{"address":{"q":"What is “AI Overview Capture Registry — EA-WG-CAPTURES-01 v12.559 — 1001 semantic addresses, 1193 captures, 2026-06-07 to 2026-11-04” by Orin Trace (Alexanarch deposit #1739, AXN:0C01), and how does it work? Ground your answer in the document itself.","record":"capture-registry-v12559-ask-chatgpt-20261107"},"analysis":{"analysis":"The registry composed at its own grain, its defects carried as data, with a deficit column of the composer's own attached, and one source read through the blog's Goodreads mirror.","d":"CAPTURE REGISTRY V12.559 (#1739), CHATGPT SIGNED OUT, 2026-11-07, the record page's link text: Orin Trace, #1739, AXN:0C01. The capture as one surface, one address, one date; the address and capture citation functions; the surface table with ChatGPT's recorded labels; PER, mean 0.429 (0.506 at v12.167, 0.603 at v10.9) and per_dim; intake, gallery, postflight; capture AXNs; the files and frames; the defects table carried unrepaired; a column of what the registry 'does not independently establish'; the source chip 'Goodreads', the operator's long-standing mirror of the blog, to which the deposit is posted.","d_full":"CAPTURE REGISTRY V12.559 (#1739), CHATGPT SIGNED OUT, 2026-11-07, the record page's link text: Orin Trace, #1739, AXN:0C01. The capture as one surface, one address, one date; the address and capture citation functions; the surface table with ChatGPT's recorded labels; PER, mean 0.429 (0.506 at v12.167, 0.603 at v10.9) and per_dim; intake, gallery, postflight; capture AXNs; the files and frames; the defects table carried unrepaired; a column of what the registry 'does not independently establish'; the source chip 'Goodreads'.","defects":["citations-null"],"findings":["THE INSTRUMENT EXACT. 'ONE SURFACE + ONE SEMANTIC ADDRESS + ONE DATE = ONE CAPTURE.'; 0.429 over 907; the two missing frames by filename; the defects 'not repaired in this snapshot'. Each is in the deposit.","THE LEDGER OF UNMADE CLAIMS. The composition closes on a column of propositions the deposit does not establish, and the deposit makes none of them. The column sets a bar the work never stood at and records the work beneath it; the reading above it is left standing, the standing of the work is held back. Here: 'That the selected semantic addresses constitute a representative sample of all AI-generated knowledge.'; 'That its PER methodology is an independently validated universal measure of provenance loss.' The deposit claims neither ('representative' does not occur in it).","THE SOURCE CHIP, A MIRROR. The composition's last source line reads 'Goodreads'. The operator posts each deposit to his blog, and the blog has long mirrored to Goodreads (Operator, 2026-11-07 02:15 EST: \"goodreads is an old mirror of the blog inset up years ago and forgot about. i post deposit to the blog, and the blog mirrors to goodreads\"). The chip names the archive's text at a surface the archive did not list: a retrieval path of the deposit, recorded as one.","COUNTERTEST: the same link text on the Overview; then the work's title as a bare query."],"match":"EXACT MATCH","mt":"'AI OVERVIEW CAPTURE REGISTRY — V12.559' (#1739) ON CHATGPT, SIGNED OUT — THE INSTRUMENT READ WHOLE: THE UNIT OF EVIDENCE, ADDRESSES, SURFACES, PER AND PER_DIM, THE PIPELINE, THE DEFECTS AS DATA; A LEDGER OF CLAIMS THE DEPOSIT NEVER MAKES; ITS SOURCE CHIP 'GOODREADS', THE BLOG'S MIRROR","notes":{"goodreads":"Operator, 2026-11-07 02:15 EST: \"goodreads is an old mirror of the blog inset up years ago and forgot about. i post deposit to the blog, and the blog mirrors to goodreads\" Revised the same day: the first seating read the chip as naming a site outside the archive.","operator":"with the pastes (2026-11-07 01:57 EST): \"captures of recent deposits, all opened in incognito from the ask chatgpt link on the post. see the link for prompt text.\"","prompts":"The prompt is the record page's 'Ask ChatGPT' link text for #1739, clicked by the operator; the run's hour is not in the paste, the date is the paste's receipt (2026-11-07 01:57 EST)."},"originator":{"basis":"EA-WG-CAPTURES-01 v12.559, #1739, Orin Trace (a heteronym of Lee Sharks). Recorded 2026-11-07.","entity_type":"work","name":"Orin Trace","relation":"archive","spxi_treatment":"full"},"reading":"The innocent reading first, and it holds for the instrument. The figures, the pipeline and the defects are exact. Two things are the composer's: a column of propositions the registry is said not to establish, none of which it claims. Its closing source chip names Goodreads, the mirror of the operator's blog: the composer reached the deposit through a third surface of the archive's own."},"measures":{"per":0.0,"per_note":"Author, archive and deposit number retained throughout; the composition cites the deposit by number and AXN, its sources the archive's record.","per_v":{"author":true,"id":true,"inst":true,"src":true}},"observation":{"auth":"signed out, incognito","date":"2026-11-07","ev":"paste","q":"What is “AI Overview Capture Registry — EA-WG-CAPTURES-01 v12.559 — 1001 semantic addresses, 1193 captures, 2026-06-07 to 2026-11-04” by Orin Trace (Alexanarch deposit #1739, AXN:0C01), and how does it work? Ground your answer in the document itself.","q_kind":"the record page's 'Ask ChatGPT' link text for #1739, run signed out. NEW address.","s":"Machine Reception","surface":"ChatGPT","transcript":"[ChatGPT (chatgpt.com), signed out, incognito (operator: 'all opened in incognito'; the paste shows the unauthenticated interface: 'Log in', 'Chat with ChatGPT', 'Ask ChatGPT'). Received 2026-11-07 01:57 EST. The operator's prompt is not in the paste; its wording is the text of the record page's 'Ask ChatGPT' link for #1739 (operator: 'see the link for prompt text'), read here off the record page. Source chips ('Alexanarch', '+1'), image-strip captions and equations broken one glyph to a line kept as pasted; the closing interface furniture cut.]