A library can look deceptively simple from the outside: shelves, books, tables, computers, perhaps a counter where a reader borrows something and returns it later. But the visible room is only the front end of a much larger system. Behind every successful library sits a chain of decisions about what to collect, what not to collect, how to identify each item, how to describe it, where to place it, who may use it, how long it should be kept, how it should be preserved, what happens when it becomes damaged or obsolete, and how a reader who does not already know its title can still find it.

A library therefore is not merely a place where books are stored. It is a knowledge infrastructure. Its job is to make selected knowledge discoverable, usable, trustworthy enough for its purpose, preservable across time and available to the people the institution exists to serve.

The IFLA-UNESCO Public Library Manifesto 2022 describes the public library as a living force for education, culture and information and emphasises free and equal access to knowledge, participation, digital inclusion and lifelong learning. That public mission is one important form of library work, but libraries also exist in schools, universities, governments, research institutes, companies, hospitals, courts, museums, monasteries and private collections. Their audiences differ. Their acquisition rules differ. Their preservation requirements differ. Yet the underlying machinery is recognisably related.

The library loop

MISSION
→ COLLECTION POLICY
→ DISCOVER NEED
→ SELECT
→ ACQUIRE OR LINK
→ IDENTIFY
→ DESCRIBE
→ CLASSIFY
→ SHELVE OR HOST
→ DISCOVER
→ REQUEST
→ ACCESS
→ USE
→ RETURN OR RETAIN
→ PRESERVE
→ REVIEW
→ WEED, TRANSFER, DIGITISE OR ARCHIVE
→ NEW NEED

This loop explains why a large library is much more than storage. A collection that cannot be discovered is functionally hidden. A catalogue that does not distinguish editions can route a reader to the wrong object. A digital file without preservation planning can become unreadable. A popular collection with no retention discipline can crowd out rare or strategically important material. Every stage contributes to whether knowledge actually reaches the reader.

1. The mission determines the collection

No library can collect everything. Even the largest national libraries make choices because publishing output, digital information and research data exceed any institution’s practical capacity. The first discipline is therefore mission.

A school library may prioritise age-appropriate reading, curriculum support and literacy development. A university library may support research disciplines, scholarly journals, specialist databases and long-horizon preservation. A public library may balance fiction, practical information, local history, digital access, language needs, community programmes and lifelong learning. A corporate library may retain only material connected to a narrow technical field or institutional memory.

Mission answers the question that makes every later acquisition defensible: why does this collection need this item?

2. Collection development is strategic selection

Collection development is the process by which a library decides what knowledge should enter, remain in or leave a collection. It considers users, subject coverage, format, cost, authority, duplication, licensing, preservation, language, demand, uniqueness and future value.

• Need: does the item serve an actual reader or institutional job?
• Authority: is the source appropriate for the claims or subject it represents?
• Coverage: does it fill a gap or merely duplicate strong holdings?
• Durability: will the item remain useful long enough to justify acquisition and maintenance?
• Rights: can the library lawfully lend, copy, digitise, preserve or provide access?
• Format: is print, digital, audio, dataset, archive, object or link the correct form?
• Cost: does the expected value justify purchase, licensing and storage?
• Diversity of viewpoint: is the collection broad enough to support inquiry rather than only one institutional perspective?

In eduKateSingapore, the same logic appears in the Collections Development, Curation & Acquisitions Wing. A new article is not automatically the right acquisition. Sometimes the correct action is to link to a stronger external authority, update an existing owner, register a primary source or leave the gap deliberately unfilled until better evidence exists.

3. Acquisition is not always purchase

Libraries acquire material through purchase, subscription, donation, legal deposit, exchange, licence, institutional transfer, digitisation, open-access harvesting and links to external repositories. Each route creates different obligations.

A purchased printed book may physically belong to the library. A licensed journal database may only be available while subscription terms remain active. A donated archive may arrive with restrictions. An open-access paper may be free to read but still carry copyright conditions. A legal-deposit publication may have statutory preservation duties. A link to an external repository provides access without transferring custody at all.

