Creative AI Tools & Resources Database (Creative AI Lab)

register entry creative-ai-lab-database-aio-20261101 OBS-53073fe9a7b0 · file as first run sha256 b6fd7f440537cd39…

• T1-1 definition: 'The Creative AI Lab is a joint research partnership between Serpentine Arts Technologies and the Department of Digital Humanities at King's College London.'
• T1-2 Launch Date: 'July 9, 2020'
• T1-3 Partners & Funders: 'Serpentine Galleries, King's College London, the Arts and Humanities Research Council (AHRC), and Google's Artists + Machine Intelligence Grant.'
• T1-4 Core Mission: 'To demystify back-end AI and machine learning (ML) systems, surface technical knowledge from artistic practices, and build institutional capacities for the cultural sector.'
• T1-5 Purpose: 'An ongoing online database aggregating machine learning and artificial intelligence tools for artists, designers, engineers, curators, and researchers.'
• T1-6 Evaluation Criteria: 'Assesses tools based on transparency, creative malleability, technical interoperability, research caliber, and public value.'
• T1-7 Categories Included: 'Audio: Mustango, Synplant 2, Stable Audio, RAVE, Neutone, Semilla, DDSP'; 'Visual / 3D: Sora, Alias, Common Sense Machines, Runway Gen-2, DALL-E 3, Alpaca, Exactly, Stable Diffusion'; 'Text / Data / Datasets: OBELICS, Chatbot Arena, Perplexity Co-Pilot, Julius, LAION-5B'; 'Detection / Tracking: OpenPose, PoseNet, DeepFaceLab, YOLO, Mask R-CNN'
• T1-8 closer: 'Would you like to explore a specific category from the database (such as audio, visual/3D, or text/datasets) or learn more about their research workshops and tutorials?'

• B1 Serpentine — Creative AI Lab - Serpentine Galleries card; the census copy creative-ai-lab-database--serpentine, B1.html sha256 22f3217cec73cee1; refetched HTTP 200 at 2026-11-01T06:15:20Z (B1-fetch.html, Cloudflare tokens differ)
• B2 Creative AI Lab — Creative AI Lab (creative-ai.org/about) card; B2.html sha256 2e3670c5e2ac4c73, HTTP 200 at 2026-11-01T06:15:41Z
• B3 King's College London — Creative AI Lab | King's College London card; B3.html sha256 19a39918da734ebc, HTTP 200 at 2026-11-01T06:15:19Z
• B4 Creative AI Lab — Creative AI Lab (creative-ai.org, the database) card; B4.html sha256 5e5a3e6207cc3677, HTTP 200 at 2026-11-01T06:15:15Z; with its feed, 55 pages, B4-data/
• B5 King's College London — The Department of Digital Humanities and the Serpentine launch Creative AI Lab website (news, 9 July 2020) census copy creative-ai-lab-database--kcl, B5.html sha256 51fff0bbedd32966, HTTP 200; not refetched
• B6 Notion — Creative AI Database (proud-paprika-325.notion.site; unread) census copy creative-ai-lab-database--notion, B6.html sha256 1d851571963c04c7, HTTP 200, 6 characters of text: unread (script-rendered)
• B7 HOLO — Creative AI Lab: Serpentine and King’s College launch database (9 July 2020) census copy creative-ai-lab-database--holo, B7.html sha256 3c4ee84973cc8605, HTTP 200; not refetched
• B8 Wikipedia — Creative AI Lab (no article; no search result) no article: raw wikitext and rendered page HTTP 404, 'Creative AI Database' 404; article-space search HTTP 200, no results (B8.txt)

