@prefix : . @prefix schema: . @prefix skos: . @prefix rdfs: . @prefix rdf: . @prefix owl: . @prefix xsd: . @prefix dbo: . @prefix org: . # ───────────────────────────────────────────── # Lightweight Ontology # ───────────────────────────────────────────── :StandardizationEra a rdfs:Class ; rdfs:label "Standardization Era"@en ; rdfs:comment "A historical period characterised by the adoption of interoperability standards."@en . :LibraryStandard a rdfs:Class ; rdfs:label "Library Standard"@en ; rdfs:comment "A bibliographic or metadata standard adopted by the library community."@en . :FragmentationRisk a rdfs:Class ; rdfs:label "Fragmentation Risk"@en ; rdfs:comment "A risk arising when independently developed implementations diverge and become incompatible."@en . :GovernanceBody a rdfs:Class ; rdfs:label "Governance Body"@en ; rdfs:comment "An institution positioned to initiate sector-wide standardization."@en . :hasAdoptionYear a rdf:Property ; rdfs:label "has adoption year"@en ; rdfs:domain :LibraryStandard ; rdfs:range xsd:gYear . :hasFragmentationStage a rdf:Property ; rdfs:label "has fragmentation stage"@en ; rdfs:range xsd:string . # ───────────────────────────────────────────── # Main Article # ───────────────────────────────────────────── :article a schema:Article ; schema:name "The Whitworth Problem: AI in Libraries Is Still Building Incompatible Bolts"@en ; schema:headline "The Whitworth Problem: AI in Libraries Is Still Building Incompatible Bolts"@en ; schema:description "Libraries spent a century solving the standardization problem for knowledge organisation. Now AI is recreating the pre-Whitworth fragmentation they already solved — and the sector risks ignoring it."@en ; schema:url "https://www.linkedin.com/pulse/whitworth-problem-ai-libraries-still-building-bolts-edelenbos-5wbze" ; schema:datePublished "2026-04-21"^^xsd:date ; schema:dateModified "2026-04-21"^^xsd:date ; schema:inLanguage "en" ; schema:author :author ; schema:publisher :publisher ; schema:about :whitworthProblem, :libraryStandardization, :aiLibrariesFragmentation, :standardizationEnablement ; schema:keywords "AI libraries, standardization, Whitworth, MARC, Dublin Core, RDA, interoperability, metadata, library science, knowledge organisation"@en ; schema:isAccessibleForFree true ; schema:interactionStatistic [ a schema:InteractionCounter ; schema:interactionType schema:LikeAction ; schema:userInteractionCount 8 ] ; schema:hasPart :faqSection, :glossarySection, :howtoSection . # ───────────────────────────────────────────── # Author & Publisher # ───────────────────────────────────────────── :author a schema:Person ; schema:name "Stuart Michael Edelenbos"@en ; schema:url "https://nl.linkedin.com/in/stuart-edelenbos" ; schema:identifier "https://nl.linkedin.com/in/stuart-edelenbos" ; schema:interactionStatistic [ a schema:InteractionCounter ; schema:interactionType schema:FollowAction ; schema:userInteractionCount 1034 ] . :publisher a schema:Organization ; schema:name "LinkedIn"@en ; schema:url "https://www.linkedin.com" ; schema:identifier "https://www.linkedin.com" . # ───────────────────────────────────────────── # Core Argument: The Whitworth Problem # ───────────────────────────────────────────── :whitworthProblem a schema:CreativeWork ; schema:name "The Whitworth Problem"@en ; schema:description """In the early 19th century every engineering workshop made its own bolts — a bolt from Manchester would not fit a nut from Birmingham. Joseph Whitworth proposed a single national standard for screw threads in 1841. The Society of Automotive Engineers codified fastener standards in 1911 because Ford's assembly line required fully interchangeable parts. The article argues that AI in libraries is recreating this pre-standardisation fragmentation."""@en ; schema:about :josephWhitworth, :automotiveStandard, :preStandardisationEra . :josephWhitworth a schema:Person ; schema:name "Joseph Whitworth"@en ; schema:description "British engineer who proposed the first national standard for screw threads in 1841."@en ; schema:birthDate "1803"^^xsd:gYear ; schema:deathDate "1887"^^xsd:gYear ; owl:sameAs ; rdfs:seeAlso . :whitworthStandard a :LibraryStandard, schema:CreativeWork ; schema:name "Whitworth Screw Thread Standard"@en ; :hasAdoptionYear "1841"^^xsd:gYear ; schema:description "First national standard for screw threads: consistent angles, consistent pitches, consistent tolerances."@en ; owl:sameAs . :automotiveStandard a :StandardizationEra ; schema:name "SAE Fastener Standards (1911)"@en ; schema:description "Society of Automotive Engineers codified fastener standards in 1911 because Ford's assembly line required fully interchangeable parts."