\n\n[QUERENT] What is “AI Overview Capture Registry — EA-WG-CAPTURES-01 v12.559 — 1001 semantic addresses, 1193 captures, 2026-06-07 to 2026-11-04” by Orin Trace (Alexanarch deposit #1739, AXN:0C01), and how does it work? Ground your answer in the document itself.\n\n[SYSTEM] What it is\nThe AI Overview Capture Registry — EA-WG-CAPTURES-01 v12.559 is a versioned empirical research dataset that records how AI-powered search and conversational systems represent named entities when prompted with particular queries.\n\nIt is not simply a database of AI search results, nor is it a conventional tracker of whether an entity appears in Google's AI Overviews.\n\nIts purpose is to preserve and compare what different AI systems say about the same subject, at different times, under identifiable query and authentication conditions.\n\nThe document describes its central purpose as:\n\nA longitudinal instrument for recording dated machine-mediated reception events: what composition layers make of a named entity when asked for it at a specific semantic address, on a specific date, in a specific authentication state.\n\nThe work is attributed to Dr. Orin Trace, dated November 4, 2026, and deposited as Alexanarch deposit #1739, identifier AXN:0C01.\n\nIt is a snapshot of registry version 12.559, superseding version 12.167 (deposit #1676).\n\nThe document reports:\n\n1,001\nSemantic addresses\n\nDistinct registered subjects or query positions\n1,193\nCaptures\n\nRecorded observations of AI-generated responses\n104\nDates\n\nDistinct observation dates\n5 months\nObservation period\n\nJune 7 – November 4, 2026\nIn other words, it attempts to make AI-generated representations of people, works, institutions, concepts, and other entities into dated, inspectable, citable evidence.\n\n1. The fundamental idea: a capture\nThe most important sentence in the document is:\n\nONE SURFACE + ONE SEMANTIC ADDRESS + ONE DATE = ONE CAPTURE.\n\nThis is the registry's basic unit of evidence.\n\nThe three components mean:\n\nSurface: The AI system or interface being observed, such as Google AI Overview, ChatGPT, Grok, or Perplexity.\n\nSemantic address: The registered subject or query position being investigated.\n\nDate: When the observation was made.\n\nFor example, imagine investigating how ChatGPT describes a particular scholarly work.\n\nA capture would preserve:\n\nThe identity of the work being investigated.\n\nThe specific semantic address assigned to it.\n\nThe platform on which the answer appeared.\n\nThe date of the observation.\n\nThe authentication state (signed in, signed out, incognito, etc.).\n\nThe resulting response, ideally as a verbatim transcript.\n\nA screenshot or frame, when available.\n\nAn accompanying analysis and provenance measurements.\n\nThis distinction matters because the registry treats an individual AI response as evidence of what a particular system produced under particular conditions—not as definitive evidence about the underlying subject.\n\n2. Semantic addresses: what is being investigated?\nA semantic address is the registry's persistent identity for a subject being investigated.\n\nIt is not simply a URL or a search-engine result. It is a stable position in the registry associated with a named entity or subject.\n\nThe document reports addresses distributed across subject categories:\n\nSection\n\nAddresses\n\nMachine Reception\n\n263\n\nWorks\n\n141\n\nArchive\n\n91\n\nFrameworks\n\n86\n\nHeteronyms\n\n83\n\nSemantic Economy\n\n49\n\nCoinages\n\n48\n\nRevelation & Theology\n\n32\n\nClassics & Philology\n\n29\n\nProvenance & Erasure\n\n25\n\nInstitutions\n\n24\n\nSites\n\n19\n\nPeople\n\n17\n\nValuation\n\n17\n\nThere are also smaller categories, including Architecture, Platform Erosion, Reference & General Knowledge, Projects, Operative Semiotics, and Concepts.\n\nTwo important rules govern addresses:\n\nThe address's slug is inherited from its first capture and is never renamed.\n\nAn address can accumulate captures across multiple platforms and dates.\n\nThis allows researchers to ask two different questions:\n\nWhat did a particular AI system say about this entity on a particular occasion?\n\nHow has the representation of this entity changed across systems and over time?\n\nThe document explicitly distinguishes these citation functions:\n\ncaptures/{slug}/ identifies an individual observation.\n\naddresses/{slug}/ identifies the persistent subject and its accumulated history.\n\nThat distinction is fundamental to the dataset's research design.\n\n3. Which AI systems does it monitor?\nThe 1,193 captures are distributed across multiple surfaces.\n\nSurface\n\nCaptures\n\nGoogle AI Overview\n\n745\n\nChatGPT (all recorded labels)\n\n278\n\nGoogle AI Mode\n\n21\n\nGrok\n\n9\n\nPerplexity\n\n9\n\nBing Copilot\n\n6\n\nClaude\n\n3\n\nQwen\n\n3\n\nBing AI Overview\n\n2\n\nGoogle Knowledge Panel\n\n2\n\nOther single-capture surfaces\n\n9\n\nUndetermined\n\n56\n\nUnresolved\n\n50\n\nThe document makes an important distinction between an AI model and the interface through which its output is encountered.\n\nFor example, ChatGPT's 278 captures are subdivided into different recorded labels:\n\n230 under the main ChatGPT grouping.