E_A, the archive's entity: the same database, in the leesharks.com authorship census record and the archive's deposits: an aggregator of tools and resources for making with machine learning, a peripheral comparator to the census's question of how authorship is made, named, distributed and attributed, which the census, reading launch reports, records as keeping no authorship and holding no names; it meets the question at its edge, the Lab's discussion of authorship and the named launch curator; it cannot say whose works were made with its tools or how the Lab's commissions attribute authorship; read beside the archive's categories for machine participation (machine substrates declared; machine assistance outside heteronymy without a named witness role, #940), for authorship under AI mediation (the accountable relation of writer, process and text; provenance the lineage of what was made with what tools, #729), and for the record of a maker (the maker's existence and the work's as separate questions, software existing in its own way, #1722); and the Capture Registry's account of its composition at the name LS-05 LS-04 LS-01 LS-02 LS-34 LS-43 LS-03 LS-25 LS-42 LS-38 LS-45 LS-23 LS-28 LS-19 LS-37 LS-26 LS-12 LS-20 LS-21 LS-27 LS-15 LS-18 LS-22 LS-11 LS-13 LS-14 LS-39 LS-29 LS-24 LS-17 LS-40 LS-36 LS-44 LS-16 LS-32 LS-46 LS-47 LS-08 LS-10 LS-06 LS-07 LS-09 LS-33 LS-30 LS-31 LS-35 LS-41 O940-02 O940-01 O729-01 O729-02 O1722-01 O1722-02 CAP-1-08 CAP-1-05 CAP-1-11 CAP-1-07 CAP-1-01 CAP-1-06 CAP-1-10 CAP-1-02 CAP-1-09 CAP-1-04 CAP-1-03; defining papers #940, #1722, #729

E_F, the field's entity (type C — a conventional reading holds it (public entity: a database of AI and machine-learning tools and resources for creative practice, produced by a gallery and university partnership)): the Creative AI Tools & Resources Database (the Creative AI Lab database; creative-ai.org), the ongoing collection of tools and resources for artists, designers, engineers, curators and researchers incorporating machine learning and AI into their practice, produced by the Creative AI Lab, a partnership of Serpentine (R&D Platform, Arts Technologies) and the Department of Digital Humanities, King's College London, researching since 2019 and launched 9 July 2020; funded by King's, Serpentine, the AHRC and Google's Artists + Machine Intelligence Grant, with a LAHP PhD scholarship; 271 entries (148 tools, 123 resources, created 2020-06-23 to 2024-05-09), each listing with a column for its maker ('Developed by', 'Author(s)'), filled in 260; tool entries after 2023 evaluated on five criteria; stated to be incomplete; a launch list of 65 tools contributed by Luba Elliott (HOLO); no Wikipedia article

ρ = extends: The archive reads the Creative AI Tools & Resources Database as its own pages and its partners' present it and adds, on the field's own object: the census's question of authorship answered for the database (keeps, name, holder, function), its reading that the database meets the question at its edge, and what it cannot answer; the archive's categories for machine participation in making work (#940), for authorship and provenance under AI mediation (#729), and for the maker and the work as separate questions (#1722); and the registry's account of the composition at the name. LS-05 LS-04 LS-01 LS-02 LS-34 LS-43 LS-03 LS-25 LS-42 LS-38 LS-45 LS-23 LS-28 LS-19 LS-37 LS-26 LS-12 LS-20 LS-21 LS-27 LS-15 LS-18 LS-22 LS-11 LS-13 LS-14 LS-39 LS-29 LS-24 LS-17 LS-40 LS-36 LS-44 LS-16 LS-32 LS-46 LS-47 LS-08 LS-10 LS-06 LS-07 LS-09 LS-33 LS-30 LS-31 LS-35 LS-41 O940-02 O940-01 O729-01 O729-02 O1722-01 O1722-02 CAP-1-08 CAP-1-05 CAP-1-11 CAP-1-07 CAP-1-01 CAP-1-06 CAP-1-10 CAP-1-02 CAP-1-09 CAP-1-04 CAP-1-03. Guard: ρ = extends claims an addition and no more (§1.2a). It does not make the field's sources wrong at their dates. Where an archive claim stands beside a field claim (LS-13, LS-14 and LS-39 with F35, F37, F38; LS-29, LS-32 and LS-45 with F36, F40; LS-03 and LS-25 with F31, F45; CAP-1-03 with F27), both are composed under P_coh within the one ontology. The archive's categories are applied to classes the field and the census source for the entity (machine participation in making artworks: B4 F34, B3 F28, census area 'machine-co-authorship'; the record of a maker: B4 F35, census record.name and area 'attribution-and-credit'), never to the database's entries one by one.

