Librarians and Media Collections Specialists

• Card catalog1876–1970

  Card catalog (Dewey Decimal, 1876 onward)

  Melvil Dewey's 1876 Dewey Decimal Classification and its companion card-catalog infrastructure — small wooden drawers of 3×5-inch cards, one per item — defined the physical workspace of the librarian for nearly a century. Every acquisition required a cataloger to type multiple cards (title, author, subject, call-number), refile them in the correct drawer positions, and maintain the sequence by hand. The card catalog was both the librarian's primary tool and the public interface to the collection. A mid-size library might maintain 500,000 cards across several hundred drawers; reclassification projects required physically pulling and retyping thousands of entries.

• Library of Congress Classification1897–1975

  Library of Congress Classification (LCC, 1897)

  The Library of Congress began developing its own alphanumeric classification system in 1897 as an alternative to Dewey for large research collections. By the mid-20th century, LCC had become the dominant scheme at academic and research libraries, while Dewey remained standard in public and school libraries. Catalogers had to be trained in whichever scheme their institution used — or in both, for hybrid collections. LC Subject Headings (LCSH), the authorized vocabulary that accompanies LCC, became the de facto standard for subject description and required continuous maintenance as language evolved and new subjects emerged.

• MARC standard1968–1990

  MARC standard (Henriette Avram, LOC, 1968) — machine-readable cataloging

  Henriette Avram at the Library of Congress published the MARC (Machine-Readable Cataloging) standard in 1968, defining a digital format for bibliographic records that allowed cataloging data to be shared between institutions via magnetic tape. MARC did not immediately displace card catalogs — it took two more decades before most libraries had OPAC systems that could serve MARC records to patrons — but it inaugurated the era of shared cataloging. Libraries that contributed records to a MARC database could download cataloging done elsewhere rather than creating original records from scratch. Per-item cataloging labor dropped dramatically for common titles.

  Effect on the work

  The OCLC shared-cataloging network (see next era) estimated that member libraries saved 60-80% of original-cataloging labor for standard titles within a decade of joining. Cataloging labor shifted from original creation to copy-cataloging and quality review.

• OCLC shared cataloging network1971–1995

  OCLC shared cataloging network (1967/1971 Ohio launch)

  OCLC was founded in 1967 by Frederick Kilgour and launched online shared cataloging in Ohio in 1971; it expanded nationally through the 1970s. By the 1980s, OCLC's WorldCat database contained tens of millions of bibliographic records contributed by member libraries, and the interlibrary loan (ILL) system built on it became the backbone of US library resource-sharing. For the first time, a cataloger in a small college library could receive a new book, search WorldCat, download a MARC record from a peer institution, and add the item to the local catalog in minutes rather than hours. OCLC fundamentally redistributed cataloging labor from per-library creation to centralized shared infrastructure.

  Effect on the work

  OCLC estimated that member libraries shared over 1 billion records by 2000, reducing original cataloging to a small fraction of total cataloging labor. Copy-cataloging — verifying and adapting a record rather than creating one — became the dominant workflow.

• OPAC1980–2000

  OPAC — Online Public Access Catalog (CLSI, Geac, DRA, Innovative Interfaces)

  Online Public Access Catalogs began replacing card catalogs in the late 1970s; by the mid-1980s, major integrated library system (ILS) vendors — CLSI, Geac, Data Research Associates (DRA), and Innovative Interfaces — had deployed OPAC terminals in most large US libraries. By the early 1990s, the card catalog had been retired in virtually all public and academic libraries. Patrons could now search by title, author, subject, keyword, and ISBN from a terminal; librarians were no longer the sole interface to the collection. Reference work shifted from pointing patrons toward the card catalog to teaching them to search the OPAC effectively and navigate its limitations (keyword truncation, Boolean operators, subject-heading structures that didn't match natural language).

  Effect on the work

  The transition from card catalog to OPAC is estimated to have eliminated most card-filing positions (often filled by paraprofessionals) while increasing the complexity of systems and cataloging work for professional librarians. Net effect on professional librarian headcount was approximately neutral; the mix of tasks shifted decisively toward technology-mediated work.