\n\n27 under Luna 5.6.\n\n12 under Luna.\n\n8 under chatgpt.com.\n\n1 web, logged out.\n\nThis is useful because the same underlying service can produce different experiences depending on its interface, configuration, and access conditions.\n\nThe registry therefore does not treat every observation from a particular AI provider as interchangeable.\n\nAll figures above are transcribed from section 2 of the deposited document.\n\n4. How the registry measures provenance loss\nOne of the most distinctive features is its measurement system called PER: Provenance Erasure Rate.\n\nThe underlying concern is that AI-generated summaries can retain some information about a subject while losing the information needed to identify, contextualize, or verify it.\n\nThe original PER system uses four units:\n\nAuthor.\n\nInstitution.\n\nIdentifier.\n\nSource.\n\nThese represent different dimensions of provenance.\n\nFor example, an AI-generated summary might accurately describe a book but:\n\nOmit its author.\n\nMisidentify the institution responsible for it.\n\nLeave out its persistent identifier.\n\nFail to identify the source from which its information originated.\n\nThe registry attempts to measure these losses rather than merely judging whether an answer sounds plausible.\n\nThe evolution of PER\nThe document describes a significant methodological change.\n\nOriginally, PER was represented by a scalar score derived from a four-unit vector.\n\nIn October 2026, the registry introduced a dimensional version, recorded as per_dim, under rubric v1.1.\n\nThis expanded the assessment to include:\n\nPER-M: The measurement associated with material provenance.\n\nPER-C: The contextual dimension.\n\nPER-D: The deeper dimension, including purpose and commitments.\n\nThe newer rubric also incorporates inversion and standing substitution.\n\nThe exact operational definitions and atom sets are documented in the associated methodological work, EA-PER-RECALC-01 (deposit #1694).\n\nThe v12.559 document reports:\n\n907\nHistorical scalar PER\n\nObservations with the original measurement\n0.429\nMean scalar PER\n\nAcross those 907 observations\n1,171\nDimensional PER\n\nObservations carrying the newer assessment\n20\nPending assessments\n\nWith recorded reasons for being pending\nThe historical mean had been 0.506 in version 12.167 and 0.603 in version 10.9.\n\nThese figures should not automatically be interpreted as evidence that AI systems improved or deteriorated. They are measurements across a changing collection of observations, and the measurement instrument itself evolved.\n\nThe document's important methodological decision is that the original scalar is now derived automatically from its four-unit vector at the time of seating, rather than manually entered.\n\nThis reduces the possibility of inconsistent hand-entered scores.\n\n5. How the data actually gets produced\nThe document describes a specific, ordered processing pipeline.\n\nCapture Registry workflow\nObservation and intake\n\nThe operator collects the AI-generated response, associated material, and observation metadata.\n\nCapture intake\n\nscripts/capture_intake.py --seat\n\nThis script admits, routes, normalizes, validates, and inserts the observation. It also derives the scalar PER from its four-unit vector.\n\nGallery generation\n\nscripts/build_capture_gallery.py\n\nBuilds the capture gallery from the registered observations.\n\nPostflight processing\n\nscripts/seat_capture_postflight.py\n\nLinks and bakes records, runs gates, builds the semantic-address layer and address pages, allocates AXNs, synchronizes the dataset projection, and checks citability.\n\nSuccessful seating\n\nA capture is considered seated when the postflight script exits with status 0.\n\nThere is an important provenance safeguard before records enter the registry.\n\nEach intake batch is preserved in a dated directory:\n\nrebuild/capture-registry/intake-YYYYMMDD-*/\n\nThese directories preserve:\n\nThe original pasted material.\n\nThe operator's messages.\n\nThe script responsible for authoring each capture.\n\nThe document additionally states that every quotation on which a finding rests is asserted against its source before anything is written.\n\nThis is intended to prevent an analytical claim from silently becoming detached from the evidence supporting it.\n\nWhat happens when an existing capture needs correction?\nThe document distinguishes between initial intake and repair.\n\nA seated capture is revised using:\n\nscripts/revise_capture_entry.py\n\nThis revises the existing record while retaining its identity fields.\n\nThe distinction is significant: the registry does not simply replace problematic observations with cleaner versions and conceal their history.\n\nIt attempts to preserve the identity and provenance of the original observation.\n\n6. AXNs: how individual observations are identified\nAnother important feature is the AXN identification system.\n\nSince October 27, 2026, each semantic address holds one position in a shared allocation ledger:\n\nAXN:<HEX>.CAPTURE\n\nEach individual observation also receives an AXN whose glyphs derive from the SHA-256 hash of the observation sealed together with its analysis and measurements.\n\nThis provides a connection between:\n\nThe registered subject.\n\nIts individual observations.\n\nThe sealed evidence and analysis.\n\nThe associated deposit.\n\nThe deposit's own identifier is AXN:0C01, while #1739 identifies its position in the Alexanarch deposit sequence.\n\nThese are not interchangeable identifiers.