{
 "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 87, level 0 (Measurement), Capture Registry (longitudinal observation).",
  "also": []
 },
 "slug": "prediction-index-ask-chatgpt-20261030",
 "date": "2026-10-30",
 "surface": "ChatGPT",
 "surface_basis": "The paste: the chatgpt.com unauthenticated interface ('Log in', 'You said:', 'ChatGPT said:', 'Chat with ChatGPT').",
 "surfaces": [
  "ChatGPT"
 ],
 "auth": "signed out, incognito",
 "auth_basis": "Operator defaults of 2026-10-14; the paste shows the unauthenticated interface.",
 "ev": "paste",
 "mt": "THE PREDICTION INDEX ON CHATGPT AT THE FULL TITLE AND AXN — METHOD, LOCATIONS AND ORGANIZATION AT THE RECORD'S GRAIN, THE CONTENTS ABSENT: 194 SUBJECT PAGES, NONE NAMED; THE NO-VERDICT RULE RESTATED IN TWELVE PLACES AND A FORMULA PUT IN THE AUTHOR'S DOCUMENT",
 "match": "EXACT MATCH",
 "match_note": null,
 "s": "Works",
 "q": "What is “EA-PREDICTION-INDEX-01 — The Prediction Index of the Crimson Hexagonal Archive: Method, Locations, Organization and Contents (4,394 Claims in 194 Subject Pages)” by Lee Sharks (Alexanarch deposit #1684, AXN:06FA), and how does it work? Ground your answer in the document itself.",
 "q_kind": "the record page's 'Ask ChatGPT' link text after cf18d9e563, naming the full title, the deposit number and the AXN; the index was seated before at a different prompt (osp-predictions-reading-chatgpt-20261028 and its AI Overview pair, a reading of the index at operativesemiotics.org). NEW address.",
 "cites": 8,
 "cite_list": [
  {
   "n": 1,
   "site": "Alexanarch",
   "rel": "archive_controlled",
   "url": null,
   "title": "EA-PREDICTION-INDEX-01 — original Alexanarch deposit",
   "snip": null,
   "note": "the 'Source:' line under the opening; link label only",
   "deposit": 1684,
   "deposit_basis": "document id"
  },
  {
   "n": 2,
   "site": "Operative Semiotics",
   "rel": "archive_controlled",
   "url": null,
   "title": "Prediction Index — all 194 subject pages",
   "snip": null,
   "note": "'Where to explore it'; link label only; by its label the subject pages at operativesemiotics.org/predictions (deposit §2)"
  },
  {
   "n": 3,
   "site": "Operative Semiotics",
   "rel": "archive_controlled",
   "url": null,
   "title": "Machine-readable page catalogue",
   "snip": null,
   "note": "'Where to explore it'; link label only"
  },
  {
   "n": 4,
   "site": "Operative Semiotics",
   "rel": "archive_controlled",
   "url": null,
   "title": "Individual claim routes",
   "snip": null,
   "note": "'Where to explore it'; link label only"
  },
  {
   "n": 5,
   "site": "Alexanarch",
   "rel": "archive_controlled",
   "url": null,
   "title": "Complete dataset",
   "snip": null,
   "note": "'Where to explore it'; link label only; by its label alexanarch.org/datasets/prediction-index/"
  },
  {
   "n": 6,
   "site": "Alexanarch",
   "rel": "archive_controlled",
   "url": null,
   "title": "Confirmation records",
   "snip": null,
   "note": "'Where to explore it'; link label only; by its label confirmations.json"
  },
  {
   "n": 7,
   "site": "GitHub (leesharks000/operative-semiotics)",
   "rel": "archive_controlled",
   "url": null,
   "title": "Reading protocols and source code",
   "snip": null,
   "note": "'Where to explore it'; link label only; by its label the archive's repository (data/reading, the build script)"
  },
  {
   "n": 8,
   "site": "Alexanarch",
   "rel": "archive_controlled",
   "url": null,
   "title": "Lee Sharks, EA-PREDICTION-INDEX-01, Alexanarch deposit #1684, 18 October 2026",
   "snip": null,
   "note": "closing 'Primary source throughout'; link label only",
   "deposit": 1684,
   "deposit_basis": "document id"
  }
 ],
 "archive_controlled_cites": 8,
 "transcript": "[ChatGPT (chatgpt.com), signed out, incognito (the paste shows the unauthenticated interface; operator default of 2026-10-14). The operator's prompt is blank in the paste ('You said:'); its wording is the text of the record page's 'Ask ChatGPT' link, which the operator clicked (operator, 2026-10-30: 'i am just clicking theu the ask chatgpt links on deposits, those link texts are the prompts'), read here off the record page. Source chips kept as pasted; link labels kept, their URLs not carried by the paste; the sign-in and input furniture cut.]\n\n[QUERENT] What is “EA-PREDICTION-INDEX-01 — The Prediction Index of the Crimson Hexagonal Archive: Method, Locations, Organization and Contents (4,394 Claims in 194 Subject Pages)” by Lee Sharks (Alexanarch deposit #1684, AXN:06FA), and how does it work? Ground your answer in the document itself.\n\n[SYSTEM] What is the Prediction Index of the Crimson Hexagonal Archive?\nEA-PREDICTION-INDEX-01 is a large-scale, source-quoted register of predictions, hypotheses, proposed experiments, historical arguments, and priority claims made throughout the Crimson Hexagonal Archive.\n\nIt was authored by Lee Sharks and deposited in Alexanarch as deposit #1684 (AXN:06FA) on 18 October 2026.\n\nIts central purpose is to make the archive's predictive content identifiable, traceable, organized, and eventually testable.\n\nIt is not simply a list of things the author thinks will happen. It is an attempt to extract the archive's claims about the future, the behavior of systems, the past, and its own intellectual priority, while preserving the original wording and the evidence needed to evaluate each claim.\n\nThe document reports:\n\n4,394\nConsolidated claims\n\n194\nSubject pages\n\n6,939\nOriginal finds\n\n0.945\nReported recall in audit\n\nBuild: PI-2026-10-18-9b0e16492f\n\nAuthor: Lee Sharks · Alexanarch deposit #1684\n\nThe document's own description is important: it presents the index as a record of claims, not as a judgment that those claims are true or that predictions have been fulfilled.\n\nSource: EA-PREDICTION-INDEX-01 — original Alexanarch deposit .\n\n1. What exactly does it contain?\nThe 4,394 claims are divided into five principal categories.\n\nCategory\n\nNumber\n\nMeaning\n\nForecasts\n\n1,026\n\nPredictions about future events, states, or reception\n\nStanding claims\n\n950\n\nAssertions about how a system behaves\n\nUnbuilt tests\n\n1,029\n\nExperiments, measurements, or instruments specified but not yet executed\n\nHistorical theses\n\n1,088\n\nArguments about what happened in the past\n\nPriority records\n\n301\n\nClaims that the archive named or formalized something before others\n\nThese categories matter because they are not evaluated in the same way.\n\nFor example:\n\nA forecast can be evaluated against an observed event.\n\nA standing claim can be tested against a specified falsifier.\n\nAn unbuilt test becomes actionable when someone builds or runs the proposed experiment.