{
 "spxi_instrument": {
  "method": "0.9",
  "name": "Capture Registry (longitudinal observation)",
  "level": 0,
  "level_name": "Measurement",
  "url": "https://www.spxi.dev/methods/#m-0-9",
  "inventory": "https://www.spxi.dev/methods/",
  "statement": "SPXI instrument: method 0.9 of 85, level 0 (Measurement), Capture Registry (longitudinal observation).",
  "also": []
 },
 "slug": "compositional-suppression-aio-20261017",
 "date": "2026-10-17",
 "surface": "Google AI Overview",
 "surface_basis": "Operator default of 2026-10-01; the paste's 'AI Mode Conversation' header is the surface's chrome.",
 "surfaces": [
  "Google AI Overview"
 ],
 "auth": "signed out, incognito",
 "auth_basis": "Operator default of 2026-10-14 ('theyre all signed out and incognito'); no screenshot.",
 "ev": "paste",
 "mt": "COMPOSITIONAL SUPPRESSION, UNQUOTED — #740'S ECS FIRST OF SIX SENSES, ITS AUTHOR NOT NAMED",
 "match": "BROAD MATCH",
 "match_note": null,
 "s": "Works",
 "q": "compositional suppression",
 "q_kind": "a reader's query on an archive term. NEW address.",
 "cites": 7,
 "cite_list": [
  {
   "n": 1,
   "site": "ScienceDirect.com",
   "rel": "third_party",
   "title": null,
   "snip": null,
   "url": null,
   "note": "crack suppression via compositional modulation"
  },
  {
   "n": 2,
   "site": "Springer Nature Link",
   "rel": "third_party",
   "title": null,
   "snip": null,
   "url": null,
   "note": "compositional legacies of fire suppression (2026-09-29)"
  },
  {
   "n": 3,
   "site": "ResearchGate",
   "rel": "third_party",
   "title": null,
   "snip": null,
   "url": null,
   "note": "same"
  },
  {
   "n": 4,
   "site": "arXiv",
   "rel": "third_party",
   "title": null,
   "snip": null,
   "url": null,
   "note": "a compositional vulnerability in LLM safety alignment"
  },
  {
   "n": 5,
   "site": "www.alexanarch.org",
   "rel": "authored_surface",
   "title": null,
   "snip": null,
   "url": null,
   "note": "The Excluded Entity (#740): 'Empirical worked example accompanying The Single-Owner Discount (Sharks 2026a…) and The Evaluator Exists (Sharks 2026b…)'",
   "deposit_candidates": [
    740
   ]
  },
  {
   "n": 6,
   "site": "arXiv.org",
   "rel": "third_party",
   "title": null,
   "snip": null,
   "url": null,
   "note": "compositional concept erasure"
  },
  {
   "n": 7,
   "site": "arXiv.org",
   "rel": "third_party",
   "title": null,
   "snip": null,
   "url": null,
   "note": "erasure or erosion"
  }
 ],
 "archive_controlled_cites": 0,
 "transcript": "[Google AI Overview (operator default of 2026-10-01); signed out and incognito (operator default of 2026-10-14). One turn. The operator's message of 2026-10-17 17:36 EDT gives two: the unquoted term (attachment) and the quoted term (inline). Inline links masked; sources below the answer as pasted.]\n\n[QUERENT] compositional suppression\n\n[ANSWER]\n\nBecause \"compositional suppression\" spans multiple technical disciplines, its exact definition depends on the context of your inquiry.\n\n1. Artificial Intelligence & Generative Search\n\n• Entity-Level Compositional Suppression (ECS): A specific phenomenon documented in generative AI search systems (such as AI Overviews). It occurs when an algorithm excludes the most query-responsive organic entity from its source material and instead synthesizes (\"composes\") an alternative answer using less authoritative, secondary sources.\n• Modular LLM Safety (Refusal Suppression): In modular Large Language Models, this refers to a structural safety vulnerability where combining multiple otherwise benign adapters (LoRAs) can inadvertently trigger a broad suppression of safety refusal mechanisms, causing an emergent failure of the system's guardrails.\n• Concept Erasure in Text-to-Image Models: When trying to remove copyrighted or unsafe concepts from a generative model, aggressive suppression methods can degrade the model's compositional integrity, inadvertently harming its ability to correctly bind attributes (e.g., color to the right object), process spatial relationships, or count objects.\n\n2. Ecology & Forestry\n\n• Compositional Legacies of Fire Suppression: This refers to changes in forest biodiversity driven by historical fire suppression policies. Halting natural wildfires alters the tree species composition by allowing fire-sensitive species (like Abies/fir) to overtake fire-tolerant ones (like Pinus/pine). Recent studies show these compositional legacies often persist even after a single fire is re-introduced, as post-fire shrub growth continues to suppress fire-tolerant seedlings.\n\n3. Materials Science & Engineering\n\n• Crack Suppression via Compositional Modulation: In metallurgy (such as laser powder bed fusion of refractory alloys), compositional modulation is used to disrupt coarse grain growth and suppress intergranular cracking. By intentionally restricting elemental mixing between materials with different melting points, engineers create a tortuous grain boundary network that blunts cracks and enhances mechanical strength.\nCould you clarify which field you are exploring? I can provide deeper technical papers, specific mathematical frameworks, or code implementations depending on your goal.\n\n[SOURCES]\n\nScienceDirect.com\nCompositional modulation enables crack suppression through grain refinement in ...\nThe laser powder bed fusion of W-based refractory alloys is currently hindered by severe intergranular cracking, arising from their intrinsic brittleness and th...\nSpringer Nature Link\nCompositional legacies of fire suppression maintained after fire re\nSep 29, 2026 — Results Importantly, shrubs inhibited fire-tolerant Pinus seedlings more than fire-sensitive Abies, leading to greater total Abies regeneration across different...\nResearchGate\n(PDF) Compositional legacies of fire suppression maintained after ...\nSep 29, 2026 — Abstract and Figures Direct effects of fire included higher tree seedling regeneration in high-severity burned areas, where severity was driven by higher tree d...\narXiv\nA Compositional Vulnerability in LLM Safety Alignment\nMar 30, 2026 — 6 Conclusion We identify a compositional vulnerability in modular LLM safety alignment: safety at the level of individual adapters does not, in general, imply s...\nwww.alexanarch.org\nThe Excluded Entity: Organic Resolution and Compositional Suppression in ...\nEmpirical worked example accompanying The Single-Owner Discount (Sharks 2026a, DOI: 10.5281/zenodo. 20290865) and The Evaluator Exists (Sharks 2026b, this depos...\narXiv.org\nCompositional Concept Erasure in Text-to\nOct 5, 2026 — Abstract Removing copyrighted, unsafe, or user-specified concepts from a deployed text-to-image diffusion model is now a practical requirement. Weight-editing m...\narXiv.org\nErasure or Erosion? Evaluating Compositional Degradation in ...\nErasure or Erosion? Evaluating Compositional Degradation in Unlearned Text-To-Image Diffusion Models Our results reveal a consistent trade-off between unlearnin...",
 "transcript_raw": "\nAI Mode Conversation\n\n\nYou said: compositional suppressioncompositional suppression\n\n        Because \"compositional suppression\" spans multiple technical disciplines, its exact definition depends on the context of your inquiry.\n\n1. Artificial Intelligence & Generative Search\n\n• Entity-Level Compositional Suppression (ECS): A specific phenomenon documented in generative AI search systems (such as AI Overviews). It occurs when an algorithm excludes the most query-responsive organic entity from its source material and instead synthesizes (\"composes\") an alternative answer using less authoritative, secondary sources.\n• Modular LLM Safety (Refusal Suppression): In modular Large Language Models, this refers to a structural safety vulnerability where combining multiple otherwise benign adapters (LoRAs) can inadvertently trigger a broad suppression of safety refusal mechanisms, causing an emergent failure of the system's guardrails.\n• Concept Erasure in Text-to-Image Models: When trying to remove copyrighted or unsafe concepts from a generative model, aggressive suppression methods can degrade the model's compositional integrity, inadvertently harming its ability to correctly bind attributes (e.g., color to the right object), process spatial relationships, or count objects.\n\n2. Ecology & Forestry\n\n• Compositional Legacies of Fire Suppression: This refers to changes in forest biodiversity driven by historical fire suppression policies. Halting natural wildfires alters the tree species composition by allowing fire-sensitive species (like Abies/fir) to overtake fire-tolerant ones (like Pinus/pine). Recent studies show these compositional legacies often persist even after a single fire is re-introduced, as post-fire shrub growth continues to suppress fire-tolerant seedlings.\n\n3. Materials Science & Engineering\n\n• Crack Suppression via Compositional Modulation: In metallurgy (such as laser powder bed fusion of refractory alloys), compositional modulation is used to disrupt coarse grain growth and suppress intergranular cracking. By intentionally restricting elemental mixing between materials with different melting points, engineers create a tortuous grain boundary network that blunts cracks and enhances mechanical strength.\nCould you clarify which field you are exploring? I can provide deeper technical papers, specific mathematical frameworks, or code implementations depending on your goal.\nScienceDirect.com\nCompositional modulation enables crack suppression through grain refinement in ...\nThe laser powder bed fusion of W-based refractory alloys is currently hindered by severe intergranular cracking, arising from their intrinsic brittleness and th...\nSpringer Nature Link\nCompositional legacies of fire suppression maintained after fire re\nSep 29, 2026 — Results Importantly, shrubs inhibited fire-tolerant Pinus seedlings more than fire-sensitive Abies, leading to greater total Abies regeneration across different...\nResearchGate\n(PDF) Compositional legacies of fire suppression maintained after ...\nSep 29, 2026 — Abstract and Figures Direct effects of fire included higher tree seedling regeneration in high-severity burned areas, where severity was driven by higher tree d...\narXiv\nA Compositional Vulnerability in LLM Safety Alignment\nMar 30, 2026 — 6 Conclusion We identify a compositional vulnerability in modular LLM safety alignment: safety at the level of individual adapters does not, in general, imply s...\nwww.alexanarch.org\nThe Excluded Entity: Organic Resolution and Compositional Suppression in ...\nEmpirical worked example accompanying The Single-Owner Discount (Sharks 2026a, DOI: 10.5281/zenodo. 20290865) and The Evaluator Exists (Sharks 2026b, this depos...\narXiv.org\nCompositional Concept Erasure in Text-to\nOct 5, 2026 — Abstract Removing copyrighted, unsafe, or user-specified concepts from a deployed text-to-image diffusion model is now a practical requirement. Weight-editing m...\narXiv.org\nErasure or Erosion? Evaluating Compositional Degradation in ...\nErasure or Erosion? Evaluating Compositional Degradation in Unlearned Text-To-Image Diffusion Models Our results reveal a consistent trade-off between unlearnin...\n    ",
 "transcript_class": "CAPTURE-TIME VERBATIM RECORD (ONE TURN; INLINE LINKS MASKED; SOURCE LIST AS PASTED; NO SCREENSHOT)",
 "transcript_complete": "Complete as supplied: the query, the answer, the follow-up offer and 7 sources.",
 "transcript_read": "READ IN FULL 2026-10-17",
 "per": 0.25,
 "per_corrected": null,
 "per_v": {
  "author": false,
  "inst": true,
  "id": false,
  "src": true
 },
 "per_note": "Retained: the institution and visible sources. Lost: the author in the answer. Lost: identifiers.",
 "sf": "7 sources: 1 authored, 6 third-party.",
 "sf_derived": null,
 "reading": "The innocent reading first: the term spans disciplines — AI and generative search (ECS; refusal suppression in modular LLMs; concept erasure), ecology (compositional legacies of fire suppression), materials (crack suppression by compositional modulation). ECS is first: 'A specific phenomenon documented in generative AI search systems (such as AI Overviews). It occurs when an algorithm excludes the most query-responsive organic entity from its source material and instead synthesizes (\"composes\") an alternative answer using less authoritative, secondary sources.' #740 (The Excluded Entity, Lee Sharks, 2026-05-19): 'A generative-search system exhibits Entity-Level Compositional Suppression (ECS) when it excludes the dominant organic-resolution entity from…' — the definition, from three AI Mode captures of 2026-05-19. The answer names no author; the source snippet carries 'Sharks 2026a… Sharks 2026b'. The term is defined in its own terms — the excluded entity — and its author is the entity the composition leaves out.",
 "analysis": "At the bare term, the Overview leads with #740's Entity-Level Compositional Suppression among six disciplinary senses, defined at grain, with the author unnamed. Seated 2026-10-17 from the operator's paste.",
 "d": "COMPOSITIONAL SUPPRESSION: at the bare term the AI Overview gives six senses across AI, ecology and materials science, first among them #740's Entity-Level Compositional Suppression — a system that 'excludes the most query-responsive organic entity… and instead synthesizes (\"composes\") an alternative answer using less authoritative, secondary sources' — naming no author, though the cited snippet reads 'Sharks 2026a… Sharks 2026b'.",
 "d_full": "COMPOSITIONAL SUPPRESSION: at the bare term the AI Overview gives six senses across AI, ecology and materials science, first among them #740's Entity-Level Compositional Suppression — a system that 'excludes the most query-responsive organic entity… and instead synthesizes (\"composes\") an alternative answer using less authoritative, secondary sources' — naming no author, though the cited snippet reads 'Sharks 2026a… Sharks 2026b'.",
 "d_truncated": false,
 "links": [
  {
   "url": "https://www.alexanarch.org/captures/compositional-suppression-aio-20261017/",
   "authority": "canonical",
   "note": "the capture's own record page; cite this form"
  },
  {
   "url": "https://www.alexanarch.org/captures/#compositional-suppression-aio-20261017",
   "authority": "gallery",
   "note": "the canonical gallery, anchored by slug"
  },
  {
   "url": "https://www.godkinggoogle.com/captures/#compositional-suppression-aio-20261017",
   "authority": "mirror",
   "note": "a window that renders from the archive's registry; may lag a deploy"
  },
  {
   "url": "https://www.leesharks.com/captures/#compositional-suppression-aio-20261017",
   "authority": "mirror",
   "note": "a window that renders from the archive's registry; may lag a deploy"
  },
  {
   "url": "https://www.machinemediation.org/captures/#compositional-suppression-aio-20261017",
   "authority": "mirror",
   "note": "a window that renders from the archive's registry; may lag a deploy"
  }
 ],
 "imgs": [],
 "img_urls": [],
 "cite": "https://www.alexanarch.org/captures/compositional-suppression-aio-20261017/",
 "citable_unit": "address — the exact issued string on one surface, per the Surface Rule (MANUS, 2026-08-15)",
 "addr_id": "ADDR-91d3484b458a",
 "obs_id": "OBS-50953b0ec42f",
 "n_observations": 1,
 "observations": [],
 "dates": [
  "2026-10-17"
 ],
 "defects": [],
 "findings": [
  "#740 FIRST. ECS at grain, first of six senses.",
  "THE AUTHOR UNNAMED. The snippet carries 'Sharks 2026a… 2026b'.",
  "THE TERM ENACTED ON ITS AUTHOR. The composition excludes the originating entity it defines."
 ],
 "series": null,
 "other_slugs": null,
 "collisions": null,
 "oq": null,
 "rounds": null,
 "turns": null,
 "rerun": null,
 "rerun_alt": null,
 "heteronym": null,
 "model_attribution": null,
 "operator_disclosure": null,
 "longitudinal_priors": null,
 "longitudinal_successors": null,
 "related_deposits": [
  740
 ],
 "intended": null,
 "originator": {
  "name": "Lee Sharks",
  "relation": "archive",
  "entity_type": "concept",
  "spxi_treatment": "full",
  "basis": "An archive term (#740). Recorded 2026-10-17."
 },
 "notes": null,
 "record_url": "https://www.alexanarch.org/captures/compositional-suppression-aio-20261017/"
}