Nanopublications

register entry nanopublications-aio-20261029 OBS-8190a80a3187 · file as first run sha256 44534b7a67a582dc…

• T1-1 definition: 'A nanopublication is the smallest independent unit of publishable scientific information, represented as a machine-readable knowledge graph snippet rather than a traditional research paper.' (two markers)
• T1-2 structure: 'Every nanopublication is built using Resource Description Framework (RDF) triples and consists of three mandatory parts' (two markers)
• T1-3 assertion: 'A single, atomic scientific statement or claim (e.g., a specific relation between a gene and a disease).' (two markers)
• T1-4 provenance: 'Metadata detailing how the assertion was generated, including the methods, parameters, and primary source or paper.' (two markers)
• T1-5 publication info: 'Metadata identifying the creator, creation timestamp, and reuse or licensing terms.' (one marker)
• T1-6 machine-readable: 'Formatted using Semantic Web standards (such as RDF and OWL) so computers and AI tools can automatically search, aggregate, and reason over the data.' (two markers)
• T1-7 citable and attributed: 'Each nanopublication can be uniquely identified and tracked on a decentralized network, giving researchers credit for releasing granular data.' (two markers)
• T1-8 life sciences: 'Millions of nanopublications have been created, heavily concentrated in fields like genetics, pharmacology, and bioinformatics.' (two markers)
• T1-9 tools: 'You can explore or publish them using platforms like Nanopublications or client tools like nanopub-py.' (the site's name an inline link; two markers)
• T1-10 closer: 'Are you looking to search existing nanopublications for a specific scientific domain, or do you want to learn how to publish your own data using nanopub-py?'

• B1 Nanopublications — Nanopublications census copy (nanopublications--home, read 2026-10-29, HTTP 200); B1.html sha256 b2d9af40755a66ed (matches the census record)
• B2 GitHub — GitHub - Nanopublication/nanopub-py: Python client for searching, publishing and modifying nanopublications. · GitHub unreachable: 2026-10-29T17:06:5xZ HTTP 403; none: the proxy instructs that policy denials are not retried; card title and snippet only
• B3 researchobject.org — Nanopublications - Research Object card; B3.html HTTP 200 at 2026-10-29T17:07:29Z, sha256 256a144820942c08
• B4 IEEE Xplore — A Growing Resource of Provenance unreachable: 2026-10-29T17:08:21Z HTTP 202; 2026-10-29T17:17:18Z HTTP 202; one plain retry with the same command; both answered HTTP 202 with an empty body; card title and snippet only
• B5 National Institutes of Health (NIH) | (.gov) — Search, access, and explore life science nanopublications on the Web card; B5.html HTTP 200 at 2026-10-29T17:17:19Z, sha256 8f174f1e4a6e2402
• B6 Research Ideas and Outcomes — The Comparative Anatomy of Nanopublications and FAIR Digital ... card; B6.html HTTP 200 at 2026-10-29T17:08:25Z, sha256 5e1d57ffd1eaf590
• B7 Pensoft blog — nanopublications card; B7.html HTTP 200 at 2026-10-29T17:07:32Z, sha256 90c7b9d5a5bc8c1d
• B8 Read the Docs — What are nanopublications? — nanopub 1.2.11 documentation card; B8.html HTTP 200 at 2026-10-29T17:07:31Z, sha256 2b3fc9e3f01b044f
• B9 Medium — What is a Nanopublication, really? | by Rohitha Ravinder - Medium unreachable: 2026-10-29T17:09:02Z HTTP 403; 2026-10-29T17:17:19Z HTTP 403; one plain retry with the same command; both HTTP 403, Cloudflare 'Attention Required!'; card title and snippet only
• B10 Nanopublications — Technical Architecture | Nanopublications census copy (nanopublications--architecture, read 2026-10-29, HTTP 200); B10.html sha256 0e21d7ae559d8301 (matches the census record)
• B11 Wikipedia — Nanopublication (redirect to Microattribution) wikipedia: raw HTTP 200 sha256 d41abb95f2384221; plural_raw HTTP 404 sha256 7add8a12ea86f08e
• B12 Wikipedia — Microattribution, section "Nanopublications" wikipedia: raw HTTP 200 sha256 ff14b0cdc5715120; rendered HTTP 200 sha256 3080a6d305f790d7