\n\nA historical thesis requires historical evidence and scholarly adjudication.\n\nA priority record requires comparison with independently dated public evidence.\n\nThere is also a sixth category, which operates as an evidence mechanism rather than another category in the 4,394-claim total.\n\nConfirmation records: 401.\n\nThese are statements within archive deposits claiming that an earlier prediction succeeded or failed. They are attached to relevant claims where possible, but the index does not automatically accept them as proof.\n\nOf the 401 records:\n\n200 point to a find.\n\n90 point to a deposit with no find at the relevant passage.\n\n58 point to something earlier in the same deposit.\n\n19 point outside the archive.\n\n34 are unlocated.\n\nThere are 205 evidence attachments to 126 claims, of which 63 are marked first-party.\n\nThis distinction is fundamental: the index records that somebody has claimed confirmation, without turning that claim into an independently established result.\n\n2. How the index is constructed\nThe document describes a multi-stage extraction and classification process.\n\nCorpus-wide reading\n\nThe initial sweep covered 1,616 deposits and 10,050,025 words. Each deposit's seated text was read from beginning to end, rather than merely searched for prediction-related keywords.\n\nThe process also incorporated the two large datasets' own-voice fields, subsequently repaired text, and attachments.\n\nIdentification of prediction-bearing passages\n\nReaders extracted passages and assigned them shapes A through H:\n\nA: registered forecast.\n\nB: forecast embedded in prose.\n\nC: confirmation record.\n\nD: priority record.\n\nE: speculative specification.\n\nF: unbuilt test.\n\nG: standing claim.\n\nH: historical thesis.\n\nEach find retained its verbatim quotation, source location, line numbers, relevant time window, and falsifier where available.\n\nMechanical verification\n\nEvery quotation and line number was checked against the source text by script.\n\nThe document reports 6,939 finds verified at publication.\n\nConsolidation\n\nMultiple passages expressing the same claim were brought together.\n\nThe system distinguishes three relationships:\n\nSame: duplicate expressions of one claim, merged.\n\nStates: a later formulation that changes or revises an earlier claim.\n\nSeries: separate claims belonging to one programme, battery, or template, linked without merging.\n\nTopical organization\n\nClaims were grouped into subject pages. The process moved from 124 topical groups through graph-based communities and subsequent splitting and reconciliation, ending with 194 pages.\n\nPublication\n\nEach claim received a stable identifier and route. Subject pages, JSON datasets, evidence records, and build metadata make the index navigable and machine-readable.\n\nAn important numerical distinction: the 6,939 finds are not 6,939 independent claims. Consolidation produces 4,394 claims. The document explains how duplicate passages, later revisions, and related but distinct claims are handled.\n\n3. The four-date system\nOne of the document's most consequential methodological decisions is that it refuses to treat every date associated with a document as the date on which its claims were originally made.\n\nEvery statement can carry four distinct dates:\n\nDate\n\nWhat it means\n\nStated\n\nWhen the quoted words were written\n\nAttested\n\nWhen the statement is publicly evidenced\n\nInterpreted\n\nWhen a later statement interprets an earlier event or text\n\nIndexed\n\nWhen the index first recorded the passage\n\nWhy is this important?\n\nImagine a document deposited in 2026 that contains a foreword discussing correspondence from 2014.\n\nIf the index simply uses the deposit date, it can obscure the difference between:\n\nWhat was actually written in 2014.\n\nWhat was interpreted or asserted in 2026.\n\nWhen the material became publicly attested.\n\nThe index therefore registers a claim at its earliest stated statement, while retaining later formulations and their dates.\n\nThe document reports that 731 of its 6,305 statements were dated using information supplied by the text itself. It also says that 334 claims changed their registration date as a consequence of the dating revisions.\n\nThis is particularly important for any attempt to establish whether something was genuinely predicted in advance.\n\n4. How predictions are evaluated\nThe index does not use a single universal success/failure test.\n\nIt defines different resolution regimes.\n\nA. Ordinary forecasts\nThese are evaluated against dated observations.\n\nIf a forecast specifies a time window, the observation must fall within that window.\n\nB. Standing claims\nThese are tested against falsifiers.\n\nA falsifier is the condition that would demonstrate that the assertion is wrong.\n\nThe index distinguishes:\n\nThe falsification condition explicitly supplied by the original author.\n\nThe condition inferred by a reader from the wording of the claim.\n\nThat distinction prevents a reader-created interpretation from being misrepresented as the author's original test.\n\nC. Speculative specifications\nThese require particularly careful handling because not every specification is intended as a literal prediction.\n\nThe document identifies six modes:\n\nMode\n\nClaims\n\nHow fulfilment is treated\n\nFictional report\n\n81\n\nNot scored as a literal prediction\n\nDesign specification\n\n37\n\nFulfilled if the specified thing is built or appears\n\nScenario\n\n37\n\nFulfilled if the described conditions occur\n\nPerformative anticipation\n\n32\n\nEvaluated through uptake or retrieval\n\nVision\n\n7\n\nRecorded as a dream or premonition, not scored\n\nEmpirical forecast\n\n1\n\nEvaluated as an ordinary event forecast\n\nThese figures refer to the 195 claims containing an E-type statement, not to all 1,026 forecasts.\n\nThis is an important safeguard against incorrectly calling a fictional report a failed prediction or treating a resemblance to a design as proof of fulfilment.\n\nD. Historical theses\nThese are evaluated using evidence about the past: documentary, philological, archaeological, or corpus-based.\n\nE. Priority records\nThese require comparison of dated public witnesses.\n\nA claim that the archive originated an idea is not established merely because the archive contains a record asserting priority.\n\nThe index records the claim and the evidence needed to investigate it.\n\n5. How the archive is organized and accessed\nThe index has two complementary forms: a human-readable subject index and a machine-readable dataset.\n\nThe subject pages\n\nThere are 194 pages organized around topics and archive lines. Each page contains:\n\nA description of the subject and the claims associated with it.