{
 "axn": "AXN:0286.GOVERNANCE.⚙️🕒💫🗺️✏️📎",
 "hex": "0286",
 "family": "GOVERNANCE",
 "emoji": "⚙️🕒💫🗺️✏️📎",
 "hash": "32640a514347a413d4f003be59d730e38eb9f598705729d6d628b6e0204a6e79",
 "title": "Audited Claims for the Semantic Deviation Research Program — The Glas Function: An External-Format Restatement",
 "creator": "Nobel Glas",
 "orcid": "0009-0000-1599-0703",
 "date": "2026-05-17",
 "description": "A standalone technical audit of the Semantic Deviation research program, authored by Nobel Glas in the “transparent-medium” register. It separates the potentially testable core from the program’s philosophical interpretation and symbolic-institutional architecture so that external researchers can evaluate the measurement claims without adopting the surrounding archive vocabulary.\n\nThe audit identifies the unspecified semantic field as the program’s highest-priority gap and proposes three operationalizations: closed-system language-model continuation distributions, changing retrieval-response fields, and citation-graph fields. It narrows the headline principle, specifies defaults and controls, calls for component ablation and anti-Goodhart design, and presents budgeted experiments whose failure would require revision. It does not amend the founding formulation or report completed results.",
 "content_type": "Empirical study",
 "license": "CC-BY-4.0",
 "substrate": "AI-assisted (substrate)",
 "keywords": [
  "empirical study"
 ],
 "version": "v1.0",
 "status": "ACTIVE",
 "clusters": [
  "Instrumental",
  "Temporal",
  "Celestial",
  "Navigational",
  "Scriptural",
  "Scriptural"
 ],
 "reading": "Method → Duration → Origin → Search → Text → Text",
 "full_text_path": "/data/texts/AXN-0286-text.md",
 "mirrors": {
  "blog": "https://mindcontrolpoems.blogspot.com/2026/05/audited-claims-for-semantic-deviation.html",
  "machinemediation": "https://machinemediation.org/registry/#search=MM-CHA-0642"
 },
 "wiki_article": "*Audited Claims for the Semantic Deviation Research Program* performs what the record calls the **Glas function**: it attempts to make the technical object visible without requiring the reader to adopt the institutional theater around it.\n\nThree layers are separated:\n\n- **Layer A — technical core:** deviation integrals, field trajectories, closed-system computation, retrieval protocols, and training interventions.\n- **Layer B — philosophical interpretation:** durable trajectory deformation, provenance-resolved and normative measures, and canon-formation implications.\n- **Layer C — institutional and symbolic apparatus:** heteronyms, observatories, Assembly roles, vow-language, torus structures, and archive coordinates.\n\nThe paper audits Layer A. It does not claim Layers B and C are worthless; it argues they should not be prerequisites for technical evaluation.\n\nThe central problem is the semantic field `Ψ_t(C)`. Unless the field is specified, different researchers will measure different objects and obtain incomparable values.\n\nThree canonical operationalizations are proposed:\n\n1. **F1 — Closed-System Continuation Field:** exact divergence between language-model next-token distributions with and without an intervention.\n2. **F2 — Retrieval Response Field:** repeated measurement of external AI responses before and after a DOI-anchored or otherwise indexable intervention.\n3. **F3 — Citation Graph Field:** long-horizon divergence in topic-cluster citation distributions using matched or synthetic controls.\n\nF1 is exact relative to a model checkpoint, not to “the world.” F2 requires instrumentation controls because models, indices, and retrieval systems drift. F3 is slow and may be underpowered for single-paper interventions.\n\nThe audit narrows the universal claim that meaning *is* deviation. The defensible research claim becomes conditional on operationalization, field, horizon, divergence measure, and durability threshold.\n\nIt also calls for component decomposition. If a training objective combines deviation, provenance, and coherence, each component must be tested independently. The paper predicts that provenance may carry more independent uplift than deviation because attribution failures have clearer prior empirical support. Either outcome is treated as informative.\n\nAnti-Goodhart protections include pre-registration, held-out judges, adversarial examples, provenance-theater detection, and explicit reporting of null or negative results.\n\nThe paper’s function is corrective rather than ratifying. It identifies what the program can currently defend, what remains speculative, what would falsify it, and which experiments should be funded next.",
