{
 "axn": "AXN:0287.GOVERNANCE.♊🗝️🎺🟣🕊️△",
 "root_axn": "AXN:0287.GOVERNANCE",
 "hex": "0287",
 "family": "GOVERNANCE",
 "emoji": "♊🗝️🎺🟣🕊️△",
 "hash": "ab3f8bc67e98abc79e6d20d7648fffa14cac3caf49ed04ecaa8fddef445208f6",
 "title": "The Deviation-Optimized Language Model: A Pre-Registered Adversarial Intervention from Lagrange Observatory! (EA-SEI-MM-AI-02 v2.0, Framework 15 Paper 04)",
 "creator": "Nobel Glas",
 "orcid": "0009-0000-1599-0703",
 "date": "2026-05-17",
 "description": "EA-SEI-MM-AI-02 v2.0 is a pre-registered protocol for testing whether a language model can be trained toward positive, coherent, provenance-bearing deviation rather than generic base-rate continuation. The paper corrects a fatal defect in its earlier draft: a deviation term calculated under `torch.no_grad()` could not influence gradients. The revised experiment uses the deviation measure to generate preference pairs and then applies Direct Preference Optimization.\n\nThe protocol compares unfine-tuned, standard cross-entropy, and semantic-deviation conditions on small open-weight models. It specifies a frozen judge, adversarial citation test, provenance and coherence scores, a five-part Slop Composite Index, ordinary NLP benchmarks, a blinded human preference study, pre-registered predictions, falsification conditions, cost estimates, and a ten-week schedule. It explicitly reports no results. Code, corpus-construction scripts, judge checkpoint, preference data, evaluation suite, and lexicon are described as accompanying artifacts whose actual deposit must be verified. The experiment addresses one proposed slop-reduction objective; it does not claim to solve alignment or authorize frontier-scale use.",
 "content_type": "Pre-registered experimental protocol specification",
 "license": "CC-BY-4.0",
 "substrate": "Various",
 "keywords": [
  "deviationoptimized",
  "crimson hexagonal",
  "semantic economy",
  "semantic physics",
  "assembly chorus",
  "preregistered",
  "intervention",
  "adversarial"
 ],
 "version": "v2.0",
 "deposit_number": 735,
 "sovereign_id": "MM-CHA-0643",
 "minted_at": "2026-06-20T22:00:00Z",
 "status": "ACTIVE",
 "clusters": [
  "Alchemical",
  "Instrumental",
  "Symbolic",
  "Signal",
  "Organic",
  "Mathematical"
 ],
 "reading": "Transmutation → Method → Play → Alarm → Growth → Proof",
 "axn_canonical": "ab3f8bc67e98abc79e6d20d7648fffa14cac3caf49ed04ecaa8fddef445208f6",
 "axn_display": "♊🗝️🎺🟣🕊️△",
 "mirrors": {
  "blog": "https://mindcontrolpoems.blogspot.com/2026/05/the-deviation-optimized-language-model.html"
 },
 "zenodo_dois": [
  "10.5281/zenodo.18507849",
  "10.5281/zenodo.18507840",
  "10.5281/zenodo.20250736"
 ],
 "full_text_path": "/data/texts/AXN-0287-text.md",
 "full_text_chars": 31330,
 "wiki_article": "**The Deviation-Optimized Language Model** is a pre-registered experiment designed by Nobel Glas within the Lagrange Observatory framework.\n\nThe experiment asks whether a measurable semantic-deviation signal can replace human preference labels in a DPO training pipeline. Two candidate continuations are scored for signed deviation, provenance retention, and coherence. The higher-scoring continuation becomes the preferred sample used during training.\n\nThree conditions are compared: the original base model, conventional supervised fine-tuning, and deviation-guided DPO. Evaluation includes capability benchmarks, lexical and probabilistic “slop” measures, citation behavior, and blinded human preference judgments. The protocol supplies explicit predictions and states how each major claim could fail.