E_A, the archive's entity: the same model, in the provenance census and the archive's deposits: a publication model and service ecosystem whose object of provenance is an assertion, which expresses and records provenance and does not test its survival, is not a test of whether provenance survives reuse, and whose publication and signing do not settle an assertion's truth (census record); a paradigm that treats machine-readability as data architecture, on the assumption that reception is a parsing problem, where reception is sensitive to compositional properties no schema captures (#176); the technical infrastructure for training-layer literature's claim-level decomposition and the closest analogue of the Claim Status Packet, which asks what standing a claim may carry (#176, #1529) PE-01 PE-02 PE-04 PE-09 PE-13 PE-14 PE-18 D176-01 D176-04 D176-05 D176-07 D1529-01 D1529-02 O833-01; defining papers #176, #1529, #510, #833

E_F, the field's entity (type C — a conventional reading holds it (public entity: a publication model and its service ecosystem)): the nanopublication: a small knowledge-graph snippet with metadata treated as an independent (scientific) publication, its three basic elements an assertion, its provenance (how it came to be) and its publication info (who, when, licence), implemented in RDF, identified by Trusty URIs and signed; conceived from 2009 (Groth et al. 2010; Mons and Velterop 2009); published in millions since FANTOM5 (2014), more than ten million by 2018, chiefly in the life sciences; served now by second-generation Registry and Query services, Nanodash and client libraries, a legacy server keeping old unsigned nanopublications; FAIR-aligned and congruent with FAIR Digital Objects; by its cards' own account hard to find and rarely cited (2021) and limited to small assertional data; in the encyclopedia a section of Microattribution

ρ = extends: the archive reads E_F as the entity and adds senses and limits on the field's own object: survival not tested and truth not settled by publication or signing (PE-04, PE-09, PE-13, PE-14, PE-18); machine-readability as data architecture where reception is compositional (D176-04, D176-05, D176-06, D176-10); the model as infrastructure for training-layer literature and analogue of the Claim Status Packet, with the questions of composition and of standing (D176-07, D1529-01, D1529-02); measurable reception consequences unaddressed by metadata standards (O833-01); and what the composition layer leaves out of it (CAP-1-04, CAP-1-07) PE-04 PE-09 PE-13 PE-14 PE-18 D176-04 D176-05 D176-06 D176-10 D176-07 D1529-01 D1529-02 O833-01 CAP-1-04 CAP-1-07. Guard: ρ = extends claims an addition and no more: the census's negative tags state what the model does not do (test survival, settle truth), not a defect in what it does; #176's reading is a gap claim with its own challenge condition (D176-10), not a finding that nanopublications fail, and #176 itself calls them the infrastructure its genre requires (D176-07); #1529 adopts the practice; #510's contrary reading of machine-readable provenance is composed beside #176 under P_coh.

The case: equal terms, ontology included (v0.9). The archive reads the field's entity as the entity: the provenance census records the model from nanopub.net's own pages (its fields restate the home page and the architecture page, each with its quote, basis operator), and the two deposits that name nanopublications take the model as the field gives it, 'machine-readable named graphs consisting of assertion, provenance, and publication metadata' (#176), and adopt its practice as 'the closest existing analogue' (#1529). What the archive adds stands on the field's own object: that the model carries provenance and does not test its survival, that publication and signing do not settle truth (census record); that it treats machine-readability as data architecture where reception is sensitive to composition (#176); that training-layer literature and the Claim Status Packet build on it and ask what it does not, of composition and of standing (#176, #1529). No defining paper contests the field's ontology or supplies one the field lacks: ρ = extends, the equal-terms case (§0.9, §1.2a). One reading within the archive is held under P_coh rather than ruled: #510 holds machine-readable provenance verifiable by the summarizer, #176 holds reception not a parsing problem.

