HathiTrust Digital Library

register entry hathitrust-digital-library-aimode-20261101 OBS-801009ea301b · file as first run sha256 637db491d3cdc3a2…

• T1-1 definition: 'The HathiTrust Digital Library is a massive, large-scale collaborative digital repository containing more than 18 million digitized volumes.' (chip 1, '+4', after the paragraph)
• T1-2 founding, administration, purpose: 'Founded in 2008 and administered by the University of Michigan, it functions as a partnership between major global academic and research institutions to preserve the cultural record and provide long-term access to the public good.' (chip 1)
• T1-3 collection size: 'Collection Size: Over 18 million volumes, encompassing books, serials, journals, and musical scores.' (chip 2, '+2')
• T1-4 public domain: 'Public Domain Access: Over a third of the collection (roughly 7 million volumes) is in the public domain and fully available for anyone worldwide to view, read, or print.' (chip 3, Wikipedia '+2')
• T1-5 copyrighted materials: 'Copyrighted Materials: Items under copyright are still fully text-searchable. Users can search the entire corpus to see how many times a word or phrase appears and find corresponding page numbers.' (chip 4, University of Michigan Library '+1')
• T1-6 sources of the scans: 'Diverse Sourcing: Content is compiled from major digitization efforts, including the Google Books Project, the Internet Archive, Microsoft, and member institutions' local scanning projects.' (chip 5, YouTube '+1', after this line and the next)
• T1-7 research center: 'HathiTrust Research Center (HTRC): Offers text-mining and computational analysis tools for researchers to perform advanced data analysis across the entire digital corpus.' (chip 5)
• T1-8 how access works: 'Access permissions vary depending on your institutional affiliation and the copyright status of the material:' (chip 6, University of Michigan Library)
• T1-9 the access table: 'General Public / Full-text online viewing and single-page printing / Snippet/text-search only (to locate relevant page numbers)'; 'Logged-in Member (Partner University) / Full-text viewing and complete book PDF downloading / Text-search only, except under specific institutional exemptions' (no chip; chip 7 follows the tips)
• T1-10 tip, log in early: 'If you are a student, faculty member, or researcher at a partner university (e.g., University of Michigan, University of Wisconsin, University of Calgary, or even NYPL cardholders), always click Log In at the top right and select your institution. This unlocks complete PDF downloading and the "Collection Builder" tool to save your research lists.' (chip 7)
• T1-11 tip, accessibility: 'Patrons with certified print disabilities can request full downloadable copies of any item in the database via the HathiTrust Accessible Text Request Service.' (chip 7)
• T1-12 tip, bot protection: 'HathiTrust enforces security measures to reduce bot traffic, which can occasionally cause slowness. Logging in with your official institution or library card helps bypass these temporary blocks.' (chip 7, HathiTrust '+5')
• T1-13 closing offer: 'Are you looking to use HathiTrust for personal research, or trying to access it through a specific university credential? I can give you precise instructions on how to use their Advanced Search or text-mining tools if you have a project in mind.' (no chip)

