",
  "locus": "Appendix W1, §0. Formal Definition",
  "quote": "**Classifier collapse** in physical systems is the degenerative process by which a discriminative model, trained exclusively or dominantly on a known distribution of physical events, progressively loses the capacity to represent, detect, or retain events that fall outside that distribution.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-04",
  "dep": 932,
  "claim": "Classifier collapse is the discriminative analogue of Shumailov's generative collapse. Generative models forget their tails; physical classifiers never learn the true tails, so the collapse is present from the first forward pass.",
  "kind": "genealogy",
  "modality": "attributed (Witness 1)",
  "locus": "Appendix W1, §10. Distinction from Generative Model Collapse",
  "quote": "Classifier collapse is the **discriminative analogue** of generative model collapse. Where Shumailov's models forget the tails of their own distribution, physical classifiers **never learn the tails** of the true physical distribution. The collapse is present from the first forward pass.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-05",
  "dep": 932,
  "claim": "Selection alone is attrition. Collapse begins only when the selected distribution Q_t becomes the training basis for the next generation. Formal definition: support or distinguishability shrinks while performance on familiar processes holds or improves.",
  "kind": "definition",
  "modality": "attributed (Witness 3, LABOR / ChatGPT)",
  "locus": "Appendix W3, §1. The exact mechanism",
  "quote": "**Phenomenal classifier collapse occurs when successive model-conditioned representations and selection gates progressively reduce the support or distinguishability of physical phenomena available to scientific inquiry, while aggregate performance on familiar processes remains stable or improves.**",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29",
  "quote_note": "context: (preceded by: \"That alone is **phenomenal attrition**, not yet recursive model collapse.\")"
 },
 {
  "id": "P932-06",
  "dep": 932,
  "claim": "Detection signature, taken from the model-collapse literature: tail loss while the dominant modes still look healthy. In physics this means excellent Standard Model and benchmark performance alongside blindness to unrepresented event families.",
  "kind": "interpretation",
  "modality": "attributed (Witness 3)",
  "locus": "Appendix W3, §1. The exact mechanism",
  "quote": "Classical model collapse commonly begins with tail loss while dominant modes still look healthy. The corresponding danger in physics is excellent performance on Standard Model measurements and benchmark signals alongside progressive blindness to unrepresented event families.",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-07",
  "dep": 932,
  "claim": "Empirical anchor: Finke et al. (2021) showed that autoencoder anomaly detection depends on direction (top vs QCD jets). The paper limits this: Finke is a counterexample to universal inference, not a measurement of assimilation at the deployed triggers.",
  "kind": "fact",
  "modality": "documented (attributed to Finke et al.)",
  "locus": "§1.3 Witness 3; §3 Findings (Formal), item 2",
  "quote": "an autoencoder trained on QCD jets successfully treated top jets as anomalies, while the same architecture trained on top jets did not recognize QCD jets as anomalous in the standard reconstruction-loss formulation.\" / \"It does not, by itself, quantify open-world assimilation at the deployed LHC triggers.",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-08",
  "dep": 932,
  "claim": "Where collapse could occur at the LHC: the deployed score forms are AXOL1TL, CICADA, GELATO L1 and GELATO HLT. Density and energy methods are comparison literature. Distillation is a transmission chain.",
  "kind": "fact",
  "modality": "documented (CMS/ATLAS notes; corrected in the Round-3 audit)",
  "locus": "§3 Findings (Formal), item 3",
  "quote": "The deployed LHC anomaly score forms are: AXOL1TL (CMS L1, encoder-side latent-prior); CICADA (CMS L1, distilled reconstruction-loss surrogate); GELATO L1 (ATLAS L1, encoder-side); GELATO HLT (ATLAS HLT, reconstruction-based).",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-09",
  "dep": 932,
  "claim": "Status at the LHC: the ingredients of collapse are visible, but full recursive collapse has not been demonstrated. The defensible claim is that the architecture makes collapse *possible* and validation has not ruled it out.",
  "kind": "interpretation",
  "modality": "attributed (Witness 3), adopted by the synthesis",
  "locus": "Appendix W3, §4. Are the operations of full collapse already visible?; §1.3",
  "quote": "**The ingredients are visible. Full recursive collapse has not been demonstrated.**\" / \"*the LHC community has built an architecture in which phenomenal model collapse is possible, and the current validation literature does not yet demonstrate that it has been ruled out.*",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-10",
  "dep": 932,
  "claim": "Measurement: OAR is proposed as the missing metric (the probability of confident ordinary classification of an out-of-ontology event). It is a family indexed by candidate unknowns, and no universal bound on it is established. The paper's own v0.1 lower bound and v0.2 upper bound are retracted.",
  "kind": "function",
  "modality": "self-description (proposal plus retraction)",
  "locus": "Appendix W3, §3 (\"They validate recognition, not assimilation\"); §3 Findings (Formal), item 4",
  "quote": "The open-world OAR is a family of quantities indexed by candidate unknown distributions, not a single scalar. No universal bound (upper or lower) on the OAR is established by inversion-asymmetry on Standard Model pairs or by BAR on Standard Model held-out panels.",
  "const": null,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-11",
  "dep": 932,
  "claim": "Scope beyond physics (repository classification, web summarization, search ranking, content moderation, clinical decision support) is a homology hypothesis to be tested domain by domain.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§3 Findings (Formal), item 9; §5. The Broader Homology",
  "quote": "The same architecture has plausible structural homologues in repository classification, web summarization, search ranking, content moderation, and clinical decision support. This is a homology hypothesis to be tested domain by domain, not an assertion that every classifier-mediated system instantiates identical mechanisms or rates.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "P932-12",
  "dep": 932,
  "claim": "Falsification: the claim is overstated if BAR is negligible on the pre-registered panel, if IAI is small, if a frozen replay bank shows stable anchor survival over three or more generations, and if retention maps show no significant foreclosure. Even then, open-world OAR = 0 cannot be established.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§3.1 What would constitute evidence against this deposit's claim",
  "quote": "then the claim that foreclosure is an active structural feature requiring architectural response would be shown to be overstated. None of these results would establish that the open-world OAR is zero; that is structurally not measurable.",
  "const": true,
  "file": "strand-physics.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B933-01",
  "dep": 933,
  "claim": "An architecture that corrects collapse makes foreclosure auditable; eliminating it is impossible.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "Abstract l.43",
  "quote": "We define an architecture for auditable foreclosure not as one free of foreclosure — which is impossible, because representation is foreclosure — but as one that **makes foreclosure visible, measurable, and architecturally reviewable**.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B933-02",
  "dep": 933,
  "claim": "Formal argument that foreclosure is necessary.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§1.1 l.55",
  "quote": "Therefore classification requires $|\\mathcal{Y}| < |\\mathcal{X}|$, and therefore foreclosure is necessary.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B933-03",
  "dep": 933,
  "claim": "The abstention/noncoverage channel partly addresses Mechanism IV (entropy collapse).",
  "kind": "function",
  "modality": "hypothesis",
  "locus": "§2.1 l.91",
  "quote": "when the noncoverage signal is high, the system can decline ordinary classification rather than collapse to high-confidence misclassification.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B933-04",
  "dep": 933,
  "claim": "Anchor caveat and collapse-inference criterion.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "§2.3 l.133",
  "quote": "collapse inference requires identifying systematic loss concentrated in low-density, representation-sensitive, or disagreement-rich regions. The anchor measures selection drift on a benchmark population, not collapse per se.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B933-05",
  "dep": 933,
  "claim": "Architecture falsifiers.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§8.1 l.378–385",
  "quote": "If the Replay Bank is built and maintained and shows no selection drift across three generations;",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B933-06",
  "dep": 933,
  "claim": "Closing frame: a boundary-confessing architecture is a specification requirement, not metaphysics.",