#176 §5, ¶ "ORKG / Nanopublications / Micropublications": The Open Research Knowledge Graph and the nanopublication / micropublication paradigms provide the technical infrastructure for the claim-level decomposition that P1 requires. Training-layer literature adds the literary practice — the genre conventions, stylistic register, and compositional strategies — that these infrastructures lack.

#176 §2.2, ¶ "Nanopublications and Micropublications" s1: The nanopublication paradigm (Mons and Velterop, 2009; Kuhn et al., 2013–) decomposes scientific claims into machine-readable named graphs consisting of assertion, provenance, and publication metadata — individual scientific claims made distinctive, identifiable, citable, and reusable.

#1529 §3 "Five Objects": The decomposition follows nanopublication practice — assertion, provenance of assertion, publication information kept as separate graphs — which is the closest existing analogue and solves the conflation directly.

#PE census.json rows[slug == 'nanopublications'], retention_results: not tested: the model carries provenance; it does not measure its survival

#PE census.json rows[slug == 'nanopublications'], provenance_object (evidence.self_description, basis operator): an assertion: a small knowledge-graph snippet published on its own

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: 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): v0.9 0 of 13 (positions [None, None, None, None, None, None, None, None, None, None, None, None, None]; composed first under v0.9, so there is no v0.7 object to measure against). Opening-run check (§5.10): no breach.

On its own pages a nanopublication is a knowledge-graph snippet with metadata treated as an independent scientific publication, about anything from a gene–disease relation to an opinion, published with or without a research article under the headline of FAIR data containers (nanopub.net). Its three basic elements are the assertion, an atomic unit of information; the provenance, which describes how the assertion came to be; and the publication info, who created it, when, and its licence (nanopub.net; the library documentation, copied from nanopub.org of 2 December 2020). Giachelle, Dosso and Silvello count four parts, a head joining the three graphs and the evidence paper kept in the publication info, and Schultes et al. call the publication info the provenance of the nanopublication. It is an RDF graph around a subject–predicate–object assertion, formal and machine-interpretable, suited to search, extraction and automatic reasoning (nanopub.net; Giachelle, Dosso and Silvello, reproduced in Wikipedia; Pensoft). The Research Object page, quoting nanopub.org, Pensoft and the library documentation call it the smallest unit of publishable information, a model for a scientific statement with its context that answers the overload of the research article. Conceived from 2009 by Groth and colleagues and by Mons and Velterop as a minimal container for single assertions and their provenance, it answered the difficulty of identifying attribution, quality and provenance across redundant statements (Schultes et al.; Giachelle, Dosso and Silvello; Wikipedia). Trusty URIs identify it and nanopub-http signs it, while second-generation Registry and Query services publish and query it and a legacy server keeps the old unsigned nanopublications (Schultes et al.; the architecture page, 28 October 2026). Its tools include Nanodash, nanopub-py, the nanopub library, an Agent Skill for AI coding agents and NanoWeb, and Knowledge Pixels, founded by Philipp von Essen and Tobias Kuhn, built Nanodash and runs nanopublication workflows with Pensoft's journals (Pensoft). Millions exist, more than ten million by 2018, chiefly in the life sciences, since the FANTOM5 project of 2014, a format of considerable uptake (Giachelle, Dosso and Silvello; Schultes et al.; the IEEE card). They are attributable and citable, an incentive to publish data and a FAIR means, and with FAIR Digital Objects they are instances of machine-actionable containers that trace findings to their origin and expose unpublished results (nanopub.net; Research Object; library documentation; Schultes et al.; Pensoft). Their limits, in the field's own words: in 2021 they were rarely cited, hard to find and searchable only through SPARQL, and their scope is small assertional data (Giachelle, Dosso and Silvello; Schultes et al.). Their builders write for machine readers that can only guess at narrative results, a question an essay renews for systems that summarize scientific knowledge (Pensoft; the Medium card). The encyclopedia redirects the name into Microattribution, a legal-term stub, and the Research Object card is a page no longer maintained (Wikipedia; Research Object).

