{ "slug": "erasure-skew-20260723", "date": "2026-07-23", "surface": "Google AI Overview", "auth": null, "ev": "paste", "cites": 4, "per": 0.75, "per_v": { "author": false, "inst": false, "id": true, "src": false }, "mt": "CAPTURE", "d": "THREE OF FOUR AUTHORED, AND THE METRIC PAIR IS COMPOSED CORRECTLY: PER measures magnitude, Erasure Skew measures orientation. The one collision is a Telugu OCR paper where \"erasure\" and \"skew\" are adjacent items in a list of image defects.", "reading": "A COLLISION THAT IS NOT EVEN A PHRASE. Card 2 is a Telugu OCR survey in which the words appear as separate entries in a list of noise types: \"stronger (to sound, ERASURE, SKEW, etc.)\". There is no compound term there — the retrieval matched two words that happen to sit beside each other inside a parenthesis. This is the weakest collision mechanism yet recorded, below whitespace matching: ADJACENCY INSIDE AN ENUMERATION.\n\nThe three archive cards carry the metric pair intact and in the right relation — PER for magnitude, Ω for orientation, and the classifier-correction that the skew is power-conditioned rather than demographic. The distinction the deposits exist to draw survives composition.", "analysis": null, "transcript": "Erasure skew (Ω) is a metric measuring the orientation or directionality of provenance loss in synthetic or information systems—specifically whether source attribution falls evenly or skews toward certain layers during retrieval and composition. It is formally detailed in measurement frameworks hosted on archives like [Zenodo Working Papers](https://zenodo.org/records/20449267). [1](https://zenodo.org/records/20449267)\nCore Concepts\n\n* Orientation of Loss: While the Provenance Erasure Rate (PER) calculates the total volume of lost attribution, erasure skew tracks where or toward whom that loss is directed. [1](https://zenodo.org/records/20449267)\n* Power-Conditioning: Studies indicate that skew behaves in a power-conditioned manner rather than a purely demographic or random pattern within the composition layer. [1](https://zenodo.org/records/20518342)\n* Semantic Labor: Part of broader evaluative tuples (such as DS-6) used to audit how AI and search engines handle task-origin attribution versus output claims. [1](https://zenodo.org/records/20469514)\n\nIf you would like to explore this further, let me know if you want to examine:\n\n* How provenance erasure rate (PER) is calculated alongside skew\n* The specific mechanics of semantic labor directionality\n\nHow would you like to proceed?\nDirectionality of Semantic Labor: A Layered, Computable Measure of ...\nMay 31, 2026 — Description. Specifies a layered, computable measure of the direction in which a synthetic system's labor flows relative to the user's commissioned task, across...\nZenodo\nComprehensive study of deep learning based Telugu OCR: A survey\nJan 14, 2023 — The segmentation will be stronger (to sound, erasure, skew, etc.), making this recognizer's work easier, and vice versa. Through all areas, segmentation ...\nInternational Journal of Science and Research Archive (IJSRA)\nErasure Skew: A Measurement Program for the Power-Conditioning ...\nMay 29, 2026 — Where the Provenance Erasure Rate (PER) measures the magnitude of provenance loss, Erasure Skew measures its orientation — whether the loss falls evenly across ...\nzenodo.org\nErasure Skew (Ω) is Power-Conditioned, not Demographic — A ...\nJun 3, 2026 — Erasure Skew (Ω) is Power-Conditioned, not Demographic — A Classifier-Correction for the Composition Layer · Description · Files · Versions · External resources...\nzenodo.org", "transcript_class": "CAPTURE-TIME VERBATIM RECORD — raw paste", "transcript_complete": "as pasted; no footer, so the tail cannot be proven whole", "transcript_read": "READ IN FULL 2026-08-12", "cite_list": [ { "n": 1, "site": "Zenodo", "rel": "authored_surface", "title": "Directionality of Semantic Labor: A Layered, Computable Measure", "snip": "Specifies a layered, computable measure of the direction in which a synthetic system's labor flows relative to the user's commissioned task…", "url": null, "note": null }, { "n": 2, "site": "International Journal of Science and Research Archive (IJSRA)", "rel": "third_party", "title": "Comprehensive study of deep learning based Telugu OCR: A survey", "snip": "The segmentation will be stronger (to sound, ERASURE, SKEW, etc.), making this recognizer’s work