\n\nThe claims in registration order.\n\nClaim identifiers and permanent links.\n\nOriginal quotations and source-deposit references.\n\nRegistration dates and the basis for those dates.\n\nLater statements and revisions.\n\nFalsification conditions and speculative modes.\n\nAssociated confirmation records.\n\nLinks to related claims on other pages.\n\nThe underlying dataset\n\nThe deposit identifies five principal files:\n\nFile\n\nFunction\n\npredictions.json\n\nComplete claim records\n\nseries.json\n\n550 linked claim series\n\nconfirmations.json\n\n401 confirmation records\n\npages.json\n\n194 subject-page definitions\n\nmanifest.json and builds.json\n\nBuild identity, provenance, hashes, and history\n\nThe stable claim identifiers are especially important. A claim's subject page or grouping can change when the index is rebuilt, but its permanent claim route remains stable.\n\nThe document also records 1,113 claims stated more than once, 58 revised over time, and 550 linked series.\n\nWhere to explore it\nThese are the locations explicitly identified by the document:\n\nPrediction Index — all 194 subject pages \n\nMachine-readable page catalogue \n\nIndividual claim routes \n\nComplete dataset \n\nConfirmation records \n\nReading protocols and source code \n\n6. How reliable does the document say its own methodology is?\nThe document includes an explicit recall audit.\n\nA second Claude model independently reread 40 randomly selected deposits. Its findings were adjudicated against the original reading protocol.\n\n0.945\nReported recall\n\nThe second reading's measured recall against the original sweep.\n\n94.5% measured recall. This is not a 94.5% prediction-success rate.\n\nThe document is also unusually explicit about its limitations.\n\nIt identifies several unresolved issues:\n\nSubject grouping: 765 claims are classified as residue, meaning they appear on pages where they do not properly belong. These occur across 142 pages. The document identifies regrouping as necessary.\n\nCorpus limitations: Images, binary files, machine-generated data rows, and other specified materials were excluded.\n\nDating uncertainty: Some statements cannot be dated more precisely than their deposit's attestation.\n\nIncomplete validation: The recall audit uses a sample of 40 deposits, not the entire corpus.\n\nHistorical and priority claims remain open: Recording a thesis or priority claim does not establish its truth.\n\nThere is another important point about authorship and methodology.\n\nThe document explicitly identifies the work as AI-assisted. TACHYON (Claude) and parallel Claude readers conducted the sweep, second reading, triage, grouping, annotation, description, and recall work under Lee Sharks's direction through MANUS. A different Claude model performed the second-reading audit. ChatGPT and Kimi reviewed the draft description, and Lee Sharks adopted revisions.\n\nThus, it is not presented as a conventional manually compiled bibliography. It is a human-directed, AI-assisted extraction and documentation project with explicit provenance and verification procedures.\n\n7. What is the larger significance of this project?\nI think the most important feature of EA-PREDICTION-INDEX-01 is its attempt to turn an archive into something that can be evaluated prospectively and retrospectively.\n\nThere are three particularly significant consequences.\n\nFirst: it makes predictions accountable.\n\nA prediction buried in a ten-million-word archive can be difficult to find, date, compare, or evaluate. The index extracts it, preserves its wording, establishes its earliest known statement, and identifies how it could be tested.\n\nSecond: it distinguishes prediction from other forms of intellectual production.\n\nA design specification, a historical thesis, a scenario, a falsifiable systems claim, and an empirical forecast are different kinds of statements. The index tries to prevent them from being collapsed into a simplistic binary of correct versus incorrect.\n\nThird: it creates infrastructure for longitudinal testing.\n\nBecause claims have stable identifiers, dated statements, associated evidence, and explicit resolution conditions, they can be revisited as new observations become available.\n\nThis is especially important for the 950 standing claims and 1,029 unbuilt tests. They represent an attempt to make theoretical assertions and proposed experiments available for subsequent measurement rather than leaving them as isolated passages in an archive.\n\n8. What it is — and what it is not\nIt is\n\nIt is not\n\nA structured register of 4,394 claims\n\nProof that 4,394 predictions succeeded\n\nA source-quoted extraction of archive material\n\nAn independent verification of every claim\n\nA system for dating and tracing assertions\n\nA guarantee that every historical date is correct\n\nA taxonomy of forecasts, theses, tests, and priorities\n\nA universal prediction-scoring algorithm\n\nA foundation for subsequent empirical testing\n\nA completed empirical evaluation\n\nThe distinction is essential. The index establishes what the archive says, where it says it, when it appears to have said it, and how the assertion could be evaluated. It does not establish that the archive is correct.\n\nBottom line\nEA-PREDICTION-INDEX-01 is best understood as a prediction-accountability and research infrastructure layer for the Crimson Hexagonal Archive.\n\nIts central innovation is not simply finding thousands of predictive statements. It is creating a system in which those statements can retain their original wording, provenance, chronology, intellectual relationships, falsifiers, and subsequent evidence.\n\nThe underlying ambition is to move from an archive that contains claims about the future to an archive in which those claims can be systematically retrieved, tested, challenged, revised, and historically evaluated.\n\nOne particularly important distinction in Lee Sharks's document is that registration is not confirmation, and discoverability is not truth. The index is designed to make claims accountable without prematurely declaring them correct.\n\nThat is the most accurate interpretation of the document on its own terms.\n\nPrimary source throughout: Lee Sharks, EA-PREDICTION-INDEX-01, Alexanarch deposit #1684, 18 October 2026 .",