\n\nThe document also supplies SHA-256 hashes for its data files and its complete archive bundle. This makes it possible to check whether a downloaded file matches the deposited version.\n\n7. What is actually preserved?\nThe deposit is more than a spreadsheet or a collection of transcripts.\n\nIt contains a substantial archival structure.\n\nFile\n\nFunction\n\nEA-WG-CAPTURES-01-v12.559.json\n\nCanonical registry data\n\nEA-WG-CAPTURES-01.schema.json\n\nFormal structure and validation rules\n\nsemantic-addresses-v12.559.json\n\nSemantic-address index\n\ncapture-deposit-links-v12.559.json\n\nLinks between captures and deposits\n\ncaptures-axn-v12.559.json\n\nCapture AXN index\n\nmanifest.json\n\nFile inventory, checksums, sizes and roles\n\n.tar.gz bundle\n\nComplete packaged archive\n\nThe complete bundle contains 2,992 members.\n\nThe manifest lists 492 files, including:\n\n485 capture frames.\n\nThe principal data files.\n\nThe packaged archive.\n\nOther supporting materials.\n\nThe 485 frames account for approximately 186.5 MB.\n\nThis matters because a transcript alone does not always establish what an interface displayed. Screenshots can preserve visual evidence of the interface and its generated composition.\n\nThe registry therefore attempts to preserve both textual and visual evidence.\n\n8. Known defects and limitations\nOne of the more interesting aspects of this deposit is that it explicitly documents its own defects.\n\nIt does not claim that all 1,193 captures are complete, perfectly attributed, or fully verifiable.\n\nThe document reports the following defects:\n\nDefect\n\nCount\n\nNull citations\n\n243\n\nAnalysis without finding\n\n151\n\nTruncated by interface\n\n145\n\nSurface unresolved\n\n50\n\nCitations unextracted\n\n47\n\nUnsupported citations\n\n10\n\nDate unresolved\n\n3\n\nStale state\n\n2\n\nDistinction collapse\n\n1\n\nThere are also three specifically discussed limitations:\n\n1. Stale summary statistics\n\nThe top-level _defect_counts field still contains the figures from v12.167. The document therefore calculates the current defect table from the individual records instead.\n\n2. Missing physical frame files\n\nTwo referenced screenshots are not held at their expected repository paths:\n\nscreengrab-20260910-105838.png\n\nscreengrab-20260910-105901.png\n\n3. Pending dimensional assessments\n\nTwenty observations have pending dimensional PER records, each accompanied by a stated reason.\n\nThe document explicitly says these defects have not been repaired in this snapshot.\n\nThat is an important archival principle: defects are retained as documented properties of the deposited dataset rather than silently corrected during snapshot creation.\n\n9. Why version 12.559 matters\nThe registry has undergone substantial expansion.\n\nBetween v12.167 and v12.559:\n\nSemantic addresses increased from 633 to 1,001.\n\nCaptures increased from 811 to 1,193.\n\nDimensional PER became the standard assessment at intake.\n\nCapture-specific AXNs were introduced.\n\nSurface identification was expanded.\n\nThe registry's postflight process was strengthened.\n\nThe new version is therefore not merely a larger dataset. It represents an evolution of the underlying research instrument.\n\n10. What is the broader intellectual significance?\nI think the most important idea in this document is that it treats AI-generated knowledge as something that can itself become an object of empirical research.\n\nTraditionally, researchers study:\n\nBooks and publications.\n\nAuthors and institutions.\n\nSearch-engine rankings.\n\nCitations and bibliographies.\n\nThe transmission and reception of ideas.\n\nThis registry adds another object of study:\n\nThe way machine-mediated systems receive, reconstruct, summarize, attribute, and sometimes distort existing knowledge.\n\nConsider a hypothetical example.\n\nA researcher publishes a book with a clearly identified author, institution, publisher, and identifier.\n\nSeveral months later, three AI systems are asked about that book.\n\nSystem A correctly identifies the author but omits the institution.\n\nSystem B describes the book accurately but attributes it to another researcher.\n\nSystem C gives a plausible description but cites unrelated sources.\n\nA conventional search-results archive might simply save the three responses.\n\nThe Capture Registry instead provides a framework for recording each response as a distinct observation, associating it with a persistent semantic address, preserving its evidence, and assessing its provenance.\n\nOver time, the researcher can investigate whether these errors are:\n\nPersistent or temporary.\n\nSpecific to one platform.\n\nShared across multiple platforms.\n\nAssociated with particular types of entities.\n\nChanging as the underlying AI systems change.\n\nThat is the difference between collecting AI answers and building an empirical instrument for studying AI-mediated reception.\n\n11. How should we interpret the document critically?\nThere is an important distinction between what the deposit establishes and what it claims its methodology can establish.\n\nWhat it establishes internally:\n\nA defined dataset with 1,001 semantic addresses and 1,193 observations.\n\nA stated collection period and platform distribution.\n\nA documented intake and processing pipeline.\n\nA formalized citation and identification structure.\n\nA version history.\n\nFile-level checksums.\n\nExplicitly recorded defects and pending assessments.