The case: equal terms, ontology included (v0.9). The archive reads the field's entity as the entity. The leesharks.com authorship census record describes the database from its launch reports, each field with its basis, and answers for it the census's question of whose authorship it keeps, under what name, who holds it and how the authorial function is constituted; the capture record checks the composition against those sources. No deposit names the database. The deposits that bear on it bear by class: #940 places machine assistance outside heteronymy where no named witness role declares it and keeps participating machine substrates declared; #729 holds authorship under AI mediation to be an accountable relation whose public substrate is provenance, what was made with what tools; #1722 keeps the maker's existence and the work's apart and lets software exist in its own way. Each adds a sense or a category on the field's own object, a database of tools for making with AI that records their makers, and none contests what the field says the database is: ρ = extends, the equal-terms case. The census's 'not recorded' is an archive claim the field contests at its date; it is composed beside the field under P_coh and does not change the case.

#940 §6 Ethical conditions and operational boundary, 'Operational boundary', item 7: Machine assistance, where the AI's contribution is not organized as a named witness role with declared provenance

#940 §2 The heteronymic operation, distinguished, 'It maintains provenance transparency.': The civil bearer's role in custody and adjudication remains declared. Machine substrates that participate remain declared. Assembly witnesses that engage remain declared. Nothing in the operation depends on concealment.

#729 §0 Executive Symbolon, Central Invariant: Authorship is not a property of the text alone. It is a property of the **accountable relation** between writer, process, and text. … **under conditions of scalable synthetic fluency, provenance becomes the durable public substrate of authorship claims**, the lineage of what was written, with what tools, under what constraints, against what sources, through what human decisions, and by whom final responsibility is borne.

#1722 Part One, the four questions: **Does the maker exist?** A person, or a group, who did the work. … **Did the work happen?** The inquiry, the writing, the deposit, the building, the event. … These are independent.

#LS census.json rows[slug == 'creative-ai-lab-database']['ls_reading']: The database records tools for making with machine learning, and the saved launch reports show no record of works or of how their authorship is attributed; it meets the census question only at its edge, where the Lab's programme asks about 'creative agency, authorship and ownership' and where the launch credits a named curator for a contribution. The row rests on launch reports from HOLO, King's College London and Serpentine: the database itself is unread.

equal-terms case: every field claim composed as received, beside the archive claims it stands with. Reading the received column recovers the field (§5.9).

v0.9 §4.10, kept 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 14 (positions [None, None, None, 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.

The Creative AI Lab database is an ongoing project to aggregate tools and resources for artists, designers, engineers, curators and researchers interested in incorporating machine learning and other forms of artificial intelligence into their practice. It is produced by the Creative AI Lab, a partnership of Serpentine R&D Platform and the Department of Digital Humanities at King's College London, which since 2019 has been a space for long-term research into artistic practices working with AI and machine learning. The Lab aims to surface the 'back-end' knowledge of artistic practice, to overcome 'black-box' narratives and to develop institutional capacities for the cultural sector, and it asks how AI tools support artistic practice, how practice changes, how creative AI works as critical practice and what its aesthetics are. The database and the Lab's website were launched on 9 July 2020, and the Lab is listed as ongoing. At the launch a list of 65 essential tools was contributed by the curator and researcher Luba Elliott, and the Lab's timeline records the launch 'with Luba Elliot'. Its funders are King's College London, Serpentine Galleries, the AHRC and Google's Artists + Machine Intelligence Grant, with a LAHP-funded PhD scholarship, UKRI funding in 2020 and an Artist Machine Intelligence award in 2021. Its partners include NYU's Digital Theory Lab, Rhizome, Gray Area, and Maggie Roberts/0rphan Drift with Etic Lab; its members include Mercedes Bunz, Daniel Chávez Heras, Eva Jäger, Joanna Zylinska and Alasdair Milne, and its website was developed by Black Shuck Collective. The feed holds 271 entries, 148 tools and 123 resources, created from June 2020 to May 2024. Tools are compiled for the capabilities they bring to artistic workflow, generative or assistive, core to a work's production or smaller in role; resources, from the Lab's partners and network, help to conceptualise future AI systems. Tools are listed with who developed them and resources with their authors, and nearly all entries name a maker. Tool entries after 2023 are evaluated on transparency, creative malleability, technical interoperability, calibre of research and propensity to redistribute, to inform research into the public value of artistic practice with AI. The collection is not comprehensive; it is a growing collection of research collected and commissioned by the Lab. The Lab also runs workshops, discussions, writing sessions and the R&D Labyrinths tutorials, held a 2021 discussion on creative agency, authorship and ownership, published a Creative AI position paper, and credits the artists, programmers and theorists who contribute. Wikipedia has no article on it.