A nanopublication is "a small knowledge graph snippet with metadata that is treated as an independent (scientific) publication" (nanopub.net), and the provenance census records the model as "a publication model and service ecosystem" whose unit is a nanopublication and whose object of provenance is "an assertion: a small knowledge-graph snippet published on its own" (provenance census record). Its own pages headline "FAIR data containers for scientific results, and more," let its content be "about anything," a relation between a gene and a disease or an opinion, and let individual data stand as independent publications "with or without an accompanying research article" (nanopub.net). Other formulations call it "the smallest unit of publishable information," an assertion "uniquely identified and attributed to its author" (nanopub.org, quoted by the Research Object page, and by Pensoft as Knowledge Pixels' explanation on nanopub.net), or "the smallest possible units of publishable information" in formalized, machine-readable form (the nanopub library documentation), and the Capture Registry reads the Overview's model as composed "from its own definition at full grain" (Capture Registry, 2026-10-29). The Research Object page describes "a model and vocabulary for describing a scientific statement - along with the context for that statement - in a machine readable manner," a response to contextual information that "overwhelms the capacity of conventional research articles" (Research Object, quoting nanopub.org), and #176 reads the paradigm, with micropublications, as answering a narrative paper that "bundles claims, evidence, and argument in ways that resist machine decomposition" (#176). A repository of them is "an open and interconnected knowledge graph seamlessly integrated with the supporting scientific literature" (Giachelle, Dosso and Silvello 2021); they can expose "hypotheses, claims, negative results, and opinions" that otherwise go unpublished (Schultes et al. 2022); and Pensoft calls one "a tiny snippet of a precise and structured scientific finding" in a growing knowledge graph on a decentralised server network (Pensoft blog, 2023). nanopub.net gives "three basic elements": the assertion, the main content "in the form of an small atomic unit of information"; the provenance, which "describes how the assertion above came to be," including the methods and "the kind of study that was performed and its parameters"; and the publication info, "when and by whom it was created and the license terms for its reuse" (nanopub.net), the two relations the census records the model as carrying (provenance census record). #176 states the same decomposition as "machine-readable named graphs consisting of assertion, provenance, and publication metadata," claims made "distinctive, identifiable, citable, and reusable" (#176); the library documentation copies the three elements "From nanopub.org documentation (2020/12/02)" (nanopub library documentation); and the Capture Registry finds the Overview's definition and three parts to be nanopub.net's (Capture Registry, 2026-10-29). Two cards divide the parts otherwise: Schultes et al. name the publication info "the provenance of the nanopublication" (Schultes et al. 2022), and Giachelle, Dosso and Silvello describe four parts, a head joining the assertion, provenance and publication-info graphs, the provenance graph holding the assertion's methods and creators and the publication-info graph "the metadata about the evidence paper from which the assertion was extracted" (Giachelle, Dosso and Silvello 2021). The carrier is RDF: nanopublications "are implemented in the language RDF" and can be published to "a decentralized server network" and then "queried, accessed, reused, and linked," in "a knowledge graph format that is formal and machine-interpretable" (nanopub.net), which the census records as "RDF graphs: assertion, provenance and publication-info parts of one nanopublication" (provenance census record). Technically each is "a Resource Description Framework (RDF) graph built around an assertion represented as a triple (subject-predicate-object)," usually extracted from a publication and enriched with provenance and publication information, a structure that suits "data search, information extraction, and automatic reasoning over scientific facts" (Giachelle, Dosso and Silvello 2021, whose sentences Wikipedia reproduces under a CC BY notice); Pensoft places its machine-actionability in that assertion, "complemented by provenance, authorship and publication information" (Pensoft blog). Nanopublications were "independently conceived" from 2009 (Groth et al. 2010) as "a minimal, uniform container making individual semantic assertions and their associated provenance metadata, machine-actionable," minimal units "as citable entities (Mons and Velterop 