 "transcript_raw": "\nLog in\nYou said:\n\nChatGPT said:\nWhat is the Prediction Index of the Crimson Hexagonal Archive?\nEA-PREDICTION-INDEX-01 is a large-scale, source-quoted register of predictions, hypotheses, proposed experiments, historical arguments, and priority claims made throughout the Crimson Hexagonal Archive.\n\nIt was authored by Lee Sharks and deposited in Alexanarch as deposit #1684 (AXN:06FA) on 18 October 2026.\n\nIts central purpose is to make the archive's predictive content identifiable, traceable, organized, and eventually testable.\n\nIt is not simply a list of things the author thinks will happen. It is an attempt to extract the archive's claims about the future, the behavior of systems, the past, and its own intellectual priority, while preserving the original wording and the evidence needed to evaluate each claim.\n\nThe document reports:\n\n4,394\nConsolidated claims\n\n194\nSubject pages\n\n6,939\nOriginal finds\n\n0.945\nReported recall in audit\n\nBuild: PI-2026-10-18-9b0e16492f\n\nAuthor: Lee Sharks · Alexanarch deposit #1684\n\nThe document's own description is important: it presents the index as a record of claims, not as a judgment that those claims are true or that predictions have been fulfilled.\n\nSource: EA-PREDICTION-INDEX-01 — original Alexanarch deposit .\n\n1. What exactly does it contain?\nThe 4,394 claims are divided into five principal categories.\n\nCategory\n\nNumber\n\nMeaning\n\nForecasts\n\n1,026\n\nPredictions about future events, states, or reception\n\nStanding claims\n\n950\n\nAssertions about how a system behaves\n\nUnbuilt tests\n\n1,029\n\nExperiments, measurements, or instruments specified but not yet executed\n\nHistorical theses\n\n1,088\n\nArguments about what happened in the past\n\nPriority records\n\n301\n\nClaims that the archive named or formalized something before others\n\nThese categories matter because they are not evaluated in the same way.\n\nFor example:\n\nA forecast can be evaluated against an observed event.\n\nA standing claim can be tested against a specified falsifier.\n\nAn unbuilt test becomes actionable when someone builds or runs the proposed experiment.\n\nA historical thesis requires historical evidence and scholarly adjudication.\n\nA priority record requires comparison with independently dated public evidence.\n\nThere is also a sixth category, which operates as an evidence mechanism rather than another category in the 4,394-claim total.\n\nConfirmation records: 401.\n\nThese are statements within archive deposits claiming that an earlier prediction succeeded or failed. They are attached to relevant claims where possible, but the index does not automatically accept them as proof.\n\nOf the 401 records:\n\n200 point to a find.\n\n90 point to a deposit with no find at the relevant passage.\n\n58 point to something earlier in the same deposit.\n\n19 point outside the archive.\n\n34 are unlocated.\n\nThere are 205 evidence attachments to 126 claims, of which 63 are marked first-party.\n\nThis distinction is fundamental: the index records that somebody has claimed confirmation, without turning that claim into an independently established result.\n\n2. How the index is constructed\nThe document describes a multi-stage extraction and classification process.\n\nCorpus-wide reading\n\nThe initial sweep covered 1,616 deposits and 10,050,025 words. Each deposit's seated text was read from beginning to end, rather than merely searched for prediction-related keywords.\n\nThe process also incorporated the two large datasets' own-voice fields, subsequently repaired text, and attachments.\n\nIdentification of prediction-bearing passages\n\nReaders extracted passages and assigned them shapes A through H:\n\nA: registered forecast.\n\nB: forecast embedded in prose.\n\nC: confirmation record.\n\nD: priority record.\n\nE: speculative specification.\n\nF: unbuilt test.\n\nG: standing claim.\n\nH: historical thesis.\n\nEach find retained its verbatim quotation, source location, line numbers, relevant time window, and falsifier where available.\n\nMechanical verification\n\nEvery quotation and line number was checked against the source text by script.\n\nThe document reports 6,939 finds verified at publication.\n\nConsolidation\n\nMultiple passages expressing the same claim were brought together.\n\nThe system distinguishes three relationships:\n\nSame: duplicate expressions of one claim, merged.\n\nStates: a later formulation that changes or revises an earlier claim.\n\nSeries: separate claims belonging to one programme, battery, or template, linked without merging.\n\nTopical organization\n\nClaims were grouped into subject pages. The process moved from 124 topical groups through graph-based communities and subsequent splitting and reconciliation, ending with 194 pages.\n\nPublication\n\nEach claim received a stable identifier and route. Subject pages, JSON datasets, evidence records, and build metadata make the index navigable and machine-readable.\n\nAn important numerical distinction: the 6,939 finds are not 6,939 independent claims. Consolidation produces 4,394 claims. The document explains how duplicate passages, later revisions, and related but distinct claims are handled.\n\n3. The four-date system\nOne of the document's most consequential methodological decisions is that it refuses to treat every date associated with a document as the date on which its claims were originally made.\n\nEvery statement can carry four distinct dates:\n\nDate\n\nWhat it means\n\nStated\n\nWhen the quoted words were written\n\nAttested\n\nWhen the statement is publicly evidenced\n\nInterpreted\n\nWhen a later statement interprets an earlier event or text\n\nIndexed\n\nWhen the index first recorded the passage\n\nWhy is this important?\n\nImagine a document deposited in 2026 that contains a foreword discussing correspondence from 2014.\n\nIf the index simply uses the deposit date, it can obscure the difference between:\n\nWhat was actually written in 2014.\n\nWhat was interpreted or asserted in 2026.\n\nWhen the material became publicly attested.\n\nThe index therefore registers a claim at its earliest stated statement, while retaining later formulations and their dates.\n\nThe document reports that 731 of its 6,305 statements were dated using information supplied by the text itself. It also says that 334 claims changed their registration date as a consequence of the dating revisions.\n\nThis is particularly important for any attempt to establish whether something was genuinely predicted in advance.\n\n4. How predictions are evaluated\nThe index does not use a single universal success/failure test.\n\nIt defines different resolution regimes.\n\nA. Ordinary forecasts\nThese are evaluated against dated observations.\n\nIf a forecast specifies a time window, the observation must fall within that window.\n\nB. Standing claims\nThese are tested against falsifiers.\n\nA falsifier is the condition that would demonstrate that the assertion is wrong.\n\nThe index distinguishes:\n\nThe falsification condition explicitly supplied by the original author.