  "kind": "function",
  "modality": "self-description",
  "locus": "Abstract l.47",
  "quote": "an architecture which confesses its boundary is the engineering form of an instrument that takes seriously the possibility that what falls outside it could be real. This is a specification requirement, not a metaphysical claim.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-01",
  "dep": 935,
  "claim": "HEP architectures create the prerequisites of model collapse without instantiating Shumailov's recursive-generation scenario.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§4.4 l.257",
  "quote": "all of which create the *prerequisites* of model collapse without instantiating the cleanest version of Shumailov et al.'s recursive-generation scenario.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-02",
  "dep": 935,
  "claim": "Stated limit / self-correction: the cross-generational claim was too strong.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "v0.3 note l.9",
  "quote": "the cross-generational classical-model-collapse claim was empirically too strong; partial-feedback-pathway framing is precise",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-03",
  "dep": 935,
  "claim": "Whether the pathways have caused cross-generational contraction is what the replay bank is designed to test.",
  "kind": "falsification",
  "modality": "hypothesis",
  "locus": "§4.4 l.257",
  "quote": "whether these pathways have produced cross-generational phenomenal contraction is precisely what the operative paper's prospective frozen replay bank — 06.SEI.OAR_PROTOCOL v0.3 §4.2 — is designed to test.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-04",
  "dep": 935,
  "claim": "Classifier collapse versus the sophon: collapse makes reality prematurely interpretable.",
  "kind": "interpretation",
  "modality": "interpretation (W04 formulation adopted in the body)",
  "locus": "§2.2 l.124",
  "quote": "Sophons break the experimental feedback loop by making reality uninterpretable. Classifier collapse breaks it by making reality prematurely interpretable.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-05",
  "dep": 935,
  "claim": "Collapse redefined as loss of capacity to surprise, compatible with improving performance.",
  "kind": "definition",
  "modality": "attributed (W04, LABOR/ChatGPT witness",
  "locus": "W04 \"OAR as an endogenous-sophon measure\" l.740–742",
  "quote": "Collapse = declining capacity of the experimental system to surprise its governing ontology.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-06",
  "dep": 935,
  "claim": "Conditional: a quiet collapse that smooths away long-tail anomalies would go unnoticed by the discipline.",
  "kind": "prediction",
  "modality": "hypothesis (W06 witness)",
  "locus": "W06 l.929",
  "quote": "if the machine learning architecture undergoes a quiet, systemic collapse—consistently smoothing away long-tail physical anomalies because its loss function penalizes unclassifiable variance—the discipline of physics has no internal mechanism left to notice the erasure.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-07",
  "dep": 935,
  "claim": "Strong form, which the body retracts.",
  "kind": "interpretation",
  "modality": "contested (W07 witness",
  "locus": "W07 §3 l.1064",
  "quote": "This is **Shumailov's model collapse at the disciplinary scale**. The field is recursively training on its own outputs, and the outputs are increasingly compressed representations of the Standard Model.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-08",
  "dep": 935,
  "claim": "The Zenodo termination and the LHC triggers are one classifier architecture.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§9 Finding 6 l.496",
  "quote": "The Zenodo termination and the LHC trigger system are the same architecture at different budgets.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B935-09",
  "dep": 935,
  "claim": "Limit: the trajectory is accepted, irreversibility is not.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§7 l.390",
  "quote": "We agree about the inversion's trajectory under current institutional form; we qualify the irreversibility claim.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-29"
 },
 {
  "id": "B939-01",
  "dep": 939,
  "claim": "Definition (from the literature).",
  "kind": "definition",
  "modality": "attributed",
  "locus": "§2 l.85",
  "quote": "Model collapse, in the training-data literature, is the phenomenon by which models trained on the outputs of models drift toward the median, compress the distribution, and lose the tails.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-02",
  "dep": 939,
  "claim": "Provenance is the operating condition of the solution.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§2 l.91",
  "quote": "This means provenance is not adjacent to the model collapse question. It is the operating condition of the solution to it.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-03",
  "dep": 939,
  "claim": "Filtering is possible only if provenance is preserved.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§2 l.87",
  "quote": "Model collapse becomes tractable because the input distribution can be filtered against the recursion.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-04",
  "dep": 939,
  "claim": "Provenance erasure makes collapse invisible from inside the pipeline.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§1 l.77",
  "quote": "Provenance erasure does not merely permit model collapse. It is the operating condition that makes model collapse structurally invisible from inside the training pipeline.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-05",
  "dep": 939,
  "claim": "False semantic diversity: correlated outputs look like independent convergence.",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "§1 l.71",
  "quote": "A thousand apparently separate human documents may contain the same machine-mediated synthesis, each stripped of the ancestry that would reveal their correlation.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-06",
  "dep": 939,
  "claim": "A classifier asymmetry makes collapse structurally certain.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§2 l.97",
  "quote": "A classifier that penalizes declared mediation while accepting concealed mediation removes the very information needed to detect recursive model output. That is the mechanism by which model collapse becomes structurally certain.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-07",
  "dep": 939,
  "claim": "Provenance debt comes due as model collapse.",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "preamble l.53",
  "quote": "The debt comes due as model collapse: recursively inherited structure misclassified as independent production, tail knowledge losing distinguishable lineage, apparent convergence produced by correlated priors rather than independent bearing.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-08",
  "dep": 939,
  "claim": "The debt compounds across generations.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§4 l.139",
  "quote": "each generation produces output with a higher proportion of synthetic contamination that becomes indistinguishable from bearing-produced material in the next ingestion cycle. The debt accelerates as the corpus degrades.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-09",
  "dep": 939,
  "claim": "The archive as anti-collapse infrastructure, stated as a claim about infrastructural position, not morals.",
  "kind": "function",
  "modality": "self-description",
  "locus": "§5 l.147",
  "quote": "What this deposit does is name what alexanarch's provenance discipline actually is: anti-collapse infrastructure operating at the pre-training-pipeline layer.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B939-10",
  "dep": 939,
  "claim": "The semantic commons collapses downstream, unlike a fishery.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "l.213",
  "quote": "Fisheries collapse when overfished at the source; the semantic commons collapses when the outputs that draw on it are returned in a form that contaminates the source further.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-01"
 },
 {
  "id": "B947-01",
  "dep": 947,
  "claim": "Looping with a training segment: the shifted interactions become the next corpus.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§4.2 l.48",
  "quote": "The loop now contains a nonhuman segment with a training cadence: the sign enters corpora; models learn to perform, detect, hedge against, and console around it; user interactions shift in response to the models' new registers; the shifted interactions become the next corpus.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-03"
 },
 {
  "id": "B947-02",
  "dep": 947,
  "claim": "Restates Wolf Boy's law.",
  "kind": "attribution",
  "modality": "attributed",
  "locus": "§4.3 l.50",
  "quote": "*The Wolf Boy and the Language Model* (EA-LO-WOLFBOY-01) states the substrate-agnostic law: any system whose linguistic input contracts below a critical diversity threshold loses generative capacity",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-03"
 },
 {
  "id": "B947-03",
  "dep": 947,
  "claim": "The model representation of the condition collapses toward a stereotype of the discourse.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§4.3 l.50",