• CD-ROM databases1985–1998

  CD-ROM databases (Encyclopedia Britannica, ProQuest, Wilson, DIALOG on Disc)

  From the mid-1980s through the late 1990s, commercial database vendors distributed full-text and index databases on CD-ROM — DIALOG on Disc, ProQuest Periodical Index, Wilson Indexes, and the Grolier/Britannica encyclopedias. Libraries installed dedicated CD-ROM workstations in reference areas; librarians trained patrons on the search syntax for each disc. CD-ROM expanded the scope of reference well beyond what print indexes could cover, but access was physical and shared — patrons had to come to the library and queue for the workstation. The librarian's role as database-search expert became a formal skill.

• Web + search engines1995–2005

  Web + search engines (Netscape 1994, Google 1998) — disintermediation of reference

  The public web and especially Google's launch in 1998 fundamentally shifted the information landscape that librarians operated in. For the first time, patrons could search for and often retrieve information independently, without a library intermediary. "Just Google it" became the cultural default for quick reference queries. Librarian reference desk traffic declined sharply through the 2000s: ARL statistics show reference transactions at research libraries fell from 11 million per year in the mid-1990s to 5 million by 2008. The professional response was a pivot toward "information literacy" instruction — teaching patrons to evaluate the quality of what they found online, not just find it.

  Effect on the work

  ARL member libraries reported a 57% decline in reference desk transactions between 1991 and 2011 (from 9.5 million to 4.1 million annually). Basic reference queries — fact-checking, quick lookups, directional questions — migrated to Google; complex, specialized, and research-depth queries remained the librarian's domain.

• OverDrive + e-books2000–2015

  OverDrive + e-books (OverDrive founded 1986; e-lending at scale 2000s onward)

  OverDrive (founded 1986 in Cleveland) began licensing e-books and audiobooks to public libraries in the early 2000s; the service expanded to hundreds of thousands of titles through agreements with major publishers. Libraries could now circulate digital titles without physical copies — but the lending model (publisher-imposed one-copy-one-user restrictions, license expiration after 26 circulations) created entirely new workflows for collection development, license negotiation, and patron communication. By 2010, OverDrive served 11,000+ public and academic libraries; the Libby app (launched 2017) made mobile e-lending mainstream.

• Discovery layers2009–2022

  Discovery layers — Summon (2009), EBSCO Discovery, Ex Libris Primo

  ProQuest's Summon, launched in 2009, was the first "web-scale discovery" layer — a Google-like single-search box that indexed a library's entire licensed corpus (catalog + databases + open-access repositories) and returned relevance-ranked results. EBSCO Discovery Service and Ex Libris Primo followed. Discovery layers supplanted the OPAC as the patron-facing search interface and required systems librarians to configure and maintain complex metadata harvesting pipelines. Reference instruction pivoted again — from "how to search OPAC + individual databases" to "how to evaluate what the discovery layer surfaces and when to go deeper."

• Generative AI2022–present

  Generative AI — ChatGPT, AI research tools (Elicit, Consensus, Scite), OCLC AI cataloging

  ChatGPT's public launch on November 30, 2022 triggered a wave of patron and student questions that librarians had never faced: "Is this AI-generated citation real?" "How do I tell if this paper summary is accurate?" "Can I use ChatGPT for my research?" Within 18 months, generative AI reshaped every pillar of library work: reference (AI tools like Elicit and Consensus handle first-pass literature synthesis), cataloging (OCLC's AI assistance auto-suggests MARC fields, cutting per-record time 40-60%), collection development (demand-driven acquisition tools rank titles algorithmically), and readers' advisory (NoveList's AI recommendation layer). The ALA issued its first AI Policy Statement in 2024. Academic libraries began formally adding "AI literacy instruction" to their strategic plans — a new professional mandate with no historical precedent.

  Effect on the work

  Routine reference triage and standard-title cataloging are the highest-displacement activities; specialized research support, community programming, and AI literacy instruction are gaining mandate. The ALA AI Policy Working Group (established 2024) is developing formal institutional guidance on AI use in libraries.

What's changing in your day

Three parts of your work where AI is already doing real lifting, and what stays yours.