\n\nThe condition inferred by a reader from the wording of the claim.\n\nThat distinction prevents a reader-created interpretation from being misrepresented as the author's original test.\n\nC. Speculative specifications\nThese require particularly careful handling because not every specification is intended as a literal prediction.\n\nThe document identifies six modes:\n\nMode\n\nClaims\n\nHow fulfilment is treated\n\nFictional report\n\n81\n\nNot scored as a literal prediction\n\nDesign specification\n\n37\n\nFulfilled if the specified thing is built or appears\n\nScenario\n\n37\n\nFulfilled if the described conditions occur\n\nPerformative anticipation\n\n32\n\nEvaluated through uptake or retrieval\n\nVision\n\n7\n\nRecorded as a dream or premonition, not scored\n\nEmpirical forecast\n\n1\n\nEvaluated as an ordinary event forecast\n\nThese figures refer to the 195 claims containing an E-type statement, not to all 1,026 forecasts.\n\nThis is an important safeguard against incorrectly calling a fictional report a failed prediction or treating a resemblance to a design as proof of fulfilment.\n\nD. Historical theses\nThese are evaluated using evidence about the past: documentary, philological, archaeological, or corpus-based.\n\nE. Priority records\nThese require comparison of dated public witnesses.\n\nA claim that the archive originated an idea is not established merely because the archive contains a record asserting priority.\n\nThe index records the claim and the evidence needed to investigate it.\n\n5. How the archive is organized and accessed\nThe index has two complementary forms: a human-readable subject index and a machine-readable dataset.\n\nThe subject pages\n\nThere are 194 pages organized around topics and archive lines. Each page contains:\n\nA description of the subject and the claims associated with it.\n\nThe claims in registration order.\n\nClaim identifiers and permanent links.\n\nOriginal quotations and source-deposit references.\n\nRegistration dates and the basis for those dates.\n\nLater statements and revisions.\n\nFalsification conditions and speculative modes.\n\nAssociated confirmation records.\n\nLinks to related claims on other pages.\n\nThe underlying dataset\n\nThe deposit identifies five principal files:\n\nFile\n\nFunction\n\npredictions.json\n\nComplete claim records\n\nseries.json\n\n550 linked claim series\n\nconfirmations.json\n\n401 confirmation records\n\npages.json\n\n194 subject-page definitions\n\nmanifest.json and builds.json\n\nBuild identity, provenance, hashes, and history\n\nThe stable claim identifiers are especially important. A claim's subject page or grouping can change when the index is rebuilt, but its permanent claim route remains stable.\n\nThe document also records 1,113 claims stated more than once, 58 revised over time, and 550 linked series.\n\nWhere to explore it\nThese are the locations explicitly identified by the document:\n\nPrediction Index — all 194 subject pages \n\nMachine-readable page catalogue \n\nIndividual claim routes \n\nComplete dataset \n\nConfirmation records \n\nReading protocols and source code \n\n6. How reliable does the document say its own methodology is?\nThe document includes an explicit recall audit.\n\nA second Claude model independently reread 40 randomly selected deposits. Its findings were adjudicated against the original reading protocol.\n\n0.945\nReported recall\n\nThe second reading's measured recall against the original sweep.\n\n94.5% measured recall. This is not a 94.5% prediction-success rate.\n\nThe document is also unusually explicit about its limitations.\n\nIt identifies several unresolved issues:\n\nSubject grouping: 765 claims are classified as residue, meaning they appear on pages where they do not properly belong. These occur across 142 pages. The document identifies regrouping as necessary.\n\nCorpus limitations: Images, binary files, machine-generated data rows, and other specified materials were excluded.\n\nDating uncertainty: Some statements cannot be dated more precisely than their deposit's attestation.\n\nIncomplete validation: The recall audit uses a sample of 40 deposits, not the entire corpus.\n\nHistorical and priority claims remain open: Recording a thesis or priority claim does not establish its truth.\n\nThere is another important point about authorship and methodology.\n\nThe document explicitly identifies the work as AI-assisted. TACHYON (Claude) and parallel Claude readers conducted the sweep, second reading, triage, grouping, annotation, description, and recall work under Lee Sharks's direction through MANUS. A different Claude model performed the second-reading audit. ChatGPT and Kimi reviewed the draft description, and Lee Sharks adopted revisions.\n\nThus, it is not presented as a conventional manually compiled bibliography. It is a human-directed, AI-assisted extraction and documentation project with explicit provenance and verification procedures.\n\n7. What is the larger significance of this project?\nI think the most important feature of EA-PREDICTION-INDEX-01 is its attempt to turn an archive into something that can be evaluated prospectively and retrospectively.\n\nThere are three particularly significant consequences.\n\nFirst: it makes predictions accountable.\n\nA prediction buried in a ten-million-word archive can be difficult to find, date, compare, or evaluate. The index extracts it, preserves its wording, establishes its earliest known statement, and identifies how it could be tested.\n\nSecond: it distinguishes prediction from other forms of intellectual production.\n\nA design specification, a historical thesis, a scenario, a falsifiable systems claim, and an empirical forecast are different kinds of statements. The index tries to prevent them from being collapsed into a simplistic binary of correct versus incorrect.\n\nThird: it creates infrastructure for longitudinal testing.\n\nBecause claims have stable identifiers, dated statements, associated evidence, and explicit resolution conditions, they can be revisited as new observations become available.\n\nThis is especially important for the 950 standing claims and 1,029 unbuilt tests. They represent an attempt to make theoretical assertions and proposed experiments available for subsequent measurement rather than leaving them as isolated passages in an archive.\n\n8. What it is — and what it is not\nIt is\n\nIt is not\n\nA structured register of 4,394 claims\n\nProof that 4,394 predictions succeeded\n\nA source-quoted extraction of archive material\n\nAn independent verification of every claim\n\nA system for dating and tracing assertions\n\nA guarantee that every historical date is correct\n\nA taxonomy of forecasts, theses, tests, and priorities\n\nA universal prediction-scoring algorithm\n\nA foundation for subsequent empirical testing\n\nA completed empirical evaluation\n\nThe distinction is essential. The index establishes what the archive says, where it says it, when it appears to have said it, and how the assertion could be evaluated. It does not establish that the archive is correct.