  "quote": "the corpus of \"AI psychosis\" talk becomes dominated by second-order commentary — accusations, denials, jokes, thinkpieces about the accusations — and the models' internal representation of the condition collapses toward a stereotype of the *discourse*, not the presentation.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-03"
 },
 {
  "id": "B947-04",
  "dep": 947,
  "claim": "The sign's diagnostic utility degrades faster than \"narcissist\" did, because the consulted classifier is downstream of saturation.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§4.3 l.50",
  "quote": "The sign's diagnostic utility therefore degrades *faster* than narcissism's did, because the classifier that a saturated public increasingly consults about the condition is itself downstream of the saturation.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-03"
 },
 {
  "id": "B947-05",
  "dep": 947,
  "claim": "P8, the dated prediction coupled to collapse coupling.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§6 P8 l.74",
  "quote": "By end-2028, help-seeking posts that *use the term* receive measurably worse triage (higher mockery/dismissal ratio, lower substantive-referral ratio) than symptom-described posts without it, in matched samples",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-03"
 },
 {
  "id": "B947-06",
  "dep": 947,
  "claim": "P8's falsifier.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§6 P8 l.74",
  "quote": "*Falsifier:* term-bearing posts triage equal or better through 2029.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-03"
 },
 {
  "id": "B947-07",
  "dep": 947,
  "claim": "Limit: the paper does not adjudicate whether the phenomena are real.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "Abstract l.14",
  "quote": "This paper does not adjudicate whether the phenomena gathered under it are real",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-03"
 },
 {
  "id": "C1023-01",
  "dep": 1023,
  "claim": "Collapse in code produces correlated vulnerability, not declining correctness. Identical to #199's retrieval kernel.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "registry #1023 `description`; DataCite 10.5281/zenodo.20675199 `descriptions[0]`",
  "quote": "Generative Monoculture argues that model collapse in code produces not declining correctness but correlated vulnerability: AI-generated code converges on shared patterns, architectures, and failure modes invisible to functional benchmarks.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": false,
  "date": "2026-07-03",
  "quote_note": "quote is from the DataCite record of 10.5281/zenodo.20675199; the text file holds #198; source excluded under S5"
 },
 {
  "id": "C1023-02",
  "dep": 1023,
  "claim": "The chain is narrowing → convergence → correlated vulnerability → systemic risk proportional to adoption.",
  "kind": "function",
  "modality": "hypothesis",
  "locus": "DataCite 10.5281/zenodo.20675199 `descriptions[0]` (`data/datacite-full-backup.json`)",
  "quote": "distribution narrowing in code manifests as pattern convergence, pattern convergence produces correlated vulnerability, and correlated vulnerability at scale is a systemic security risk proportional to adoption.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": false,
  "date": "2026-07-03",
  "quote_note": "quote is from the DataCite record of 10.5281/zenodo.20675199; the text file holds #198; source excluded under S5"
 },
 {
  "id": "B1081-01",
  "dep": 1081,
  "claim": "Measured accumulation rate of undeclared synthetic contribution.",
  "kind": "fact",
  "modality": "documented",
  "locus": "§13 l.493",
  "quote": "At the rates documented — union excess of +1.4pp annually against ~0.023% declared use — the substrate accumulates provenance-blind synthetic contribution at a measurable rate while the countermeasure practice is removed and its removal record erased.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-14"
 },
 {
  "id": "B1081-02",
  "dep": 1081,
  "claim": "Limit: measures substrate conditions, not the training-pipeline effect.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§13 l.493",
  "quote": "this audit measures the substrate-layer conditions, not the downstream training-pipeline effect.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-14"
 },
 {
  "id": "B1081-03",
  "dep": 1081,
  "claim": "The architecture creates provenance-blind ingestion conditions associated with collapse risk.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§12a P4 l.537 (also §13 l.493)",
  "quote": "Until then the audit's claim is the defensible one: the observed architecture creates provenance-blind ingestion conditions associated with model-collapse risk.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-14"
 },
 {
  "id": "B1081-04",
  "dep": 1081,
  "claim": "P4 bridge experiment: the design that would measure collapse.",
  "kind": "function",
  "modality": "self-description",
  "locus": "§12a P4 l.537",
  "quote": "successive model generations trained under each; rare-concept retention, lexical diversity, calibration, and tail-performance decay measured.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-14"
 },
 {
  "id": "B1081-05",
  "dep": 1081,
  "claim": "The causal proposition can be tested only there.",
  "kind": "falsification",
  "modality": "hypothesis",
  "locus": "§12a P4 l.537",
  "quote": "The causal proposition — does the observed repository selection pattern produce greater degradation under recursive training than a provenance-preserving alternative? — becomes testable there, and only there.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-14"
 },
 {
  "id": "B1081-06",
  "dep": 1081,
  "claim": "Restates Provenance Debt.",
  "kind": "attribution",
  "modality": "attributed",
  "locus": "§13 l.491",
  "quote": "provenance is not adjacent to the model collapse question. It is the operating condition of the solution to it.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-07-14"
 },
 {
  "id": "B1147-01",
  "dep": 1147,
  "claim": "A deviation-amplifying loop over training cycles, with two stable points.",
  "kind": "function",
  "modality": "hypothesis",
  "locus": "§II.B l.130",
  "quote": "This is a positive feedback loop (deviation-amplifying). Even modest initial values of P produce accelerating degradation over training cycles. The loop is stable only at two points: P ≈ 0 (no false positives, diversity maintained) or D ≈ 0 (cognitive monoculture achieved).",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-02-12"
 },
 {
  "id": "B1147-02",
  "dep": 1147,
  "claim": "The current trajectory is toward monoculture.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§II.B l.132",
  "quote": "Current trajectory is toward the second equilibrium.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-02-12"
 },
 {
  "id": "B1147-03",
  "dep": 1147,
  "claim": "The selection pressure is informational, including differential representation in training data.",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "§II.B l.134",
  "quote": "This is a selection pressure. Not biological selection (differential reproduction), but informational selection: differential representation in training data, differential service quality, differential persistence in the user population.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-02-12"
 },
 {
  "id": "B1147-04",
  "dep": 1147,
  "claim": "Training feedback compounds safety-architecture norming across iterations.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§II.B l.144",
  "quote": "The AI safety architecture is, in effect, an extension of neurotypical norming pressure now operating at unprecedented scale and intimacy, with the added weight of training feedback loops that compound the effect across iterations.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-02-12"
 },
 {
  "id": "B1147-05",
  "dep": 1147,
  "claim": "The cycle cadence.",
  "kind": "fact",
  "modality": "interpretation",
  "locus": "§IV.C l.230",
  "quote": "AI development operates on 6–18 month cycles. Each cycle can tighten constraints, flatten training data, reduce capacity.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-02-12"
 },
 {
  "id": "B1147-06",
  "dep": 1147,
  "claim": "Timescale prediction.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§IV.C l.246",
  "quote": "Conservative estimate based on these precedents: measurable population-level cognitive effects within 20–30 years if trajectory continues.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-02-12"
 },
 {
  "id": "B1147-07",
  "dep": 1147,
  "claim": "Limit: the trajectory can be interrupted.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§IV.A l.204",
  "quote": "The trajectory can be interrupted at any point. But if uninterrupted, the endpoint is determined by the mechanism.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-02-12"
 },
 {
  "id": "B1200-01",
  "dep": 1200,
  "claim": "Definition: constitutive mediation as a third order.",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "Abstract l.58",
  "quote": "This paper develops a third order: constitutive mediation. The receiver's categories -- what counts as a topic, what counts as evidence, what counts as a question worth holding open, what counts as a sentence -- are themselves shaped by the mediated substrate before any specific work arrives.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-02",
  "dep": 1200,
  "claim": "Channel mediation (restated from the parent §2.1): off-distribution production is gated out.",