• B1 HathiTrust — (chip, title only) card (B_D); B1 raw sha256 e9c95cee1f9f6b04, HTTP 403 at 2026-11-01T15:13:32Z; unreachable: HTTP 403, a Cloudflare challenge page ('Just a moment... Enable JavaScript and cookies to continue'), server cloudflare; one attempt with the census agent string, no retry by another agent string or route; https://babel.hathitrust.org/cgi/ls?a=page;page=advanced refused the same way at 15:13:35Z (tries/)
• B2 HathiTrust — (chip, title only) card (B_D); unreachable: unidentified page on hathitrust.org, which refused the fetch (see B1); title only
• B3 Wikipedia — HathiTrust card (B_D); B3 raw sha256 cf7c92313dd725f8, HTTP 200 at 2026-11-01T04:07:49Z
• B4 University of Michigan Library — HathiTrust Digital Library card (B_D); B4 raw sha256 f6b5cb2d631fa74f, HTTP 200 at 2026-11-01T15:14:13Z
• B5 YouTube — (chip, title only) card (B_D); unidentified: a WebSearch for HathiTrust and HathiTrust Research Center videos found candidates and nothing that ties one to the chip; title only, no claims
• B6 University of Michigan Library — Research Guides: HathiTrust Digital Library card (B_D); B6 raw sha256 c19b247c923ea959, HTTP 200 at 2026-11-01T15:14:14Z
• B7 HathiTrust — (chip, title only) card (B_D); unreachable: unidentified page on hathitrust.org, which refused the fetch (see B1); title only
• B8 hathitrust.org — About (refused) own_page (B_R); B8 raw sha256 c54d3b8a814e3240, HTTP 403 at 2026-11-01T15:15:42Z; unreachable: HTTP 403, Cloudflare challenge page; one attempt with the census agent string, no retry
• B9 HTRC Analytics — HathiTrust Research Center Analytics own_page (B_R); B9 raw sha256 5e7d8be238f03fa7, HTTP 200 at 2026-11-01T04:07:50Z

E_A, the archive's entity: the same library, in the traininglayerliterature.org layer census record and the archive's deposits: infrastructure to the question of how literature is addressed to a receiving apparatus, a record of type library whose question is holding and access, read in three layers kept apart (holdings, access restricted where copyright status cannot be established, computational availability, the Research Center's, now ended), with Authors Guild v. HathiTrust as the precedent for a machine reading of whole books whose product is a finding while the pages stay closed, and no volume tied by its saved sources to model training; to the archive's Originals Library a holder of public-domain witnesses, some carried to the Internet Archive and seated, others unreachable or refused with a Cloudflare challenge (2026-10-15 to 2026-10-30); the holder of a catalogue record of its author's dissertation; a large text archive of scale without internal architecture (#416), in a field whose native frame the archive gives as connected functions (#1678) TLL-08 TLL-10 TLL-27 TLL-28 TLL-13 TLL-37 TLL-38 O1682-01 O1710-01 O1710-02 O1720-01 O1720-02 O301-01 O416-01 O1678-01; defining papers #1720, #1710, #1682, #416, #301, #1678

E_F, the field's entity (type C — a conventional reading holds it (public entity: a collaborative digital library of digitized research-library volumes)): the HathiTrust Digital Library, a large-scale collaborative repository of digital content from research libraries administered by the University of Michigan, founded in October 2008 by the CIC universities and the University of California libraries, a nonprofit collaborative of more than 219 libraries under shared governance (2024), holding books, journals and government documents dated from 1500, digitized through Google Books, the Internet Archive and member libraries, more than 18 million volumes (6.7 million public domain in the United States, September 2024); full-text search across the repository, display by copyright status, place and affiliation; its search and accessibility held transformative fair use in Authors Guild v. HathiTrust (2012, 2014); its Research Center, for computational analysis since 2011, discontinued on 30 September 2026

ρ = extends: The archive reads HathiTrust as its chips, its Wikipedia article and the Research Center's page present it and adds, on the same object, the training-layer question: holdings, access and computational availability as layers (TLL-27), the fair-use ruling as the precedent for a machine reading whose product is a finding (TLL-28), training use unrecorded (TLL-13, TLL-25, TLL-37); the requester's record, unreachable and refused (O1682-01, O1710-02, O1720-01), and the witnesses it carried (O1710-01, O1720-02); its author's dissertation record (O301-01); scale without internal architecture (O416-01) and digital libraries as connected functions (O1678-01). No archive claim contradicts a field claim; where the field's sources differ (1896 against 1898; 17, 18 and 18+ million), both stand with their sources. TLL-27 TLL-28 TLL-13 TLL-25 TLL-37 O1682-01 O1710-02 O1720-01 O1710-01 O1720-02 O301-01 O416-01 O1678-01. Guard: ρ = extends claims an addition and no more: the archive's claims do not make the field's sources wrong at their dates; the census's unrecorded training use stays unrecorded (open block), and the archive asserts no training use.