@en ; schema:about :fordAssemblyLine ; rdfs:seeAlso . :fordAssemblyLine a schema:Product ; schema:name "Ford Assembly Line"@en ; schema:description "Required fully interchangeable parts; a bolt that 'almost fitted' was useless."@en ; owl:sameAs . :preStandardisationEra a :StandardizationEra ; :hasFragmentationStage "Exploratory"@en ; schema:name "Pre-Standardisation Era"@en ; schema:description "Early-stage technology adoption where variety of implementations is normal and useful — but tips into compounding fragmentation if a standard is not established."@en . # ───────────────────────────────────────────── # Library Standardization History # ───────────────────────────────────────────── :libraryStandardization a schema:CreativeWork ; schema:name "Library Standardisation History"@en ; schema:description "Libraries have a century-long track record of solving the Whitworth Problem for knowledge organisation: MARC (1966), Dublin Core (1995), RDA (2010)."@en ; schema:hasPart :marc, :dublinCore, :rda . :marc a :LibraryStandard ; schema:name "MARC (Machine-Readable Cataloging)"@en ; :hasAdoptionYear "1966"^^xsd:gYear ; schema:description "Introduced in 1966. A cataloging record created at the Library of Congress could be read by any library system anywhere in the world that implemented the standard — the first time bibliographic data could travel across institutional borders without losing meaning."@en ; schema:identifier "https://www.loc.gov/marc/" ; owl:sameAs . :dublinCore a :LibraryStandard ; schema:name "Dublin Core"@en ; :hasAdoptionYear "1995"^^xsd:gYear ; schema:description "Fifteen simple metadata elements applicable to any resource, any system, globally interoperable. Extended MARC logic to the web."@en ; schema:identifier "https://www.dublincore.org" ; owl:sameAs . :rda a :LibraryStandard ; schema:name "RDA (Resource Description and Access)"@en ; :hasAdoptionYear "2010"^^xsd:gYear ; schema:description "Rebuilt the conceptual framework from the ground up, aligning library description practice with the entity-relationship models that linked data requires."@en ; schema:identifier "https://www.rdatoolkit.org" ; owl:sameAs . # ───────────────────────────────────────────── # AI Libraries Fragmentation # ───────────────────────────────────────────── :aiLibrariesFragmentation a :FragmentationRisk ; schema:name "AI in Libraries Fragmentation Problem"@en ; schema:description """The library sector is developing AI implementations — for cataloging, reference services, collection development, accessibility, preservation — without establishing shared output specifications or validation criteria. Each institution builds its own bolt."""@en ; schema:hasPart :taskReplacementFrame, :strategicFrame, :innovationTheater, :whitworthQuestion . :taskReplacementFrame a schema:CreativeWork ; schema:name "Task Replacement Frame"@en ; schema:description "The dominant but insufficient frame: what can AI do instead of a librarian? Which workflow steps can be automated? Where can we reduce costs or handle backlogs? A reasonable operational question — not a strategic one."@en . :strategicFrame a schema:CreativeWork ; schema:name "Strategic Frame"@en ; schema:description "The correct frame: what standardised, interoperable, domain-wide AI-assisted workflows can the library sector establish such that expertise accumulates rather than fragments?"@en . :innovationTheater a schema:CreativeWork ; schema:name "Innovation Theater"@en ; schema:description "Impressive demonstrations that do not survive contact with production workflows. Identified as part of the current AI-for-libraries discourse."@en . :whitworthQuestion a schema:Question ; schema:name "The Whitworth Question"@en ; schema:text "What is the shared specification that any library, anywhere, implementing AI-assisted cataloging, should produce outputs compatible with? What are the validation criteria that travel across systems?"@en ; schema:description "The strategic question the sector has not yet answered — contrasted with pre-Whitworth questions about individual tool choices and local workflow metrics."@en . :preWhitworthQuestions a schema:Question ; schema:name "Pre-Whitworth Questions"@en ; schema:text "What tool does your institution use? What workflow did you develop? What percentage of catalog records does your AI assistant handle correctly?"@en ; schema:description "Operational questions with their place, but insufficient — the questions of the pre-standardisation era."@en . # ───────────────────────────────────────────── # Governance Bodies # ───────────────────────────────────────────── :ifla a :GovernanceBody, org:Organization ; schema:name "IFLA"@en ; schema:description "International Federation of Library Associations and Institutions. Identified as positioned to initiate AI standardisation for the sector."@en ; schema:identifier "https://www.ifla.org" ; owl:sameAs . :liber a :GovernanceBody, org:Organization ; schema:name "LIBER"@en ; schema:description "Association of European Research Libraries. Identified as positioned to initiate AI standardisation for the sector."@en ; schema:identifier "https://libereurope.eu" . :nationalLibraryAssociations a :GovernanceBody ; schema:name "National Library Associations"@en ; schema:description "National bodies that drove previous standardisation cycles (MARC, Dublin Core, RDA) and need to initiate the AI standardisation conversation."@en . # ───────────────────────────────────────────── # Standardization Enables Diversity # ───────────────────────────────────────────── :standardizationEnablement a schema:CreativeWork ; schema:name "Standardization Enables Diversity"@en ; schema:description """The Whitworth insight usually forgotten: the standard was the foundation for diversity, not a constraint on it. Standardised thread specifications let specialized tools proliferate because they all operated from a shared base. Library AI standardisation is not an argument for uniformity — specialised applications for rare books, oral history archives, multilingual collections can flourish precisely because they operate from a shared foundation."""@en ; schema:hasPart :sharedLearning, :islandGeography . :sharedLearning a schema:CreativeWork ; schema:name "Shared Learning from Common Output Specification"@en ; schema:description "When the output specification is common, every institution's implementation contributes to collective knowledge. A cataloger in Leiden and a cataloger in Lagos using AI-assisted subject analysis against the same output standard can compare results, identify systematic errors, and improve together."@en . :islandGeography a :FragmentationRisk ; schema:name "Island Geography of Fragmentation"@en ; schema:description "Without standardisation, the gap between well-resourced and under-resourced institutions grows rather than staying stable. Institutions with resources build sophisticated implementations; institutions without resources fall further behind."@en . # ───────────────────────────────────────────── # FAQ # ───────────────────────────────────────────── :faqSection a schema:FAQPage ; schema:name "FAQ: The Whitworth Problem in AI Libraries"@en ; schema:mainEntity :q1, :q2, :q3, :q4, :q5, :q6, :q7, :q8, :q9, :q10, :q11, :q12 . :q1 a schema:Question ; schema:text "What is the Whitworth Problem as applied to AI in libraries?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "Just as 19th-century engineering workshops built incompatible bolts until Whitworth proposed a common screw-thread standard in 1841, libraries are currently deploying AI implementations without any shared output specification — creating a modern fragmentation where every institution is building its own bolt."@en ] . :q2 a schema:Question ; schema:text "Why are libraries well-positioned to solve this problem?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "Libraries have solved the Whitworth Problem repeatedly for knowledge organisation: MARC (1966) for machine-readable cataloging, Dublin Core (1995) for web metadata interoperability, and RDA (2010) for linked-data-aligned description. This is institutional expertise of the first order."@en ] . :q3 a schema:Question ; schema:text "What is the difference between pre-Whitworth and Whitworth questions for AI?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "Pre-Whitworth questions are operational: which tool does your institution use, what percentage accuracy does your AI achieve? The Whitworth question is strategic: what shared specification should any library anywhere produce, such that outputs are interoperable across systems and institutions?"@en ] . :q4 a schema:Question ; schema:text "What is 'innovation theater' in the context of AI for libraries?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "Innovation theater refers to impressive AI demonstrations that do not survive contact with production workflows — technically rigorous-seeming showcases that create the impression of progress without establishing the shared infrastructure needed for sector-wide advancement."@en ] . :q5 a schema:Question ; schema:text "Why is task replacement framing insufficient?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "Task replacement — asking what AI can do instead of a librarian — is a reasonable operational question but not a strategic one. It produces local efficiency gains while allowing institutional gaps to widen. Without a common output standard, expertise fragments rather than accumulates."@en ] . :q6 a schema:Question ; schema:text "Which institutions are positioned to lead AI standardisation for libraries?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "IFLA, LIBER, and national library associations — the same bodies that drove previous standardisation cycles producing MARC, Dublin Core, and RDA. The conversation needs to move from 'here is what we are building' to 'here is what we should all be building toward.'"@en ] . :q7 a schema:Question ; schema:text "Does standardisation constrain diversity in library AI implementations?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "No — the Whitworth observation is the opposite: the standard was the foundation for diversity. Standardised thread specifications let specialised tools proliferate because they operated from a shared base. Similarly, library AI standardisation enables specialised applications for rare books, oral history archives, and multilingual collections to flourish."@en ] . :q8 a schema:Question ; schema:text "What is the 'island geography' problem in library AI?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "Without common standards, institutions with resources build sophisticated AI implementations while under-resourced institutions fall further behind. The gap between islands grows rather than staying stable — the pre-standardisation era reasserts itself, but with compounding inequality."@en ] . :q9 a schema:Question ; schema:text "How did the automotive industry learn from the Whitworth Problem?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "Ford's assembly line made the consequences of non-standardization impossible to ignore: a bolt that 'almost fitted' was useless. The Society of Automotive Engineers codified fastener standards in 1911 not for philosophical reasons but because the industry literally could not function otherwise."@en ] . :q10 a schema:Question ; schema:text "What would a Whitworth standard for AI-assisted cataloging look like?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "The author acknowledges this is unanswered: a minimum viable standard for AI-assisted subject analysis such that a record produced in Helsinki and one from Cape Town are interoperable — shared validation criteria, consistent output specifications, and benchmarks that travel across systems."@en ] . :q11 a schema:Question ; schema:text "Is early fragmentation in AI tool adoption inherently bad?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "No. Fragmentation in early-stage technology adoption is normal and even useful — you need variety of implementations to discover what works before usefully standardising. The pre-Whitworth era was exploratory. The danger comes when exploration tips into fragmentation that compounds rather than resolves."@en ] . :q12 a schema:Question ; schema:text "What does 'shared learning' mean in the context of library AI standardisation?"@en ; schema:acceptedAnswer [ a schema:Answer ; schema:text "When all institutions work against a common output specification, every implementation contributes to collective knowledge — systematic errors become visible and fixable across institutions. Without the common standard, identical problems are solved independently and invisibly, with no mechanism for experience to connect."@en ] . # ───────────────────────────────────────────── # Glossary # ───────────────────────────────────────────── :glossarySection a skos:ConceptScheme, schema:DefinedTermSet ; schema:name "Glossary: The Whitworth Problem in AI Libraries"@en ; skos:hasTopConcept :termWhitworthProblem, :termMARC, :termDublinCore, :termRDA, :termInnovationTheater, :termInteroperability, :termTaskReplacement, :termIslandGeography, :termSharedLearning, :termOutputSpecification . :termWhitworthProblem a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Whitworth Problem"@en ; schema:description "The fragmentation arising when independent actors each build incompatible solutions to the same problem, requiring a common standard to enable interoperability."@en ; skos:inScheme :glossarySection . :termMARC a skos:Concept, schema:DefinedTerm ; skos:prefLabel "MARC (Machine-Readable Cataloging)"@en ; schema:description "Adopted 1966. The first standard enabling bibliographic records to travel across institutional borders without losing meaning."@en ; skos:inScheme :glossarySection . :termDublinCore a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Dublin Core"@en ; schema:description "Adopted 1995. Fifteen metadata elements enabling globally interoperable resource description on the web."@en ; skos:inScheme :glossarySection . :termRDA a skos:Concept, schema:DefinedTerm ; skos:prefLabel "RDA (Resource Description and Access)"@en ; schema:description "Adopted 2010. Rebuilt library description practice from the ground up to align with entity-relationship models required by linked data."@en ; skos:inScheme :glossarySection . :termInnovationTheater a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Innovation Theater"@en ; schema:description "Impressive AI demonstrations that do not survive contact with production workflows — creating the appearance of progress without durable infrastructure."@en ; skos:inScheme :glossarySection . :termInteroperability a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Interoperability"@en ; schema:description "The ability of outputs from different systems or institutions to work together without loss of meaning — the goal of standardisation."@en ; skos:inScheme :glossarySection . :termTaskReplacement a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Task Replacement Frame"@en ; schema:description "Framing AI adoption as substitution of librarian tasks — a reasonable operational question but insufficient as a strategic framework for sector-wide progress."@en ; skos:inScheme :glossarySection . :termIslandGeography a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Island Geography"@en ; schema:description "The pre-standardisation condition where each institution is an island, unable to exchange compatible outputs with others; compounding inequality when well-resourced and under-resourced institutions diverge."@en ; skos:inScheme :glossarySection . :termSharedLearning a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Shared Learning"@en ; schema:description "The collective improvement enabled when institutions work against a common output specification — systematic errors become visible and fixable across the sector."@en ; skos:inScheme :glossarySection . :termOutputSpecification a skos:Concept, schema:DefinedTerm ; skos:prefLabel "Output Specification"@en ; schema:description "A shared, formal definition of what AI-assisted cataloging or subject analysis should produce — the equivalent of Whitworth's screw-thread specification for library AI."@en ; skos:inScheme :glossarySection . # ───────────────────────────────────────────── # HowTo # ───────────────────────────────────────────── :howtoSection a schema:HowTo ; schema:name "How to Initiate AI Standardisation for Libraries"@en ; schema:description "Steps for governance bodies and institutions to move from fragmented local implementations to a Whitworth-style common standard for AI in libraries."@en ; schema:step :step1, :step2, :step3, :step4, :step5, :step6, :step7 . :step1 a schema:HowToStep ; schema:name "Reframe the Conversation"@en ; schema:position 1 ; schema:text "Shift discourse from 'here is what we are building' to 'here is what we should all be building toward.' Replace task-replacement framing with the strategic question: what standardised, interoperable workflows can the sector establish?"@en . :step2 a schema:HowToStep ; schema:name "Convene Governance Bodies"@en ; schema:position 2 ; schema:text "Engage IFLA, LIBER, and national library associations — the institutions that drove MARC, Dublin Core, and RDA — to initiate the AI standardisation conversation at the sector level."@en . :step3 a schema:HowToStep ; schema:name "Define the Whitworth Question for AI Cataloging"@en ; schema:position 3 ; schema:text "Establish the minimum viable output specification: what should AI-assisted subject analysis produce such that a record from Helsinki and one from Cape Town are interoperable? Define validation criteria that travel across systems."@en . :step4 a schema:HowToStep ; schema:name "Distinguish Exploratory from Compounding Fragmentation"@en ; schema:position 4 ; schema:text "Monitor early adoption variety as valuable exploration, but identify the tipping point — comparable to the automotive industry circa 1905 — where fragmentation begins to compound rather than resolve, and act before it sets."@en . :step5 a schema:HowToStep ; schema:name "Build Shared Benchmarking Infrastructure"@en ; schema:position 5 ; schema:text "Create common benchmarks against the shared output specification so any institution's AI implementation can be compared, validated, and improved against the same criteria — enabling the shared learning that standardisation unlocks."@en . :step6 a schema:HowToStep ; schema:name "Design for Diversity Within the Standard"@en ; schema:position 6 ; schema:text "Frame standardisation as the foundation for specialisation, not uniformity. Ensure the shared base is minimal enough that specialised applications for rare books, oral history archives, and multilingual collections can innovate above it."@en . :step7 a schema:HowToStep ; schema:name "Address the Equity Dimension"@en ; schema:position 7 ; schema:text "Explicitly address the island geography risk: under-resourced institutions must be included in the standardisation process from the start to prevent the gap from compounding as sophisticated implementations proliferate at well-resourced institutions."@en .