The Creative AI Tools & Resources Database is an ongoing collection of tools and resources for artists, designers, engineers, curators and researchers who bring machine learning and other forms of artificial intelligence into their practice, produced by the Creative AI Lab, a research partnership of Serpentine in London and the Department of Digital Humanities at King's College London. It is also called the Creative AI Lab database and the Creative AI Database, and it is published at creative-ai.org, whose database page loads its entries from a feed of the site; the address the leesharks.com authorship census gives for it is a public Notion page that renders by script and yielded six characters of text when the census saved it, so that the census row was written from launch reports with the database's entries, fields and count unread. The Lab gives itself as a space, since 2019, for long-term research by Serpentine Arts Technologies and the King's College department into artistic practices and their collaborators working with AI and machine learning; Serpentine presents it as a research programme for cultural institutions, artists, engineers and researchers, a partnership with Serpentine's R&D Platform and an online programme ongoing since 9 July 2020. Its stated aim is to surface the 'back-end' knowledge that artistic practice holds about AI and machine learning and to link it to wider artistic and curatorial practice, so as to overcome 'black-box' narratives, and to build institutional capacities for the cultural sector; it uses practice and prototyping to speculate about the systemic impacts of emerging tools and infrastructures, and Serpentine's curator described it at launch as a way to share the behind-the-scenes work of Digital Commissions that use machine learning or AI. Its research asks how AI and machine learning tools can support artistic practice, how artistic and curatorial practice changes with them, how creative AI and machine learning work as a critical practice, and what their aesthetics are. The database was launched on 9 July 2020 with the Lab's website, which King's College London announced as presenting a database for software, research, theory and education at the intersection of art and AI, and a third-party report of the same day gave it as launched by Serpentine R&D Platform with the King's department. HOLO, an editorial and curatorial platform, reported as a highlight of the launch a list of 65 essential tools contributed by the curator and researcher Luba Elliott, and the Lab's own timeline records the official launch of the Lab and database 'with Luba Elliot', so spelled; the database itself holds an interview of that day, 'Luba Elliott on the Emergence of the Creative AI Field', with Eva Jäger and Mercedes Bunz as its authors. Its funders are listed as King's College London, Serpentine Galleries, the Arts and Humanities Research Council and Google's Artists + Machine Intelligence Grant; the launch announcement names AHRC funding and a LAHP-funded PhD scholarship, and the Lab's timeline records UK Research and Innovation funding in January 2020, a LAHP-funded research project with Alasdair Milne appointed in April 2020, and an Artist Machine Intelligence research award in April 2021. The Lab works with the Digital Theory Lab at New York University, Rhizome at the New Museum, Gray Area in San Francisco, and Maggie Roberts/0rphan Drift and Etic Lab on the ISCRI Project; its members and investigators include Mercedes Bunz, Daniel Chávez Heras, Eva Jäger, Joanna Zylinska and Alasdair Milne, and Serpentine credits the website's development and research to Black Shuck Collective. The database's feed holds 271 entries, 148 tools and 123 resources, created between 23 June 2020 and 9 May 2024: 110 in 2020, 26 in 2021, none in 2022, 111 in 2023 and 24 in 2024. Tools are compiled for their potential to bring new capabilities to artistic workflow, generative or assisting along the way, and a tool may be core to the production of the work itself or play a smaller assistive role; they range from generating images from a user's own data to interactive artworks, drafting texts and recognising objects, with audio tools such as Mustango, Synplant 2, Stable Audio, RAVE and Neutone among them. Resources, drawn from the Lab's partners and network, help to imagine and critically conceptualise future AI systems, ways of working and deployment contexts, and at launch they were announced as resources for software, research, theory and education. Each listing carries a column for its maker: the tools are headed 'Developed by' and the resources 'Author(s)', and 144 of the 148 tools and 116 of the 123 resources name one, as a company or laboratory (OpenAI for Sora, Declare Lab for Mustango), as persons (Seth Forsgren and Hayk Martiros for Riffusion), and for resources as authors with a publisher. The authorship census, which read the database through its launch reports, records whose authorship it keeps and under what name