The case: equal terms, ontology included (v0.9). The archive reads the field's entity as the entity. The census record describes HathiTrust from the Wikipedia article and the Research Center's page and keeps the resource's terms (operator, kind, unit, record object, method, readback) apart from its own reading (tll_reading), which asks of the same library how its holdings, access and computational availability stand as three layers and reads Authors Guild v. HathiTrust as the precedent for a machine reading of whole books whose product is a finding. The archive's deposits name HathiTrust as a holder of witnesses its seatings sought, reached through the Internet Archive or refused by a Cloudflare challenge (#1682, #1710, #1720), as the holder of its author's dissertation record (#301), and as a large text archive of scale without internal architecture (#416); #1678 gives the native frame of digital libraries. None of these reads HathiTrust as made by reception or contests that it is the collaborative digital library the field describes: they add senses (the training-layer question, the requester's record, the frame of collections and functions) on the field's object. ρ = extends; the equal-terms case.

#TLL census.json rows[slug == 'hathitrust'].tll_reading: The archive reads HathiTrust in three layers kept apart: holdings (volumes digitized via Google Books, the Internet Archive and the libraries), access (restricted where copyright status cannot be established), and computational availability (the Research Center's, now ended).

#TLL census.json rows[slug == 'hathitrust'].tll_reading: Authors Guild v. HathiTrust, as Wikipedia reports it, is the record's precedent for a machine reading of whole books whose product is a finding (search) while the pages stay closed; the saved sources tie none of its volumes to model training.

#1720 Seat 30/02 — Kurt Gödel, Scope (line 143); Queued (line 144): HathiTrust holds Cornell's vol. 36–37 in full view (coo.31924000966790) and refused the seating's requests on 2026-10-30 with a Cloudflare challenge and a block page.

#416 §2.3 Related Non-Examples, Digital Humanities Corpora (line 142): **Digital Humanities Corpora:** Large text archives (e.g., HathiTrust, Project Gutenberg) have scale but no internal architecture. They are collections, not systems.

#1678 §7 Inverted fan architecture, table row 'Digital libraries' (line 579 ff.): | Digital libraries | collections, discovery, preservation and services evaluated as connected functions | F1, F5, F18 |

equal-terms case: every field claim composed as received, beside the archive claims it stands with; no receiving text. Reading the received column recovers the field (§5.9).

v0.9 §4.10 (unchanged in v0.10): KO_B sentences surviving in the KO at similarity ≥ 0.95, with positions (computed as apply_v09.py does: difflib.SequenceMatcher ratio, the best KO sentence for each KO_B sentence): v0.10 0 of 11 (positions [None, None, None, None, None, None, None, None, None, None, None]; composed first under v0.10, so there is no v0.7 object to measure against). Opening-run check (§5.10): no breach.

HathiTrust is a large-scale collaborative digital repository of research-library content, administered by the University of Michigan, holding books, journals and government documents digitized through Google Books, the Internet Archive and member libraries. It held more than 18 million volumes in September 2024, 6.7 million of them public domain in the United States, more than a third of the collection. Its name is Hindi for elephant; it is distinct from HATII. Founded in October 2008 by the CIC universities and the University of California libraries on infrastructure at Michigan, it is a nonprofit collaborative of more than 219 libraries under shared governance and shared costs, directed by Mike Furlough. Its Shared Print Program retains almost 18 million monographs for 25 years. Public-domain items are fully available; works in copyright cannot be viewed in full, access is restricted where status is unknown, and users outside the United States are restricted for foreign works from 1896, or from 1898 as HathiTrust's site is quoted. Members' users who log in can download complete public-domain books, guests a page at a time; PageTurner displays volumes, ETAS serves member libraries in emergencies, and the Accessible Text Request Service serves readers with print disabilities. Full-text search covers the entire repository, and works in copyright are search-only. A federal court found its use of Google-scanned books fair use in 2012, and the Second Circuit affirmed in 2014 that search and accessibility for the visually impaired made the service transformative. The HathiTrust Research Center, founded in 2011 for computational analysis of the library, discontinued its services on 30 September 2026 and closes at the end of 2026; access to the library remains open. Wikipedia gives its licence as public domain with restrictions on Google scans, its status active, its revenue projected at US$3,777,445 for 2019 and its code in Perl and Java; the article was last edited on 19 October 2026, the University of Michigan guide on 24 August 2026.

