Magazine article American Libraries

Google Print vs. Onsite Collections: Don't Send Your Paper Copies off to Remote Storage Just Yet

Magazine article American Libraries

Google Print vs. Onsite Collections: Don't Send Your Paper Copies off to Remote Storage Just Yet

Article excerpt

When the topic of Google Print came up at a recent meeting, one librarian made an interesting comment. His supervisor, he said, looked forward to having 15 million electronic books so he could send to remote storage every paper copy with an online equivalent. That struck me as unwise, and it ties in with another proposed scheme advocated by some prominent members of our profession--that libraries ought to use shared warehouses, not merely to house little-used older materials, but to avoid "costly duplication" of current purchases.

In this vision, the primary function of a library is to offer access to the internet and commercial databases and, beyond that, to serve mainly as a community social center with meeting rooms, shared study facilities, and coffee bars. But let's consider the virtues of focused browsing in onsite collections, as well as the adequacy of Google's keyword access--either in Google Print itself or in plans to make library catalogs "more like Google."

Researchers in this country have long been accustomed to having direct access to large book collections shelved in subject-classified stacks. What benefits arise from this configuration, as opposed to having digital texts at hand? What difference is there to scholarship in the way books are stored and presented?

Let me give an example. A graduate student writing on Paul Valery had a problem with unverified information regarding the French Symbolist poet's connection to the famous Dreyfus affair of the 1890s. The student had hearsay information from Valery's relatives that he had signed some kind of petition on the issue, but no specifics. The ARTFL database of digitized French texts did not help; neither did biographies, several subscription databases, or two massive bibliographies on Valery. Finally, I had to go into the Library of Congress stacks, where there are 186 volumes on six shelves in the classes DC354-354.9 ("Dreyfus affair").

I was looking for one book of primary sources that I located in the online catalog with the subject heading "Dreyfus, Alfred, 1859-1935--Trials, litigation, etc.--Sources"; however, it did not include the Valery petition. But on the shelf above it I noticed another book (Patrice Boussel, L'affaire Dreyfus et la presse, 1960) which, it turned out, did indeed have the necessary information. As an extra serendipitous bonus, the same source contained additional information about one of Valery's close friends--information that solved another problem the researcher hadn't specifically asked about.

Could she have found this source if the same 186 books had been shelved in separate tubs in a remote warehouse? Could she have found it if all of the texts were digitized and fully keyword searchable? The answer, in either case, is "no." This may seem counterintuitive to information theorists, but I suspect working scholars will immediately know the reason: Classified shelving enables researchers to recognize sources whose keywords they could never specify in advance.

If the books were stored offsite, they could not have been retrieved as a group. Even if the library were willing to deliver all 186 volumes, the researcher would have had to identify, and fill out a separate request for, each volume individually and then wait weeks for all to be retrieved--at the same time not having any clue in advance which one, if any, might repay such extraordinary effort. That degree of hassle puts such persistence beyond the realm of any behavior that can reasonably be expected of any researchers, even that of senior scholars.

Keyword quandary

By contrast, when the same books on a common subject are shelved immediately adjacent to each other, arranged for in-depth inspection, a researcher can rapidly recognize valuable information, even when it's found on a single page. The difference is not simply that of timeliness of access; it's more a matter of recognition versus prior specification. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.