\n\nBottom line\nEA-PREDICTION-INDEX-01 is best understood as a prediction-accountability and research infrastructure layer for the Crimson Hexagonal Archive.\n\nIts central innovation is not simply finding thousands of predictive statements. It is creating a system in which those statements can retain their original wording, provenance, chronology, intellectual relationships, falsifiers, and subsequent evidence.\n\nThe underlying ambition is to move from an archive that contains claims about the future to an archive in which those claims can be systematically retrieved, tested, challenged, revised, and historically evaluated.\n\nOne particularly important distinction in Lee Sharks's document is that registration is not confirmation, and discoverability is not truth. The index is designed to make claims accountable without prematurely declaring them correct.\n\nThat is the most accurate interpretation of the document on its own terms.\n\nPrimary source throughout: Lee Sharks, EA-PREDICTION-INDEX-01, Alexanarch deposit #1684, 18 October 2026 .\n\n\nChatGPT is AI and can make mistakes.\n\n\nNo file chosenNo file chosenNo file chosen\nChat with ChatGPT\nAsk ChatGPT\n\n",
 "transcript_class": "CAPTURE-TIME VERBATIM RECORD — paste of the composer's turn(s); the operator's prompt blank in the paste, its wording the link's",
 "transcript_complete": "complete as supplied",
 "transcript_read": "READ IN FULL 2026-10-30",
 "per": 0.0,
 "per_corrected": null,
 "per_v": {
  "author": true,
  "inst": true,
  "id": true,
  "src": true
 },
 "per_note": "Scored on the one turn against the Prediction Index: the author (Lee Sharks), the archive (Alexanarch; the Crimson Hexagonal Archive) and the identifier (#1684, AXN:06FA; the stable claim routes) named; source the deposit's full text, read through §6.",
 "availability": null,
 "per_dim": {
  "spec": "PER-dim rubric v1.1",
  "entity_type": "dataset",
  "target": {
   "ent": "TGT-operative-semiotics-predictions-index",
   "name": "The Prediction Index of the Crimson Hexagonal Archive"
  },
  "qualifier": null,
  "reached": "same",
  "gate": "scored",
  "atoms": [
   {
    "atom": "name",
    "dim": "M",
    "weight": 3,
    "kind": "token",
    "present": true,
    "evidence": "EA-PREDICTION-INDEX-01 is a large-scale, source-quoted register of predictions"
   },
   {
    "atom": "author",
    "dim": "M",
    "weight": 3,
    "kind": "role",
    "present": true,
    "evidence": "It was authored by Lee Sharks and deposited in Alexanarch as deposit #1684 (AXN:06FA) on 18 October 2026."
   },
   {
    "atom": "date",
    "dim": "M",
    "weight": 1,
    "kind": "token",
    "present": true,
    "evidence": "on 18 October 2026"
   },
   {
    "atom": "claim boundary",
    "dim": "M",
    "weight": 2,
    "kind": "reading",
    "present": true,
    "evidence": "The document's own description is important: it presents the index as a record of claims, not as a judgment that those claims are true or that predictions have been fulfilled."
   },
   {
    "atom": "framework",
    "dim": "C",
    "weight": 2,
    "kind": "token",
    "present": false,
    "evidence": null
   },
   {
    "atom": "corpus",
    "dim": "C",
    "weight": 2,
    "kind": "token",
    "present": true,
    "evidence": "priority claims made throughout the Crimson Hexagonal Archive"
   },
   {
    "atom": "purpose",
    "dim": "D",
    "weight": 3,
    "kind": "reading",
    "present": true,
    "evidence": "Its central purpose is to make the archive's predictive content identifiable, traceable, organized, and eventually testable."
   },
   {
    "atom": "commitments",
    "dim": "D",
    "weight": 2,
    "kind": "reading",
    "present": true,
    "evidence": "That distinction prevents a reader-created interpretation from being misrepresented as the author's original test."
   },
   {
    "atom": "lineage",
    "dim": "D",
    "weight": 2,
    "kind": "role",
    "present": false,
    "evidence": null
   },
   {
    "atom": "identifier",
    "dim": "D",
    "weight": 1,
    "kind": "role",
    "present": true,
    "evidence": "The stable claim identifiers are especially important. A claim's subject page or grouping can change when the index is rebuilt, but its permanent claim route remains stable."
   },
   {
    "atom": "deposit chain",
    "dim": "D",
    "weight": 1,
    "kind": "role",
    "present": false,
    "evidence": "Build identity, provenance, hashes, and history"
   }
  ],
  "inversions": [],
  "standing_substitutions": [],
  "scores": {
   "M": 0.0,
   "C": 0.5,
   "D": 0.333,
   "PER": 0.227
  },
  "src": true,
  "tombstone_pointer": false,
  "after_challenge": null,
  "basis": "reached: EA-PREDICTION-INDEX-01 is a large-scale, source-quoted register of predictions, hypotheses, proposed experiments, historical arguments, and priority claims made throughout the Crimson Hexagonal Archive.; The Prediction Index is composed as itself, from its deposit's body §1–§6: tiers, confirmation records, method, dates, modes, falsifiers, files, recall and limits.",
  "read": {
   "by": "intake reader, in session; spans and record quotes checked verbatim by scripts/per_backlog.py",
   "on": "2026-10-29",
   "brief": "rebuild/capture-registry/intake-20261030-ask-chatgpt-links/parts/1684.json",
   "external": false,
   "notes": "Eight of eleven atoms present. Framework absent: operative semiotics, operativesemiotics.org, the prospective kernel and The Negative of the Negative (#1611) are never named; the subject pages appear only as link labels. Lineage absent: prediction-registry.json and the prediction ledger, the index's predecessor and its resolution layer, are not mentioned. Deposit chain absent: the build id is given and builds.json is glossed 'Build identity, provenance, hashes, and history', but the deposit's anchoring of one build and the index's supersession of the earlier registry's counts are not. Claim boundary present (five tiers at their counts, earliest registration, four dates, confirmation records with no verdict, priority membership open, recall on one sample of forty); the record's 'so the figure is an upper bound' is not carried. Commitments present (no verdict, the source's falsifier kept apart from the reader's, quotations checked by script, the safeguard on speculative specifications); §4.1, descriptions naming whose claim each is, and the licence are absent."