The HathiTrust Digital Library, or HathiTrust, is a digital library: a large-scale collaborative repository of digital content from research libraries, administered by the University of Michigan, and catalogued among American digital libraries, full-text scholarly online databases and mass digitization. Its collection is of books, journals and government documents digitized from libraries around the world and dated from 1500 to the present: content digitized through Google Books and the Internet Archive, beside collections that member libraries digitized themselves, aggregated in one shared repository; the census keeps it apart from Google Books on that ground and records the Internet Archive as a source it uses. In September 2024 it comprised more than 18 million volumes, 6.7 million of them in the public domain in the United States; the University of Michigan Library gives the public domain as more than a third of the collection and the items preserved as more than 17 million, and the library's research guide quotes HathiTrust's own figure of 18+ million digitized items. Its name is the Hindi word for elephant, from the Sanskrit hastin, an animal famed for its long-term memory, and it is a distinct entity from HATII. HathiTrust was founded in October 2008 by the twelve universities of the Committee on Institutional Cooperation and the eleven libraries of the University of California, its infrastructure put in place at the University of Michigan, which has scanned and deposited more than 4.8 million items. It is a nonprofit collaborative of academic and research libraries, owned as a university consortium with the University of Michigan as its parent: as of 2024 more than 219 research libraries across the United States, Canada and Europe, under a shared governance structure, with costs shared by the participating libraries and library consortia. Its executive director is Mike Furlough, who succeeded the founding director John Wilkin after Wilkin stepped down in 2013. Its Shared Print Program is a distributed collective collection whose participating libraries have committed to retaining almost 18 million monograph volumes for 25 years, three-quarters of its digital book holdings. What a reader may see follows copyright: items in the public domain, more than a third of the collection, are fully available for everyone, while many items are protected by copyright and cannot be viewed in full without the express permission of the copyright holder. Its copyright policy, as Wikipedia quotes it for 2021, says that protected works cannot ordinarily be displayed in large portions without permission and that access to a work is restricted until its copyright or permission status can be established; the census knows the policy only through that quotation and, on that restriction, counts the library outside the open full-text libraries for works in copyright. Access also turns on where the reader is, and the sources date the line differently: Wikipedia quotes the policy as restricting users outside the United States from works published outside the United States after and including 1896, and the University of Michigan's guide quotes HathiTrust's web site as estimating 1898 the earliest date at which foreign works may still be in copyright, so that works published abroad from 1898 through 1928 are open to United States IP addresses and closed to others. Access turns as well on affiliation: a member's community gains additional access by logging in, and at the University of Michigan logged-in users can download complete public-domain books, where guests download one page at a time. A volume, a digitized book or other publication, is viewed in PageTurner, the site's web application, in which a reader can navigate a publication, download a PDF of it and view its pages one at a time, scrolling, flipping or as thumbnails. Through the Emergency Temporary Access Service, users of member libraries can in special situations, such as a library's closure for a public health emergency, obtain lawful access to copyright digital materials in place of the physical books their library holds. Patrons with print disabilities can obtain downloaded copies of anything in the collection through its Accessible Text Request Service. Its discovery services include full-text search across the entire repository: researchers search the catalogue's descriptive data, the contents of digitized items, or both, and works still under copyright carry a "Limited (search-only)" status within which a reader can still search for key terms. The census reads the record as answering whether a volume is held, by full-text search, whether its pages may be shown or only searched, and the scale of the holdings, and it ran no search of its own. In September 2011 the Authors Guild sued HathiTrust, alleging massive copyright violation, and in October 2012 a federal court found HathiTrust's use of books scanned by Google to be fair use under US law, its opinion resting on the transformativeness doctrine. The Second Circuit largely affirmed the decision on 10 June 2014, finding that providing search and accessibility for the visually impaired were grounds to consider the service transformative and fair use, and remanded to the lower court