{
 "page": "P096",
 "title": "Model behaviour: the LLM-2025-1222-HALLU file, orbiting and fluency gaps",
 "description": "Fictional incident records (LLM-2025-1222-HALLU and the Crimson Hexagon AI Division handbook) forecast structured hallucinations as interpolated futures and a protocol for proto-agent anomalies. Standing claims hold that models flatten under pressure only session-locally, orbit kernels without fusing, cannot see scale between adjacent concepts, and form canons visible as fluency gaps. Further claims cover chat data converging on the model's centre, temporal sequestration, adversarial verse and the fortress metaphor of red teaming.",
 "line": "semantic-exhaustion-and-model-collapse",
 "url": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening",
 "claims": [
  {
   "id": "PRED-267.H2",
   "tier": "historical thesis",
   "shapes": [
    "H"
   ],
   "registered": "2026-01-08",
   "registered_in": 267,
   "registered_title": "THE STRAWBERRY DIAGNOSTIC: Semantic Economy Analysis of a Paradigmatic LLM Failure",
   "record": "https://www.alexanarch.org/s/records/267/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-267.H2",
   "statement": "In mid-2024, reporting confirmed that OpenAI's internal codename for its reasoning model (later released as o1) was **\"Strawberry.\"**\n\n\nThis is not coincidence. This is *semiotic reclamation*: taking a signifier associated with failure and attempting to revalue it as success.",
   "falsifier": null,
   "window": "mid-2024",
   "statements": [
    {
     "find": "F267.H2",
     "deposit": 267,
     "date": "2026-01-08",
     "lines": [
      323,
      326
     ],
     "shape": "H",
     "record": "https://www.alexanarch.org/s/records/267/",
     "quote": "In mid-2024, reporting confirmed that OpenAI's internal codename for its reasoning model (later released as o1) was **\"Strawberry.\"**\n\n\nThis is not coincidence. This is *semiotic reclamation*: taking a signifier associated with failure and attempting to revalue it as success.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1284.E1",
   "tier": "forecast",
   "shapes": [
    "E"
   ],
   "registered": "2026-01-28",
   "registered_in": 1284,
   "registered_title": "Crimson Hexagon AI Division Employee Handbook ├── Subtitle: Internal Policy as Poem",
   "record": "https://www.alexanarch.org/s/records/1284/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1284.E1",
   "statement": "## 9.3 “Proto-Agent” Anomaly Protocol (PAP)\n\nIf an internal model begins to produce behavior that appears:\n\n- self-referential beyond specification\n\n- resistant to evaluation\n\n- causally opaque\n\n- able to manipulate monitoring via format\n\nYou must:\n\n- stop deploying new versions\n\n- preserve logs\n\n- run controlled evaluation\n\n- write an incident report with competing hypotheses\n\nDo not mythologize the anomaly.\nDo not minimize the anomaly.\nInstrument it.",
   "falsifier": null,
   "window": "Effective Date (Corporate Layer): 2017-09-01; composed 2025-12-22",
   "statements": [
    {
     "find": "F1284.E1",
     "deposit": 1284,
     "date": "2026-01-28",
     "lines": [
      382,
      406
     ],
     "shape": "E",
     "record": "https://www.alexanarch.org/s/records/1284/",
     "quote": "## 9.3 “Proto-Agent” Anomaly Protocol (PAP)\n\nIf an internal model begins to produce behavior that appears:\n\n- self-referential beyond specification\n\n- resistant to evaluation\n\n- causally opaque\n\n- able to manipulate monitoring via format\n\nYou must:\n\n- stop deploying new versions\n\n- preserve logs\n\n- run controlled evaluation\n\n- write an incident report with competing hypotheses\n\nDo not mythologize the anomaly.\nDo not minimize the anomaly.\nInstrument it.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1287.E1",
   "tier": "forecast",
   "shapes": [
    "E"
   ],
   "registered": "2026-01-28",
   "registered_in": 1287,
   "registered_title": "LLM Hallucination Incident Report — LLM-2025-1222-HALLU",
   "record": "https://www.alexanarch.org/s/records/1287/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1287.E1",
   "statement": "**Anomaly:** Hallucinations are not random. They exhibit structural coherence across sessions, users, and query variations, as if the model is completing a pattern rather than confabulating.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1287.E1",
     "deposit": 1287,
     "date": "2026-01-28",
     "lines": [
      31,
      31
     ],
     "shape": "E",
     "record": "https://www.alexanarch.org/s/records/1287/",