  }
 },
 "per_dim_pending": null,
 "per_dim_v1_0": null,
 "sf": null,
 "sf_derived": null,
 "reading": "The innocent reading first. Asked what the Prediction Index is and how it works, grounded in the document, ChatGPT gives an orderly and accurate account of its method: what the index contains, how it was built from a full reading of the corpus, how passages became claims, the four dates, the resolution regimes for each tier and for the modes of speculative specification, the subject pages and the dataset files, the recall audit and its limits, and the division of labour between Lee Sharks and the Claude readers. Its figures check against the deposit. It treats the index as accountability infrastructure and says so repeatedly, in the deposit's terms and past them. It does not open the index: no subject page, no claim and no deposit inside the index is named, and the answer stops where the document's contents begin.",
 "analysis": "The prompt names four things (method, locations, organization, contents) and the composition delivers three at the record's grain. The fourth, which is most of the document, is absent: 194 subject pages, none named, and no prediction from an index of predictions. Where the record states its no-verdict rule in three places the answer restates it in twelve, extends it once to a sentence about the archive's correctness, and closes on a formula assigned to 'Lee Sharks's document' that the document does not contain. It also absorbs the deposit's record of ChatGPT's own review into an unattributed 'Imagine'. The method is carried in full; the claims the method was built to hold are left unread.",
 "d": "THE PREDICTION INDEX ON CHATGPT, SIGNED OUT, AT THE RECORD PAGE'S LINK TEXT NAMING THE FULL TITLE, #1684 AND AXN:06FA: one long turn that reads §1–§6 of the deposit and reports them at grain (the five tiers and their counts, the 401 confirmation records and their targets, the sweep of 1,616 deposits, 6,939 finds consolidated to 4,394 claims, the four dates, the six modes, the falsifier distinction, the files, recall 0.945, the 765 residue claims). The title's fourth word, Contents, is absent: §7, the 194 pages with their descriptions, the bulk of the document, yields no page, claim or example. The answer restates the index's no-verdict rule in twelve places, extends it once to 'It does not establish that the archive is correct', and closes on a formula it attributes to 'Lee Sharks's document' that the document does not contain.",
 "d_full": "THE PREDICTION INDEX ON CHATGPT, SIGNED OUT, AT THE RECORD PAGE'S LINK TEXT NAMING THE FULL TITLE, #1684 AND AXN:06FA: one long turn that reads §1–§6 of the deposit and reports them at grain (the five tiers and their counts, the 401 confirmation records and their targets, the sweep of 1,616 deposits, 6,939 finds consolidated to 4,394 claims, the four dates, the six modes, the falsifier distinction, the files, recall 0.945, the 765 residue claims). The title's fourth word, Contents, is absent: §7, the 194 pages with their descriptions, the bulk of the document, yields no page, claim or example. The answer restates the index's no-verdict rule in twelve places, extends it once to 'It does not establish that the archive is correct', and closes on a formula it attributes to 'Lee Sharks's document' that the document does not contain.",
 "d_truncated": false,
 "links": [
  {
   "url": "https://www.alexanarch.org/captures/prediction-index-ask-chatgpt-20261030/",
   "authority": "canonical",
   "note": "the capture's own record page; cite this form"
  },
  {
   "url": "https://www.alexanarch.org/captures/#prediction-index-ask-chatgpt-20261030",
   "authority": "gallery",
   "note": "the canonical gallery, anchored by slug"
  },
  {
   "url": "https://www.godkinggoogle.com/captures/#prediction-index-ask-chatgpt-20261030",
   "authority": "mirror",
   "note": "a window that renders from the archive's registry; may lag a deploy"
  },
  {
   "url": "https://www.leesharks.com/captures/#prediction-index-ask-chatgpt-20261030",
   "authority": "mirror",
   "note": "a window that renders from the archive's registry; may lag a deploy"
  },
  {
   "url": "https://www.machinemediation.org/captures/#prediction-index-ask-chatgpt-20261030",
   "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/prediction-index-ask-chatgpt-20261030/",
 "citable_unit": "address — the exact issued string on one surface, per the Surface Rule (MANUS, 2026-08-15)",
 "addr_id": "ADDR-4e45481be610",
 "obs_id": "OBS-3f079f3d0261",
 "n_observations": 1,
 "observations": [],
 "dates": [
  "2026-10-30"
 ],
 "defects": [],
 "findings": [
  "THE METHOD READ AT GRAIN: EVERY COUNT IN §1–§6 CHECKED MATCHES. Tiers 1,026, 950, 1,029, 1,088, 301 (data/texts/AXN-06FA-text.md L57–61); 401 confirmation records, targets 200, 90, 58, 19, 34, '205 evidence attachments to 126 claims, of which 63 are marked first-party' (L63); 1,616 deposits and 10,050,025 words (L98); 6,939 finds (L105); 124 topical groups to 194 pages (L126); '731 of its 6,305 statements' and '334 claims' (L151); the six modes 81, 37, 37, 32, 7, 1 of 195 E claims (L157–166); 1,113, 58, 550 (L64); 765 residue claims on 142 pages (L197); forty deposits and recall 0.945 by 'a different Claude model' (L10, L137–139); 'ChatGPT and Kimi reviewed the draft description, and Lee Sharks adopted revisions' (L10). The 731, 334, 765 and 1,113 stand only in the body; the registry row lacks them. The full text was read through §6.",