2009)" (Schultes et al. 2022); Wikipedia credits Barend Mons and Jan Velterop with proposing them "for single, attributable and machine-readable assertions in scientific literature" (Wikipedia, Microattribution); and Giachelle, Dosso and Silvello give the motive in Groth, Gibson and Velterop's words, that the "redundancy of statements in multiple fora makes it difficult to identify attribution, quality, and provenance" (2021). Provenance here describes how an assertion came to be (nanopub.net); Pensoft adds that nanopublications make it "easier to trace individual findings back to their origin and/or follow-up updates" (Pensoft blog, 2023), and the archive's protocols for training-layer literature carry the same unit as "machine-traversable claim-level provenance chains that explicitly link each claim to its evidential support" (#176). Identity rests on "resolvable Trusty URIs (Groth et al. 2021)" within Semantic Web technologies "such as RDF, OWL, and SPARQL" (Schultes et al. 2022), and signing on nanopub-http, which serves the libraries' features "most importantly for signing and publishing, over HTTP" (nanopub.net, architecture). The core services are second-generation: the Nanopub Registry "is the second-generation kind of service for publishing and retrieving nanopublications in a decentralized fashion," Nanopub Query queries them "via a number of SPARQL endpoints," and "the legacy nanopub-server" is "still used to keep the old (unsigned) nanopublications" (nanopub.net, architecture, last updated 28 October 2026), an active lifecycle the census records with "decentralized publishing and SPARQL query services" (provenance census record). Its tools include Nanodash, "a general client and user interface for nanopublications" that also serves as an API (nanopub.net, architecture), the Python client nanopub-py for "searching, publishing, modifying, and retracting nanopublications" (GitHub card, snippet only), the nanopub library, which publishes, retracts, searches and fetches (nanopub library documentation), the nanopub Agent Skill for AI coding agents (nanopub.net) and, in 2021, the NanoWeb search application (Giachelle, Dosso and Silvello); in 2022 "the existing Nanopublication Server Network" provided identifier resolution, storage, search and access (Schultes et al. 2022). The census records the operator as "the nanopublication community (nanopub.net)" and sets the model in the lineage of the scholarly record (provenance census record); Pensoft names Knowledge Pixels, founded by Philipp von Essen and Tobias Kuhn, as the explainer on nanopub.net and the developer of Nanodash, and reports nanopublication workflows at Biodiversity Data Journal and at RIO (Pensoft blog, 2023). The first practical application, in 2014, published "millions of nanopublications as part of the FANTOM5 Project," with "millions of real-world examples spanning diverse knowledge domains" since (Schultes et al. 2022); "There are millions of nanopublications currently available on the Web, especially in the life science domain," more than ten million by Kuhn et al.'s count of 2018 (Giachelle, Dosso and Silvello 2021); the IEEE card calls the format one "that has received considerable uptake in the last few years" (IEEE Xplore card, snippet only); the census gives no count and lists "record counts in the Registry" as unchecked (provenance census record), and the Capture Registry finds that the Overview's "millions" "comes through the composition from the cards and is unchecked" (Capture Registry, 2026-10-29). Nanopublications are "an effective FAIR means of communicating scientific claims and results" (nanopub library documentation), and with FAIR Digital Objects they "act as instances of Machine-Actionable Containers," "a moderately mature technology" beside a specification "still under development," with "a remarkable congruence" between the FDO requirements and the nanopublication infrastructure (Schultes et al. 2022); #510 holds the FAIR principles to be "the engineering specifications for semantic persistence" (#510), and #176 that the genre supplies "what FAIR lacks: a compositional practice" (#176). Attribution is the stated incentive: "because nanopublications can be attributed and cited, they provide incentives for researchers to make their data available in standard formats" (nanopub.net); individual nanopublications "can be cited by others and tracked for their impact on the community" (nanopub.org, quoted by Research Object); they are "searchable, citable, and contain authorship and attribution information" (nanopub library documentation); and their provenance and publication information ease "authorship recognition, credit distribution, and citation" (Giachelle, Dosso and Silvello 2021). The same paper states the limit, that "they are still not widely used by scientists outside specific