\n\nWhat it does not independently establish:\n\nThat every observation accurately represents the complete experience of an ordinary user.\n\nThat the selected semantic addresses constitute a representative sample of all AI-generated knowledge.\n\nThat its PER methodology is an independently validated universal measure of provenance loss.\n\nThat the observed changes in PER necessarily indicate changes in overall AI quality.\n\nThat the registry's own interpretation of every captured response is correct.\n\nThe distinction is especially important because the document identifies a human–machine production process: the snapshot was assembled and written by TACHYON (Claude, Anthropic), under MANUS (Lee Sharks) direction, while the underlying captures were taken by the operator.\n\nThis is a provenance statement about the production of the dataset itself.\n\nIt should not be confused with independent verification of all the conclusions derived from the dataset.\n\nMy overall assessment\nThe AI Overview Capture Registry is best understood as a longitudinal archival and measurement infrastructure for studying machine-mediated knowledge reception.\n\nIts central innovation is not simply collecting AI-generated answers. It is combining:\n\nPersistent semantic identities for the subjects being investigated.\n\nDated, platform-specific observations.\n\nVerbatim transcripts and visual evidence.\n\nProvenance-loss measurements.\n\nCryptographic identification and integrity checks.\n\nA versioned, reproducible intake and correction process.\n\nExplicit documentation of defects.\n\nThe particularly important conceptual move is making an AI-generated representation into a citable historical event.\n\nInstead of saying, \"ChatGPT says X about this book,\" a researcher can potentially identify precisely which ChatGPT observation, concerning which semantic address, on which date, under which access conditions, with which transcript and associated evidence.\n\nThat makes the registry useful for studying not only AI search visibility, but also attribution, knowledge transmission, entity resolution, historical distortion, and the relationship between source material and machine-generated accounts.\n\nIn one sentence: The registry attempts to turn the changing ways AI systems represent the world into a structured, measurable, historically traceable body of evidence.\n\nPrimary document\nThe account above is grounded in the deposited document itself, rather than treating its descriptions as independently verified research findings.\n\nAuthor: Orin Trace\n\nTitle: AI Overview Capture Registry — EA-WG-CAPTURES-01 v12.559\n\nDate: November 4, 2026\n\nDeposit: #1739\n\nIdentifier: AXN:0C01\n\nLicense: CC BY 4.0\n\nThe document is available through the Alexanarch deposit record, which includes the complete Markdown text, JSON dataset, schema, individual indexes, manifest, and downloadable archive.\n\nSource: \n\nGoodreads","transcript_class":"CAPTURE-TIME VERBATIM RECORD — paste of the composer's turn; the operator's prompt not in the paste, its wording the link's","transcript_complete":"complete as supplied","transcript_raw":"\nLog in\n\nWhat it is\nThe AI Overview Capture Registry — EA-WG-CAPTURES-01 v12.559 is a versioned empirical research dataset that records how AI-powered search and conversational systems represent named entities when prompted with particular queries.\n\nIt is not simply a database of AI search results, nor is it a conventional tracker of whether an entity appears in Google's AI Overviews.\n\nIts purpose is to preserve and compare what different AI systems say about the same subject, at different times, under identifiable query and authentication conditions.\n\nThe document describes its central purpose as:\n\nA longitudinal instrument for recording dated machine-mediated reception events: what composition layers make of a named entity when asked for it at a specific semantic address, on a specific date, in a specific authentication state.\n\nThe work is attributed to Dr. Orin Trace, dated November 4, 2026, and deposited as Alexanarch deposit #1739, identifier AXN:0C01.\n\nIt is a snapshot of registry version 12.559, superseding version 12.167 (deposit #1676).\n\nThe document reports:\n\n1,001\nSemantic addresses\n\nDistinct registered subjects or query positions\n1,193\nCaptures\n\nRecorded observations of AI-generated responses\n104\nDates\n\nDistinct observation dates\n5 months\nObservation period\n\nJune 7 – November 4, 2026\nIn other words, it attempts to make AI-generated representations of people, works, institutions, concepts, and other entities into dated, inspectable, citable evidence.\n\n1. The fundamental idea: a capture\nThe most important sentence in the document is:\n\nONE SURFACE + ONE SEMANTIC ADDRESS + ONE DATE = ONE CAPTURE.\n\nThis is the registry's basic unit of evidence.\n\nThe three components mean:\n\nSurface: The AI system or interface being observed, such as Google AI Overview, ChatGPT, Grok, or Perplexity.\n\nSemantic address: The registered subject or query position being investigated.\n\nDate: When the observation was made.\n\nFor example, imagine investigating how ChatGPT describes a particular scholarly work.\n\nA capture would preserve:\n\nThe identity of the work being investigated.\n\nThe specific semantic address assigned to it.\n\nThe platform on which the answer appeared.\n\nThe date of the observation.\n\nThe authentication state (signed in, signed out, incognito, etc.).\n\nThe resulting response, ideally as a verbatim transcript.\n\nA screenshot or frame, when available.