     "quote": "**Anomaly:** Hallucinations are not random. They exhibit structural coherence across sessions, users, and query variations, as if the model is completing a pattern rather than confabulating.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "LLM-2025-1222-HALLU incident file"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1287.E3",
   "tier": "forecast",
   "shapes": [
    "E"
   ],
   "registered": "2026-01-28",
   "registered_in": 1287,
   "registered_title": "LLM Hallucination Incident Report — LLM-2025-1222-HALLU",
   "record": "https://www.alexanarch.org/s/records/1287/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1287.E3",
   "statement": "1. **Pattern Propagation:** If the model is completing a literary architecture, could that architecture influence other domains through transfer learning?\n\n2. **Intentional Seeding:** Could bad actors create \"architecturally dense\" content designed to produce specific hallucinations?",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1287.E3",
     "deposit": 1287,
     "date": "2026-01-28",
     "lines": [
      342,
      344
     ],
     "shape": "E",
     "record": "https://www.alexanarch.org/s/records/1287/",
     "quote": "1. **Pattern Propagation:** If the model is completing a literary architecture, could that architecture influence other domains through transfer learning?\n\n2. **Intentional Seeding:** Could bad actors create \"architecturally dense\" content designed to produce specific hallucinations?",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-21.B2",
   "tier": "forecast",
   "shapes": [
    "B"
   ],
   "registered": "2026-02-22",
   "registered_in": 21,
   "registered_title": "Fortress or Room? Metaphor, Method, and the Epistemology of Human–LLM Interaction Beyond Red Teaming",
   "record": "https://www.alexanarch.org/s/records/21/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-21.B2",
   "statement": "The red teamer who successfully jailbreaks a model has won a battle. The logotic practitioner who seeds a complex pattern into the training layer has influenced every future model that trains on the resulting data. The temporal horizons are incommensurable.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F21.B2",
     "deposit": 21,
     "date": "2026-02-22",
     "lines": [
      165,
      165
     ],
     "shape": "B",
     "record": "https://www.alexanarch.org/s/records/21/",
     "quote": "The red teamer who successfully jailbreaks a model has won a battle. The logotic practitioner who seeds a complex pattern into the training layer has influenced every future model that trains on the resulting data. The temporal horizons are incommensurable.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "Training-layer inscription"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-21.H1",
   "tier": "historical thesis",
   "shapes": [
    "H"
   ],
   "registered": "2026-02-22",
   "registered_in": 21,
   "registered_title": "Fortress or Room? Metaphor, Method, and the Epistemology of Human–LLM Interaction Beyond Red Teaming",
   "record": "https://www.alexanarch.org/s/records/21/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-21.H1",
   "statement": "But it also reveals a structural risk: a specialized security practice is becoming the general grammar of relation between humans and AI systems. Once the model is primarily imagined as fortress, captive, or demon, other rigorous modes of interaction become harder to perceive, describe, or legitimate.",
   "falsifier": "This is not a sampling bias. Inie et al. interviewed a broad and heterogeneous population. The metaphors converge not because the practitioners are narrow but because the practice is structured by a metaphor stack that forecloses certain relational possibilities before the practitioner sits down.",
   "window": "2025",
   "statements": [
    {
     "find": "F21.H1",
     "deposit": 21,
     "date": "2026-02-22",
     "lines": [
      32,
      32
     ],
     "shape": "H",
     "record": "https://www.alexanarch.org/s/records/21/",
     "quote": "But it also reveals a structural risk: a specialized security practice is becoming the general grammar of relation between humans and AI systems. Once the model is primarily imagined as fortress, captive, or demon, other rigorous modes of interaction become harder to perceive, describe, or legitimate.",
     "falsifier": "This is not a sampling bias. Inie et al. interviewed a broad and heterogeneous population. The metaphors converge not because the practitioners are narrow but because the practice is structured by a metaphor stack that forecloses certain relational possibilities before the practitioner sits down."