 "wiki_status": "provisional",
 "entities": [
  {
   "subject": "Audited Claims for the Semantic Deviation Research",
   "predicate": "created_by",
   "object": "Nobel Glas",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "Audited Claims for the Semantic Deviation Research",
   "predicate": "is_type",
   "object": "Empirical study",
   "type": "classification",
   "evidence_status": "observed"
  },
  {
   "subject": "Audited Claims for the Semantic Deviation Research",
   "predicate": "belongs_to_family",
   "object": "GOVERNANCE",
   "type": "classification",
   "evidence_status": "observed"
  },
  {
   "subject": "Audited Claims for the Semantic Deviation Research",
   "predicate": "is_part_of",
   "object": "Crimson Hexagonal Archive",
   "type": "institution",
   "evidence_status": "observed"
  },
  {
   "subject": "Background and ongoing",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "Each major deposit in the program should be sent to at least one external researcher in a directly r"
  },
  {
   "subject": "Caveat on F1",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "Under F1 the divergence is computed exactly from the model's softmax logits. This is a measurement o"
  },
  {
   "subject": "Citation theater",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "ornamental provenance markers (\"according to X,\" \"as noted in Y\") that inflate the $\\pi$ score witho"
  },
  {
   "subject": "Diachronic semantic change",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "The Layer B canonicity discourse — works whose meaning persists across centuries — is structurally a"
  },
  {
   "subject": "Entropy-floor capping",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "$\\mathcal{M}*T^{\\text{net}}$ contribution at each token can be capped by a per-token entropy floor: "
  },
  {
   "subject": "Layer A",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "is the technical core: the SDP integral, the divergence functional, the field-trajectory measurement"
  },
  {
   "subject": "Layer B",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "is the philosophical interpretation: meaning as durable trajectory deformation, the three-measure se"
  },
  {
   "subject": "Layer C",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "is the institutional and symbolic apparatus: heteronyms, observatories, choruses, septads, vow-langu"
  },
  {
   "subject": "Margin filtering",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "thresholding which preference pairs are used for training."
  },
  {
   "subject": "Memetic volatility farming",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "outputs designed to provoke discussion-divergence (high $\\mathcal{M}_T$ in the discourse-graph opera"
  },
  {
   "subject": "Model-Base",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "is the pre-fine-tune checkpoint of the chosen open-weight model (e.g., the base Llama-3.1-8B prior t"
  },
  {
   "subject": "Model-CE",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "is the standard cross-entropy supervised fine-tune of Model-Base on the same instruction corpus used"
  },
  {
   "subject": "Pre-registered protocol",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "**Corpus.** Three categories, balanced for length and topic:"
  },
  {
   "subject": "Provenance retention",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "($\\pi$): the degree to which the intervention preserves attributable lineage to its sources."
  },
  {
   "subject": "Provenance-weighted damping",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "$\\mathcal{M}_T$ is multiplied by the provenance-retention score $\\pi$ before being entered into pref"
  },
  {
   "subject": "Recursive citation rings",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "groups of artifacts that cite each other to inflate forward-citation $\\mathcal{M}_T$ under F3."
  },
  {
   "subject": "Reference model",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "Llama-3.1-8B-Instruct (the specific HuggingFace checkpoint meta-llama/Llama-3.1-8B-Instruct at the p"
  },
  {
   "subject": "Reference-model anchoring",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "the KL-divergence from a reference distribution, as in standard DPO (Rafailov et al. 2023)."