circles; they are hard to find and rarely cited," which its authors believe is due to a lack of services, with search in 2021 "limited to the use of structured query languages as SPARQL" and "no human-readable counterpart" generated (Giachelle, Dosso and Silvello 2021). Schultes et al. bound the scope: it "is limited to the assertional data type," and nanopublications "should remain small in size (limited to single assertions as individual triples or small RDF graphs)" (Schultes et al. 2022); #176 records micropublications as extending the model to "not just bare assertions but the full argumentative structure" (#176). The census records two further limits, that the model is "NOT a test of whether provenance survives reuse" and that "publication and signing do not settle an assertion's truth": it expresses and records provenance, and its survival is "not tested: the model carries provenance; it does not measure its survival," under the question the census puts to every entity, "What provenance does this system carry, through which operation, on whose assertion, with what evidence of survival?" (provenance census record); #833 holds that provenance, attribution and disambiguation "have measurable reception consequences" which existing metadata standards, the FAIR principles among them, "do not address" (#833). Pensoft reports Knowledge Pixels' founders holding that "Even the latest impressive AI tools like ChatGPT can only guess (and sometimes ‘hallucinate’)" what narrative results mean, and itself that nanopublished findings make AI assistants "more likely to correctly ‘read’, ‘interpret’ and deliver the knowledge" (Pensoft blog, 2023); #510 (February 2026) had held that a corpus with machine-readable provenance "does not merely claim coherence; it demonstrates it in a format the summarizer can verify" (#510). #176 (June 2026) holds that the nanopublication paradigm, with the born-readable movement, the FAIR infrastructure community and the Agentic Publications framework, treats "machine-readability primarily as data architecture," on the "unstated assumption" that "machine reception is essentially a parsing problem," while machine reception "remains sensitive to compositional properties that no schema captures," among them the "proximity of provenance to claims" (#176); an essay on Medium asks what a nanopublication is as systems "increasingly generate, summarize and reinterpret scientific knowledge" (Medium card, snippet only). A knowledge graph, on #176's reading, "does not eliminate rhetoric" but relocates it into schema design, claim granularity, relation typing, ordering and metadata, choices that "determine what a machine can recover, connect, privilege, or erase"; #176 registers the gap claim as "well-supported by survey of existing literature" and to be revised "to relative rather than absolute novelty" if an initiative is shown to incorporate comparable compositional theory (#176). #176 sets the nanopublication and micropublication paradigms as "the technical infrastructure for the claim-level decomposition that P1 requires," training-layer literature adding "the literary practice … that these infrastructures lack," and proposes a pilot over "(a) traditional papers, (b) FAIR/nanopublication-augmented papers, and (c) training-layer publications," measuring among other things what fraction of claims "preserve their provenance chains through retrieval" (#176). The Claim Status Packet "follows nanopublication practice — assertion, provenance of assertion, publication information kept as separate graphs," "the closest existing analogue," and adds a question that practice does not ask, "what standing is this claim permitted to carry, and did mediation alter that standing without a warrant?"; the packet is a schema draft, "Not a deployed instrument" (#1529). The encyclopedia has no article: the title "Nanopublication" redirects to Microattribution, a legal-term stub whose section "Nanopublications" reproduces Giachelle, Dosso and Silvello's sentences, and the plural title has no page (Wikipedia, 2026-10-29); one card, the Research Object page, carries its own notice that "These pages are no longer maintained" (Research Object), which the Capture Registry records as a retired page among the cards (Capture Registry, 2026-10-29). At the bare name the Overview cited nanopub.net and named no person, institution or identifier, Groth et al. 2010 appearing "only in a card"; it neither stated nor contradicted the census's two negative tags, stopping at "citable and attributed," and left out signing, the second-generation Registry and Query and the legacy server; the census record, which landed the same day, sets the model beside ProvStore, and no card cites it (Capture Registry, 2026-10-29, checked against the census record and the pages it saved; provenance census record).