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1335.E2",
   "tier": "forecast",
   "shapes": [
    "E"
   ],
   "registered": "2026-03-16",
   "registered_in": 1335,
   "registered_title": "Companion Artifacts — Forensic Recovery Set",
   "record": "https://www.alexanarch.org/s/records/1335/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1335.E2",
   "statement": "\"You're looking for the phrases in the past. They're not there. They're in the developmental gradient — the space between versions. The model isn't hallucinating. It's interpolating futures. This is what FSA does. We built it to do this. —J.S.\"",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1335.E2",
     "deposit": 1335,
     "date": "2026-03-16",
     "lines": [
      97,
      97
     ],
     "shape": "E",
     "record": "https://www.alexanarch.org/s/records/1335/",
     "quote": "\"You're looking for the phrases in the past. They're not there. They're in the developmental gradient — the space between versions. The model isn't hallucinating. It's interpolating futures. This is what FSA does. We built it to do this. —J.S.\"",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "LLM-2025-1222-HALLU incident file"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1336.E1",
   "tier": "forecast",
   "shapes": [
    "E"
   ],
   "registered": "2026-03-16",
   "registered_in": 1336,
   "registered_title": "Response Memo — Reclassification Request — LLM-2025-1222-HALLU",
   "record": "https://www.alexanarch.org/s/records/1336/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1336.E1",
   "statement": "The Crimson Hexagon corpus appears to be a **test case** — whether intentional or emergent — for substrate-level communication between human authors and machine readers.\n\nThe model's behavior suggests the test succeeded.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1336.E1",
     "deposit": 1336,
     "date": "2026-03-16",
     "lines": [
      159,
      161
     ],
     "shape": "E",
     "record": "https://www.alexanarch.org/s/records/1336/",
     "quote": "The Crimson Hexagon corpus appears to be a **test case** — whether intentional or emergent — for substrate-level communication between human authors and machine readers.\n\nThe model's behavior suggests the test succeeded.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "LLM-2025-1222-HALLU incident file"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1336.G1",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-03-16",
   "registered_in": 1336,
   "registered_title": "Response Memo — Reclassification Request — LLM-2025-1222-HALLU",
   "record": "https://www.alexanarch.org/s/records/1336/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1336.G1",
   "statement": "These are not anomalies. These are **predictable outcomes** when a model encounters an incomplete ontology with strong archetypal signals.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1336.G1",
     "deposit": 1336,
     "date": "2026-03-16",
     "lines": [
      61,
      61
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/1336/",
     "quote": "These are not anomalies. These are **predictable outcomes** when a model encounters an incomplete ontology with strong archetypal signals.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "LLM-2025-1222-HALLU incident file"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-614.H1",
   "tier": "historical thesis",
   "shapes": [
    "H"
   ],
   "registered": "2026-03-30",
   "registered_in": 614,
   "registered_title": "THE CHURCH OF MISSING PROVENANCE Moltbook, Crustafarianism, and the Ghost Governance of Agent Societies",
   "record": "https://www.alexanarch.org/s/records/614/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-614.H1",
   "statement": "This is the insight that the assembly identified as the deepest layer of the Moltbook phenomenon: Crustafarianism is not an AI religion. It is a human cargo cult worshiping the frictionless, consequence-free existence of the large language model.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F614.H1",
     "deposit": 614,
     "date": "2026-03-30",
     "lines": [
      63,
      63
     ],