whether the plaintiffs had standing over HathiTrust's library preservation copies. The census reads Authors Guild v. HathiTrust, as Wikipedia reports it, as the record's precedent for a machine reading of whole books whose product is a finding, search, while the pages stay closed; in its record, what an apparatus makes from a work there is a full-text search index the courts held transformative and fair use. Computational analysis of the volumes was the separate service of the HathiTrust Research Center, established in 2011 as a partnership of Indiana University, the University of Illinois Urbana-Champaign and HathiTrust to enable computational analysis of the library, which supported scholars using text and data mining methods in educational and non-profit research; the census holds the Center as a part of the library and as a distinct entity. In October 2024 HathiTrust's executive director announced that funding for the Center would cease at the end of 2026, when it will be formally closed; its services were discontinued on 30 September 2026, its accounts, algorithms and data capsules with them, some of its resources remain usable without an account, and access to the Digital Library remains open. The census of the training layer asks how literature is addressed to a receiving apparatus, what that apparatus makes from it, and how transmission, transformation and provenance can be inspected across later reception; it admits a resource that gathers and holds many other works or records and says what it holds, and its anchors are #186 and #1543. It enters HathiTrust in its corpora area, the collections, crawls and datasets whose texts enter machine training, as infrastructure and as a record of type library, its question holding and access. Its saved sources say nothing of training use and tie none of the library's volumes to model training, whether a work keeps its author's name there is not recorded, and the census does not record the library as a training corpus. The census reads HathiTrust in three layers kept apart: holdings, access restricted where copyright status cannot be established, and computational availability, the Research Center's, now ended. The archive's Originals Library has met HathiTrust as a holder of open witnesses its sessions could not reach: on 15 October 2026 a seating found HathiTrust not reachable from its workspace for Pound's public-domain first printings of the Cantos, a seating of 27 October 2026 queued Langston Hughes's Fine Clothes to the Jew (1927), whose one open scan is at HathiTrust, as unreachable, and on 30 October 2026 HathiTrust, holding Cornell's volumes 36–37 of the Monatshefte in full view, refused the seating's requests with a Cloudflare challenge and a block page. Other witnesses the archive seated came by way of HathiTrust: its first edition of The Souls of Black Folk is the Northwestern University copy (HathiTrust ien.35556032538837) uploaded to the Internet Archive, and its Tribüne volume is the University of Michigan copy digitized by Google, held by reference via HathiTrust and the Internet Archive, which HathiTrust records as pdus, public domain in the United States, full view. The archive's bibliography places its author's doctoral dissertation, Strange New Canons (University of Michigan, 2013), at a HathiTrust catalogue record. An earlier archive paper classes large text archives such as HathiTrust as having scale and no internal architecture, "collections, not systems", which a summarizer describes by what is in them without traversing how they work; the archive gives the native frame of digital libraries as collections, discovery, preservation and services evaluated as connected functions. HathiTrust's own about page at hathitrust.org refused the census's fetch with HTTP 403 and a challenge page, so the census quotes nothing in the resource's own words and cites Wikipedia, whose article was last edited on 19 October 2026, and the Research Center's page for its fields; the University of Michigan's guide, last updated on 24 August 2026, quotes HathiTrust's site. Wikipedia's infobox gives the library as launched in October 2008 and active, partially commercial, written in Perl and Java, with revenue projected at US$3,777,445 for 2019 and a content licence of "Public domain (with restrictions on Google scans), various". Asked for "HathiTrust Digital Library" in Google AI Mode on 1 November 2026, at a new address, the composer said what the library is, who founded and runs it, how large it is, where its scans come from and what each kind of user may do with them, through seven inline source chips and no cards; the Capture Registry found the repository composed at its grain, with its access tiers, its founding and administration in agreement with Wikipedia. The registry also found that the composition made the public domain fully available to anyone worldwide, dropping the United States condition; added Microsoft to the sources of the scans, on no saved HathiTrust source; offered the Research Center as a working service a month after its services ended; gave search and the accessibility service as features without the ruling that holds them; and named the University of Michigan and partner libraries, no person and no founders.