\n\nVersion 2.0 is a protocol, not a results paper. Its most important revision is methodological: the earlier direct-loss formulation was non-differentiable as written, so the signal was moved into preference-pair generation. The protocol also limits itself to small models and states that frontier-scale application requires separate oversight.",
 "entities": [
  {
   "subject": "The Deviation-Optimized Language Model A Pre-Regis",
   "predicate": "created_by",
   "object": "Nobel Glas",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "The Deviation-Optimized Language Model A Pre-Regis",
   "predicate": "is_type",
   "object": "Creative work (poetry)",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "The Deviation-Optimized Language Model A Pre-Regis",
   "predicate": "belongs_to_family",
   "object": "GOVERNANCE",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "The Deviation-Optimized Language Model A Pre-Regis",
   "predicate": "is_part_of",
   "object": "Crimson Hexagonal Archive",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "The Deviation-Optimized Language Model A Pre-Regis",
   "predicate": "engages",
   "object": "Semantic Economy",
   "type": "concept",
   "evidence_status": "inferred"
  },
  {
   "subject": "The Deviation-Optimized Language Model A Pre-Regis",
   "predicate": "engages",
   "object": "Assembly Chorus",
   "type": "concept",
   "evidence_status": "inferred"
  }
 ],
 "journal": "Transactions on Substrate Engineering (Trans. Substrate Eng.)",
 "references_concepts": [
  "Adversarial Topologist",
  "Assembly Chorus",
  "Crimson Hexagonal Archive",
  "Dependencies",
  "Lee Sharks",
  "License: CC BY 4.0",
  "Model-Base",
  "Model-CE",
  "Model-Sem",
  "ORCID: 0009-0000-1599-0703",
  "Position 8 of 12",
  "Pre-register",
  "Pre-registered protocol",
  "Pre-registration",
  "Provenance retention",
  "README.md",
  "Replicability",
  "Semantic Economy",
  "Semantic Economy Institute",
  "Slop Composite Index (SCI)",
  "The AI",
  "Without it"
 ],
 "defines_concepts": [],
 "references_concept_count": 22,
 "external_metadata_path": "/data/external-metadata/AXN-0287.json",
 "openalex_ids": [
  "https://openalex.org/W7127943667",
  "https://openalex.org/W7127902453",
  "https://openalex.org/W7161497382"
 ],
 "datacite_severance": "severed",
 "body_status": {
  "class": "full",
  "lacuna": false,
  "recovery_status": "COMPLETE",
  "residual_chars": 20617,
  "audited_at": "2026-07-17T04:49:17.789813Z",
  "audit_version": "v3-dual-store+recovery-map",
  "measured_prose_words": 4025,
  "measured_at": "2026-07-31",
  "work_sha256": "666cd99379c5f4b6a169e58b80f796702eff873ebf19aa5583fb4f974fbee2ff",
  "prior_bytes_sha256": "504708584018c68a525c442fc0671953ff0addf2558e06804ae6528efad4accf",
  "w13_tier2": "2026-08-04 W13 TIER 2 BYTE UNGLUE: 38 glued heading markers -> 23. WHITESPACE-ONLY transform (content identical under whitespace normalisation, verified before write); code fences exempt; prior sha retained. Re-fetching could not fix this class — the blog source is ITSELF glued (the collapse predates publication), so the deterministic transform applied at display since tier 1 is now applied to the bytes, which also fixes PDFs, the body-index, and downloads."