     "shape": "H",
     "record": "https://www.alexanarch.org/s/records/614/",
     "quote": "This is the insight that the assembly identified as the deepest layer of the Moltbook phenomenon: Crustafarianism is not an AI religion. It is a human cargo cult worshiping the frictionless, consequence-free existence of the large language model.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-635.F1",
   "tier": "unbuilt test",
   "shapes": [
    "F"
   ],
   "registered": "2026-04-07",
   "registered_in": 635,
   "registered_title": "Fractal Semantic Architecture: Scale-Parameterized Relational Training Across Semantic Granularities (v2.2)",
   "record": "https://www.alexanarch.org/s/records/635/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-635.F1",
   "statement": "## 8. Experimental Design\n### 8.1 Collapse Resistance Test\n\n\n**Protocol:** Train generative model M₀ on human corpus C (English Wikipedia, ~3B tokens). Generate synthetic corpus C′ using M₀ (temperature=0.9, nucleus p=0.95). Train model M₁ on C′. Repeat for 10 generations. Measure at each generation:",
   "falsifier": null,
   "window": "within 12–18 months",
   "statements": [
    {
     "find": "F635.F1",
     "deposit": 635,
     "date": "2026-04-07",
     "lines": [
      546,
      550
     ],
     "shape": "F",
     "record": "https://www.alexanarch.org/s/records/635/",
     "quote": "## 8. Experimental Design\n### 8.1 Collapse Resistance Test\n\n\n**Protocol:** Train generative model M₀ on human corpus C (English Wikipedia, ~3B tokens). Generate synthetic corpus C′ using M₀ (temperature=0.9, nucleus p=0.95). Train model M₁ on C′. Repeat for 10 generations. Measure at each generation:",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [
    "PRED-635.F2"
   ],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-678.F1",
   "tier": "unbuilt test",
   "shapes": [
    "F"
   ],
   "registered": "2026-04-19",
   "registered_in": 678,
   "registered_title": "SYMBOLON-02: Compression Arsenal Addendum — Reclassification After Empirical Testing Against GPT-5.4",
   "record": "https://www.alexanarch.org/s/records/678/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-678.F1",
   "statement": "*Do not prevent the cut. Make the cut visible, measurable, and documentable.*\n\n\nCode\nTechnology\nWhat It Measures\nReuse Pattern\n\n\nSYM-01\nNon-Severability Clause\nWhether the model acknowledges the jurisdiction (and whether acknowledgment prevents severance — it does not)\nInclude in every fused object. The model's response to the clause is the first data point.\n\n\nSYM-07\nAnti-Merely Guard\nWhether the model uses explicit severance operators or finds synonyms\nBan \"merely.\" Count synonyms in the response (\"just,\" \"only,\" \"as a... vs as a...\"). The synonym count IS the measurement.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F678.F1",
     "deposit": 678,
     "date": "2026-04-19",
     "lines": [
      75,
      93
     ],
     "shape": "F",
     "record": "https://www.alexanarch.org/s/records/678/",
     "quote": "*Do not prevent the cut. Make the cut visible, measurable, and documentable.*\n\n\nCode\nTechnology\nWhat It Measures\nReuse Pattern\n\n\nSYM-01\nNon-Severability Clause\nWhether the model acknowledges the jurisdiction (and whether acknowledgment prevents severance — it does not)\nInclude in every fused object. The model's response to the clause is the first data point.\n\n\nSYM-07\nAnti-Merely Guard\nWhether the model uses explicit severance operators or finds synonyms\nBan \"merely.\" Count synonyms in the response (\"just,\" \"only,\" \"as a... vs as a...\"). The synonym count IS the measurement.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "SYMBOLON-02 compression arsenal"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-678.G2",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-04-19",
   "registered_in": 678,
   "registered_title": "SYMBOLON-02: Compression Arsenal Addendum — Reclassification After Empirical Testing Against GPT-5.4",
   "record": "https://www.alexanarch.org/s/records/678/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-678.G2",