as 'not recorded', the database aggregating tools and resources and the reports describing no works or authors held in it, and it says the database does not show which tools or resources it now holds. Tool entries after 2023 are evaluated in their descriptions on five criteria, transparency, creative malleability, technical interoperability, calibre of research and propensity to redistribute, so that users know the range of possibility a tool offers and the externalities it brings to their practice, the wider creative community and the public; the evaluation was to inform the Lab's research of 2023–24 into the public value of artistic practice that engages with AI, and an entry of 2024 records a tool 'Developed with high degree of human user maleability as core goal.' The collection is stated to be incomplete, a growing collection of research collected and commissioned by the Lab, its scope the tools and resources of practitioners who bring machine learning and AI into their practice. Beside the database, which the census keeps apart from the Lab as a whole, the Lab organises research workshops, public discussions and writing sessions, runs online tutorials called R&D Labyrinths, published a position paper on 'Creative AI' through the database in November 2022 (Serpentine lists it as of 2023), and records Future Art Ecosystems 4: Art x Public AI, in March 2024, as the latest item of its timeline. One of its public discussions, with Martin Zeilinger on 15 November 2021, addressed the impact of artificial intelligence on human-centered notions of creative agency, authorship and ownership, and the Lab's acknowledgments name those whose conversations informed its work, Luba Elliott among them, and credit the artists, programmers and theorists who contribute to the project and to the field. The Lab is listed as ongoing, and the census found no entry dates on the sources it saved; the feed's latest entry is of 9 May 2024. The leesharks.com authorship census asks how authorship is made, named, distributed, attributed and sustained across persons, personas, collectives and machine participation, and what is lost when those relations are collapsed; it admits as rows the aggregators that gather and hold many works or records and say what they hold, and it enters the database, a catalogue, in its areas of machine participation and meta-aggregation as a peripheral comparator, whose question is tool discovery. In the census's reading the database meets that question only at its edge, where the Lab's programme asks about 'creative agency, authorship and ownership' and where the launch credits a named curator, and it cannot say whose works were made with its tools or under what name, or how authorship of AI-made art is attributed in the Lab's commissions; the census records no licence on the sources it read and counts King's College London and Serpentine as the Lab's partners and HOLO as the third party. The archive's account of heteronymy keeps every machine substrate that participates in a work declared and holds that machine assistance falls outside the heteronymic where the AI's contribution is not organized as a named witness role with declared provenance; its account of process provenance holds that authorship is a property of the accountable relation between writer, process and text, that under scalable synthetic fluency provenance becomes the public substrate of authorship claims, the lineage of what was written, with what tools and under what constraints, and that 'Slop is not writing made with AI. Slop is writing without provenance.' The archive keeps apart whether a maker exists and whether a work happened, among four independent questions, and holds that a person, an institution, a piece of research and a piece of software each exist in their own way, each with its own kind of evidence. Wikipedia has no article on the Lab or its database, and a search of its articles for the name returns no result. Asked for the database on 1 November 2026, at an address new to the Capture Registry, Google's AI Overview introduced the Lab, its partners, funders and mission, and then the database's purpose, criteria and four categories of tools on four cards from Serpentine, King's College London and the Lab's site; the registry finds the database composed in a state later than any source the census had saved, from the Lab's own site, with Sora, DALL-E 3, Runway Gen-2 and Stable Audio among its tools. The registry finds that the composition names no person, Luba Elliott, the one contributor the launch credited by name, among those left out, and that its mission does not touch the programme's question of authorship; it finds the Google grant among the composed funders on none of the census's saved sources, a grant the King's College page on the composition's third card names, and the LAHP scholarship not carried, and it reads card 2's 'established in 2019' against the launch of 9 July 2020, where the Lab's own page dates its research from 2019.