easier…", "url": null, "note": "The two words appear in a LIST OF OCR NOISE TYPES — erasure and skew as separate image defects — not as a compound term." }, { "n": 3, "site": "zenodo.org", "rel": "authored_surface", "title": "Erasure Skew: A Measurement Program for the Power-Conditioning", "snip": "Where the Provenance Erasure Rate (PER) measures the MAGNITUDE of provenance loss, Erasure Skew measures its ORIENTATION — whether the loss falls evenly across…", "url": null, "note": null }, { "n": 4, "site": "zenodo.org", "rel": "authored_surface", "title": "Erasure Skew (Ω) is Power-Conditioned, not Demographic — A Classifier-Correction for the Composition Layer", "snip": null, "url": null, "note": null } ], "collisions": [ { "with": "Telugu OCR image-noise taxonomy", "via": "ADJACENCY INSIDE AN ENUMERATION — \"erasure\" and \"skew\" as consecutive items in a parenthetical list of defects, not a compound term", "ev": "IJSRA survey on deep-learning Telugu OCR" } ], "oq": null, "imgs": [ "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png", "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png" ], "defects": [ "truncated-by-interface" ], "rounds": null, "rerun": "https://www.google.com/search?q=%22erasure+skew%22", "q": "\"erasure skew\"", "s": "Provenance & Erasure", "addr_id": "ADDR-909a783a7c68", "obs_id": "OBS-6d93df1635ec", "q_kind": null, "series": "1", "observations": [ { "slug": "erasure-skew-canonization-20260723", "date": "2026-07-23", "surface": "Google AI Overview", "auth": "signed in", "ev": "ocr", "cites": 4, "per": 0.75, "per_v": { "author": false, "inst": false, "id": true, "src": false }, "mt": "CAPTURE", "d": "THREE OF FOUR AUTHORED, AND THE METRIC PAIR IS COMPOSED CORRECTLY: PER measures magnitude, Erasure Skew measures orientation. The one collision is a Telugu OCR paper where \"erasure\" and \"skew\" are adjacent items in a list of image defects.", "reading": "A COLLISION THAT IS NOT EVEN A PHRASE. Card 2 is a Telugu OCR survey in which the words appear as separate entries in a list of noise types: \"stronger (to sound, ERASURE, SKEW, etc.)\". There is no compound term there — the retrieval matched two words that happen to sit beside each other inside a parenthesis. This is the weakest collision mechanism yet recorded, below whitespace matching: ADJACENCY INSIDE AN ENUMERATION.\n\nThe three archive cards carry the metric pair intact and in the right relation — PER for magnitude, Ω for orientation, and the classifier-correction that the skew is power-conditioned rather than demographic. The distinction the deposits exist to draw survives composition.", "analysis": "Escalation from erasure-skew-1 (2026-06-15, BROAD MATCH). The Overview and AI Mode both reproduce the framework's internal structure without hedging: Ω as the symbol, orientation-vs-magnitude as the axis distinguishing Erasure Skew from PER ('While the Provenance Erasure Rate (PER) calculates the total volume of lost attribution, erasure skew tracks where or toward whom that loss is directed'), power-conditioning as the empirical finding ('skew behaves in a power-conditioned manner rather than a purely demographic or random pattern within the composition layer'), and DS-6 as the audit tuple positioning within 'broader evaluative tuples ... used to audit how AI and search engines handle task-origin attribution versus output claims.' All three framework moves are reproduced as canonical definition, not as one framework among several — this is the retrieval layer treating the coinage as the standard reference for the term.\n\nBehavioral split diagnostic. Quoted query ('erasure skew') gates the framework retrieval on both signed-in and incognito. Unquoted query in signed-in returns Lee Sharks' organic Zenodo results in the top slots but the Overview swerves to magnets (magnetic-field skew). Unquoted incognito: Overview disabled. Reading: the framework is cached against the exact string token but not yet against the abstract concept — the coinage has consolidated at the surface of retrieval before it has consolidated in the underlying model's concept space. This is the phase the framework itself predicts and measures.