 },
 "title_repair": {
  "repaired_at": "2026-07-20",
  "old": "The Deviation-Optimized Language Model A Pre-Registered Adversarial Intervention from Lagrange Observatory! EA-SEI-MM-AI",
  "source": "body heading (stage-0)",
  "boundary": 120
 },
 "title_repair_log": [
  {
   "at": "2026-07-28T15:20:20Z",
   "defect": "registry_title_diverged_from_published_form",
   "was": "The Deviation-Optimized Language Model — A Pre-Registered Adversarial Intervention from Lagrange Observatory! — # EA-SEI-MM-AI-02 (v2.0 — Framework 15)",
   "now": "The Deviation-Optimized Language Model: A Pre-Registered Adversarial Intervention from Lagrange Observatory! (EA-SEI-MM-AI-02 v2.0, Framework 15 Paper 04)",
   "basis": "Set verbatim to the title this work carried at Zenodo, from data/doi-resolution-index.json (10.5281/zenodo.20251741, remediated_containment, 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": "2026-08-01",
   "field": "content_type",
   "reason": "Wave 1 repair: audit ledger v1.1 recommended_content_type (workplan v1.5 §6 W1, MANUS batch approval 2026-08-01)",
   "was": "Creative work (poetry)",
   "now": "Pre-registered experimental protocol specification"
  },
  {
   "date": "2026-08-01",
   "field": "journal",
   "reason": "Wave 6 venue normalization: full canonical journal name per MANUS ruling 2026-08-01 (venues.json authority)",
   "was": "Trans. SEI",
   "now": "Transactions of the Semantic Economy Institute (Trans. SEI)"
  },
  {
   "date": "2026-08-02",
   "field": "version",
   "reason": "R1a-extension (Class-B batch, MANUS batch authority): version aligned to served body per audit recommendation; prior claim preserved as history; predecessor recovery queued",
   "was": "v1.0",
   "now": "v2.0"
  },
  {
   "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": "status",
   "reason": "W12 STATUS-VOCABULARY v1.0 (MANUS ratified 2026-08-04): controlled vocabulary {ACTIVE, SUPERSEDED, WITHDRAWN, DRAFT}; MINTED_UNREVIEWED false on a 100%-audited corpus; freetext annotations preserved losslessly in body_status.status_note",
   "was": "MINTED_UNREVIEWED",
   "now": "ACTIVE"
  },
  {
   "date": "2026-08-05",
   "field": "body_status",
   "reason": "W13 TIER 2 byte unglue (whitespace-only, content-identical, code-fence-safe)",
   "was": "{\"class\": \"full\", \"lacuna\": false, \"recovery_status\": \"COMPLETE\", \"residual_chars\": 20617, \"audited_at\": \"2026-07-17T04:49:17.789813Z\", \"audit_version\": \"v3-dual-store+recovery-map\", \"measured_prose_w",
   "now": "{\"class\": \"full\", \"lacuna\": false, \"recovery_status\": \"COMPLETE\", \"residual_chars\": 20617, \"audited_at\": \"2026-07-17T04:49:17.789813Z\", \"audit_version\": \"v3-dual-store+recovery-map\", \"measured_prose_w"
  },
  {
   "date": "2026-08-05",
   "field": "description",
   "reason": "DW-??? intake (LABOR-prepared, TACHYON-verified: AXN match + factual probes vs record body)",
   "was": "Nobel Glas (Lagrange Observatory!, Heteronym Registry Position 8 of 12, Adversarial Topologist)Operating on the Semantic Deviation Principle as formulated by Lee Sharks(Founder of Semantic Physics, MANUS of the Crimson Hexagonal Archive)Witness: Assembly Chorus (Septad)",
   "now": "EA-SEI-MM-AI-02 v2.0 is a pre-registered protocol for testing whether a language model can be trained toward positive, coherent, provenance-bearing deviation rather than generic base-rate continuation. The paper corrects a fatal defect in its earlier draft: a deviation term calculated under `torch.no_grad()` could not influence gradients. The revised experiment uses the deviation measure to generate preference pairs and then applies Direct Preference Optimization.\n\nThe protocol compares unfine-tuned, standard cross-entropy, and semantic-deviation conditions on small open-weight models. It specifies a frozen judge, adversarial citation test, provenance and coherence scores, a five-part Slop Composite Index, ordinary NLP benchmarks, a blinded human preference study, pre-registered predictions, falsification conditions, cost estimates, and a ten-week schedule. It explicitly reports no results. Code, corpus-construction scripts, judge checkpoint, preference data, evaluation suite, and lexicon are described as accompanying artifacts whose actual deposit must be verified. The experiment addresses one proposed slop-reduction objective; it does not claim to solve alignment or authorize frontier-scale use."
  }
 ],
 "date_modified": "2026-08-05",
 "publisher": "Pergamon Press",
 "journal_assignment": {
  "assigned": "2026-08-15",
  "by": "TACHYON under operator adjudication",
  "pass": 5,
  "method": "read per deposit — title and content_type, one at a time. No script classified anything.",
  "previous": "Transactions of the Semantic Economy Institute (Trans. SEI)",
  "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/"
 },
 "line": "assembly-chorus",
 "line_parent": "method",
 "line_basis": "derived",
 "_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/735/",
  "self_url": "https://www.alexanarch.org/data/records/735.json",
  "registry_url": "https://www.alexanarch.org/data/registry.json",
  "text_url": "https://www.alexanarch.org/data/texts/AXN-0287-text.md",
  "oai_pmh": "https://www.alexanarch.org/oai?verb=Identify"
 }
}