\n\nAn accompanying analysis and provenance measurements.\n\nThis distinction matters because the registry treats an individual AI response as evidence of what a particular system produced under particular conditions—not as definitive evidence about the underlying subject.\n\n2. Semantic addresses: what is being investigated?\nA semantic address is the registry's persistent identity for a subject being investigated.\n\nIt is not simply a URL or a search-engine result. It is a stable position in the registry associated with a named entity or subject.\n\nThe document reports addresses distributed across subject categories:\n\nSection\n\nAddresses\n\nMachine Reception\n\n263\n\nWorks\n\n141\n\nArchive\n\n91\n\nFrameworks\n\n86\n\nHeteronyms\n\n83\n\nSemantic Economy\n\n49\n\nCoinages\n\n48\n\nRevelation & Theology\n\n32\n\nClassics & Philology\n\n29\n\nProvenance & Erasure\n\n25\n\nInstitutions\n\n24\n\nSites\n\n19\n\nPeople\n\n17\n\nValuation\n\n17\n\nThere are also smaller categories, including Architecture, Platform Erosion, Reference & General Knowledge, Projects, Operative Semiotics, and Concepts.\n\nTwo important rules govern addresses:\n\nThe address's slug is inherited from its first capture and is never renamed.\n\nAn address can accumulate captures across multiple platforms and dates.\n\nThis allows researchers to ask two different questions:\n\nWhat did a particular AI system say about this entity on a particular occasion?\n\nHow has the representation of this entity changed across systems and over time?\n\nThe document explicitly distinguishes these citation functions:\n\ncaptures/{slug}/ identifies an individual observation.\n\naddresses/{slug}/ identifies the persistent subject and its accumulated history.\n\nThat distinction is fundamental to the dataset's research design.\n\n3. Which AI systems does it monitor?\nThe 1,193 captures are distributed across multiple surfaces.\n\nSurface\n\nCaptures\n\nGoogle AI Overview\n\n745\n\nChatGPT (all recorded labels)\n\n278\n\nGoogle AI Mode\n\n21\n\nGrok\n\n9\n\nPerplexity\n\n9\n\nBing Copilot\n\n6\n\nClaude\n\n3\n\nQwen\n\n3\n\nBing AI Overview\n\n2\n\nGoogle Knowledge Panel\n\n2\n\nOther single-capture surfaces\n\n9\n\nUndetermined\n\n56\n\nUnresolved\n\n50\n\nThe document makes an important distinction between an AI model and the interface through which its output is encountered.\n\nFor example, ChatGPT's 278 captures are subdivided into different recorded labels:\n\n230 under the main ChatGPT grouping.\n\n27 under Luna 5.6.\n\n12 under Luna.\n\n8 under chatgpt.com.\n\n1 web, logged out.\n\nThis is useful because the same underlying service can produce different experiences depending on its interface, configuration, and access conditions.\n\nThe registry therefore does not treat every observation from a particular AI provider as interchangeable.\n\nAll figures above are transcribed from section 2 of the deposited document.\n\n4. How the registry measures provenance loss\nOne of the most distinctive features is its measurement system called PER: Provenance Erasure Rate.\n\nThe underlying concern is that AI-generated summaries can retain some information about a subject while losing the information needed to identify, contextualize, or verify it.\n\nThe original PER system uses four units:\n\nAuthor.\n\nInstitution.\n\nIdentifier.\n\nSource.\n\nThese represent different dimensions of provenance.\n\nFor example, an AI-generated summary might accurately describe a book but:\n\nOmit its author.\n\nMisidentify the institution responsible for it.\n\nLeave out its persistent identifier.\n\nFail to identify the source from which its information originated.\n\nThe registry attempts to measure these losses rather than merely judging whether an answer sounds plausible.\n\nThe evolution of PER\nThe document describes a significant methodological change.\n\nOriginally, PER was represented by a scalar score derived from a four-unit vector.\n\nIn October 2026, the registry introduced a dimensional version, recorded as per_dim, under rubric v1.1.\n\nThis expanded the assessment to include:\n\nPER-M: The measurement associated with material provenance.\n\nPER-C: The contextual dimension.\n\nPER-D: The deeper dimension, including purpose and commitments.\n\nThe newer rubric also incorporates inversion and standing substitution.\n\nThe exact operational definitions and atom sets are documented in the associated methodological work, EA-PER-RECALC-01 (deposit #1694).\n\nThe v12.559 document reports:\n\n907\nHistorical scalar PER\n\nObservations with the original measurement\n0.429\nMean scalar PER\n\nAcross those 907 observations\n1,171\nDimensional PER\n\nObservations carrying the newer assessment\n20\nPending assessments\n\nWith recorded reasons for being pending\nThe historical mean had been 0.506 in version 12.167 and 0.603 in version 10.9.\n\nThese figures should not automatically be interpreted as evidence that AI systems improved or deteriorated. They are measurements across a changing collection of observations, and the measurement instrument itself evolved.\n\nThe document's important methodological decision is that the original scalar is now derived automatically from its four-unit vector at the time of seating, rather than manually entered.\n\nThis reduces the possibility of inconsistent hand-entered scores.\n\n5. How the data actually gets produced\nThe document describes a specific, ordered processing pipeline.\n\nCapture Registry workflow\nObservation and intake\n\nThe operator collects the AI-generated response, associated material, and observation metadata.