OCLC, the Dublin, Ohio-based cataloging source and bibliographic utility for thousands of libraries worldwide, has apparently become one of the many streams feeding into Amazon's book title database. Other sources—such as book vendors like Baker & Taylor, direct feeds from publishers, and ONIX and other metadata standards—may be better known, but the OCLC content provides a depth that was missing from what is becoming the world's premier publicly available book database. When comparing records in the Amazon and OCLC WorldCat databases in narrow subject areas, it appears the records OCLC provided to Amazon cover older book titles (19th and early-to-mid 20th century English-language titles were readily identifiable). The discovery of relatively unique items held by only a few libraries worldwide triggered the research reported here.
The book search feature of A9.com (Amazon's search engine; see http://newsbreaks.infotoday.com/wndreader.asp?ArticleID=16370) uses Amazon as a source. The "book" search guides you to the content in Amazon through current imprints and includes "search inside the book" depth as well as the older bibliographic data. Sometimes it even offers the opportunity to purchase older items. It's a real plus to find original source material listed alongside current works as a result of searching a specific phrase or subject.
If catering to third-party sellers lies behind Amazon's decision to import bibliographic data, subject headings, and imprint information from OCLC, the addition of older OCLC records makes selling simpler for out-of-print (OOP) material. Amazon touts its ease-of-use for sellers—I've noticed some claims that you can set up a title to sell in 60 seconds.
The Discovery Process
Titles of scarce items in 19th and early 20th century religious history were searched in Amazon. The Amazon records of items that the searcher knew were held by only one or two libraries worldwide were compared to the OCLC WorldCat records for the same titles. Because people manually input data in the original records, the typos that appeared in WorldCat carried over into Amazon. The 1874 pamphlet identified in standard bibliographic sources as Restoration of the Gentile Apostleship has various typos. One library that cataloged the item apparently dropped the "i" in "gentile," resulting in a kinder, gentler apostleship. Another library incorrectly keyed in "Gorden Square" (instead of "Gordon Square") in a record for a different pamphlet from the same church. These entries have the same errors in the Amazon and OCLC records. (When asked about this, an Amazon customer representative indicated these records were "seller" input records. However, I suspect that not even OCLC tracks when libraries hold book sales, much less what they plan to sell.)
The records for these items are easily retrievable with a keyword search for the church or for the authors. With the additional material from OCLC, researchers lacking direct access to OCLC's WorldCat may find Amazon a major source of retrospective bibliographic data for many subject areas for English-language books. In an interesting development, OCLC's Open WorldCat offerings through Google, Yahoo! Search, etc., do not offer all the same citations. Apparently, licensing data directly from OCLC guarantees a more complete collection of records than the Open WorldCat offerings from which libraries can opt out. However, the Amazon load does not include a "find in a library" bridge. Nonetheless, searching Amazon's database does not require knowing much more than general keywords to find titles. Serials and multipart volumes seem to be excluded.
Librarians and bibliographic experts will realize that many of these titles are unlikely to ever become available on the open market. John Ulmschneider, university librarian at Virginia Commonwealth University (VCU) and an OCLC Members Council delegate, thought this content in the Amazon database could enhance serendipitous searching. He also mentioned print-on-demand (POD) possibilities. Staffers at the James Branch Cabell library at VCU collaborate with POD specialists at Wildside Press to keep Cabell's works in print in hardcover. They can succeed even when individual titles sell very few copies annually—in part because the cost of re-keying has plummeted.
A search for Mormon material in Amazon brings up unique source items too. Joseph Wright, Mormon immigrant to Utah in 1849: his ancestry and posterity," which was scanned from a manuscript, is held only by the Washington County Public Library in St. George, Utah. (It is also identifiable through the "find in a library" function in Google and Yahoo!.)
Amazon handles subject headings on OCLC records by de-linking dashed entries. For Wilford Woodruff's journal, 1837, a 7-page Mormon pamphlet published in 1919 and held by the State Library of Maine, the OCLC member library contributed copy has these subject headings: Mormon Church - Missions - Maine; Woodruff, Wilford, 1807-1898. Amazon's record only gives the subjects (Maine; Missions; Mormon Church; Woodruff, Wilford); it doesn't include the dates.
The Bottom Line
Amazon might find some advantages in supporting access to OCLC's Open WorldCat project, despite its clear preference that people buy, rather than borrow, books. Already individual libraries can participate in A9.com and can register to make their catalogs available as a search option. There's a somewhat accidental quid pro quo between libraries and Amazon. Many libraries have links from their catalogs to Amazon (to provide additional information or for patrons interested in purchasing books). At the same time, the libraries receive potential readers from people who use Amazon or its A9.com search engine, which completes the circle and—at the same time—enhances Amazon's image as an all-in-one resource.
OCLC member libraries and individual members of the Users Council were contacted for this story. Some were aware of experiments and projects with Amazon; others were not. Attempts to acquire details from OCLC were limited due to a non-disclosure agreement. Attempts to contact Amazon to discuss OCLC records and their inclusion and use were equally unsuccessful.