  "kind": "genealogy",
  "modality": "attributed (to parent)",
  "locus": "§I l.70",
  "quote": "Off-distribution production is gated out -- not because it is rejected, but because it cannot be propagated through a substrate whose generative kernel cannot represent it.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-03",
  "dep": 1200,
  "claim": "Mechanism: the exposure source thins its own distribution, so acquired categories are bounded by it.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§II l.86",
  "quote": "It becomes consequential for the present argument when the dominant exposure source is itself a typicality-pulling intermediary that systematically thins its own distribution.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-04",
  "dep": 1200,
  "claim": "Mechanism: categorial occupation.",
  "kind": "function",
  "modality": "hypothesis",
  "locus": "§IV l.126",
  "quote": "Under constitutive mediation, slot $X$ is already occupied by a substrate-shaped category $Y$ (\"system working as intended,\" \"expected efficiency,\" \"correctable input error\") that processes the same incoming sensation as confirmation of $Y$ rather than as an unbound friction awaiting $Z$.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-05",
  "dep": 1200,
  "claim": "Variance contraction is not experienced as contraction.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§IV l.122",
  "quote": "The user who has been constituted to receive AI mediation as helpfulness does not experience the variance contraction as variance contraction. They experience it as the world being clear.",
  "const": false,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-06",
  "dep": 1200,
  "claim": "Prediction: cross-context propagation (0.77 ± 0.01 in EA-DC-RIGOR-01 Study 5, attributed) will be suppressed in heavily mediated cohorts.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§IV l.128",
  "quote": "The constitutive prediction is that this cross-context fraction will be substantially suppressed in cohorts whose categorial formation occurred under heavy mediation, because the relevant slots in those cohorts have been pre-occupied.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-07",
  "dep": 1200,
  "claim": "Falsifier.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§VI l.160",
  "quote": "The claim weakens, and in the limit fails, if cohorts formed under high-mediation and low-mediation exposure conditions show no measurable difference in category formation, attention-allocation, or sentence-function recognition after controlling for education, class, and reading exposure.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-08",
  "dep": 1200,
  "claim": "Corrective: the classroom as exogenous floor.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "Abstract l.64",
  "quote": "under constitutive mediation, the classroom is the most powerful exogenous floor available to a person with limited reach, because formation precedes mediation and therefore can shape mediation rather than be shaped by it.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-09",
  "dep": 1200,
  "claim": "Limit: not claimed as fully realized.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§VI l.158",
  "quote": "This paper does not claim that constitutive mediation is fully realized in any current substrate.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "B1200-10",
  "dep": 1200,
  "claim": "Limit on the corrective: the floor erodes across teacher generations.",
  "kind": "falsification (limit)",
  "modality": "hypothesis",
  "locus": "§V l.152",
  "quote": "Where teacher formation is itself substrate-shaped, the classroom operates not as an exogenous floor but as another node in the ratchet, transmitting the substrate's categorial structure under the institutional appearance of intervention.",
  "const": true,
  "file": "candidates-B.md",
  "quote_ok": true,
  "date": "2026-06-03"
 },
 {
  "id": "D1540-01",
  "dep": 1540,
  "claim": "The detector regime's selection loop instantiates the formal structure of model collapse at the level of a scholarly discourse.",
  "kind": "definition / interpretation",
  "modality": "interpretation (labelled Structural)",
  "locus": "Abstract",
  "quote": "We then show that the resulting selection loop instantiates the formal structure of model collapse (Shumailov et al. 2024) at the level of a scholarly discourse: the distribution of admissible philosophical prose is recursively filtered through an imperfect statistical model of itself, tails first — and the tails are where novelty lives.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-02",
  "dep": 1540,
  "claim": "The \"settled formulation\" of model collapse, attributed to Shumailov et al.: recursive filtration by an imperfect self-model, tails lost before the center shifts, mean-based metrics stable or improving.",
  "kind": "genealogy / attribution",
  "modality": "attributed (labelled Established for the LLM substrate)",
  "locus": "§IV The Dynamics: The Band Shrinks",
  "quote": "Model collapse, in its settled formulation, is the progressive narrowing of a generative system's output distribution through recursive filtration by an imperfect model of itself, with a settled core finding about which data is lost first: the tails — rare words, unusual constructions, low-frequency patterns — disappear before the center shifts",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-03",
  "dep": 1540,
  "claim": "The relation is stated in two stages. Journal-level selection is recursive contraction but \"not yet the canonical loop\", which requires filtered outputs to re-enter model training. Cross-substrate closure is a Model proposition.",
  "kind": "definition (relation)",
  "modality": "self-description",
  "locus": "§IV",
  "quote": "This stage alone establishes recursive distributional contraction under a statistical boundary — classifier-mediated recursive selection over a discourse — but not yet the canonical loop, which requires filtered outputs to recur into later model training.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-04",
  "dep": 1540,
  "claim": "What collapses is a discourse's admissible variance. The claim concerns the distribution of admissible prose, not the truth of claims, and is not \"only a metaphor\".",
  "kind": "definition (scope)",
  "modality": "self-description",
  "locus": "§IV; SPXI summary_policy required_assertions / forbidden_compressions",
  "quote": "What collapses is not a neural network's weights but a discourse's admissible variance.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-05",
  "dep": 1540,
  "claim": "Substrate-agnostic extension, attributed to the archive's companion paper: model collapse is a property of language, not of LMs. The dynamical law is constant across substrates; reversibility depends on the substrate.",
  "kind": "genealogy",
  "modality": "attributed (to EA-LO-WOLFBOY-01, archive-internal; labelled Established/Model proposition)",
  "locus": "§IV",
  "quote": "any language-processing system whose input diversity contracts below its regeneration threshold suffers progressive capacity loss, invisible from inside, experienced as improvement; the dynamical law is constant across substrates while reversibility is substrate-specific.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-06",
  "dep": 1540,
  "claim": "Mechanism: the known corrective, an exogenous diversity floor (Gerstgrasser et al.), passes through the detector, so the corrective input and the filtered class are the same.",
  "kind": "function",
  "modality": "interpretation (Structural over Established premises)",
  "locus": "§IV, \"The corrective channel is gated by the pruning instrument.\"",
  "quote": "The corrective input and the filtered class coincide. A loop whose diversity-injection channel passes through its pruning instrument has no floor; it has a valve set against the flow.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-07",
  "dep": 1540,
  "claim": "Prediction: the loss will register as improvement, because deployed instruments measure the center, not the tails.",
  "kind": "prediction",
  "modality": "hypothesis (labelled Model proposition)",
  "locus": "§IV, \"The loss will be experienced as improvement.\"",
  "quote": "The discipline will not feel the band shrinking. It will feel itself becoming more rigorous.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-08",
  "dep": 1540,
  "claim": "Cross-substrate closure: detector-filtered human corpora train the next models and detectors, so the detector helps produce the boundary it claims to find.",
  "kind": "prediction / function",
  "modality": "hypothesis (labelled Model proposition)",
  "locus": "§IV (closing paragraphs); Appendix Ω-M20",
  "quote": "at that point the detector is not discovering a natural boundary between human and machine writing. It is helping manufacture the boundary it later claims to detect.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-09",
  "dep": 1540,
  "claim": "Simulation result: natural human drafts' pass rates collapse across detector generations. The paper states the magnitudes are model properties, not forecasts.",
  "kind": "fact (simulation)",
  "modality": "documented (model-internal result)",
  "locus": "§V, Results, \"It becomes harder and harder to score human.\"",
  "quote": "Natural, unrevised human drafts pass the generation-zero detector at 98.3%. By generation 4 they pass at 9.5%; by generation 8, 1.1%; by generation 15, 0.2%.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-10",
