{
 "axn": "AXN:0335.COMPOSITIONAL.🀄💡⚓🌹🌈🪧",
 "hex": "0335",
 "family": "COMPOSITIONAL",
 "emoji": "🀄💡⚓🌹🌈🪧",
 "hash": "8f3c2c731e4dd62bff552e509be72f1edf5c5c93677b29ad9004509e3f43b282",
 "title": "The Threat Model Is Backwards: On Classifying High-Perplexity Text as a Security Threat in an Era of Model Collapse",
 "creator": "Lee Sharks (primary), with Nobel Glas and Talos Morrow",
 "orcid": "0009-0000-1599-0703",
 "date": "2026-06-11",
 "description": "A v1.1 structural and normative companion critique by Lee Sharks, with Nobel Glas and Talos Morrow. It grants the empirical framing of *AI Bleeding* for argument’s sake and analyzes the proposed mitigation: pre-inference rejection of languages deemed unexpected for a deployment.\n\nThe paper argues that a detector rejecting high-perplexity, low-coverage input functions as an **input-layer tail-pruning instrument**, selecting against the same rare-language variation that model-collapse literature treats as especially vulnerable and valuable. It distinguishes direct service exclusion from the further, conditional claim that widespread filtering may shape inference logs and future training distributions. The proposed alternative is content-neutral cost control plus language-aware routing and preserved demand records. Findings about the reviewed paper and the wider literature remain claims of this deposited critique and its cited sources.",
 "content_type": "Theoretical paper",
 "license": "CC-BY-4.0",
 "substrate": "AI-assisted (substrate)",
 "keywords": [
  "theoretical paper"
 ],
 "version": "v1.1",
 "status": "ACTIVE",
 "clusters": [
  "Symbolic",
  "Instrumental",
  "Architectural",
  "Organic",
  "Elemental",
  "Scriptural"
 ],
 "reading": "Play → Method → Foundation → Growth → Force → Text",
 "full_text_path": "/data/texts/AXN-0335-text.md",
 "mirrors": {
  "blog": "https://mindcontrolpoems.blogspot.com/2026/06/the-threat-model-is-backwards-on.html",
  "machinemediation": "https://machinemediation.org/registry/#search=MM-CHA-0817"
 },
 "wiki_article": "The paper’s argument proceeds in two levels.\n\n**Structural claim**\n\nIf a filter:\n\n- identifies high-perplexity or low-coverage language;\n- rejects it before inference;\n\nthen it is, by function, pruning the linguistic tail from the input path.\n\n**Coupling hypothesis**\n\nInference and training are distinct layers, but the paper proposes that they can couple through:\n\n- retained interaction logs;\n- standardized procurement patterns;\n- population-scale deployment;\n- mediated feedback into later model updates.\n\nThe paper contrasts two frames for the same properties:\n\n- high perplexity as attack signal;\n- high perplexity as rare, informative tail data.\n\nIt argues that the reviewed mitigation turns model-relative distance from the training distribution into a security category.\n\nThe alternative controls proposed are:\n\n- rate limiting;\n- token and output budgets;\n- language-aware routing;\n- content-neutral resource controls;\n- preservation of tail-language demand in logs.",
 "wiki_status": "provisional",
 "entities": [
  {
   "subject": "The Threat Model Is Backwards",
   "predicate": "created_by",
   "object": "Lee Sharks",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "The Threat Model Is Backwards",
   "predicate": "is_type",
   "object": "Theoretical paper",
   "type": "classification",
   "evidence_status": "observed"
  },
  {
   "subject": "The Threat Model Is Backwards",
   "predicate": "belongs_to_family",
   "object": "COMPOSITIONAL",
   "type": "classification",
   "evidence_status": "observed"
  },
  {
   "subject": "The Threat Model Is Backwards",
   "predicate": "is_part_of",
   "object": "Crimson Hexagonal Archive",
   "type": "institution",
   "evidence_status": "observed"
  }
 ],
 "entity_status": "provisional",
 "sovereign_id": "MM-CHA-0817",
 "word_count": 3586,
 "download_md": "/data/deposits/AXN-0335.md",
 "zenodo_dois": [
  "10.5281/zenodo.20644761",
  "10.5281/zenodo.20644757",
  "10.5281/zenodo.20616418",
  "10.5281/zenodo.20644767",
  "10.5281/zenodo.20644769",
  "10.5281/zenodo.20644765",
  "10.5281/zenodo.20518338"
 ],
 "citations": [
  {
   "title": "Zenodo record 20518338",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20518338",
   "url": "https://doi.org/10.5281/zenodo.20518338",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20616418",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20616418",
   "url": "https://doi.org/10.5281/zenodo.20616418",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20644757",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20644757",
   "url": "https://doi.org/10.5281/zenodo.20644757",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20644761",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20644761",
   "url": "https://doi.org/10.5281/zenodo.20644761",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20644765",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20644765",
   "url": "https://doi.org/10.5281/zenodo.20644765",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20644767",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20644767",
   "url": "https://doi.org/10.5281/zenodo.20644767",
   "role": "Cross-referenced work"
  },
  {
   "title": "Zenodo record 20644769",
   "authors": [
    "Crimson Hexagonal Archive"
   ],
   "year": "2026",
   "doi": "10.5281/zenodo.20644769",
   "url": "https://doi.org/10.5281/zenodo.20644769",
   "role": "Cross-referenced work"