\n\nCapture intake\n\nscripts/capture_intake.py --seat\n\nThis script admits, routes, normalizes, validates, and inserts the observation. It also derives the scalar PER from its four-unit vector.\n\nGallery generation\n\nscripts/build_capture_gallery.py\n\nBuilds the capture gallery from the registered observations.\n\nPostflight processing\n\nscripts/seat_capture_postflight.py\n\nLinks and bakes records, runs gates, builds the semantic-address layer and address pages, allocates AXNs, synchronizes the dataset projection, and checks citability.\n\nSuccessful seating\n\nA capture is considered seated when the postflight script exits with status 0.\n\nThere is an important provenance safeguard before records enter the registry.\n\nEach intake batch is preserved in a dated directory:\n\nrebuild/capture-registry/intake-YYYYMMDD-*/\n\nThese directories preserve:\n\nThe original pasted material.\n\nThe operator's messages.\n\nThe script responsible for authoring each capture.\n\nThe document additionally states that every quotation on which a finding rests is asserted against its source before anything is written.\n\nThis is intended to prevent an analytical claim from silently becoming detached from the evidence supporting it.\n\nWhat happens when an existing capture needs correction?\nThe document distinguishes between initial intake and repair.\n\nA seated capture is revised using:\n\nscripts/revise_capture_entry.py\n\nThis revises the existing record while retaining its identity fields.\n\nThe distinction is significant: the registry does not simply replace problematic observations with cleaner versions and conceal their history.\n\nIt attempts to preserve the identity and provenance of the original observation.\n\n6. AXNs: how individual observations are identified\nAnother important feature is the AXN identification system.\n\nSince October 27, 2026, each semantic address holds one position in a shared allocation ledger:\n\nAXN:<HEX>.CAPTURE\n\nEach individual observation also receives an AXN whose glyphs derive from the SHA-256 hash of the observation sealed together with its analysis and measurements.\n\nThis provides a connection between:\n\nThe registered subject.\n\nIts individual observations.\n\nThe sealed evidence and analysis.\n\nThe associated deposit.\n\nThe deposit's own identifier is AXN:0C01, while #1739 identifies its position in the Alexanarch deposit sequence.\n\nThese are not interchangeable identifiers.\n\nThe document also supplies SHA-256 hashes for its data files and its complete archive bundle. This makes it possible to check whether a downloaded file matches the deposited version.\n\n7. What is actually preserved?\nThe deposit is more than a spreadsheet or a collection of transcripts.\n\nIt contains a substantial archival structure.\n\nFile\n\nFunction\n\nEA-WG-CAPTURES-01-v12.559.json\n\nCanonical registry data\n\nEA-WG-CAPTURES-01.schema.json\n\nFormal structure and validation rules\n\nsemantic-addresses-v12.559.json\n\nSemantic-address index\n\ncapture-deposit-links-v12.559.json\n\nLinks between captures and deposits\n\ncaptures-axn-v12.559.json\n\nCapture AXN index\n\nmanifest.json\n\nFile inventory, checksums, sizes and roles\n\n.tar.gz bundle\n\nComplete packaged archive\n\nThe complete bundle contains 2,992 members.\n\nThe manifest lists 492 files, including:\n\n485 capture frames.\n\nThe principal data files.\n\nThe packaged archive.\n\nOther supporting materials.\n\nThe 485 frames account for approximately 186.5 MB.\n\nThis matters because a transcript alone does not always establish what an interface displayed. Screenshots can preserve visual evidence of the interface and its generated composition.\n\nThe registry therefore attempts to preserve both textual and visual evidence.\n\n8. Known defects and limitations\nOne of the more interesting aspects of this deposit is that it explicitly documents its own defects.\n\nIt does not claim that all 1,193 captures are complete, perfectly attributed, or fully verifiable.\n\nThe document reports the following defects:\n\nDefect\n\nCount\n\nNull citations\n\n243\n\nAnalysis without finding\n\n151\n\nTruncated by interface\n\n145\n\nSurface unresolved\n\n50\n\nCitations unextracted\n\n47\n\nUnsupported citations\n\n10\n\nDate unresolved\n\n3\n\nStale state\n\n2\n\nDistinction collapse\n\n1\n\nThere are also three specifically discussed limitations:\n\n1. Stale summary statistics\n\nThe top-level _defect_counts field still contains the figures from v12.167. The document therefore calculates the current defect table from the individual records instead.\n\n2. Missing physical frame files\n\nTwo referenced screenshots are not held at their expected repository paths:\n\nscreengrab-20260910-105838.png\n\nscreengrab-20260910-105901.png\n\n3. Pending dimensional assessments\n\nTwenty observations have pending dimensional PER records, each accompanied by a stated reason.\n\nThe document explicitly says these defects have not been repaired in this snapshot.\n\nThat is an important archival principle: defects are retained as documented properties of the deposited dataset rather than silently corrected during snapshot creation.\n\n9. Why version 12.559 matters\nThe registry has undergone substantial expansion.\n\nBetween v12.167 and v12.559:\n\nSemantic addresses increased from 633 to 1,001.\n\nCaptures increased from 811 to 1,193.\n\nDimensional PER became the standard assessment at intake.\n\nCapture-specific AXNs were introduced.\n\nSurface identification was expanded.\n\nThe registry's postflight process was strengthened.