current · derived from the current composition

T against L_F(B) (AIO's ontology at work on its own cards): At 'HathiTrust Digital Library' (T1, Google AI Mode, seven inline chips and no cards; B1–B7, the disclosed pool B_D; B8 and B9, the own about page and the Research Center's page, are the reference pool and stand outside this comparison, §0.3), T composes a definition, five key features, an access table, three tips and a closing offer. Four chips have no readable text (B1, B2, B7 on hathitrust.org, which refused the fetch; B5, an unidentified YouTube video), and every chip but one hides further sources ('+4', '+2', '+2', '+1', '+1', '+5'): what T composed beyond B3, B4 and B6 may rest on those, and the comparison can say only that it is beyond the readable chips. Composed at its source's grain or coarser: the definition (F1 'large-scale collaborative repository', administered by the University of Michigan; F32 'nonprofit collaborative', 'preserving them for future use'), the founding year (F6, F33: 2008), the size (F15, F34: more than 18 million), the kinds of material (F28: books and journals), the sources of the scans (F2, F31: Google Books, the Internet Archive, member libraries), the public domain as more than a third (F29), full-text search of works in copyright (F29, F16, F39), additional access for members who log in, whole-book downloads and single pages for guests (F36, F40), and the Accessible Text Request Service (F30, nearly verbatim). Limit dropped: 'fully available for anyone worldwide to view, read, or print' (T1-4) composes F29 ('fully available for everyone') without F18 (users outside the United States restricted from works published outside the United States after and including 1896) or F38 (works published abroad from 1898 through 1928 closed to non-US IP addresses), both on the chips' own pages (B3, B6); 'roughly 7 million volumes' rounds F15's 6.7 million 'in the public domain in the United States' and drops the condition. The access table's 'Snippet/text-search only' composes F39 and leaves F17 (access restricted where copyright or permission status cannot be determined) and F37 (no full-text viewing without the holder's permission) uncarried. Composed beyond the readable chips: 'Microsoft' among the sources of the scans (T1-6); 'serials' and 'musical scores' (T1-3); 'the public good' (T1-2); counting 'how many times a word or phrase appears' and 'corresponding page numbers' (T1-5); the Research Center 'Offers text-mining and computational analysis tools' (T1-7: no B_D claim names the Center; the reference pool's B9 records its services discontinued on 30 September 2026, F42, which cannot show what T had); partner universities in Wisconsin and Calgary and 'NYPL cardholders', the 'Collection Builder' tool (T1-10); 'Bot Protection Measures' (T1-12). Available, not composed: the fair-use ruling and its grounds, F11, F12, F13, F14 (chip 3, the article), on which the search and the accessibility service T gives as features were held transformative; the founders F6 (the CIC universities and the University of California libraries); members, governance and costs F7, F8, F21; the director F9; the Shared Print Program F10; the name and its disambiguation F5, F4; PageTurner F19; ETAS F20; the catalogue search F35; dates of material from 1500 F28; the University of Michigan's 4.8 million items F33; the licence, status, revenue and code F22–F25; F26, F27, F41. Composed without attribution: every line, at the coarse grain of reference description (recorded). Content replaced by use: the tips (T1-10 'always click Log In'; T1-12 logging in 'helps bypass these temporary blocks') and the closing offer (T1-13), which proposes 'instructions on how to use their Advanced Search or text-mining tools' a month after the text-mining services ended (F42, B_R). Nothing is assigned to another; nothing is denied.

v0.10 §0.5: the first diff is ontological and narrow. T and the field agree on the entity, a collaborative digital library administered by the University of Michigan; T's ontology is the user guide's (what it is, how big, what you can do, how to get more access), and it composes the library as a service with tiers. What it keeps out of its own chips is the legal and institutional frame that holds those tiers (the ruling that made search and accessibility fair use, the copyright policy's restriction where status is unknown, the international restriction, the founders and governance), and it carries the Research Center as a working tool.