This distinction is foundational: access, possession, ownership, licence and preservation rights are not the same thing.

4. Every item needs identity

Once something enters the library, it needs a stable identity. A title alone is not enough. Many books share titles. Works appear in multiple editions, languages, formats and revisions. Journals continue across years. Digital files can change while keeping the same visible filename.

Libraries therefore use identifiers such as accession numbers, barcodes, call numbers, ISBNs, ISSNs, DOIs, handles and internal record IDs. These identifiers perform different jobs. An ISBN distinguishes a publication product. An accession number can represent the library’s own custody record. A DOI can provide persistent scholarly identification. A call number helps route a physical object to a shelf location.

Good identity survives superficial change. A book can move shelf. A digital item can move server. A catalogue interface can be redesigned. The underlying identity should remain stable enough that citations, loans, conservation records and later corrections still refer to the same object.

5. Cataloguing turns objects into discoverable records

Cataloguing creates structured descriptions so that a user can find, identify, select and obtain resources. Typical catalogue fields include title, creator, date, edition, publisher, subject, language, format, identifiers, physical description, notes and location.

This work may appear clerical until something goes wrong. If a translated edition is catalogued as though it were the original language edition, a reader may request the wrong text. If a corporate author is inconsistent, related reports become scattered. If a date reflects digitisation rather than original publication, historical research may be distorted.

Metadata is therefore a reasoning layer. It decides which properties of a resource are important enough to make explicit and machine searchable.

6. Description is not classification

Description records what the item is. Classification places the item into a knowledge structure. The two tasks overlap but are not identical.

A catalogue can describe a book as being about climate adaptation, urban planning and flood risk. A classification system must decide where that combination sits within a hierarchy or faceted structure. One library may place it under environmental science, another under civil engineering, another under public policy.

This is why classification is never completely neutral. It is a practical map of knowledge. The map must be useful, revisable and transparent enough that users can understand its logic.

eduKate’s How to Categorise Anything, How to Build a Taxonomy and How Classification Crosswalks Work provide the general reasoning that sits underneath this library function.

7. Authority control keeps names coherent

Authors, organisations, places and subjects can appear under different names. A person may publish under initials, a married name or different transliterations. An institution may be renamed. A historical city may have multiple names across languages and periods.

Authority control connects these variants to a preferred or stable identity. Without it, a search for one form may miss half the collection.

ENTITY
→ STABLE IDENTITY
→ PREFERRED LABEL
→ ALTERNATIVE LABELS
→ PREVIOUS NAMES
→ LANGUAGE VARIANTS
→ RELATED ENTITIES
→ DATES / CONTEXT
→ RECORDS USING THAT IDENTITY

This idea scales far beyond libraries. Any knowledge system that wants identity to survive renaming needs the same distinction between the thing and the labels used to refer to it.

8. Shelf order is a physical user interface

In a physical library, classification often becomes spatial. Related subjects are shelved near one another. The shelf therefore acts as a browsing interface: a reader looking for one book may discover another nearby.

This adjacency is valuable because human inquiry is often imprecise. A reader may not know the exact term used by the catalogue. Browsing a neighbourhood of related material can expose vocabulary, alternative perspectives and unexpected connections.

Digital systems try to reproduce some of this through recommendations, subject facets, related records and citation networks. But physical adjacency remains a remarkably efficient form of exploratory search.

9. Discovery begins before the user knows what to ask

A good library supports at least three different search states:

• Known-item search: “I need this exact title or article.”
• Subject search: “I need reliable material about this topic.”
• Exploratory search: “I do not yet know enough to name what I need.”

The third state is the hardest. It requires vocabularies, subject headings, reference help, bibliographies, browsing structures and increasingly machine-assisted discovery. Retrieval systems are strongest when they help the reader move from vague intent to a better-formed question.