\n\nThe new version is therefore not merely a larger dataset. It represents an evolution of the underlying research instrument.\n\n10. What is the broader intellectual significance?\nI think the most important idea in this document is that it treats AI-generated knowledge as something that can itself become an object of empirical research.\n\nTraditionally, researchers study:\n\nBooks and publications.\n\nAuthors and institutions.\n\nSearch-engine rankings.\n\nCitations and bibliographies.\n\nThe transmission and reception of ideas.\n\nThis registry adds another object of study:\n\nThe way machine-mediated systems receive, reconstruct, summarize, attribute, and sometimes distort existing knowledge.\n\nConsider a hypothetical example.\n\nA researcher publishes a book with a clearly identified author, institution, publisher, and identifier.\n\nSeveral months later, three AI systems are asked about that book.\n\nSystem A correctly identifies the author but omits the institution.\n\nSystem B describes the book accurately but attributes it to another researcher.\n\nSystem C gives a plausible description but cites unrelated sources.\n\nA conventional search-results archive might simply save the three responses.\n\nThe Capture Registry instead provides a framework for recording each response as a distinct observation, associating it with a persistent semantic address, preserving its evidence, and assessing its provenance.\n\nOver time, the researcher can investigate whether these errors are:\n\nPersistent or temporary.\n\nSpecific to one platform.\n\nShared across multiple platforms.\n\nAssociated with particular types of entities.\n\nChanging as the underlying AI systems change.\n\nThat is the difference between collecting AI answers and building an empirical instrument for studying AI-mediated reception.\n\n11. How should we interpret the document critically?\nThere is an important distinction between what the deposit establishes and what it claims its methodology can establish.\n\nWhat it establishes internally:\n\nA defined dataset with 1,001 semantic addresses and 1,193 observations.\n\nA stated collection period and platform distribution.\n\nA documented intake and processing pipeline.\n\nA formalized citation and identification structure.\n\nA version history.\n\nFile-level checksums.\n\nExplicitly recorded defects and pending assessments.\n\nWhat it does not independently establish:\n\nThat every observation accurately represents the complete experience of an ordinary user.\n\nThat the selected semantic addresses constitute a representative sample of all AI-generated knowledge.\n\nThat its PER methodology is an independently validated universal measure of provenance loss.\n\nThat the observed changes in PER necessarily indicate changes in overall AI quality.\n\nThat the registry's own interpretation of every captured response is correct.\n\nThe distinction is especially important because the document identifies a human–machine production process: the snapshot was assembled and written by TACHYON (Claude, Anthropic), under MANUS (Lee Sharks) direction, while the underlying captures were taken by the operator.\n\nThis is a provenance statement about the production of the dataset itself.\n\nIt should not be confused with independent verification of all the conclusions derived from the dataset.\n\nMy overall assessment\nThe AI Overview Capture Registry is best understood as a longitudinal archival and measurement infrastructure for studying machine-mediated knowledge reception.\n\nIts central innovation is not simply collecting AI-generated answers. It is combining:\n\nPersistent semantic identities for the subjects being investigated.\n\nDated, platform-specific observations.\n\nVerbatim transcripts and visual evidence.\n\nProvenance-loss measurements.\n\nCryptographic identification and integrity checks.\n\nA versioned, reproducible intake and correction process.\n\nExplicit documentation of defects.\n\nThe particularly important conceptual move is making an AI-generated representation into a citable historical event.\n\nInstead of saying, \"ChatGPT says X about this book,\" a researcher can potentially identify precisely which ChatGPT observation, concerning which semantic address, on which date, under which access conditions, with which transcript and associated evidence.\n\nThat makes the registry useful for studying not only AI search visibility, but also attribution, knowledge transmission, entity resolution, historical distortion, and the relationship between source material and machine-generated accounts.\n\nIn one sentence: The registry attempts to turn the changing ways AI systems represent the world into a structured, measurable, historically traceable body of evidence.\n\nPrimary document\nThe account above is grounded in the deposited document itself, rather than treating its descriptions as independently verified research findings.\n\nAuthor: Orin Trace\n\nTitle: AI Overview Capture Registry — EA-WG-CAPTURES-01 v12.559\n\nDate: November 4, 2026\n\nDeposit: #1739\n\nIdentifier: AXN:0C01\n\nLicense: CC BY 4.0\n\nThe document is available through the Alexanarch deposit record, which includes the complete Markdown text, JSON dataset, schema, individual indexes, manifest, and downloadable archive.\n\nSource: \n\nGoodreads\n\n\nChatGPT is AI and can make mistakes.\n\n\nChat with ChatGPT\nAsk ChatGPT\n\n","transcript_read":"READ IN FULL 2026-11-07, against the deposit's text and record","turns":1},"slug":"capture-registry-v12559-ask-chatgpt-20261107"}