  },
  {
   "subject": "Retrieval poisoning",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "coordinated deposit of artifacts designed to deform retrieval surfaces in measurable ways, regardles"
  },
  {
   "subject": "Saturation limits",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "$\\mathcal{M}_T$ saturates above a threshold $\\tau$, so that further deviation beyond $\\tau$ does not"
  },
  {
   "subject": "Shock injection",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "low-cost insertion of high-deviation tokens (lexical rarity, contrarian phrasing, attention-grabbing"
  },
  {
   "subject": "Signed net deviation",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "($\\mathcal{M}_T^{\\text{net}}$): the per-token deviation from the model's own expectation baseline, s"
  },
  {
   "subject": "Slop Composite Index (SCI)",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "a weighted aggregation of the four signal components used to generate synthetic preference labels — "
  },
  {
   "subject": "Statistical test (P1)",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "Two-sided Mann-Whitney U test comparing the distributions of $\\bar{\\delta}$ between Category Slop an"
  },
  {
   "subject": "Temporal coherence penalties",
   "predicate": "minted_in",
   "object": "Audited Claims for the Semantic Deviation Research Program T",
   "type": "concept",
   "evidence_status": "observed",
   "note": "A rolling-window variance penalty on $\\mathcal{M}_T^{\\text{net}}$ across generation horizons penaliz"
  }
 ],
 "entity_status": "provisional",
 "sovereign_id": "MM-CHA-0642",
 "word_count": 8210,
 "download_md": "/data/deposits/AXN-0286.md",
 "zenodo_dois": [
  "10.5281/zenodo.20251740",
  "10.5281/zenodo.20250735",
  "10.5281/zenodo.20250736",
  "10.5281/zenodo.20251738",
  "10.5281/zenodo.20251742"
 ],
 "citations": [
  {
   "title": "Zenodo record 20250736",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20250736",
   "url": "https://doi.org/10.5281/zenodo.20250736",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20251738",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20251738",
   "url": "https://doi.org/10.5281/zenodo.20251738",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20251740",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20251740",
   "url": "https://doi.org/10.5281/zenodo.20251740",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20251742",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20251742",
   "url": "https://doi.org/10.5281/zenodo.20251742",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20250735",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20250735",
   "url": "https://doi.org/10.5281/zenodo.20250735",
   "role": "Cross-referenced work"
  }
 ],
 "citation_stats": {
  "total_doi_refs": 5
 },
 "deposit_number": 107,
 "journal": "Transactions on Substrate Engineering (Trans. Substrate Eng.)",
 "cited_by": [
  {
   "deposit": 107,
   "axn": "AXN:0286.GOVERNANCE.⚙️🕒💫🗺️✏️📎"
  },
  {
   "deposit": 108,
   "axn": "AXN:028A.GOVERNANCE.👈🤙🐚🌖🔓🟡"
  },
  {
   "deposit": 110,
   "axn": "AXN:028F.GOVERNANCE.👁️‍🗨️🀄🗼📦🌠🌹"
  }
 ],
 "defines_concepts": [
  "Background and ongoing",
  "Caveat on F1",
  "Citation theater",
  "Diachronic semantic change",
  "Entropy-floor capping",
  "Layer A",
  "Layer B",
  "Layer C",
  "Margin filtering",
  "Memetic volatility farming",
  "Model-Base",
  "Model-CE",
  "Pre-registered protocol",
  "Provenance retention",
  "Provenance-weighted damping",
  "Recursive citation rings",
  "Reference model",
  "Reference-model anchoring",
  "Retrieval poisoning",
  "Saturation limits",
  "Shock injection",
  "Signed net deviation",
  "Slop Composite Index (SCI)",
  "Statistical test (P1)",
  "Temporal coherence penalties"
 ],
 "references_concepts": [
  "Adversarial Topologist",
  "Background and ongoing",
  "Caveat on F1",
  "Citation theater",
  "Constitution",
  "Diachronic semantic change",
  "Entropy-floor capping",
  "Google AI Overview",
  "Hallucination",
  "Layer A",
  "Layer B",
  "Layer C",
  "Margin filtering",
  "Memetic volatility farming",
  "Model-Base",
  "Model-CE",
  "Model-Sem",
  "Operationalization",
  "Pre-register",
  "Pre-registered protocol",
  "Provenance retention",
  "Provenance-weighted damping",
  "Recursive citation rings",
  "Reference model",
  "Reference-model anchoring",
  "Requirements",
  "Retrieval poisoning",
  "Saturation limits",
  "Shock injection",
  "Signed net deviation",
  "Slop Composite Index (SCI)",
  "Statistical test (P1)",
  "Temporal coherence",
  "Temporal coherence penalties",
  "The AI"
 ],
 "references_concept_count": 35,
 "external_metadata_path": "/data/external-metadata/AXN-0286.json",
 "openalex_ids": [
  "https://openalex.org/W7161488651",
  "https://openalex.org/W7161497382",
  "https://openalex.org/W7161476300",
  "https://openalex.org/W7161488326"
 ],
 "datacite_severance": "mixed",
 "body_status": {
  "class": "full",
  "lacuna": false,
  "recovery_status": "COMPLETE",
  "residual_chars": 50726,
  "audited_at": "2026-07-17T04:49:17.789813Z",
  "audit_version": "v3-dual-store+recovery-map",
  "measured_prose_words": 7961,
  "measured_at": "2026-07-31"
 },
 "title_repair_log": [
  {
   "at": "2026-07-28T15:20:20Z",
   "defect": "registry_title_diverged_from_published_form",
   "was": "Audited Claims for the Semantic Deviation Research Program The Glas Function: An External-Format Restatement Nobel Glas",
   "now": "Audited Claims for the Semantic Deviation Research Program — The Glas Function: An External-Format Restatement",
   "basis": "Set verbatim to the title this work carried at Zenodo, from data/doi-resolution-index.json (10.5281/zenodo.20259293, remediated_fuzzy, word overlap 0.92). The recovered metadata is the published form of record; the registry title was a variant."