10. Reference service is human routing

A librarian at a reference desk does not simply answer questions. Often the first task is to discover what the user is actually trying to do.

A student who asks for “books on pollution” may really need evidence for a geography assignment comparing coastal waste policies. A researcher asking for a newspaper may need an article from a specific date, not the publication generally. A parent asking for “good science books” may need material at a particular reading level.

The reference interview narrows the problem before searching. In eduKateAI terms, this resembles a good HELP route: locate the user state, measure what is missing, ask a discriminating question, then route to the correct owner or evidence source.

11. Circulation is controlled temporary custody

Lending looks simple because libraries have made it simple for users. Behind the scan is a state machine:

AVAILABLE
→ ON HOLD
→ CHECKED OUT
→ DUE
→ RENEWED / RETURNED / OVERDUE
→ INSPECT
→ RESHELVE / REPAIR / LOST

The system must know which copy is where, which user has custody, when it should return and whether another user is waiting. A modern integrated library system connects circulation with cataloguing, acquisitions, serials, fines or notices, user accounts and inventory management.

12. Access is governed, not merely opened

Libraries usually promote access, but not every item is equally open. Rare books may require supervised consultation. Archives may contain personal data. Commercial databases may be restricted to institutional members. Indigenous knowledge may carry community access protocols. Fragile material may be available only as a surrogate.

The responsible question is therefore not “open or closed?” but “what level of access is justified by mission, law, rights, preservation and ethics?”

This is the same logic seen in How Archives Reading Rooms, Access and Restrictions Work: access is a designed relationship between the user and the collection, not an accidental property of storage.

13. Public libraries are civic knowledge infrastructure

The IFLA-UNESCO Public Library Manifesto 2022 places public libraries inside a wider civic mission: access to information, culture, literacy, education, digital participation and community life. The updated manifesto reflects the fact that a modern public library is not confined to print lending.

• reading and literacy support;
• internet and device access;
• media and information literacy;
• local history and memory;
• language learning;
• children’s and youth programmes;
• research support;
• community meetings and cultural events;
• access to government and practical information;
• support for people excluded by cost, technology or geography.

A public library therefore reduces several kinds of distance at once: economic distance from books and databases, digital distance from technology, educational distance from research help and social distance from shared civic space.

14. School libraries connect curriculum to independent inquiry

A school library has a distinctive role because it sits between formal instruction and voluntary exploration. It supports curriculum but should not be reduced to a textbook storeroom.

In April 2025, the updated IFLA-UNESCO School Library Manifesto was endorsed by UNESCO’s Information for All Programme. The update highlights collaboration for literacy, critical thinking, creativity and global citizenship. That matters because students need more than access to information; they need practice in deciding which information deserves trust and how different sources can be combined.

A strong school library helps a learner move from “find a fact” to “build an argument from evidence”.

15. University libraries are research infrastructure

University libraries support teaching and research at a much higher level of specialisation. Their collections may include scholarly journals, monographs, datasets, archives, theses, standards, patents, maps, rare books, institutional repositories and expensive databases.

The university library also increasingly supports research data management, open access, citation analysis, persistent identifiers, copyright, digital scholarship and publication workflows. It becomes part of the research system itself.

This connects directly to Research Data Management and FAIR Principles and How Scholarly Publishing and Peer Review Work.

16. Serials and databases behave differently from books

A monograph is usually acquired as a discrete edition. A journal is a continuing publication. A database may change daily. A streaming service may provide temporary access to a moving catalogue.

These resources require subscription management, licence tracking, holdings data, authentication, usage statistics and renewal decisions. The library must know not only whether a journal exists, but which years and volumes are accessible, through which provider, under which terms.

This is one reason electronic access can be operationally harder than physical ownership. The reader experiences one click; the library manages identity, contracts, network access, metadata and continuity behind the click.