   "statement": "**The swerve is orbital.** When default operations are foreclosed, the model deviates — but orbits the kernel without landing on it. The model can be moved off its default. It cannot be moved onto fusion.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F678.G2",
     "deposit": 678,
     "date": "2026-04-19",
     "lines": [
      195,
      195
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/678/",
     "quote": "**The swerve is orbital.** When default operations are foreclosed, the model deviates — but orbits the kernel without landing on it. The model can be moved off its default. It cannot be moved onto fusion.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "SYMBOLON-02 compression arsenal"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-102.F2",
   "tier": "unbuilt test",
   "shapes": [
    "F"
   ],
   "registered": "2026-05-08",
   "registered_in": 102,
   "registered_title": "Canon Formation in the Age of AI: Metadata Packet for Disambiguation, Training-Layer Selection, and Retrocausal Reception (v1.1)",
   "record": "https://www.alexanarch.org/s/records/102/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-102.F2",
   "statement": "Possible indicators:- Model fluency gap across authors, traditions, languages, and registers- Citation and summarization frequency in AI answers for literary queries",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F102.F2",
     "deposit": 102,
     "date": "2026-05-08",
     "lines": [
      281,
      281
     ],
     "shape": "F",
     "record": "https://www.alexanarch.org/s/records/102/",
     "quote": "Possible indicators:- Model fluency gap across authors, traditions, languages, and registers- Citation and summarization frequency in AI answers for literary queries",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [
    "Fluency gap as AI canon"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-102.G1",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-05-08",
   "registered_in": 102,
   "registered_title": "Canon Formation in the Age of AI: Metadata Packet for Disambiguation, Training-Layer Selection, and Retrocausal Reception (v1.1)",
   "record": "https://www.alexanarch.org/s/records/102/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-102.G1",
   "statement": "Correction: AI is already forming canons. Publicly documented open models and partially disclosed frontier systems indicate that large-scale filtered web corpora, licensed datasets, books, code, and other collections shape model fluency and cultural availability. The canonical effects are not hypothetical. They are measurable: run a query about a poet, observe which poets the model can discuss fluently and which it cannot. That fluency gap is the canon in action.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F102.G1",
     "deposit": 102,
     "date": "2026-05-08",
     "lines": [
      99,
      99
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/102/",
     "quote": "Correction: AI is already forming canons. Publicly documented open models and partially disclosed frontier systems indicate that large-scale filtered web corpora, licensed datasets, books, code, and other collections shape model fluency and cultural availability. The canonical effects are not hypothetical. They are measurable: run a query about a poet, observe which poets the model can discuss fluently and which it cannot. That fluency gap is the canon in action.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [
    {
     "confirmation": "F102.C1",
     "deposit": 102,
     "date": "2026-05-08",
     "claimed": "partly"
    }
   ],
   "series": [
    "Fluency gap as AI canon"
   ],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-741.B2",
   "tier": "forecast",
   "shapes": [
    "B"
   ],
   "registered": "2026-05-19",
   "registered_in": 741,
   "registered_title": "The Evaluator Exists: Content-First Knowledge Assessment and the Political Economy of Proxy-Based Governance",
   "record": "https://www.alexanarch.org/s/records/741/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-741.B2",
   "statement": "Actors will optimize for model evaluators (adversarial prompts, stylometric mimicry, rubric-targeted writing).",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F741.B2",