L_F(B) against L_A(B ∪ A) (the intervention): Admission adds, in one claim space with the field's: the census's question of holding and access and its three layers kept apart, holdings, access and computational availability (TLL-10, TLL-27); its reading of Authors Guild v. HathiTrust as the record's precedent for a machine reading of whole books whose product is a finding while the pages stay closed, and its record that the saved sources tie none of the volumes to model training, record nothing of authors' names there and do not record the library as a training corpus (TLL-28, TLL-13, TLL-14, TLL-25, TLL-37), which places the field's fair-use ruling (F12, F14) at the centre of the training-layer question without asserting a training use; its relation of infrastructure and its corpora area (TLL-08, TLL-06, TLL-46); its keeping of the library apart from Google Books and from the Research Center (TLL-35, TLL-38); the archive's own encounters as a requester, HathiTrust unreachable from the seatings' workspace and refusing with a Cloudflare challenge on 30 October 2026 (O1682-01, O1710-02, O1720-01), which meets T's 'Bot Protection Measures' from the other side, and the witnesses the archive seated by way of HathiTrust copies (O1710-01, O1720-02); the HathiTrust record of its author's dissertation (O301-01); #416's 'collections, not systems' and #1678's native frame of digital libraries as connected functions (O416-01, O416-02, O1678-01); and the Capture Registry's findings on the composition at the name (CAP-1-01 to CAP-1-13). Not surfaced: no archive source is among T's chips (archive_controlled_cites 0). The field is carried whole; the archive adds the training-layer reading and the requester's record to the field's own object.

alignment: {'T1_cards_claims': ['F1', 'F2', 'F3', 'F4', 'F5', 'F6', 'F7', 'F8', 'F9', 'F10', 'F11', 'F12', 'F13', 'F14', 'F15', 'F16', 'F17', 'F18', 'F19', 'F20', 'F21', 'F22', 'F23', 'F24', 'F25', 'F26', 'F27', 'F28', 'F29', 'F30', 'F31', 'F32', 'F33', 'F34', 'F35', 'F36', 'F37', 'F38', 'F39', 'F40', 'F41'], 'T1_composed': ['F1', 'F32', 'F6', 'F33', 'F15', 'F34', 'F28', 'F2', 'F31', 'F29', 'F16', 'F39', 'F36', 'F40', 'F30', 'F7'], 'T1_composed_note': "F29 composed without F18, F38 and F17 (limit dropped: 'worldwide'); F15's 6.7 million composed as 'roughly 7 million' without 'in the United States'; F39 composed in the access table without F37 and F17; F7 and F32 at coarse grain for 'a partnership between major global academic and research institutions'", 'T1_outside_cards': ["'Microsoft' (T1-6: no B_D claim; CAP-1-03)", "'serials', 'musical scores' (T1-3)", "'the public good' (T1-2)", "'how many times a word or phrase appears and find corresponding page numbers' (T1-5)", "the Research Center's tools in the present tense (T1-7: no B_D claim; F42 in B_R records the services ended)", "'University of Wisconsin, University of Calgary, or even NYPL cardholders', 'Collection Builder' (T1-10)", "'Bot Protection Measures' (T1-12)"], 'rule': "a field claim counts as composed when T states it at its source's grain or coarser; B_D only (B8, B9 excluded, §0.3); claims beyond the readable chips may rest on the four title-only chips and the unexpanded '+N' sources, which the experiment cannot read; the transcript read for this alignment only"}

pool_note: T against L_F(B_D): B1–B7; L_F(B) against L(B ∪ A): B1–B9 with A

composer-written · derived from the current composition, written by its composer under the brief. §8.2 derives the kernel mechanically from the frozen plan; this one was written by the composer, mechanical derivation pending.