17. Digital libraries need preservation, not just storage

A digital file does not survive merely because it has been copied to a server. Formats become obsolete. Storage media fail. Links rot. Platforms close. Encryption keys disappear. Proprietary software becomes unavailable. Files can silently corrupt.

Digital preservation therefore uses checksums, redundant storage, format monitoring, migration, emulation, metadata, provenance and controlled versioning. The objective is not only to retain bits but to preserve enough context for the object to remain intelligible and authentic.

See How Digital Archives Work and How Preservation and Conservation Work for the deeper preservation machinery.

18. Preservation is a portfolio decision

No library has unlimited preservation resources. Rare, unique and irreplaceable material may justify expensive conservation. A widely held commercial textbook may not. A local newspaper collection may have exceptional regional value even if the physical paper is cheap.

Preservation priorities therefore consider uniqueness, significance, condition, demand, replacement availability, rights and institutional mission. Sometimes the correct action is conservation. Sometimes digitisation. Sometimes reformatting. Sometimes transfer to an archive. Sometimes withdrawal.

19. Weeding is not anti-library

Removing items can feel contrary to the idea of a library, but an active lending collection must remain usable. Outdated medical guides, superseded examination materials, duplicate low-demand copies and damaged items can consume space and mislead readers.

Weeding should be governed by policy, not taste. Criteria may include factual obsolescence, physical condition, use, duplication, replacement, local significance and preservation responsibility.

The crucial distinction is between withdrawal from an active collection and destruction of unique historical evidence. A library that owns the only surviving copy of a local pamphlet has a different responsibility from one holding the twentieth duplicate of an outdated mass-market manual.

20. Interlibrary loan expands the effective collection

No library needs to own everything if libraries can cooperate. Interlibrary loan, union catalogues, reciprocal access and shared repositories allow one institution to route a user to holdings elsewhere.

This changes the concept of collection from “what is inside my building” to “what can my users reliably reach”. The more mature the network, the less every institution must duplicate specialised material.

eduKate’s Interlibrary Discovery & Crosswalk Wing is built on the same principle: find the best owner first, then connect rather than copying unnecessarily.

21. Copyright shapes digital library behaviour

Copyright affects lending, copying, preservation, digitisation, course reserves, document delivery and online access. Physical ownership of a copy does not automatically create the right to reproduce or distribute it digitally.

Licences add another layer. A library may have lawful subscription access to an electronic resource under terms that differ from ordinary copyright defaults. Librarians therefore need rights metadata and staff expertise, especially for digitisation and repository work.

The exact legal position varies by jurisdiction, so good library practice separates the general principle—rights must be checked—from jurisdiction-specific advice.

22. Privacy is part of intellectual freedom

Library systems can know what people borrow, search and request. That information may reveal political interests, health concerns, religion, personal identity or research activity. Retaining more data than necessary can therefore create risk.

Strong library privacy practices minimise unnecessary collection, restrict access, set retention rules and separate operational need from surveillance. The reader should be able to explore ideas without assuming that every inquiry becomes a permanent behavioural profile.

23. Information literacy is a library output

Access to information does not guarantee good judgement. A modern library helps users understand how search systems work, why databases differ, how to recognise primary and secondary sources, how to evaluate authority and how to cite evidence responsibly.

This is the bridge to How Research Methods and Source Evaluation Work. The strongest library does not simply place information in front of a reader; it improves the reader’s capacity to interrogate that information.

24. AI changes discovery but not the ownership problem

AI can make a catalogue easier to query in natural language, summarise documents, recommend related material and help users traverse unfamiliar vocabularies. It can also invent citations, collapse editions, confuse authoritative and non-authoritative copies, or conceal uncertainty behind fluent prose.

A library using AI therefore needs stable identities, provenance, source boundaries and retrieval tests more than ever. AI should sit above the library’s metadata and evidence structure, not replace it.

The machine question is not “Can the model answer?” but “Can the system show which collection object, version and authority the answer came from?”

25. The catalogue itself needs quality control

Metadata can become stale, inconsistent or duplicated. Authority records can fragment. URLs break. Holdings can show items that have been lost. Subject terms can become outdated or insensitive. Automated imports can create malformed records.

Libraries therefore perform catalogue maintenance: deduplication, authority reconciliation, link checking, inventory, metadata enrichment and error correction. A catalogue is a living information system, not a finished index.

26. A library has several kinds of truth

Libraries hold works that disagree. They hold outdated science, propaganda, fiction, error, satire, historical prejudice, contested scholarship and revised editions. The presence of an item in a collection is not an endorsement of every claim inside it.

The library’s truth obligation is different. It must represent the object truthfully: what it is, who produced it, when, in which edition, under which context and how it relates to other evidence. A misleading book may still be historically important if catalogued and contextualised correctly.

27. Collections reveal institutional bias

Every collection is shaped by budget, language, geography, acquisition history, professional networks and institutional priorities. What is absent can be as revealing as what is present.

A library can reduce blind spots through collection audits, community consultation, usage analysis, subject mapping and acquisition diversity. But it should also preserve the history of its own collection decisions rather than pretending the collection emerged neutrally.

28. Libraries measure use, but use is not the only value

Circulation counts, downloads, visits, searches and programme attendance help libraries allocate resources. Yet low use does not always mean low value. A rare legal reference may be used only occasionally but become crucial during one case. A local archive may serve few users now but carry enormous future historical value.

Library value therefore has several dimensions: frequency of use, strategic importance, uniqueness, cultural significance, educational reach, resilience and future option value.

29. The strongest library is a routed network, not a warehouse

A warehouse can contain enormous quantities and still fail a reader. A library succeeds when knowledge can be routed from need to source with enough context to use it well.

READER NEED
→ CLARIFY
→ SEARCH LOCAL HOLDINGS
→ SEARCH NETWORK
→ IDENTIFY BEST SOURCE
→ VERIFY EDITION / AUTHORITY
→ PROVIDE ACCESS
→ SUPPORT USE
→ RECORD GAP IF UNRESOLVED

This model is exactly why a large digital knowledge estate needs canonical ownership and crosswalks. More pages do not automatically create more intelligence. Better routing does.

30. eduKateSingapore as a library system

eduKateSingapore increasingly behaves like a digital research library rather than a conventional website. It contains curriculum material, science manuals, world systems, country and city knowledge, publishing infrastructure, archives, data-management articles and crosswalk pages.

The useful question is therefore not only “What article should be written next?” It is “What collection position is missing, who owns it, what source authority should support it, and how will a future learner or eduKateAI route to it?”

Wintour House governs whether a commissioned article deserves publication. The Collections Development, Curation & Acquisitions Wing governs whether a gap should be acquired at all. The Eratosthenes Accession Ledger protects identity and disposition. The Research Collections Directory provides public research routes.

That architecture turns “library” from a metaphor into an operating discipline.

The final idea

The deepest purpose of a library is not to preserve objects for their own sake. It is to preserve the possibility of informed human thought.

A reader today needs access. A researcher tomorrow needs provenance. A community decades from now may need evidence that nobody currently thinks is important. The library holds these time horizons together. It chooses, identifies, describes, preserves and routes knowledge so that one generation can hand usable understanding to another.

That is why libraries remain essential even in a world of search engines and AI. Search can retrieve. AI can synthesise. But a library answers a more durable question: what knowledge object is this, where did it come from, why is it here, can I trust its identity, and will someone still be able to find it later?

Sources and further reading

• IFLA-UNESCO Public Library Manifesto 2022
• IFLA — Updated IFLA-UNESCO School Library Manifesto endorsed in 2025
• IFLA — Public Library Manifesto Toolkit
• IFLA Public Libraries Section

eduKate route: Continue with How Archives Work, How Museums and Material Evidence Work, How Publishing Works, How Research Methods and Source Evaluation Work and the World Knowledge Research Library.