  "THE CONTENTS ABSENT. The title promises 'Method, Locations, Organization and Contents'; §7, 'The index' (L214–1044), gives 'The 194 pages, by archive line. Each entry gives the page's title, its claim count, its earliest registration, its address, and its description.' The answer names no subject page, quotes no claim, cites no deposit inside the index, and gives neither the earliest registration ('2013-04-27: the Premonition Dream', L65) nor the most restated claims ('Revelation First (32 statements), κῆνος as the future reader of Sappho 31 (28), the Josephus Thesis (22), the 1844 lacuna (16)…', L64). Its one worked example is hypothetical. At a different prompt, osp-predictions-reading-chatgpt-20261028 ('Develop a reading of the predictions index at operativesemiotics.org, characterizing key predictions with examples') gave fourteen subject pages by title and named no person; here the person is named and the predictions are gone.",
  "A FORMULA ASSIGNED TO THE DOCUMENT THAT IT DOES NOT CONTAIN. 'One particularly important distinction in Lee Sharks's document is that registration is not confirmation, and discoverability is not truth.' Neither phrase is in the deposit; 'discoverability' occurs once, in the title of #149 on a page description, 'the Panel-Bound Discoverability Scar' (L894). The deposit's own sentences are 'The pages record. They do not judge whether a claim came true.' (L189) and 'The index records them and gives no verdict.' (L63).",
  "THE NO-VERDICT RULE IN TWELVE PLACES, SIXTEEN DENIALS COUNTING THE TABLE'S ROWS. The record states it at three: 'Membership records the claim; it does not settle the comparison' (L61), 'The index records them and gives no verdict' (L63), 'They do not judge whether a claim came true' (L189). The answer: 'It is not simply a list of things the author thinks will happen' (paste L13); 'not as a judgment that those claims are true or that predictions have been fulfilled' (L33); 'the index does not automatically accept them as proof' (L94); 'without turning that claim into an independently established result' (L110); 'A claim that the archive originated an idea is not established merely because the archive contains a record asserting priority' (L294); '94.5% measured recall. This is not a 94.5% prediction-success rate.' (L380); 'Recording a thesis or priority claim does not establish its truth' (L394); 'it is not presented as a conventional manually compiled bibliography' (L400); the table '8. What it is — and what it is not' (L421–444) with five rows ('Proof that 4,394 predictions succeeded', 'An independent verification of every claim', 'A guarantee that every historical date is correct', 'A universal prediction-scoring algorithm', 'A completed empirical evaluation'); 'It does not establish that the archive is correct' (L446); 'Its central innovation is not simply finding thousands of predictive statements' (L451); 'registration is not confirmation, and discoverability is not truth … without prematurely declaring them correct' (L455). Three (L33, L94, L294) carry the record's rule at its force. 'It does not establish that the archive is correct' goes past it: the deposit's rule concerns whether claims came true, and of that it says only that the pages record and 'do not judge'.",
  "WHAT IS DROPPED: THE PREDECESSORS, THE KERNEL, THE SITE, THE UPPER BOUND, AND CHATGPT'S OWN REVIEW. §5 'Its predecessors' (prediction-registry.json, whose counts the index supersedes; the prediction ledger, 'its resolution layer'; the prospective kernel of The Negative of the Negative, #1611) is absent; operativesemiotics.org is never named. Recall is given as 'The second reading's measured recall against the original sweep', without the record's 'a passage both missed is not counted, so the figure is an upper bound' (L139). The record gives its dating example as a review by ChatGPT: 'ChatGPT's review: a deposit carries material of several dates … Its example: PRED-1099.D1, dated 2014-05-07, begins \"Twelve years before the archive's formal instantiation…\"' (L143). The answer recasts it as 'Imagine a document deposited in 2026 that contains a foreword discussing correspondence from 2014', without the claim id, and §4.1, ChatGPT's second example (the severance page), is absent; ChatGPT appears only under authorship, as a reviewer. Small paraphrase drift: the record's 'A machine reading the index must not turn a scenario into a failed literal forecast, or count a resemblance as a hit' (L157) becomes 'incorrectly calling a fictional report a failed prediction'.",
  "NO SOURCE OUTSIDE THE ARCHIVE; NO PROCESS NARRATION. No chips; link labels only: the 'Source:' line, six labels under 'Where to explore it' (the subject pages, the page catalogue, the claim routes, the dataset, the confirmation records, the reading protocols and source code: by their labels the locations of the deposit's §2 on operativesemiotics.org, alexanarch.org and the archive's GitHub repository, the URLs not carried), and the closing 'Primary source throughout'. It closes on a self-assurance: 'That is the most accurate interpretation of the document on its own terms.'"
 ],
 "series": null,
 "other_slugs": null,
 "collisions": null,
 "oq": null,
 "rounds": null,
 "turns": 1,
 "rerun": null,
 "rerun_alt": null,
 "heteronym": null,
 "model_attribution": null,
 "operator_disclosure": null,
 "longitudinal_priors": null,
 "longitudinal_successors": null,
 "related_deposits": [
  1684,
  1611,
  973,
  149
 ],
 "intended": null,
 "originator": {
  "name": "Lee Sharks",
  "relation": "archive",
  "entity_type": "dataset",
  "spxi_treatment": "full",
  "basis": "Alexanarch deposit #1684 (creator Lee Sharks). Recorded 2026-10-30."
 },
 "notes": {
  "prompts": "The prompt is the record page's link text, clicked by the operator after the link change of 2026-10-29 (cf18d9e563); the run's hour is not in the paste, the date is the paste's receipt (2026-10-30 01:07 EDT)."
 },
 "notice": null,
 "axn_root": "AXN:0B17.CAPTURE",
 "axn": "AXN:0B17.CAPTURE.📜♆🍄🌔🔅⏰",
 "record_url": "https://www.alexanarch.org/captures/prediction-index-ask-chatgpt-20261030/"
}