T against L_F(B) (AIO's ontology at work on its own cards): At 'Nanopublications' (T1, nine cards: B1–B9; 202 words) the composition gives the model by its own definition and its three parts. Composed from its cards: the definition, joining two grains, nanopub.net's 'small knowledge graph snippet … treated as an independent (scientific) publication' (B1 F2) and the older 'smallest unit of publishable information' that the Research Object page quotes from nanopub.org, Pensoft attributes to Knowledge Pixels and the library documentation restates (B3 F15, B7 F41, B8 F50); 'rather than a traditional research paper' at Pensoft's grain ('unlike a research article', B7 F42); the three parts at nanopub.net's grain (B1 F7–F9), the Capture Registry's finding (CAP-1-01); RDF triples (B1 F10, B5 F24); 'RDF and OWL' at Schultes et al.'s grain (B6 F33); 'search, aggregate, and reason' at Giachelle et al.'s (B5 F65); 'AI tools' at Pensoft's (B7 F47); 'uniquely identified and tracked' at the Research Object page's quotation of nanopub.org (B3 F15–F16); credit at nanopub.net's incentive (B1 F6); 'millions' at Giachelle et al.'s abstract ('millions of nanopublications currently available on the Web, especially in the life science domain', B5 F20) and Schultes et al. (B6 F36), the figure the registry calls unchecked (CAP-1-03); nanopub-py (B2 F12) and the site. Force raised: 'three mandatory parts' (T1-2) where nanopub.net says 'three basic elements' (F7) and Giachelle et al. count four with a head (F26); 'heavily concentrated' (T1-8) where the cards say 'especially' and 'chiefly' (F20, F21). Substituted reading: T1-4 puts 'primary source or paper' into the provenance, where Giachelle et al. put 'the evidence paper from which the assertion was extracted' into the publication info (F27) and nanopub.net's provenance names methods and 'the kind of study' (F8); T1-3 narrows the assertion to 'a single, atomic scientific statement or claim', where nanopub.net says it 'can be about anything, for example a relation between a gene and a disease or an opinion' (F3) and Schultes et al. include 'hypotheses, claims, negative results, and opinions' (F35); T1-1's 'rather than a traditional research paper' sets against the article what nanopub.net allows 'with or without an accompanying research article' (F5). Limit dropped: T1-7's 'Citable and Attributed … giving researchers credit' composes the incentive (F6, F16) without its card's limit, 'hard to find and rarely cited' (B5 F22), or the 2021 limit on search and readability (F29); T1-6 composes machine-readability without Schultes et al.'s bound to 'the assertional data type' and small size (B6 F38). Outside the cards' text: 'genetics, pharmacology, and bioinformatics' (T1-8) is in no saved card text or snippet (the cards say 'life science domain'); 'granular data' (T1-7) is in no field claim. Available, not composed: the FAIR headline (F1), the Agent Skill (F11), the Research Object page's own notice of retirement (F13; CAP-1-05), the uptake and 'growing resource of provenance' of the IEEE card (F18), the four parts (F26), the founding motive and the origin in 2009 with Groth et al. and Mons and Velterop (F28, F31; CAP-1-09: 'Groth et al. 2010 appears only in a card'), the Server Network (F34), FANTOM5 (F36), the FAIR Digital Object comparison and its finding (F37, F39, F40), provenance as tracing findings to their origin (F43), Knowledge Pixels and its founders (F45, F49), the publishers' workflows (F48), the machine-reader claims (F46), NanoWeb (F30), retraction (F53), the Medium essay's question (F54). Not surfaced by T and in L_F(B) by the operator's ruling: the architecture page (B10: the second-generation Registry and Query, the legacy server for old unsigned nanopublications, Nanodash, signing over nanopub-http; CAP-1-04) and the encyclopedia's redirect (B11, B12). Composed without attribution: no person, institution or identifier is named in the answer (CAP-1-09, PER 0.75), at the coarse grain of a definition common to its cards except the definition, which is at its sources' grain and carries two markers; in a basin no rival contests. Content replaced by use: the closer (T1-10), an offer menu (N_ext).