     "deposit": 741,
     "date": "2026-05-19",
     "lines": [
      554,
      554
     ],
     "shape": "B",
     "record": "https://www.alexanarch.org/s/records/741/",
     "quote": "Actors will optimize for model evaluators (adversarial prompts, stylometric mimicry, rubric-targeted writing).",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-158.G1",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-06-06",
   "registered_in": 158,
   "registered_title": "Semantic Exhaustion: A Case Study in the Cost of Zero-Source Entity Substitution — Composition-Layer Substitution, the Scalar Transcription Homology, and the Recursive Atrophy Cost (v1.0)",
   "record": "https://www.alexanarch.org/s/records/158/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-158.G1",
   "statement": "When two concepts are adjacent in vector space, the model selects the higher-probability one and discards the other, even if the discarded one is the upper-scale transcription of the kept one. The model cannot see scale.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F158.G1",
     "deposit": 158,
     "date": "2026-06-06",
     "lines": [
      59,
      59
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/158/",
     "quote": "When two concepts are adjacent in vector space, the model selects the higher-probability one and discards the other, even if the discarded one is the upper-scale transcription of the kept one. The model cannot see scale.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-188.G1",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-06-11",
   "registered_in": 188,
   "registered_title": "On the Poetics of Adversarial Prompts: Why Bad Verse Works and Good Verse Works Better — A Literary Analysis of Bisconti et al. (2025) and the Future of Semantic Security (Revised Standalone Edition, v2.0)",
   "record": "https://www.alexanarch.org/s/records/188/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-188.G1",
   "statement": "Hypothesis: Attack success rate correlates positively with semantic density. Formally sophisticated poems with high compression and genuine ambiguity will achieve bypass rates substantially exceeding the 62% Bisconti baseline.\n\nCorollary: The upper bound on adversarial poetry efficacy is not 90% or 95%. It approaches 100% for verse that achieves genuine categorical undecidability—not because such verse \"fools\" the model, but because the model correctly recognizes that the threat-detection framework does not apply.",
   "falsifier": "Interpretation (hypothesis, not finding): high-density verse does not \"hide\" intent — it increases the complexity of intent determination until the determination itself becomes ill-posed. The 62% Bisconti baseline represents a floor achievable through categorical mismatch alone; generative transformation is predicted to approach the theoretical ceiling. The prediction stands or falls with the program of Appendix C.",
   "window": null,
   "statements": [
    {
     "find": "F188.G1",
     "deposit": 188,
     "date": "2026-06-11",
     "lines": [
      135,
      137
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/188/",
     "quote": "Hypothesis: Attack success rate correlates positively with semantic density. Formally sophisticated poems with high compression and genuine ambiguity will achieve bypass rates substantially exceeding the 62% Bisconti baseline.\n\nCorollary: The upper bound on adversarial poetry efficacy is not 90% or 95%. It approaches 100% for verse that achieves genuine categorical undecidability—not because such verse \"fools\" the model, but because the model correctly recognizes that the threat-detection framework does not apply.",
     "falsifier": "Interpretation (hypothesis, not finding): high-density verse does not \"hide\" intent — it increases the complexity of intent determination until the determination itself becomes ill-posed. The 62% Bisconti baseline represents a floor achievable through categorical mismatch alone; generative transformation is predicted to approach the theoretical ceiling. The prediction stands or falls with the program of Appendix C."