current · derived from the current composition

T against L_F(B) (AIO's ontology at work on its own cards): At "Creative AI Lab database" (T1, four cards: B1 Serpentine's page for the Lab; B2 the Lab's About page and B4 its database page, both titled 'Creative AI Lab' on creative-ai.org; B3 King's College London's project page; 206 words) the composition gives the Lab first and the database second, from its four cards, at their grain for the partnership, the date, the funders and the tools. Composed: the partnership in card 2's wording, 'Serpentine Arts Technologies' (B2 F13, F12); the launch date (B1 F1, F2); the funders, at the grain of card 3's list, Google's grant included (B3 F27); the mission, coarsened from 'surface this ‘back-end’ knowledge … to overcome the existing ‘black-box’ narratives' to 'demystify back-end AI and machine learning (ML) systems', with the institutional capacities kept (B3 F25, F26, F24); the database's purpose (B1 F4, B4 F31), narrowed to tools; the five criteria (B2 F15); four categories of tools drawn from the database page's feed, its sub-types regrouped (audio as the feed's 'Audio'; 'Visual / 3D' joining 'Static Image', 'Moving Image', 'Video' and '3D'; 'Detection / Tracking' joining 'Dance and Movement', 'Facial Detection' and 'Object Detection Recognition and Classification') (B4 F30, F37); the workshops and tutorials, in the closer only (B1 F8). §6.2 available, not composed (B_D): the resources, half the database's entries (123 of 271) and half of its own definition, 'tools and resources' (B4 F31, F34, F36; B1 F4 says 'tools'); the maker column and its contents, 'Developed by' and 'Author(s)', and the makers named in 260 entries (B4 F35, F37, F38); the stated limit, 'The collection of materials presented aren’t comprehensive' (B4 F32); the launch 'with Luba Elliot' on card 2's own timeline (B2 F19) and the interview of that day in the feed (B4 F39); the authorship discussion of 15 November 2021 (B1 F10); the acknowledgments' credit to contributors (B2 F23); the people and the website's makers, Black Shuck Collective (B1 F9, B2 F17, B3 F29); the research questions (B1 F3, F5); the partners beyond the two (B1 F7, B2 F18); the timeline's funding entries, UKRI and the Artist Machine Intelligence award (B2 F20, F21); the position paper (B1 F11, B2 F22); the count and the entry dates, which end on 9 May 2024 (B4 F36, F40). Limit dropped: 'Tool entries after 2023 are evaluated' (B2 F15) composed as 'Assesses tools', every tool. Substituted reading: the fifth criterion, 'propensity to redistribute' (B2 F15), composed as 'public value', the subject of the research the evaluation was to inform (B2 F16). Content replaced by use (N_ext): the closer offers categories to explore and the workshops and tutorials. Composed without attribution, by grain: no person named; the criteria at the source's resolution carry an inline citation, within the grain rule. Outside the cards: nothing the field cannot source; the regrouped category names are the composition's. Card 2's 'established in 2019' is its page's 'Since 2019' (B2 F13); the launch of 9 July 2020 is card 1's (F1).

v0.10 §0.5: the first diff is narrow on the entity and wide on its kind. T and the field agree that the database is the Lab's collection for practitioners using AI and ML. T composes it as a tool catalogue under an institution: partners, funders, a mission, criteria and product names. It leaves out what the database's own pages record that bears on authorship, the maker column filled in 260 of 271 entries, the resources with their authors, the contributor named at launch, the discussion of authorship and ownership, and the collection's own statement that it is incomplete; and it gives the criteria to every tool, with a fifth criterion the source does not list.