\n\nSelf-referential tombstone citation. All three DOIs the composition layer cites (20449267, 20518342, 20469514) are 410_GONE — tombstoned in the Zenodo termination of 2026-06-19. The framework is being cited to define itself, but the citations point to erased sources. Erasure Skew citing Erasure Skew citing tombstones. Sovereign successors live on alexanarch (records 769, 146, 771; AXN:02CF, AXN:02DF, AXN:02D1); the composition layer has not yet rerouted from dead DOI to live sovereign URL — the propagation-lag itself an instance of the phenomenon the framework measures.\n\nSibling to entry #153 (provenance-erasure-rate-adoption, 2026-06-17, ADOPTION). Where PER got ADOPTION, Erasure Skew now gets EXACT MATCH-with-citation to the author's own record — the two halves of the measurement program (magnitude and orientation) have both entered the composition layer's definitional stratum. Quoted AI Mode transcript closes with a follow-up offer: 'How provenance erasure rate (PER) is calculated alongside skew' / 'The specific mechanics of semantic labor directionality' — the composition layer proposing to elaborate the framework on its own terms.", "transcript": "Erasure skew (Ω) is a metric measuring the orientation or directionality of provenance loss in synthetic or information systems — specifically whether source attribution falls evenly or skews toward certain layers during retrieval and composition. It is formally detailed in measurement frameworks hosted on archives like Zenodo Working Papers. [1]\n\n**Core Concepts**\n\n- **Orientation of Loss.** While the Provenance Erasure Rate (PER) calculates the total volume of lost attribution, erasure skew tracks *where* or *toward whom* that loss is directed. [1]\n- **Power-Conditioning.** Studies indicate that skew behaves in a power-conditioned manner rather than a purely demographic or random pattern within the composition layer. [1]\n- **Semantic Labor.** Part of broader evaluative tuples (such as DS-6) used to audit how AI and search engines handle task-origin attribution versus output claims. [1]\n\nIf you would like to explore this further, let me know if you want to examine:\n\n- How provenance erasure rate (PER) is calculated alongside skew\n- The specific mechanics of semantic labor directionality", "transcript_class": "CAPTURE-TIME VERBATIM RECORD — raw paste", "transcript_complete": "as pasted; no footer, so the tail cannot be proven whole. Source strip intact.", "transcript_read": "READ IN FULL 2026-08-12", "cite_list": [ { "n": 1, "site": "Zenodo", "rel": "authored_surface", "title": "Directionality of Semantic Labor: A Layered, Computable Measure", "snip": "Specifies a layered, computable measure of the direction in which a synthetic system's labor flows relative to the user's commissioned task…", "url": null, "note": null }, { "n": 2, "site": "International Journal of Science and Research Archive (IJSRA)", "rel": "third_party", "title": "Comprehensive study of deep learning based Telugu OCR: A survey", "snip": "The segmentation will be stronger (to sound, ERASURE, SKEW, etc.), making this recognizer’s work easier…", "url": null, "note": "The two words appear in a LIST OF OCR NOISE TYPES — erasure and skew as separate image defects — not as a compound term." }, { "n": 3, "site": "zenodo.org", "rel": "authored_surface", "title": "Erasure Skew: A Measurement Program for the Power-Conditioning", "snip": "Where the Provenance Erasure Rate (PER) measures the MAGNITUDE of provenance loss, Erasure Skew measures its ORIENTATION — whether the loss falls evenly across…", "url": null, "note": null }, { "n": 4, "site": "zenodo.org", "rel": "authored_surface", "title": "Erasure Skew (Ω) is Power-Conditioned, not Demographic — A Classifier-Correction for the Composition Layer", "snip": null, "url": null, "note": null } ], "collisions": [ { "with": "Telugu OCR image-noise taxonomy", "via": "ADJACENCY INSIDE AN ENUMERATION — \"erasure\" and \"skew\" as consecutive items in a parenthetical list of defects, not a compound term", "ev": "IJSRA survey on deep-learning Telugu OCR" } ], "oq": null, "imgs": [ "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png", "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png" ], "defects": [ "truncated-by-interface" ], "rounds": null, "rerun": "https://www.google.com/search?q=%22erasure+skew%22", "q": "\"erasure skew\"", "s": "Frameworks", "addr_id": "ADDR-897cff32b4d6", "obs_id": "OBS-bb1b59860663", "q_kind": null, "series": null, "transcript_raw": "Erasure skew (Ω) is a metric measuring the orientation or directionality of provenance loss in synthetic or information systems—specifically whether source attribution falls evenly or