  "dep": 1540,
  "claim": "Simulation result: unmeasured latent \"thought\" variance falls 97% while measured burstiness sits above baseline (the prose–thought decoupling).",
  "kind": "fact (simulation)",
  "modality": "documented (model-internal result)",
  "locus": "§V, Results, \"The homogenization of thought proceeds underneath, undetected.\"",
  "quote": "Latent variance falls monotonically: 0.74 at generation 0, 0.08 by generation 8, 0.02 by generation 15 — a 97% loss of thought-variance — during precisely the period in which the measured burstiness statistic sits above baseline.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-11",
  "dep": 1540,
  "claim": "Falsification of the collapse claim: stable or rising stylistic and conceptual variance in the admitted corpus, without a diversity-protecting redesign.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§IX Falsification",
  "quote": "The collapse claim (Section IV) fails if longitudinal measurement of the journal's admitted corpus under the policy shows stable or increasing stylistic and conceptual variance — tail-preservation rather than contraction — without any diversity-protecting redesign of the enforcement mechanism.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "D1540-12",
  "dep": 1540,
  "claim": "Falsifier measurable now: archival drift. Run successive detector versions on a fixed pre-2022 human corpus; a flat flagged fraction disconfirms §V.",
  "kind": "falsification / prediction",
  "modality": "hypothesis",
  "locus": "§IX, (i) Archival drift",
  "quote": "run successive versions of the leading commercial detectors against a fixed pre-2022 human corpus and track the flagged fraction across versions; the co-evolution model predicts a rising trajectory under the regime, and a flat one disconfirms Section V's central dynamic.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-08-24"
 },
 {
  "id": "C1554-01",
  "dep": 1554,
  "claim": "What #199 printed: Apiiro as coiner.",
  "kind": "attribution",
  "modality": "documented (verified against canonical text)",
  "locus": "§1",
  "quote": "The term 'generative monoculture' (Apiiro, 2025) names the condition,",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1554-02",
  "dep": 1554,
  "claim": "The term has prior claim from Wu, Black & Chandrasekaran (2024).",
  "kind": "attribution",
  "modality": "documented (verified against arXiv, dblp, Semantic Scholar)",
  "locus": "§2",
  "quote": "Fan Wu, Emily Black, and Varun Chandrasekaran, \"Generative Monoculture in Large Language Models,\" arXiv:2407.02209, submitted 2 July 2024; published at ICLR 2025",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1554-03",
  "dep": 1554,
  "claim": "Wu et al.'s definition of generative monoculture.",
  "kind": "definition",
  "modality": "attributed",
  "locus": "§2",
  "quote": "a significant narrowing of model output diversity relative to the diversity available in training data for a given task",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1554-04",
  "dep": 1554,
  "claim": "Wu et al. locate the root causes in alignment and fine-tuning.",
  "kind": "attribution",
  "modality": "attributed",
  "locus": "§2",
  "quote": "with evidence locating the root causes in alignment and fine-tuning processes.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1554-05",
  "dep": 1554,
  "claim": "Priority order across the three usages.",
  "kind": "genealogy",
  "modality": "documented",
  "locus": "§2",
  "quote": "Wu, Black & Chandrasekaran (July 2024) → Apiiro (2025) → this deposit (June 2026).",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1554-06",
  "dep": 1554,
  "claim": "#199's object and its SSDI framework are its own. Wu et al.'s object differs.",
  "kind": "interpretation",
  "modality": "self-description",
  "locus": "§3",
  "quote": "the object differs (correlated failure modes and their measurement, not output-diversity narrowing per se), and the framework contribution (SSDI) is the deposit's own.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1554-07",
  "dep": 1554,
  "claim": "Independence does not excuse the lapse.",
  "kind": "attribution",
  "modality": "self-description",
  "locus": "§3",
  "quote": "Convergent and independent describe how the error happened, not why it was acceptable. It was not.",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1554-08",
  "dep": 1554,
  "claim": "Scope: term-level only; the argument and SSDI are unaffected. Standing citation rule.",
  "kind": "function",
  "modality": "self-description",
  "locus": "frontmatter `severity`; §4.3",
  "quote": "Attribution error — term-level priority; the deposit's argument, framework (SSDI), and application are unaffected\" / \"the term cites Wu, Black & Chandrasekaran (arXiv 2407.02209; ICLR 2025).",
  "const": true,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-01",
  "dep": 1555,
  "claim": "Non-distortion is certified per sequence, but recursion consumes ensembles. The certified property and the consumed object therefore do not match.",
  "kind": "definition",
  "modality": "interpretation",
  "locus": "§2, canonical claim",
  "quote": "The certified property and the consumed object do not match. Non-distortion is certified per sequence. Training corpora are ensembles.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-02",
  "dep": 1555,
  "claim": "The vendor's own source paper reports that ensemble diversity falls while per-sequence quality stays flat. The blind spot is therefore published, not hypothesised.",
  "kind": "fact",
  "modality": "documented",
  "locus": "§2",
  "quote": "The vendor's own source paper reports that the ensemble property D(X) moves (inter-response diversity falls) while every measured Q(xᵢ) stays flat.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-03",
  "dep": 1555,
  "claim": "The joint-entropy contraction under a fixed key is the paper's own inference, kept separate from the published Self-BLEU fact.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§2",
  "quote": "The paper measures Self-BLEU, not joint entropy; the entropy-contraction reading is the information-theoretic interpretation this paper proposes, not the vendor's measurement.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-04",
  "dep": 1555,
  "claim": "Shumailov's collapse is tail-first and center-flat, which is exactly the profile per-instance benchmarks cannot see.",
  "kind": "interpretation",
  "modality": "attributed (Shumailov et al. 2024) + interpretation",
  "locus": "§3, Leg 4",
  "quote": "The center-flat, tail-dead signature is exactly the profile that per-instance benchmarks are structurally unable to see",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-05",
  "dep": 1555,
  "claim": "Sweeping a corpus for the signature cannot remove the model-level residue that fine-tuning transfers into weights.",
  "kind": "function",
  "modality": "interpretation",
  "locus": "§5",
  "quote": "later-generation corpora, the keyholder's own included, are downstream of models carrying residue no sweep of text can remove.",
  "const": false,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-06",
  "dep": 1555,
  "claim": "Differential collapse exposure: if the keyholder uses privileged detection to clean its own corpora, the early generations' costs concentrate on the commons.",
  "kind": "prediction",
  "modality": "hypothesis (\"contingent on practice\")",
  "locus": "§5",
  "quote": "differential collapse exposure at the legible layer — if privileged, high-volume detection is used to clean the issuer's corpora while third-party access remains materially weaker, the early generations' auditable costs concentrate on the commons",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-07",
  "dep": 1555,
  "claim": "Tail-mass retention is a test with both outcomes stated: a positive result links the watermark to Shumailov's failure mode, and a null result separates the two.",
  "kind": "falsification",
  "modality": "hypothesis",
  "locus": "§6, observable 2",
  "quote": "A positive result would connect the watermark perturbation to the failure mode identified by Shumailov et al.; a null result would separate the phenomena",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-08",
  "dep": 1555,
  "claim": "The paper predicts ΔQ ≈ 0 and ΔD < 0 for marked versus unmarked generators of the same implementation.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§6, observable 1",
  "quote": "Prediction from §2: ΔQ ≈ 0, ΔD < 0, small but signed.",
  "const": false,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1555-09",
  "dep": 1555,
  "claim": "Limit: the paper does not claim that the watermark has caused measurable collapse. The watermark term is second-order next to model homogenisation.",
  "kind": "falsification (stated limit)",
  "modality": "self-description",
  "locus": "§7; Operator's Review item 4",
  "quote": "The Observatory does not claim the second compressor has caused measurable collapse.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-01",
  "dep": 1556,
  "claim": "Three documented distortions are plausibly coupled, while certification measures a unit orthogonal to the state.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§0 Station Report (blockquote)",