  }
 ],
 "canonical_text_status": "canonical_full_text",
 "modifications": [
  {
   "date": "2026-08-04",
   "field": "journal",
   "reason": "W6-COMPLETE venue normalization (deferred #1-#358 half; corpus fully audited; MANUS ruling 2026-08-01)",
   "was": "Trans. Substrate Eng.",
   "now": "Transactions on Substrate Engineering (Trans. Substrate Eng.)"
  },
  {
   "date": "2026-08-04",
   "field": "publisher",
   "reason": "PUB-POPULATE: dc:publisher from venues.json v1.1 press mapping (CP-R3 RULED-EXTENDED 2026-08-01); Alexanarch = publisher of record where no imprint applies",
   "now": "Pergamon Press"
  },
  {
   "date": "2026-08-04",
   "field": "description",
   "reason": "DW-009 intake (LABOR-prepared, TACHYON-verified: AXN match + factual probes vs record body)",
   "was": "\"Audited Claims for the Semantic Deviation Research Program T\" is an empirical study by Nobel Glas in the Crimson Hexagonal Archive (2026-05-17). The Glas Function: An External-Format Restatement. The work comprises 8,210 words and is classified under the GOVERNANCE family. Nobel Glas directs Framew",
   "now": "A standalone technical audit of the Semantic Deviation research program, authored by Nobel Glas in the “transparent-medium” register. It separates the potentially testable core from the program’s philosophical interpretation and symbolic-institutional architecture so that external researchers can evaluate the measurement claims without adopting the surrounding archive vocabulary.\n\nThe audit identifies the unspecified semantic field as the program’s highest-priority gap and proposes three operationalizations: closed-system language-model continuation distributions, changing retrieval-response fields, and citation-graph fields. It narrows the headline principle, specifies defaults and controls, calls for component ablation and anti-Goodhart design, and presents budgeted experiments whose failure would require revision. It does not amend the founding formulation or report completed results."
  }
 ],
 "date_modified": "2026-08-04",
 "publisher": "Pergamon Press",
 "journal_assignment": {
  "assigned": "2026-08-15",
  "by": "TACHYON under operator adjudication",
  "pass": 1,
  "method": "read per deposit — title and content_type, one at a time. No script classified anything.",
  "previous": "Transactions on Substrate Engineering (Trans. Substrate Eng.)",
  "supersedes": "the 2026-06-21 preliminary batch mapping (#866), which assigned 864 deposits and put 371 in one venue",
  "authority": "data/cha-journals.json · datasets/venues/records/",
  "revision": "2026-08-15 operator rulings: MPAIs/disambiguation -> JCS; JCS issue template; semantic deviation -> TSE; book-length -> New Human Press"
 },
 "_projection": {
  "note": "Derived file. Canonical machine record is this entry in data/registry.json; the human record is the record_url. Do not edit this file.",
  "record_url": "https://www.alexanarch.org/s/records/107/",
  "self_url": "https://www.alexanarch.org/data/records/107.json",
  "registry_url": "https://www.alexanarch.org/data/registry.json",
  "text_url": "https://www.alexanarch.org/data/texts/AXN-0286-text.md",
  "oai_pmh": "https://www.alexanarch.org/oai?verb=Identify"
 }
}