Recently I went down to B-level in search of four books—titles I had come across in my reading that I was sure were going to explain the mysteries of the universe, or at least, help me answer one part of my research question. I soon found myself carrying ten books up to Circulation. Three of them were from my original list of four—but one of those I rejected as soon as I looked at it; the other seven were books that I hadn’t known of, from the same general call number range. And one of these seems like it’s going to be the best book of all, taking me in an unanticipated but profitable direction.
I’m sure the experience I’m describing is familiar to many “printophiles.” There’s a magical element to the encounter with a printed text on the shelf—something about the clues you pick up from the combination of title, table of contents, index, maybe even the cover design or the size of the print—that tells you if it’s relevant to your interests. This goes for journals as well, both single issues and bound volumes. Even with the wonders of database searches that lead to full-text digital articles or pdfs from Interlibrary Services, I still have the experience on rare occasions of reluctantly trudging down to the shelves to find a particular article in a particular bound journal, only to become excited by an article in a different issue of the journal.
Why does this happen? How is this browsing experience different online? And, does serendipity mean you’re a bad researcher?
Why does it happen? Well, for the simple fact that some clues about relevance are not always visible in the online catalog record and are harder to see in relationship in an e-book or pdf. Some of our catalog records in the Sheridan Libraries are very old. Many of the books in the George Peabody Library, for example, were cataloged in the late nineteenth century using the less detailed Dewey Decimal System. Not only do these catalog records often lack subject descriptors and chapter titles and abstracts—all of which offer language for your keyword searches to land on—but the call number ranges are baggier, making it harder to zone in on an area specific to your question. A more profound problem is that, as knowledge evolves, inherited systems of classification become less relevant. Interdisciplinary content is especially hard to catalog, and since material objects like books and journals and dvds cannot inhabit more than one physical space at a time… well, sometimes catalogers have to make hard calls about what subject to prioritize, which determines where a book lives in the library, which affects what you find next to it.
Of course, library bookshelves are never comprehensive (some books are checked out, some are shelved in a different library, some we don’t have) and catalog records—not to mention database results, internet searches, and the e-texts themselves—often contain other clues about relevance… which brings us to the question about online browsing. Perhaps one answer comes from John Palfrey’s Born Digital: Understanding the First Generation of Digital Natives. Palfrey points out that readers of online news employ a variety of reading techniques that are not identical with print reading techniques, and which seem to rely more on visual scanning. His description definitely applies to my own behaviors in print versus digital reading environments, which transfer to print versus online browsing. (I am not sure if it matters if you’re technically a “digital native” or not—we all spend a lot of time online these days.) More broadly, the usefulness of online discovery varies a lot because there are many varieties of online experience. As everyone knows who has ever actually drilled down to page nine of a list of Google results, there is a difference between browsing and searching. Suffice to say, for now, that serendipity happens, but it can be harder to find that sweet spot of “unexpected but relevant.”
It’s precisely for this reason that the twelve participants in this summer’s One Week One Tool Institute, at the Roy Rosenzweig Center for History in New Media, decided to build the “Serendip-o-matic,” a search tool that finds images and documents from the Digital Public Library of America, Flickr Commons, Europeana, the National Library of Australia and other large collections. Try it out and see what you think.
Another vote in favor of the value of serendipity comes from the Harvard Library Innovation Lab “StackView” interface, a virtual browsing tool that emulates the experience of looking at format-integrated library shelves—but also indicates how popular a particular source is by translating circulation statistics into color.
I think we can safely say that, yes, serendipity plays a legitimate and important role in research; you can stop hiding your search engine results page from your friends and colleagues. But what is that role, exactly? There’s still a lot we don’t know about how to design serendipity, partly because there’s still a lot we don’t know about how people interact with different kinds of research environments. Both of these experiments, perhaps inadvertently, demonstrate the influence of the visual layout of the interface on your browsing outcomes. The “Serendip-o-matic” searches I tried returned a bunch of images representing linked sources, arranged a bit like a Google Image search page of results. “StackView” gives you a color-coded cartoon version of an imaginary library shelf, turned on its side for ease of spine-reading.
Is my aptitude at “reading” a book in the stacks—interpreting its clues—simply a product of my familiarity with the conventions of book design? As I do more and more research in exclusively online environments, will I lose my “bilingual” ability in print and digital forms? Maybe. All I know is, print and/or digital, I've gotta keep on browsin,' baby.