WorldCat Identities

» see worldcat.org/identities as it lived, on the wayback machine

~ the goodbye ~

Thom Hickey was chief scientist at OCLC, the cooperative that runs the union catalog of nearly every library on earth, and he asked a question that turned out to have a good answer. If you already hold a record of what has been written by a person and what has been written about them, how much of a biography can you assemble without a human writing any of it?

WorldCat Identities went live in 2007 with pages for twenty million identities and grew to around thirty million. Not all of them were people. Organizations had pages, and so did fictional characters, because a catalog does not distinguish between a name that lived and a name that was invented. Each page carried a publication timeline, the works held by the largest number of libraries, and every alternate spelling of the name that had ever appeared on a title page. Underneath that sat the subject headings the person was associated with and links out to their co-authors. Nobody typed any of it.

If you have ever scrolled to the bottom of an English Wikipedia biography and looked at the authority control box, you have seen one of these. That template linked out to WorldCat Identities from a very large number of articles for over a decade, sitting between the German national library and the Library of Congress identifiers. That is how most people met it, without ever knowing the thing had a name.

The data stopped being refreshed in July 2022. A banner went up on the site saying the research project was ending, that the application would be retired in the coming months, and that the most recent version of the data was from that July. The application itself answered its last request in late March 2023.

~ the chronicle ~

~ a biography with nobody's name on it ~

The pages were good in a very particular way and bad in a very particular way. They were good at scale: an obscure nineteenth-century agricultural writer got the same treatment as Tolstoy, because the algorithm did not know which was which and did not care. They were bad at edges. Two people with the same name got welded into one identity. One person who published under a married name and a maiden name got split into two.

Nobody working in libraries mistook it for authoritative. It was labeled a research prototype from the first day and stayed one for sixteen years, which is a long time to run an experiment that half the reference desks in the world had quietly started depending on.

~ the link rot ~

When the pages went, every link into them went with them. Wikipedia's template for these identifiers broke, and editors spent 2023 working out what to do about it, because the successor does not do the same job. WorldCat Entities gives you a persistent identifier for a person, which is what a linked-data project wants. It does not give you the bibliography, which is what the person clicking the link wanted.

Sixteen years of citations, catalog records and reading lists pointed at worldcat.org/identities URLs. Those addresses now resolve to a page explaining that the project has concluded.

~ the last capture ~

OCLC never announced a shutdown date. The banner said "in the coming months" and then the site simply stopped. The Internet Archive has a working page from 26 March 2023 and a 404 from the 27th, and from May 2023 onward the URL is a permanent redirect to a research summary on oclc.org. So the date on this grave is the crawler's date, not anybody's press release, which is a reasonable ending for a service whose entire premise was that machines can write the record when people do not.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »