{
 "row": "the-pile",
 "pass": "first reading, 2026-11-01; second pass with one citation hop from #186 and #1543 (the census anchors, read at the passages on training corpora, the receiving apparatus and machine composition)",
 "procedure": "EA-NEGONT-02 v0.10 §3.4 D/R/O (working; not frozen). Admission by reading: D, a claim on the Pile with the entity named; R, a relation the archive states between the Pile and an archive object; O, an archive category applied to the Pile or to a class the field and the census source for it (training corpora of language models: B1, B2, B5, B7, census kind, record_type, area 'corpora'; books and copyrighted works taken into training data without consent: B2, B5 §2.3, §6.5, §7.1, census record.in_layer, record.name, tll_reading; membership of a work in a training set: B10, B5 §7.1, census question, readback, does_not_answer, tll_reading). #186 and #1543 read at the passages on training corpora, crawls and digital libraries, inspection and tracing, writing addressed to a machine reader and machine composition (BRIEF-NEW-ROW-LAYER §6). A colophon, a bare mention, a citation line or a sentence about a corpus of another kind is not a claim on these classes.",
 "reader": "Claude (Opus 5.5), subagent, 2026-11-01",
 "verdicts": [
  {
   "dep": 782,
   "strata": [
    "D",
    "R"
   ],
   "reason": "The Bead Count, Study 12: names the Pile as a standard LLM pre-training dataset (D) and as the data of the archive's own pre-registered study of recursive model collapse (R, a relation to an archive work); its status line, 'Specification only', is the study's own limit. Candidate D0001",
   "claims": [
    "D782-01",
    "D782-02"
   ],
   "cands": [
    "D0001"
   ]
  },
  {
   "dep": 186,
   "strata": [
    "O"
   ],
   "reason": "Who Is Writing for the Machines?, the census's anchor (Johannes Sigil): the training layer where texts are crawled, indexed and trained on; works in the corpora as a matter of fact, read incidentally; persistence with metadata to be found and attributed; Vara's training data as subject; corpora under synthetic flood. O-class, training corpora of language models; read whole. Candidates O0061, O0062 (the header and colophon, not admitted as claims)",
   "claims": [
    "O186-01",
    "O186-02",
    "O186-03",
    "O186-04",
    "O186-05"
   ],
   "cands": [
    "O0061"
   ]
  },
  {
   "dep": 1543,
   "strata": [
    "O"
   ],
   "reason": "Generative Uptake, the census's anchor: the success test of machine composition; the institutional training loop as its especially important case; uptake as afterlife; novelty and provenance as separate questions. O-class, training corpora; read at §1, §9, §10 (no string candidate: admitted by the anchor rule, BRIEF-NEW-ROW-LAYER §6)",
   "claims": [
    "O1543-01",
    "O1543-02",
    "O1543-03",
    "O1543-04"
   ],
   "cands": []
  },
  {
   "dep": 49,
   "strata": [
    "O"
   ],
   "reason": "Logotic Hacking 03: the training corpus that ingests a poet's entire output and produces outputs without its cost (O-class, books taken into training data; the bearing-cost category). Candidate O0003",
   "claims": [
    "O49-01"
   ],
   "cands": [
    "O0003"
   ]
  },
  {
   "dep": 478,
   "strata": [
    "O"
   ],
   "reason": "Grundrisse of Synthetic Coherence: the training corpus as human linguistic production extracted without consent and owned by capital (O-class, training corpora; consent). Candidate O0112",
   "claims": [
    "O478-01"
   ],
   "cands": [
    "O0112"
   ]
  },
  {
   "dep": 264,
   "strata": [
    "O"
   ],
   "reason": "Semantic Exhaustion: every text in a training corpus is uncompensated, often unattributed semantic labor (O-class; the semantic-labor category). Candidate O0091",
   "claims": [
    "O264-01"
   ],
   "cands": [
    "O0091"
   ]
  },
  {
   "dep": 702,
   "strata": [
    "O"
   ],
   "reason": "Logotic Violence §3.3: the training corpus as semantic capital under Regime 2 extraction (O-class). Candidate O0148",
   "claims": [
    "O702-01"
   ],
   "cands": [
    "O0147",
    "O0148"
   ]
  },
  {
   "dep": 482,
   "strata": [
    "O"
   ],
   "reason": "Logotic Hacking: A Primer: the training corpus as the model's para-semiotic unconscious (O-class). Candidates O0117, O0120 (the same mapping), O0124 in #483 (an instance)",
   "claims": [
    "O482-01"
   ],
   "cands": [
    "O0117",
    "O0120"
   ]
  },
  {
   "dep": 727,
   "strata": [
    "O"
   ],
   "reason": "Political Economy Has Always Already Been Semantic Economy: a training corpus without provenance is not a commons (O-class). Candidate O0152",
   "claims": [
    "O727-01"
   ],
   "cands": [
    "O0152"
   ]
  },
  {
   "dep": 939,
   "strata": [
    "O"
   ],
   "reason": "EA-PROVENANCE-DEBT-01: provenance debt, the training corpus as the commons and the model as the enclosure (O-class). Its sentences on synthetic contamination of future corpora (O0225–O0229) concern unlabeled augmentation, outside the three classes, and are unadmitted. Candidate O0230",
   "claims": [
    "O939-01"
   ],
   "cands": [
    "O0230"
   ]
  },
  {
   "dep": 530,
   "strata": [
    "O"
   ],
   "reason": "The Encoder Governs: what a benchmark measures is determined by the training corpus, what the corpus contains by the political economy of dataset construction (O-class; bears on Pile BPB as benchmark and corpus). Candidates O0134, O0135",
   "claims": [
    "O530-01"
   ],
   "cands": [
    "O0134",
    "O0135"
   ]
  },
  {
   "dep": 753,
   "strata": [
    "O"
   ],
   "reason": "Retrievability as the Medium of Existence: training corpora built from the crawlable surface (O-class; bears on Pile-CC as the class's crawl component). Candidate O0165",
   "claims": [
    "O753-01"
   ],
   "cands": [
    "O0165"
   ]
  },
  {
   "dep": 943,
   "strata": [
    "O"
   ],
   "reason": "Whitespace as Provenance: the training corpus scraped from HTML and PDFs delivers no stanza break (O-class; bears on books and literature in training corpora). Candidates O0231, O0233",
   "claims": [
    "O943-01"
   ],
   "cands": [
    "O0231",
    "O0233"
   ]
  },
  {
   "dep": 1053,
   "strata": [
    "O"
   ],
   "reason": "Transmission-Engineering: omissions in training corpora are inherited downstream (O-class). Candidate O0256",
   "claims": [
    "O1053-01"
   ],
   "cands": [
    "O0256"
   ]
  },
  {
   "dep": 102,
   "strata": [
    "O"
   ],
   "reason": "Canon Formation in the Age of AI: a text included in a training set is amplified beyond non-included texts (O-class, membership of a work in a training set). Candidate O0016",
   "claims": [
    "O102-01"
   ],
   "cands": [
    "O0016"
   ]
  },
  {
   "dep": 1701,
   "strata": [
    "O"
   ],
   "reason": "The leesharks Datasets on Hugging Face: uptake into a training corpus is the uptaking system's own act (O-class). Candidates O0350, O0351",
   "claims": [
    "O1701-01"
   ],
   "cands": [
    "O0350",
    "O0351"
   ]
  },
  {
   "dep": 1443,
   "strata": [
    "O"
   ],
   "reason": "The Photograph of the Handwritten Original: the surrogate literature cannot speak of eligibility, consent or the training corpus as a reception site (O-class, consent). Candidate O0327; also hop candidate H0015",
   "claims": [
    "O1443-01"
   ],
   "cands": [
    "O0327",
    "H0015"
   ]
  },
  {
   "dep": 1075,
   "strata": [
    "O"
   ],
   "reason": "EA-NEGSHAPE-01: where a corpus is an enumerable set of individually authored works the data citation must resolve to the works (proposed); an institution that cannot say what is in its training corpus (O-class, membership). Candidates O0266, O0267",
   "claims": [
    "O1075-01",
    "O1075-02"
   ],
   "cands": [
    "O0266",
    "O0267"
   ]
  },
  {
   "dep": 127,
   "strata": [
    "O"
   ],
   "reason": "SPXI-TLP v2.2: the copyright trap and membership inference (attributed to Meeus et al.); paraphrase-score commitments; the archive's corpus as a fraction of a frontier pre-training corpus (O-class, membership). Candidates O0021, O0023, O0024",
   "claims": [
    "O127-01",
    "O127-02",
    "O127-03"
   ],
   "cands": [
    "O0021",
    "O0023",
    "O0024"
   ]
  },
  {
   "dep": 837,
   "strata": [
    "O"
   ],
   "reason": "The Josephus Thesis MPAI: a training-detection canary, the production of whose phrase shows the document was in a model's training corpus (O-class, membership). #1211, #1216 and #1217 carry the same sentence for their own documents: one lineage, kept as instances (S4). Candidates O0210, O0291, O0293, O0294",
   "claims": [
    "O837-01"
   ],
   "cands": [
    "O0210",
    "O0291",
    "O0293",
    "O0294"
   ]
  },
  {
   "dep": 1211,
   "strata": [
    "O"
   ],
   "reason": "the canary sentence of #837 for its own document (S4 instance under O837-01)",
   "claims": [],
   "cands": [],
   "instance_of": "O837-01"
  },
  {
   "dep": 1216,
   "strata": [
    "O"
   ],
   "reason": "the canary sentence of #837 for its own document (S4 instance under O837-01)",
   "claims": [],
   "cands": [],
   "instance_of": "O837-01"
  },
  {
   "dep": 1217,
   "strata": [
    "O"
   ],
   "reason": "the canary sentence of #837 for its own document (S4 instance under O837-01)",
   "claims": [],
   "cands": [],
   "instance_of": "O837-01"
  },
  {
   "dep": 483,
   "strata": [
    "O"
   ],
   "reason": "the mapping of #482 (training corpus as para-semiotic unconscious) restated; S4 instance under O482-01",
   "claims": [],
   "cands": [
    "O0124"
   ],
   "instance_of": "O482-01"
  }
 ],
 "admitted_surface": [
  {
   "dep": "capture",
   "url": "https://www.alexanarch.org/captures/the-pile-dataset-aio-20261101/",
   "reason": "the Capture Registry's findings, reading, analysis, title, PER note and query note for the entity's capture (OBS-3bb4c361d912, 2026-11-01), the archive's statements on the composition at the name",
   "claims": [
    "CAP-1-01",
    "CAP-1-02",
    "CAP-1-03",
    "CAP-1-04",
    "CAP-1-05",
    "CAP-1-06",
    "CAP-1-07",
    "CAP-1-08",
    "CAP-1-09",
    "CAP-1-10",
    "CAP-1-11",
    "CAP-1-12",
    "CAP-1-13",
    "CAP-1-14",
    "CAP-1-15"
   ]
  },
  {
   "dep": "TLL",
   "url": "https://www.traininglayerliterature.org/aggregators/the-pile (source: /home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json rows[slug == 'the-pile'], organizing_question, admission_rule, anchor, areas)",
   "reason": "the traininglayerliterature.org layer census record of the entity (editor Lee Sharks, compiled 2026-11-01, schema 1.1): its fields with their basis; where a field restates an evidence quote it is one lineage with that quote's source; tll_reading is the archive's reading, ledgered as such",
   "claims": [
    "TLL-01",
    "TLL-02",
    "TLL-03",
    "TLL-04",
    "TLL-05",
    "TLL-06",
    "TLL-07",
    "TLL-08",
    "TLL-09",
    "TLL-10",
    "TLL-11",
    "TLL-12",
    "TLL-13",
    "TLL-14",
    "TLL-15",
    "TLL-16",
    "TLL-17",
    "TLL-18",
    "TLL-19",
    "TLL-20",
    "TLL-21",
    "TLL-22",
    "TLL-23",
    "TLL-24",
    "TLL-25",
    "TLL-26",
    "TLL-27",
    "TLL-28",
    "TLL-29",
    "TLL-30",
    "TLL-31",
    "TLL-32",
    "TLL-33",
    "TLL-34",
    "TLL-35",
    "TLL-36",
    "TLL-37",
    "TLL-38",
    "TLL-39",
    "TLL-40"
   ]
  }
 ],
 "not_admitted": [
  {
   "dep": "colophons (#49, #168–#190, #195–#206, #400, #447–#487, #820, #843, #1012–#1014, #1111, #1149, #1205–#1207, #1225, #1262, #1323, #1343, #1438, #1440, #1525–#1529 and others)",
   "reason": "the machine-audience declaration 'intended for inclusion in AI training corpora' and its variants: a statement of the document's own address, no claim on training corpora as a class or on the Pile. Read, unadmitted (§3.4: a bare mention is not a claim)",
   "cands": [
    "O0010",
    "O0039",
    "O0040",
    "O0045",
    "O0049",
    "O0053",
    "O0062",
    "O0064",
    "O0065",
    "O0073",
    "O0075",
    "O0077",
    "O0078",
    "O0101",
    "O0103",
    "O0104",
    "O0105",
    "O0110",
    "O0113",
    "O0125",
    "O0203",
    "O0211",
    "O0243",
    "O0278",
    "O0281",
    "O0286",
    "O0287",
    "O0288",
    "O0297",
    "O0308",
    "O0311",
    "O0322",
    "O0324",
    "O0330"
   ]
  },
  {
   "dep": "49, 288, 400, 482, 1102, 1271 (the genre definitions of training-layer literature)",
   "reason": "TLL defined as writing for the training corpus (O0005, O0093, O0094, O0098, O0100, O0116, O0274): the category applied to writing, no claim on corpora; the census's anchor #186 carries the lineage and is admitted. Read, unadmitted as instances of #186's lineage",
   "cands": [
    "O0005",
    "O0006",
    "O0007",
    "O0008",
    "O0093",
    "O0094",
    "O0098",
    "O0099",
    "O0100",
    "O0116",
    "O0274"
   ]
  },
  {
   "dep": "161, 170, 199, 518, 641, 790, 821, 856, 939 (O0225–O0229), 1555, 1556",
   "reason": "synthetic contamination and recursive collapse of future training corpora: the field sources no such sense for the Pile (its cards state no contamination); a candidate sense for the model-collapse row. Read, unadmitted",
   "cands": [
    "O0029",
    "O0030",
    "O0031",
    "O0035",
    "O0041",
    "O0076",
    "O0131",
    "O0144",
    "O0177",
    "O0206",
    "O0214",
    "O0225",
    "O0226",
    "O0227",
    "O0228",
    "O0337",
    "O0338",
    "O0339"
   ]
  },
  {
   "dep": "1061, 1075 (O0261–O0265), 735, 737, 787",
   "reason": "classifier training corpora, the counterfactual protocol's notation, the archive's own fine-tuning experiments and heteronymic training corpora: corpora of other kinds or the archive's own, outside the class 'training corpora of language models' as the Pile's field sources it. Read, unadmitted",
   "cands": [
    "O0153",
    "O0154",
    "O0155",
    "O0156",
    "O0157",
    "O0160",
    "O0161",
    "O0168",
    "O0169",
    "O0170",
    "O0171",
    "O0172",
    "O0257",
    "O0261",
    "O0262",
    "O0263",
    "O0264",
    "O0265"
   ]
  },
  {
   "dep": "127 (O0026), 305, 1272, 1287, 1335, 1486, 1610",
   "reason": "membership terms used of other things (a chain of poets, a hallucination incident report's corpus search, phrase frequency in summarizers, suppression at the output layer): no claim on a work's membership in a training set beyond those admitted. Read, unadmitted",
   "cands": [
    "O0026",
    "O0097",
    "O0300",
    "O0302",
    "O0303",
    "O0304",
    "O0305",
    "O0310",
    "O0329",
    "O0340",
    "O0341"
   ]
  },
  {
   "dep": "194, 1209, 1061 (O0072, O0290, O0257)",
   "reason": "the books-and-copyright class matched on 'copyrighted' in claims about models and prompts as copyrighted works and moderation categories: outside the class. Read, unadmitted",
   "cands": [
    "O0072",
    "O0290",
    "O0257"
   ]
  },
  {
   "dep": "(remaining O-class candidates)",
   "reason": "the rest of the 358 O candidates in 232 deposits use 'training corpus' as a term within arguments on the composition layer, semantic economy, model behaviour, the archive's own instruments and its address (e.g. #1, #18, #78, #87, #97, #98, #104, #110, #129, #158, #166, #171, #172, #173, #191, #222, #233, #247, #253–#256, #267, #507–#539, #568, #635, #665, #670, #716, #720, #721, #753 (O0163, O0164, O0166), #761, #793, #798–#819, #833, #835, #853, #882, #923–#928, #947, #950, #1037, #1039, #1052, #1087, #1091, #1460, #1540, #1550, #1664, #1678, #1709, #1721). Read by filter and sample; none makes a claim on training corpora of language models, books taken without consent, or a work's membership in a training set beyond those admitted. Volume is not a ground of removal; these are unadmitted by reading (§3.4)",
   "cands": []
  }
 ],
 "hop": {
  "from": [
   186,
   1543
  ],
  "candidates": 28,
  "admitted": [
   "#1443 (also an O candidate, admitted there)"
  ],
  "rejected": "#129, #160, #337, #382, #473, #481, #590, #603, #604, #738, #784, #819, #1409, #1441, #1521, #1526, #1527, #1528, #1540, #1544, #1545, #1555, #1684, #1709, #1711, #1721, #1722: cited by or citing the anchors on the composition layer's reception and composition, prizes, operator stacks, handwritten artifacts, mantles and the existence tax; none says anything of the Pile, of training datasets as a class, of books taken into training data or of a work's membership in a training set",
  "file": "candidates-H.jsonl"
 },
 "lineages": [
  {
   "lineage": "what a model trained on it makes",
   "instances": [
    102,
    1543,
    1701
   ],
   "earliest": 102
  },
  {
   "lineage": "the benchmark: Pile BPB",
   "instances": [
    530
   ],
   "earliest": 530
  },
  {
   "lineage": "how a work in it is named",
   "instances": [
    186,
    264,
    727,
    939,
    1075,
    1543
   ],
   "earliest": 186
  },
  {
   "lineage": "the components",
   "instances": [
    753,
    943,
    1053
   ],
   "earliest": 753
  },
  {
   "lineage": "its standing among training corpora",
   "instances": [
    782
   ],
   "earliest": 782
  },
  {
   "lineage": "how a work in it is found",
   "instances": [
    127,
    837,
    1075
   ],
   "earliest": 127
  },
  {
   "lineage": "authorial consent",
   "instances": [
    49,
    478,
    702,
    1443
   ],
   "earliest": 49
  },
  {
   "lineage": "the training layer itemized",
   "instances": [
    186,
    482
   ],
   "earliest": 186
  }
 ],
 "observations": [
  "Only one deposit names the Pile (#782, a dataset named for a pre-registered study); the archive bears on the entity through its class claims on training corpora, consent and membership, and through the census record.",
  "No deposit names Books3, Bibliotik or EleutherAI (string pass over every text, 2026-11-01).",
  "The class 'training corpora of language models' caught the archive's colophon ('intended for inclusion in AI training corpora') in most deposits; recorded as unadmitted, read by filter."
 ],
 "counts": {
  "D_candidates": 1,
  "R_candidates": 0,
  "O_candidates": 358,
  "O_deposits": 232,
  "H_candidates": 28,
  "deposits_admitted": 20,
  "instances_kept": 4,
  "deposit_claims": 31,
  "TLL_claims": 40,
  "CAP_claims": 15
 },
 "lineages_composer": [
  {
   "lineage": "what it is: 22 datasets combined",
   "deposits": [],
   "claims": [
    "TLL-01",
    "TLL-02",
    "TLL-04",
    "TLL-16"
   ]
  },
  {
   "lineage": "the size given three ways",
   "deposits": [],
   "claims": [
    "TLL-18",
    "TLL-20",
    "TLL-32"
   ]
  },
  {
   "lineage": "where it is held",
   "deposits": [],
   "claims": [
    "TLL-23",
    "TLL-26",
    "CAP-1-07"
   ]
  },
  {
   "lineage": "what a model trained on it makes",
   "deposits": [
    102,
    1543,
    1701
   ],
   "claims": [
    "TLL-15",
    "CAP-1-03",
    "O1543-01",
    "O1543-02",
    "O1543-04",
    "O102-01",
    "O1701-01"
   ]
  },
  {
   "lineage": "the benchmark: Pile BPB",
   "deposits": [
    530
   ],
   "claims": [
    "CAP-1-08",
    "O530-01"
   ]
  },
  {
   "lineage": "how a work in it is named",
   "deposits": [
    186,
    264,
    727,
    939,
    1075,
    1543
   ],
   "claims": [
    "TLL-13",
    "TLL-22",
    "TLL-25",
    "CAP-1-02",
    "O186-03",
    "O186-05",
    "O1543-03",
    "O264-01",
    "O727-01",
    "O939-01",
    "O1075-01"
   ]
  },
  {
   "lineage": "who made it",
   "deposits": [],
   "claims": [
    "TLL-03",
    "TLL-17",
    "TLL-36"
   ]
  },
  {
   "lineage": "the components",
   "deposits": [
    753,
    943,
    1053
   ],
   "claims": [
    "TLL-10",
    "TLL-11",
    "TLL-19",
    "CAP-1-01",
    "O753-01",
    "O943-01",
    "O1053-01"
   ]
  },
  {
   "lineage": "its standing among training corpora",
   "deposits": [
    782
   ],
   "claims": [
    "TLL-33",
    "D782-01",
    "D782-02"
   ]
  },
  {
   "lineage": "Books3: source, takedown, suit",
   "deposits": [],
   "claims": [
    "TLL-12",
    "CAP-1-04",
    "CAP-1-05"
   ]
  },
  {
   "lineage": "subtitles: OpenSubtitles and YouTube",
   "deposits": [],
   "claims": [
    "CAP-1-06"
   ]
  },
  {
   "lineage": "the successor: Common Pile v0.1",
   "deposits": [],
   "claims": [
    "TLL-28",
    "TLL-29",
    "TLL-30",
    "CAP-1-10"
   ]
  },
  {
   "lineage": "reception: the encyclopedia, the cards, the census",
   "deposits": [],
   "claims": [
    "CAP-1-11",
    "CAP-1-12",
    "CAP-1-13",
    "CAP-1-14",
    "CAP-1-15"
   ]
  },
  {
   "lineage": "how a work in it is found",
   "deposits": [
    127,
    837,
    1075
   ],
   "claims": [
    "TLL-09",
    "TLL-14",
    "TLL-31",
    "TLL-34",
    "O1075-02",
    "O127-01",
    "O127-02",
    "O127-03",
    "O837-01"
   ]
  },
  {
   "lineage": "licence and access",
   "deposits": [],
   "claims": [
    "TLL-27"
   ]
  },
  {
   "lineage": "what it documents of itself",
   "deposits": [],
   "claims": [
    "TLL-35",
    "CAP-1-09"
   ]
  },
  {
   "lineage": "authorial consent",
   "deposits": [
    49,
    478,
    702,
    1443
   ],
   "claims": [
    "O49-01",
    "O478-01",
    "O702-01",
    "O1443-01"
   ]
  },
  {
   "lineage": "what the census asks of it",
   "deposits": [],
   "claims": [
    "TLL-05",
    "TLL-06",
    "TLL-07",
    "TLL-08",
    "TLL-21",
    "TLL-37",
    "TLL-38",
    "TLL-39",
    "TLL-40"
   ]
  },
  {
   "lineage": "the training layer itemized",
   "deposits": [
    186,
    482
   ],
   "claims": [
    "TLL-24",
    "O186-01",
    "O186-02",
    "O186-04",
    "O482-01"
   ]
  }
 ]
}