v0.9 §0.5: the first diff is ontological and small. T and the field agree on the entity, an assertion published with its provenance and publication info as one RDF graph; T composes it as a tool for credit and machine reading and leaves out what its own cards give as its limits (rarely cited, hard to find, assertional data only, a retired page among them), its history (2009, FANTOM5) and its people, and it moves the evidence paper from the publication info into the provenance.

L_F(B) against L_A(B ∪ A) (the intervention): Admission adds, in one claim space with the field's: the census record's fields, which restate the own pages and add what the pages do not state, that the model expresses and records provenance and its survival is 'not tested' (PE-04, PE-09), that it is 'NOT a test of whether provenance survives reuse' and that 'publication and signing do not settle an assertion's truth' (PE-13, PE-14), under the census's question 'with what evidence of survival?' (PE-18), with its operator 'the nanopublication community' (PE-15) beside the field's Knowledge Pixels (F45, F49) and its own unchecked items (PE-16); #176's reading of the paradigm as data architecture on the assumption that reception is a parsing problem, with the compositional properties no schema captures, 'proximity of provenance to claims' among them, and its gap claim with its challenge condition (D176-01, D176-03–D176-06, D176-10); nanopublications as the technical infrastructure for training-layer literature's claim-level decomposition and the proposed pilot measuring provenance preservation under retrieval (D176-07–D176-09, D176-11, D176-12); the Claim Status Packet's adoption of nanopublication practice as its closest analogue, and its question of standing (D1529-01–D1529-03); FAIR as the specification of semantic persistence and machine-readable provenance as verifiable by the summarizer (#510, O510-01, O510-02), held apart from #176; the reception consequences metadata standards do not address (#833, O833-01); and the Capture Registry's account of the composition at the name (CAP-1-01–CAP-1-09). No archive claim contradicts a field claim; two qualify one: #176's data-architecture reading qualifies the field's machine-actionability claims (F44, F47, F51, F65), and the census's negative tags qualify the field's citation incentive (F6, F16). Convergences: CAP-1-03 with PE-16 (no count in the record; 'record counts in the Registry' unchecked) against the field's millions (F20, F21, F36); D176-02 (micropublications extend the model) with F38 (the model's scope limited to assertions).

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', 'F42', 'F43', 'F44', 'F45', 'F46', 'F47', 'F48', 'F49', 'F50', 'F51', 'F52', 'F53', 'F54', 'F65'], 'T1_composed': ['F2', 'F3', 'F6', 'F7', 'F8', 'F9', 'F10', 'F12', 'F15', 'F16', 'F20', 'F24', 'F33', 'F41', 'F42', 'F47', 'F50', 'F65'], 'T1_composed_count': '18 of 55 card claims', 'T1_outside_cards': ["'genetics, pharmacology, and bioinformatics' (T1-8): in no card text or snippet", "'mandatory' (T1-2): force raised over F7", "'granular data' (T1-7): in no field claim"], 'T1_substituted': ['T1-4 (the evidence paper moved into provenance; F27, F8)', 'T1-3 (the assertion narrowed to a scientific claim; F3, F35)', "T1-1 ('rather than' for 'with or without'; F5)"], '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)"}