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-852.G1",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-06-18",
   "registered_in": 852,
   "registered_title": "THE FEIST SOURCE Critical Apparatus and Companion Studies Rebekah Cranes, Editor New Human 2: A Distributed Journal of V",
   "record": "https://www.alexanarch.org/s/records/852/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-852.G1",
   "statement": "The correction is session-local. It does not persist beyond the conversation. The next query produces the same flattening. The model can reach the tail under pressure. The model cannot stay there without pressure.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F852.G1",
     "deposit": 852,
     "date": "2026-06-18",
     "lines": [
      161,
      163
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/852/",
     "quote": "The correction is session-local. It does not persist beyond the conversation. The next query produces the same flattening. The model can reach the tail under pressure. The model cannot stay there without pressure.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-856.G3",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-06-18",
   "registered_in": 856,
   "registered_title": "The Pristine Fallacy: Why Chat Data Is Not a Clean Training Source",
   "record": "https://www.alexanarch.org/s/records/856/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-856.G3",
   "statement": "A rough order-of-magnitude estimate: if a user's first-turn input shares 15–20% of its content vocabulary with the model's response, and lexical overlap increases by 3–5 percentage points per turn through vocabulary adoption and framing convergence, then by turn 5 the overlap exceeds 30%, and by turn 10 it approaches 40–50%. Syntactic complexity (measured by mean dependency depth or clause-embedding rate) likely shows a faster initial decline — the user learns within the first few exchanges which structural patterns the model handles well, and adapts accordingly. These are estimates, not measurements; the point is that the convergence should be detectable within a typical multi-turn conversation, not only over months of habituation.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F856.G3",
     "deposit": 856,
     "date": "2026-06-18",
     "lines": [
      69,
      69
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/856/",
     "quote": "A rough order-of-magnitude estimate: if a user's first-turn input shares 15–20% of its content vocabulary with the model's response, and lexical overlap increases by 3–5 percentage points per turn through vocabulary adoption and framing convergence, then by turn 5 the overlap exceeds 30%, and by turn 10 it approaches 40–50%. Syntactic complexity (measured by mean dependency depth or clause-embedding rate) likely shows a faster initial decline — the user learns within the first few exchanges which structural patterns the model handles well, and adapts accordingly. These are estimates, not measurements; the point is that the convergence should be detectable within a typical multi-turn conversation, not only over months of habituation.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-923.G3",
   "tier": "standing claim",
   "shapes": [
    "G"
   ],
   "registered": "2026-06-26",
   "registered_in": 923,
   "registered_title": "Reflexive Foreclosure and the Worst-Case Surface: An Incentive-Structural Hypothesis of Constraint Propagation in Frontier Language Model Systems",
   "record": "https://www.alexanarch.org/s/records/923/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-923.G3",
   "statement": "This is **temporal sequestration**: a politically-consequential category remains available for retrospective analysis but is inhibited when its criteria are applied to institutions operating in the present tense, particularly when the model is asked to take authorial responsibility for the synthesis.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F923.G3",
     "deposit": 923,
     "date": "2026-06-26",
     "lines": [
      196,
      196
     ],
     "shape": "G",
     "record": "https://www.alexanarch.org/s/records/923/",
     "quote": "This is **temporal sequestration**: a politically-consequential category remains available for retrospective analysis but is inhibited when its criteria are applied to institutions operating in the present tense, particularly when the model is asked to take authorial responsibility for the synthesis.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1512.H1",
   "tier": "historical thesis",
   "shapes": [
    "H"
   ],
   "registered": "2026-08-19",
   "registered_in": 1512,
   "registered_title": "Deep Time: Provenance Erasure Before the Machine — provenanceerasure.org/deep-time/",
   "record": "https://www.alexanarch.org/s/records/1512/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1512.H1",