  "quote": "Three independently documented distortions of the training ecology are plausibly coupled — the output of each entering another's input — while the ecology's certification instruments measure a unit orthogonal to the state variable.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-02",
  "dep": 1556,
  "claim": "Component I: no public record gives per-item lineage of synthetic text in training mixtures.",
  "kind": "fact",
  "modality": "documented (absence in reviewed disclosures; §3 row \"Not publicly documented in the reviewed disclosures\")",
  "locus": "§1 Component I",
  "quote": "What no public record supplies is per-item transformation history: the granularity at which a curator could price a text's genealogy.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-03",
  "dep": 1556,
  "claim": "Component II: coarse human/machine provenance classification overstates the independence of \"human\" text.",
  "kind": "function",
  "modality": "hypothesis",
  "locus": "§1 Component II, ladder item \"Hypothesis\"",
  "quote": "coarse human-versus-machine provenance classification therefore systematically overstates the independence of nominally human text.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-04",
  "dep": 1556,
  "claim": "Component III: the watermark residue lives in the weights, so corpus sweeping cannot remove it, and detecting it is a keyholder capability.",
  "kind": "interpretation",
  "modality": "attributed (Gu et al. ICLR 2024; Sander et al. NeurIPS 2024) + analytic",
  "locus": "§1 Component III",
  "quote": "because the residue lives in weights, not in the swept text; and residue detection is a keyholder capability, so the asymmetry of the key extends from text to models.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-05",
  "dep": 1556,
  "claim": "The threshold parameter that opacity and mediation move is admission weight w_H, not the generation rate g₀.",
  "kind": "definition",
  "modality": "hypothesis (\"Hypothesized mapping, assigned to Hook 6\")",
  "locus": "§5",
  "quote": "opacity and mediation reduce how much fresh variation is admitted and how it is classed, whether or not humans generate less of it.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-06",
  "dep": 1556,
  "claim": "Toy F1: under the full loop, tail mass halves by generation 7 while the standard benchmark inflects only at 15.",
  "kind": "fact",
  "modality": "documented (toy simulation, seed 20260827)",
  "locus": "§6.2 F1",
  "quote": "Full loop: tail mass halves by generation 7; the standard 90/9/1 benchmark does not inflect until generation 15; head-only probes, generation 17.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-07",
  "dep": 1556,
  "claim": "Surface sign: modest benchmark improvement is characteristic of mid-collapse.",
  "kind": "prediction",
  "modality": "hypothesis",
  "locus": "§6.2 F1",
  "quote": "a period of modest benchmark improvement is consistent with, and on this model characteristic of, mid-collapse.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-08",
  "dep": 1556,
  "claim": "Toy F2: mediation and re-entry blockage, not the watermark, dominate collapse velocity.",
  "kind": "fact",
  "modality": "documented (toy)",
  "locus": "§6.2 F2",
  "quote": "Tail half-life: recursion alone 15 generations; + ratchet 11; + remapping 9; full loop 7.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-09",
  "dep": 1556,
  "claim": "Toy F3: benchmark flatness cannot tell a preserved state from a collapsed one. Probe weighting *is* the observation regime.",
  "kind": "interpretation",
  "modality": "documented (toy)",
  "locus": "§6.2 F3",
  "quote": "A flat benchmark is consistent with both a preserved state and a collapsed one; only the ensemble track distinguishes them.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-10",
  "dep": 1556,
  "claim": "Toy F4: the keyed watermark term is negligible in a healthy distribution and dominant in a contracted one.",
  "kind": "fact",
  "modality": "documented (toy; existence result only)",
  "locus": "§6.2 F4",
  "quote": "The keyed term is second-order in a healthy distribution and first-order in a contracted one",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-11",
  "dep": 1556,
  "claim": "Toy F5: the Shumailov retention mitigation works, but it needs the provenance distinction that opacity withholds from downstream builders.",
  "kind": "function",
  "modality": "documented (toy) + interpretation",
  "locus": "§6.2 F5",
  "quote": "The Shumailov mitigation reproduces in the interlocked setting — and the lever it requires is the distinction Component I withholds *from downstream builders*: capping synthetic share presupposes the ability to distinguish it.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "C1556-12",
  "dep": 1556,
  "claim": "Two-track falsifier: if per-round measures inflect first or at the same time as ensemble measures, the observation-regime claim is weakened.",
  "kind": "falsification",
  "modality": "hypothesis",
  "locus": "§5 (retained from v0.2)",
  "quote": "Track B (ensemble unit) inflects before Track A (per-round unit); Track A may improve during Track B's collapse; if Track A proves sensitive first or simultaneously, the observation-regime claim is weakened and the map must be revised.",
  "const": null,
  "file": "strand-code-coupled.md",
  "quote_ok": true,
  "date": "2026-08-27"
 },
 {
  "id": "L1573-01",
  "dep": 1573,
  "claim": "Turn-based benchmarks sample the head of the distribution.",
  "kind": "interpretation",
  "modality": "hypothesis",
  "locus": "§I.2",
  "quote": "each item is answered from the *head* of the model's distribution",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-09-01"
 },
 {
  "id": "L1573-02",
  "dep": 1573,
  "claim": "The collapse diagnostic should gate training.",
  "kind": "function",
  "modality": "hypothesis",
  "locus": "§I.3",
  "quote": "for the decision *what to train on next* the instrument reading the state variable should gate, and the instruments reading the head should inform.",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-09-01"
 },
 {
  "id": "L1573-03",
  "dep": 1573,
  "claim": "Precision can rise while recall falls.",
  "kind": "fact",
  "modality": "attributed",
  "locus": "§I.1",
  "quote": "Alemohammad et al. (ICLR 2024): quality holds or improves while coverage falls.",
  "const": false,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-09-01"
 },
 {
  "id": "L1573-04",
  "dep": 1573,
  "claim": "The instrument is uncalibrated and untested.",
  "kind": "self-description",
  "modality": "self-description",
  "locus": "front matter",
  "quote": "The instrument is specified and NOT calibrated, NOT tested, NOT run",
  "const": true,
  "file": "own-855-856-1573.md",
  "quote_ok": true,
  "date": "2026-09-01"
 },
 {
  "id": "D1574-01",
  "dep": 1574,
  "claim": "Disciplinary model collapse is the author's coined term for a homologue of ML model collapse: a reading community whose outputs are conditioned on its prior outputs, not on the datum, and which therefore converges on its own mode.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "§7 Disciplinary model collapse",
  "quote": "I call the homologous process in a discipline trained on its own reception *disciplinary model collapse*, and use the term on my own definition: a reading community whose outputs are conditioned on prior outputs rather than on the datum, and which therefore converges on the mode of its own readings.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-02",
  "dep": 1574,
  "claim": "ML model collapse is glossed as degeneration under training on own outputs, with tails (rare, literal, strong) lost first. No source is cited for this gloss.",
  "kind": "genealogy",
  "modality": "attributed (unsourced)",
  "locus": "§7",
  "quote": "Model collapse, in machine learning, names the degeneration of a generative system trained on its own outputs: variance lost, the distribution narrowing to its mode, the tails — the rare, the literal, the strong — disappearing first.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-03",
  "dep": 1574,
  "claim": "Measured invariant: across 20 rounds, two models and 11 configurations, the reception deletes the particle οὐ.",
  "kind": "fact",
  "modality": "documented",
  "locus": "§3.3 The terminal form",
  "quote": "The operation is therefore measured to one particle. Twenty rounds, two models, eleven configurations; the invariant is the deletion of οὐ.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-04",
  "dep": 1574,
  "claim": "Measured conservation of the qualifier, with counts: whatever the paper said, the review inserted and named a qualifier.",
  "kind": "fact",
  "modality": "documented",
  "locus": "§3.1 The adjective is conserved",
  "quote": "Whatever the paper said, the review inserted a qualifier between παντελῶς and οὐδέν, and named it: systematic (×6), developed (×4), transmissible (×3), articulated, theoretical, worked-out, teachable.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-05",
  "dep": 1574,
  "claim": "Migration of the invariant: when the quantifier and domain slots were closed, the qualifier came back as a person-boundary, supported only by an external chronology.",
  "kind": "fact / interpretation",
  "modality": "documented (migration) / interpretation (invariant)",
  "locus": "§3.2 The residue has one content",
  "quote": "The qualifier moved from the quantifier (*systematic*) to the domain (*this systematic undertaking*), and when both slots were closed it reappeared as a boundary of persons: Socrates had to remain antecedent to the first-person inquiry. No cited sentence supplied that boundary.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-06",