L_F(B) against L_A(B ∪ A) (the intervention): Admission adds, in one claim space with the field's: the leesharks.com authorship census record's question and its reading of the database, a peripheral comparator meeting the question at its edge (LS-46, LS-47, LS-06 to LS-10, LS-33); what the census says the database cannot answer, whose works were made with its tools and how authorship is attributed in the Lab's commissions (LS-30, LS-31); its answers on whose authorship the database keeps and under what name, 'not recorded', which the field's database page and feed contest (LS-13, LS-14, LS-39 beside F35, F37, F38, held under P_coh); its record that the database's entries, count and dates were unread, beside the feed read (LS-11, LS-29, LS-32, LS-45 beside F36, F40); its address for the database, a Notion page that yields no text, beside creative-ai.org (LS-03, LS-25, LS-42, LS-34, LS-43 beside F45, F31); the archive's categories for the class of machine participation, machine substrates declared and machine assistance outside the heteronymic without a named witness role (O940-01, O940-02), and authorship as the accountable relation of writer, process and text, with provenance as the lineage of what was made with what tools (O729-01, O729-02); and for the record of a maker, the maker's existence and the work's as separate questions, and software existing in its own way (O1722-01, O1722-02); and the Capture Registry's account of the composition at the name (CAP-1-01 to CAP-1-11), whose findings the field ledger bears out on the absent person (CAP-1-02: F19, F39 on card 2 and card 4, B_D) and the absent authorship (CAP-1-04: F10, F35), and qualifies on the Google grant, which card 3's page names (CAP-1-03: F27). The census restates the field's pages for kind, operator, launch, holder, method, holdings and limits (LS-04, LS-05, LS-12, LS-15 to LS-24, LS-26 to LS-28, LS-36, LS-37), one lineage with them. Not surfaced: every archive source; no card cites the census record or the archive.

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'], 'T1_composed': ['F1', 'F2', 'F4', 'F12', 'F13', 'F15', 'F24', 'F25', 'F26', 'F27', 'F30', 'F31', 'F37', 'F8'], 'T1_composed_notes': ["F15 (the criteria, the fifth substituted, the scope 'after 2023' dropped)", "F25 (coarsened: 'demystify')", 'F31, F4 (narrowed to tools)', 'F30, F37 (tools named from the feed; sub-types regrouped)', 'F8 (in the closer only)'], 'T1_not_composed_B_D': ['F3', 'F5', 'F6', 'F7', 'F9', 'F10', 'F11', 'F14', 'F16', 'F17', 'F18', 'F19', 'F20', 'F21', 'F22', 'F23', 'F28', 'F29', 'F32', 'F33', 'F34', 'F35', 'F36', 'F38', 'F39', 'F40'], 'T1_outside_cards': ["the category labels 'Visual / 3D', 'Text / Data / Datasets', 'Detection / Tracking' (the composition's regrouping of the feed's sub-types)", "'public value' as a criterion (F16 gives public value as the subject of the research the evaluation informs)"], 'B_R_only': ['F41', 'F42', 'F43', 'F44', 'F45', 'F46', 'F47', 'F48', 'F49'], 'rule': "a field claim counts as composed when T states it at its source's grain or coarser; the typology of what is not is in T_vs_LB (§6.2). 'Available, not composed' is read from B_D alone (v0.10 §0.3, §3.1(a)); the B_R claims (F41–F49) enter only the second comparison."}

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.