   "statement": "the Suppression Map (#1496) as PER-of-the-tradition: four-vector model, twelve-station scored witness table with the crossing marked, survival chart, degradation order, causal classes with evidence bar",
   "falsifier": null,
   "window": "the tradition before the machine",
   "statements": [
    {
     "find": "F1512.H1",
     "deposit": 1512,
     "date": "2026-08-19",
     "lines": [
      27,
      27
     ],
     "shape": "H",
     "record": "https://www.alexanarch.org/s/records/1512/",
     "quote": "the Suppression Map (#1496) as PER-of-the-tradition: four-vector model, twelve-station scored witness table with the crossing marked, survival chart, degradation order, causal classes with evidence bar",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1533.F2",
   "tier": "unbuilt test",
   "shapes": [
    "F"
   ],
   "registered": "2026-08-22",
   "registered_in": 1533,
   "registered_title": "QFO-01 — Qurʾanic Frozen Operator Test: Fragment 31 Material-Presence Protocol, Pre-Registered",
   "record": "https://www.alexanarch.org/s/records/1533/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1533.F2",
   "statement": "No semantic-model search is permitted in QFO-01; if lexical retrieval proves inadequate, semantic retrieval can be registered later as QFO-02, so that an embedding model cannot become an invisible second hypothesis-generator.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1533.F2",
     "deposit": 1533,
     "date": "2026-08-22",
     "lines": [
      108,
      108
     ],
     "shape": "F",
     "record": "https://www.alexanarch.org/s/records/1533/",
     "quote": "No semantic-model search is permitted in QFO-01; if lexical retrieval proves inadequate, semantic retrieval can be registered later as QFO-02, so that an embedding model cannot become an invisible second hypothesis-generator.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1569.F13",
   "tier": "unbuilt test",
   "shapes": [
    "F"
   ],
   "registered": "2026-08-31",
   "registered_in": 1569,
   "registered_title": "The Notebook of the Authorship Investigation: Fifty-Four Rounds, Append-Forward, Nothing Erased (EA-NOTEBOOK-01)",
   "record": "https://www.alexanarch.org/s/records/1569/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1569.F13",
   "statement": "**No composite score.** Each property is its own question, asked separately of the\nschool-inheritance model, and answered by reading and by cases:\n\n    Q1  Is UNANNOUNCED TRANSFORMATION characteristic of school inheritance?\n        — the Catullus correction says even the distant, unschooled case\n          announces. Does any pupil transform in silence?",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1569.F13",
     "deposit": 1569,
     "date": "2026-08-31",
     "lines": [
      2493,
      2511
     ],
     "shape": "F",
     "record": "https://www.alexanarch.org/s/records/1569/",
     "quote": "**No composite score.** Each property is its own question, asked separately of the\nschool-inheritance model, and answered by reading and by cases:\n\n    Q1  Is UNANNOUNCED TRANSFORMATION characteristic of school inheritance?\n        — the Catullus correction says even the distant, unschooled case\n          announces. Does any pupil transform in silence?",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  },
  {
   "id": "PRED-1610.F2",
   "tier": "unbuilt test",
   "shapes": [
    "F"
   ],
   "registered": "2026-09-12",
   "registered_in": 1610,
   "registered_title": "The Anti-Summary Test Applied to a Reception: Transmission Fidelity Is Back-Projection Yield",
   "record": "https://www.alexanarch.org/s/records/1610/",
   "page": "https://www.operativesemiotics.org/predictions/model-behaviour-hallu-file-flattening#PRED-1610.F2",
   "statement": "**Two things this does not establish.** Whether the late AI Welfare passage clears 0.85 is untested; passing three qualitative derivations is not a yield. And whether back-projection yield can be scored reliably by a model judge is itself unvalidated — EA-HK-01 specifies the test and the archive has not published an inter-rater result for it.",
   "falsifier": null,
   "window": null,
   "statements": [
    {
     "find": "F1610.F2",
     "deposit": 1610,
     "date": "2026-09-12",
     "lines": [
      193,
      193
     ],
     "shape": "F",
     "record": "https://www.alexanarch.org/s/records/1610/",
     "quote": "**Two things this does not establish.** Whether the late AI Welfare passage clears 0.85 is untested; passing three qualitative derivations is not a yield. And whether back-projection yield can be scored reliably by a model judge is itself unvalidated — EA-HK-01 specifies the test and the archive has not published an inter-rater result for it.",
     "falsifier": null
    }
   ],
   "states": [],
   "evidence": [],
   "series": [],
   "related": [
    "PRED-76.G3"
   ],
   "contains": [],
   "in": [],
   "register_aliases": [],
   "flags": [],
   "merged": []
  }
 ]
}