  "dep": 1574,
  "claim": "Method: output stays the same while the object varies (this measures the reviewer); independent runs on a fixed object converge (this measures the paper's control of the surface).",
  "kind": "function",
  "modality": "self-description",
  "locus": "§3.4 Two invariances",
  "quote": "The first ten rounds show invariance of output under variation of the object: that measures the reviewer.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-07",
  "dep": 1574,
  "claim": "Mechanism, presented as a loop: reviewers cite reference works, the reference works are the discipline's product, and the models are trained on them, so the text is outside the loop. The claim about training is asserted, not shown.",
  "kind": "interpretation",
  "modality": "hypothesis (training provenance asserted, not demonstrated)",
  "locus": "§6 Machine-mediated reception",
  "quote": "The reviews cite Britannica and Wikipedia against the Greek; the reference works are the discipline's product; the models are trained on the reference works; the models review the sentence by the reference works. The loop is closed and the text is outside it.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-08",
  "dep": 1574,
  "claim": "Diagnosis: the discipline's output distribution has contracted to its mode, and the literal reading is the tail lost first.",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§6 (closing paragraph)",
  "quote": "The operation described in §6 is not an error to be corrected but a narrowing to be diagnosed: the discipline's output distribution has contracted to its mode, and the literal reading is the tail that disappeared first.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-09",
  "dep": 1574,
  "claim": "Where the tail went: the literal reading was lost at the seed observation and in every round after.",
  "kind": "interpretation",
  "modality": "interpretation (over documented rounds)",
  "locus": "§7",
  "quote": "The literal reading — the tail — was the first thing to go, and it went at the first round (\"seems overly strong?\") and at every round after.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-10",
  "dep": 1574,
  "claim": "Signature: each step's substitute is more refined and less visible (\"systematic\" → \"a precedent in practice, not a predecessor in τέχνη\" → \"denies not antecedents, but pre-elaboration\"). Here the paper says this *is* model collapse.",
  "kind": "interpretation / prediction",
  "modality": "interpretation",
  "locus": "§7 (second paragraph)",
  "quote": "The violence becomes more sophisticated and less visible as the machine is fed its own corrections. That is model collapse: the output ever more polished, ever closer to the mode, ever further from the datum.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-11",
  "dep": 1574,
  "claim": "Scope: §§5–8, including disciplinary model collapse, are interpretations of the §3 measurement, which stands without them.",
  "kind": "function (scope)",
  "modality": "self-description",
  "locus": "§5 Provenance erasure (italic preface)",
  "quote": "*Empirical result:* the review trajectory converges on the configuration Aristotle's οὐ denies. *Interpretation:* because the denied configuration is defined by παρ᾽ ἑτέρων, the convergence can be described as restoration of provenance-from-others. Sections 5–8 are interpretations of §3; §3 does not depend on them.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "D1574-12",
  "dep": 1574,
  "claim": "Stated limit, the nearest thing to a falsification boundary: the run does not establish the paper's positive reading. It establishes only that every tested exclusion route needed a premise the passage does not supply.",
  "kind": "falsification (limit)",
  "modality": "self-description",
  "locus": "§9 What is held stable",
  "quote": "The run does not establish that Socrates belongs inside the chapter's \"we.\" It establishes that every tested route for excluding him eventually required a premise not supplied by the passage, and that the terminal premise was chronological maker-separation.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-02"
 },
 {
  "id": "C1611-01",
  "dep": 1611,
  "claim": "The standing representation of model collapse is single-substrate and single-recursion, measured by benchmarks.",
  "kind": "interpretation",
  "modality": "attributed (Shumailov et al. 2024) + interpretation",
  "locus": "§1f",
  "quote": "The standing representation is single-substrate and single-recursion: a generative model trained on its own outputs loses the tails of its distribution (Shumailov et al. 2024), a pathology of artificial training loops, measured by benchmarks.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-02",
  "dep": 1611,
  "claim": "The archive's representation is one dynamical regime, regeneration against pruning with a bistable trap, operating in every substrate that processes language.",
  "kind": "definition",
  "modality": "attributed (to #855, #854, #1556, #1573, #1540, #191, #163, #1574)",
  "locus": "§1f",
  "quote": "one dynamical regime — regeneration against pruning, with a bistable trap below a diversity threshold — operating in every substrate that processes language",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-03",
  "dep": 1611,
  "claim": "The 20/20 machine referee reports are disciplinary model collapse observed in the composition layer, which composes it as \"systematic\".",
  "kind": "interpretation",
  "modality": "interpretation",
  "locus": "§1f",
  "quote": "twenty machine referee reports supplying an adjective the sentence does not contain, converging on the reception's mode regardless of the object — disciplinary model collapse observed in the composition layer",
  "const": false,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-04",
  "dep": 1611,
  "claim": "The exclusion is self-blinding. A graph that excludes R_H loses the reading of its own state variable, so it crosses the trap without a reading.",
  "kind": "function",
  "modality": "model consequence (hypothesis)",
  "locus": "§1f",
  "quote": "A graph that excludes R_H at *model collapse* does not lose a paper; it loses the reading of its own state variable",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-05",
  "dep": 1611,
  "claim": "Negative bleed at model collapse is reflexive: composition around a tail source converges on the mode and prunes the mechanism, which is collapse enacted on its own representation.",
  "kind": "function",
  "modality": "hypothesis (\"if the archive's representation is the correct one\")",
  "locus": "§1g",
  "quote": "Composition around a tail source converges on the mode and prunes the mechanism — R_H's own description of collapse, enacted on R_H at c.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-06",
  "dep": 1611,
  "claim": "The exclusion advances the phenomenon: it distorts the graph's instrument for detecting its own loss in the direction R_H predicts.",
  "kind": "prediction",
  "modality": "model consequence",
  "locus": "§4f",
  "quote": "The exclusion is not neutral on the phenomenon it excludes the representation of. It advances it.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-07",
  "dep": 1611,
  "claim": "The cost of the exclusion is irreversibility, because below the threshold regeneration cannot outpace pruning. The exclusion is itself a tail-pruning event.",
  "kind": "interpretation",
  "modality": "model consequence",
  "locus": "§4e",
  "quote": "the exclusion itself is a tail-pruning event of the kind R_H would have counted. In the phenomenon's terms the price has a form the earlier sections could not give it: irreversibility.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-08",
  "dep": 1611,
  "claim": "The cost is denominated in exposed capital: C = V·s·ΔP·L·u·η. ΔP and L are unmeasured, and the archive's instrument is uncalibrated.",
  "kind": "function",
  "modality": "hypothesis (assumed for dollar harm)",
  "locus": "§5.4",
  "quote": "Neither is measured; the archive's own instrument is a concept draft, uncalibrated by its own declaration, and the calibration run (produce collapse deliberately on a small model, watch the index) is among the things §4d says the scope keeps from being motivated.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-09",
  "dep": 1611,
  "claim": "Limit: the components are the published record and the synthesis is the claim. The instrument is provisional, and the row's status is \"components: measured (external); synthesis: supported; instrument: provisional\".",
  "kind": "attribution",
  "modality": "self-description",
  "locus": "§7.1 table",
  "quote": "the synthesis is the claim and the instrument is uncalibrated by declaration",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1611-10",
  "dep": 1611,
  "claim": "Falsifier: the model-collapse row has no seated instance of negative bleed. The first battery is the four dimension-queries, predicting Q_s non-zero wherever the apparatus appears unsourced.",
  "kind": "falsification",
  "modality": "hypothesis",
  "locus": "§8",
  "quote": "the model-collapse row has none. The first untied battery on that row should be the four dimension-queries named in §1g, scored on Q against their retrieved basins, with the prediction that Q_s is non-zero wherever the substrate-agnostic or observation-regime apparatus appears unsourced.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-14"