skews toward certain layers during retrieval and composition. It is formally detailed in measurement frameworks hosted on archives like [Zenodo Working Papers](https://zenodo.org/records/20449267). [1](https://zenodo.org/records/20449267)\nCore Concepts\n\n* Orientation of Loss: While the Provenance Erasure Rate (PER) calculates the total volume of lost attribution, erasure skew tracks where or toward whom that loss is directed. [1](https://zenodo.org/records/20449267)\n* Power-Conditioning: Studies indicate that skew behaves in a power-conditioned manner rather than a purely demographic or random pattern within the composition layer. [1](https://zenodo.org/records/20518342)\n* Semantic Labor: Part of broader evaluative tuples (such as DS-6) used to audit how AI and search engines handle task-origin attribution versus output claims. [1](https://zenodo.org/records/20469514)\n\nIf you would like to explore this further, let me know if you want to examine:\n\n* How provenance erasure rate (PER) is calculated alongside skew\n* The specific mechanics of semantic labor directionality\n\nHow would you like to proceed?\nDirectionality of Semantic Labor: A Layered, Computable Measure of ...\nMay 31, 2026 — Description. Specifies a layered, computable measure of the direction in which a synthetic system's labor flows relative to the user's commissioned task, across...\nZenodo\nComprehensive study of deep learning based Telugu OCR: A survey\nJan 14, 2023 — The segmentation will be stronger (to sound, erasure, skew, etc.), making this recognizer's work easier, and vice versa. Through all areas, segmentation ...\nInternational Journal of Science and Research Archive (IJSRA)\nErasure Skew: A Measurement Program for the Power-Conditioning ...\nMay 29, 2026 — Where the Provenance Erasure Rate (PER) measures the magnitude of provenance loss, Erasure Skew measures its orientation — whether the loss falls evenly across ...\nzenodo.org\nErasure Skew (Ω) is Power-Conditioned, not Demographic — A ...\nJun 3, 2026 — Erasure Skew (Ω) is Power-Conditioned, not Demographic — A Classifier-Correction for the Composition Layer · Description · Files · Versions · External resources...\nzenodo.org", "transcript_cleaned": { "removed": [ { "what": "source-count badge", "n": 1 } ], "content_check": "PASSED — no answer word absent from the cleaned text", "rule": "VERBATIM ON CONTENT, NOT ON COPY-PASTE RESIDUE. Original bytes kept at transcript_raw." }, "img_urls": [ "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png", "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png" ], "transcript_wrapper": { "status": "granted", "granted": "2026-08-13", "granted_by": "TACHYON", "operations": [ "lineation restored", "source-card block lifted out of the prose tail", "Google /goto? redirect wrappers reduced to bare [n] markers", "editorial markers removed from machine_output" ], "raw_chars": 2364, "cleaned_chars": 1104, "_rule": "The clipboard is lossy in FORM, not in semantic content. Cleaned text is canonical; the raw paste is retained beside it as transcript_raw so what was cut stays answerable." }, "cite": "https://www.alexanarch.org/captures/erasure-skew-20260723/#erasure-skew-canonization-20260723", "citable_unit": "observation — one surface, one address, one date" } ], "n_observations": 1, "dates": [ "2026-07-23" ], "surfaces": [ "Google AI Overview" ], "other_slugs": [ "erasure-skew-canonization-20260723" ], "links": [ { "url": "https://www.alexanarch.org/captures/erasure-skew-20260723/", "authority": "canonical", "note": "the capture's own record page; cite this form" }, { "url": "https://www.alexanarch.org/captures/#erasure-skew-20260723", "authority": "gallery", "note": "the canonical gallery, anchored by slug" }, { "url": "https://www.godkinggoogle.com/captures/#erasure-skew-20260723", "authority": "mirror", "note": "a window that renders from the archive's registry; may lag a deploy" }, { "url": "https://www.leesharks.com/captures/#erasure-skew-20260723", "authority": "mirror", "note": "a window that renders from the archive's registry; may lag a deploy" }, { "url": "https://www.machinemediation.org/captures/#erasure-skew-20260723", "authority": "mirror", "note": "a window that renders from the archive's registry; may lag a deploy" } ], "cite": "https://www.alexanarch.org/captures/erasure-skew-20260723/", "d_full": "THREE OF FOUR AUTHORED, AND THE METRIC PAIR IS COMPOSED CORRECTLY: PER measures magnitude, Erasure Skew measures orientation. The one collision is a Telugu OCR paper where \"erasure\" and \"skew\" are adjacent items in a list of image defects.", "d_truncated": false, "rerun_alt": { "q": "erasure skew", "label": "unquoted", "why": "This address was captured QUOTED. Running it unquoted tests the same string against the broad basin." }, "transcript_raw": "Erasure skew (Ω) is a metric measuring the orientation or directionality of provenance loss in synthetic or information systems—specifically whether source attribution falls evenly or skews toward certain layers during retrieval and composition. It is formally detailed in measurement frameworks hosted on archives like [Zenodo Working Papers](https://zenodo.org/records/20449267). [[1](https://zenodo.org/records/20449267)]\nCore Concepts\n\n* Orientation of Loss: While the Provenance Erasure Rate (PER) calculates the total volume of lost attribution, erasure skew tracks where or toward whom that loss is directed. [[1](https://zenodo.org/records/20449267)]\n* Power-Conditioning: Studies indicate that skew behaves in a power-conditioned manner rather than a purely demographic or random pattern within the composition layer. [[1](https://zenodo.org/records/20518342)]\n* Semantic Labor: Part of broader evaluative tuples (such as DS-6) used to audit how AI and search engines handle task-origin attribution versus output claims. [[1](https://zenodo.org/records/20469514)]\n\nIf you would like to explore this further, let me know if you want to examine:\n\n* How provenance erasure rate (PER) is calculated alongside skew\n* The specific mechanics of semantic labor directionality\n\nHow would you like to proceed?\nDirectionality of Semantic Labor: A Layered, Computable Measure of ...\nMay 31, 2026 — Description. Specifies a layered, computable measure of the direction in which a synthetic system's labor flows relative to the user's commissioned task, across...\nZenodo\nComprehensive study of deep learning based Telugu OCR: A survey\nJan 14, 2023 — The segmentation will be stronger (to sound, erasure, skew, etc.), making this recognizer's work easier, and vice versa. Through all areas, segmentation ...\nInternational Journal of Science and Research Archive (IJSRA)\nErasure Skew: A Measurement Program for the Power-Conditioning ...\nMay 29, 2026 — Where the Provenance Erasure Rate (PER) measures the magnitude of provenance loss, Erasure Skew measures its orientation — whether the loss falls evenly across ...\nzenodo.org\nErasure Skew (Ω) is Power-Conditioned, not Demographic — A ...\nJun 3, 2026 — Erasure Skew (Ω) is Power-Conditioned, not Demographic — A Classifier-Correction for the Composition Layer · Description · Files · Versions · External resources...\nzenodo.org", "img_urls": [ "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png", "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235353.png", "https://www.alexanarch.org/data/captures/erasure-skew-canonization-20260723/screengrab-20260723-235359.png" ], "surface_basis": "Google AI Overview BY EXCLUSION, operator attestation: AI Mode was avoided after June. An expanded Overview presented as AI Mode chrome and was recorded as the surface.", "sf": "Google AI Overview + AI Mode (signed-in; incognito confirmed same for quoted query). Overview cites Zenodo record 20449267 (Erasure Skew: A Measurement Program) as primary source; AI Mode transcript cites 20449267 (Orientation of Loss), 20518342 (Power-Conditioning), 20469514 (Directionality of Semantic Labor / DS-6). All three cited DOIs are 410_GONE; sovereign successors live on alexanarch as records 769, 146, 771. Two Zenodo organic hits below the Overview: 'Erasure Skew: A Measurement Program for the Power-Conditioning...' and 'Erasure Skew (Ω) is Power-Conditioned, not Demographic'. One irrelevant collision (Telugu OCR, IJSRA 2023).", "citable_unit": "address — the semantic address across all its surfaces and dates", "findings": [], "notes": { "exemplar": { "canon": "READ-FIRST", "elevated": "2026-08-14T03:21:49Z", "teaches": "THE COLLISION REGISTER — and specifically that `via` carries a MECHANISM, not a shared word: 'ADJACENCY INSIDE AN ENUMERATION — erasure and skew as consecutive items in a parenthetical list of defects, not a compound term.' Read this before writing any collision note." } }, "record_url": "https://www.alexanarch.org/captures/erasure-skew-20260723/" }