  }
 ],
 "citation_stats": {
  "total_doi_refs": 7
 },
 "deposit_number": 191,
 "journal": "Transactions on Substrate Engineering (Trans. Substrate Eng.)",
 "defines_concepts": [],
 "references_concepts": [
  "Claim types",
  "Companion to",
  "Crimson Hexagonal Archive",
  "Google AI Overview",
  "Lee Sharks",
  "MANUS ruling",
  "Semantic Exhaustion",
  "Substrate"
 ],
 "references_concept_count": 8,
 "external_metadata_path": "/data/external-metadata/AXN-0335.json",
 "openalex_ids": [
  "https://openalex.org/W7164419734",
  "https://openalex.org/W7164412748",
  "https://openalex.org/W7163987438",
  "https://openalex.org/W7164357053",
  "https://openalex.org/W7164364893",
  "https://openalex.org/W7164355019",
  "https://openalex.org/W7163337742"
 ],
 "datacite_severance": "severed",
 "body_status": {
  "class": "full",
  "lacuna": false,
  "recovery_status": "COMPLETE",
  "residual_chars": 24829,
  "audited_at": "2026-07-17T04:49:17.789813Z",
  "audit_version": "v3-dual-store+recovery-map",
  "measured_prose_words": 3549,
  "measured_at": "2026-07-31",
  "work_sha256": "c14ff9d43471674482e69e63a8f7a16719e5ab9dd1fa2d36136a014b87097981",
  "prior_bytes_sha256": "ad4beb6c42df39d84d9225df6a1a28f4ed455708f408b26b577073761a1f2791",
  "w13_tier2": "2026-08-04 W13 TIER 2 BYTE UNGLUE: 12 glued heading markers -> 0. 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_log": [
  {
   "at": "2026-07-28T15:09:45Z",
   "defect": "title_did_not_read_as_a_title",
   "was": "The Threat Model Is Backwards: On Classifying High-Perplexity Text as a Security Threat in an Era of Model Collapse The AI_Bleeding Mitigation as an Input-Layer Tail-Pruning Instrument Companion to: A",
   "now": "The Threat Model Is Backwards: On Classifying High-Perplexity Text as a Security Threat in an Era of Model Collapse",
   "basis": "Read; corrected against the deposit body headings."
  }
 ],
 "canonical_text_status": "canonical_full_text",
 "modifications": [
  {
   "date": "2026-08-03",
   "field": "creator",
   "reason": "Wave C-0358 role projection (MANUS ruling: project declared roles): creator set to the body's observed byline per sealed row",
   "was": "Lee Sharks",
   "now": "Lee Sharks (primary), with Nobel Glas and Talos Morrow"
  },
  {
   "date": "2026-08-04",
   "field": "journal",
   "reason": "W6-COMPLETE venue normalization (deferred #1-#358 half; corpus fully audited; MANUS ruling 2026-08-01)",
   "was": "MMRS",
   "now": "Machine-Mediated Reception Studies (MMRS)"
  },
  {
   "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": "version",
   "reason": "AUDIT-DIRECTED VERSION REPAIR: version projected per the audit's explicit recommendation / observed body declaration",
   "was": "v1.0",
   "now": "v1.1"
  },
  {
   "date": "2026-08-05",
   "field": "description",
   "reason": "DW-016 intake (LABOR-prepared, TACHYON-verified: AXN match + factual probes vs record body)",
   "was": "\"The Threat Model Is Backwards\" is a theoretical paper by Lee Sharks in the Crimson Hexagonal Archive (2026-06-11). Mitigation as an Input-Layer Tail-Pruning Instrument. The work comprises 3,586 words and is classified under the COMPOSITIONAL family. The work was removed from Zenodo on June 19, 2026",
   "now": "A v1.1 structural and normative companion critique by Lee Sharks, with Nobel Glas and Talos Morrow. It grants the empirical framing of *AI Bleeding* for argument’s sake and analyzes the proposed mitigation: pre-inference rejection of languages deemed unexpected for a deployment.\n\nThe paper argues that a detector rejecting high-perplexity, low-coverage input functions as an **input-layer tail-pruning instrument**, selecting against the same rare-language variation that model-collapse literature treats as especially vulnerable and valuable. It distinguishes direct service exclusion from the further, conditional claim that widespread filtering may shape inference logs and future training distributions. The proposed alternative is content-neutral cost control plus language-aware routing and preserved demand records. Findings about the reviewed paper and the wider literature remain claims of this deposited critique and its cited sources."
  },
  {
   "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\": 24829, \"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\": 24829, \"audited_at\": \"2026-07-17T04:49:17.789813Z\", \"audit_version\": \"v3-dual-store+recovery-map\", \"measured_prose_w"
  }
 ],
 "date_modified": "2026-08-05",
 "publisher": "Pergamon Press",
 "version_basis": {
  "ruling": "AUDIT: project v1.1",
  "prior_registry_claim": "v1.0",
  "observed_body_version": "v1.1",
  "directive": "Project v1.1 and the full authorship ruling: Lee Sharks primary, with Nobel Glas and Talos Morrow.",
  "audit_date": "2026-08-02"
 },
 "journal_assignment": {
  "assigned": "2026-08-15",
  "by": "TACHYON under operator adjudication",
  "pass": 2,
  "method": "read per deposit — title and content_type, one at a time. No script classified anything.",
  "previous": "Machine-Mediated Reception Studies (MMRS)",
  "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/"
 },
 "_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/191/",
  "self_url": "https://www.alexanarch.org/data/records/191.json",
  "registry_url": "https://www.alexanarch.org/data/registry.json",
  "text_url": "https://www.alexanarch.org/data/texts/AXN-0335-text.md",
  "oai_pmh": "https://www.alexanarch.org/oai?verb=Identify"
 }
}