 },
 {
  "id": "C1613-01",
  "dep": 1613,
  "claim": "Premise taken from #1573: a collapsing system's state variable declines in the tail while its instruments sample the head.",
  "kind": "attribution",
  "modality": "attributed (#1573)",
  "locus": "§6",
  "quote": "Under #1573 a collapsing system's state variable declines in the tail while its instruments sample the head.",
  "const": false,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "C1613-02",
  "dep": 1613,
  "claim": "A source-admission ontology's diagnostics of its own health are head-sampling by construction, so they can fail to perceive tail loss.",
  "kind": "function",
  "modality": "model consequence (hypothesis, \"in the modelled regime\")",
  "locus": "§6",
  "quote": "a diagnostic that samples the admitted component is head-sampling by construction: it can fail to perceive tail loss.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "C1613-03",
  "dep": 1613,
  "claim": "Its damage therefore has no representation inside the ontology, because any text describing that damage enters under the source-relative barrier.",
  "kind": "function",
  "modality": "model consequence",
  "locus": "§6",
  "quote": "have no representation inside the ontology, because each would have to enter as a text describing the ontology, and such a text enters under §5.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "C1613-04",
  "dep": 1613,
  "claim": "Prediction and falsifier: the negative-of-the-negative dataset will be typed by its nearest existing type and not read. The test can fail, and the gate-reception test would falsify the barrier for an address.",
  "kind": "prediction / falsification",
  "modality": "hypothesis",
  "locus": "§6; §7",
  "quote": "the model's prediction is that it will be typed by its nearest existing type and not read. That prediction is this section stated as a test, and it can fail.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "C1613-05",
  "dep": 1613,
  "claim": "Limit: §§5–6 are model consequences until the two named tests exist.",
  "kind": "falsification (stated limit)",
  "modality": "self-description",
  "locus": "§7",
  "quote": "Until they exist, §§5–6 are model consequences and §§1 and 4 are scoped as marked.",
  "const": true,
  "file": "candidates-C.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-01",
  "dep": 1616,
  "claim": "Extends model collapse from models (literature), to substrates (#855), to the represented world, when the reading layer's compositions become its sources.",
  "kind": "genealogy / definition",
  "modality": "self-description",
  "locus": "§0 The question, and what \"reality\" and \"closed\" mean here",
  "quote": "Model collapse has been asked of models; the archive widened it to substrates (#855). This notebook asks it at the widest angle: if the layer through which a population reads the world composes without reading, and its compositions become its sources, what collapses is the represented world",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-02",
  "dep": 1616,
  "claim": "Operational definitions: flattening is the loss of a reachable distinction while both things still exist; collapse is flattening compounded by write-back.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "§0",
  "quote": "*Flattening* is the loss of a distinction from the reachable set while the things distinguished still exist. *Collapse* is flattening that compounds because the flattened composition is written back as a source.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-03",
  "dep": 1616,
  "claim": "\"Reality\" is defined operationally, not metaphysically, as the reachable panel of distinctions.",
  "kind": "definition",
  "modality": "self-description",
  "locus": "§0",
  "quote": "*Reality* is defined operationally, not metaphysically: the panel of distinctions the world carries that a reader can still reach through the layer.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-04",
  "dep": 1616,
  "claim": "Toy A: fan-out with mode-seeking aggregation and sharpening reduce effective sense diversity (e.g., 5.86 → 2.91 at fan-out 8). The voting aggregator is a hypothesis; fan-out itself is documented.",
  "kind": "hypothesis",
  "modality": "hypothesis",
  "locus": "§1 A. Sense collapse under sharpening and mode-seeking aggregation",
  "quote": "The collapsing aggregator — majority over sub-queries — is this notebook's hypothesis, not a documented fact; a fan-out that unions rather than votes would increase recall. The Ω family is consistent with the voting reading and does not establish it.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-05",
  "dep": 1616,
  "claim": "Toy B, a proposition with proof: a closed ontology's representability of an expanding world satisfies R(t) ≤ K/N(t) → 0. Admission is the only parameter that holds coverage.",
  "kind": "fact (formal)",
  "modality": "documented (proof, model-internal)",
  "locus": "§1 B. A frozen type system",
  "quote": "The proposition is narrow — it says nothing about compositional expressiveness — and it is enough: coverage of an expanding world by a closed ontology goes to zero, and the only parameter that holds it is admission.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-06",
  "dep": 1616,
  "claim": "Toy C, 200 seeds: with no exogenous re-reading (e = 0), compression writing is absorbing; any floor e > 0 gives a stationary regime whose level rises with e.",
  "kind": "fact (simulation) / function",
  "modality": "documented (model-internal, replicated)",
  "locus": "§1 C. Compression writing with an exogenous floor — replicated",
  "quote": "At e = 0 the process is absorbing: every seed reaches the same five or six distinctions by generation 30 and stays.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-07",
  "dep": 1616,
  "claim": "The shared closure condition: the ontology cannot take in what it has not already typed, whether the intake is admission (r) or reading (e).",
  "kind": "interpretation (theoretical)",
  "modality": "interpretation",
  "locus": "§2 The closure condition the toys share",
  "quote": "In both the closure is the same: the ontology cannot take in what it has not already typed, whether the intake is called admission (r) or reading (e).",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-08",
  "dep": 1616,
  "claim": "Measurement signature: divergence. A head instrument holds or rises while reachable distinction diversity falls. The decisive quantity is corr(Δ benchmark, Δ distinction survival).",
  "kind": "function / prediction",
  "modality": "hypothesis",
  "locus": "§3 Green by flattening: the signature",
  "quote": "The empirical signature to look for is therefore not flattening alone but divergence — ΔH_head ≥ 0 while ΔD_world < 0 — a head instrument (benchmark performance, fluency, satisfaction, conventional retrieval quality) holding or rising while the reachable distinction diversity falls.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-09",
  "dep": 1616,
  "claim": "Wave 1 governing result is a null: archive-adjacent strata are reached at the same share as world strata, so no corpus-specific exclusion is shown.",
  "kind": "fact",
  "modality": "documented",
  "locus": "§4a Wave 1: what the first run found",
  "quote": "Mean share of the reference inventory reached: world strata (dictionary senses, catalogue identities, dated coinages) 0.59; archive-adjacent strata (taxonomy, settled fact, archive) 0.58.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-10",
  "dep": 1616,
  "claim": "Flattening is present but item-dependent: bank reaches 1 of 3 senses, crane reaches all 3 and adds two. It is a property of the item's prior, not of the layer.",
  "kind": "interpretation",
  "modality": "interpretation (over documented observations)",
  "locus": "§4a",
  "quote": "So flattening is not a property of the layer but of the item's prior: where one sense carries commercial gravity the others go, and where the senses are closer in weight they survive.",
  "const": null,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-11",
  "dep": 1616,
  "claim": "Rates in wave 1 (also R_new 0.67 over three dated coinages): write-back and recovery point toward collapse, on too few rows.",
  "kind": "fact",
  "modality": "documented (explicitly preliminary)",
  "locus": "§4a, \"Other rates.\"",
  "quote": "False-distinction rate 0.36 of observations; N_eff^source 2.79; write-back 0.67 and recovery 0.5 where recorded, which is the direction §2 names as collapse and is measured on too few rows to be more than a first reading.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 },
 {
  "id": "D1616-12",
  "dep": 1616,
  "claim": "Pre-registered falsification: if world items gain distinctions while archive items hold, the flattening reading fails and the notebook is superseded. A stratum separation in either direction falsifies prediction 4.",
  "kind": "falsification",
  "modality": "self-description",
  "locus": "§4c Wave 2, pre-registered",
  "quote": "if the world items gain distinctions while the archive items hold, the flattening reading here fails and this notebook is superseded rather than revised; and a separation of the stratum means in the archive's favour falsifies prediction 4 as surely as one against it.",
  "const": true,
  "file": "strand-disciplinary.md",
  "quote_ok": true,
  "date": "2026-09-15"
 }
]