Planning for Serendipity
Posted by Dan Cohen in February 7, 2014.
[Another post in our behind-the-scenes series.]
One of the great aspects of bookstores and libraries is their uncanny ability to produce serendipity, as people bump unexpectedly into works as they walk through aisles and scan shelves. Everyone seems to have a story of a serendipitous find. I know a professor who was hit on the head by a book falling off a shelf as he reached for a different one; that book ended up being a key part of his future work.
Unfortunately—er, fortunately—we cannot engineer for book-to-head injuries online. But we can try to create environments in which serendipity is able to flourish. And truth be told, bookstores and libraries have their own forms of “serendipity engineering,” from storefront staff picks to behind-the-scenes cataloguing and shelving methods that make for happy accidents.
At DPLA, we’ve been thinking a lot about what’s involved with serendipitous discovery. Since we started from scratch and didn’t need to create a standard online library catalog experience, we were free to experiment and provide novel ways into our collection of over five million items. How to arrange a collection of that scale so that different users can bump into items of unexpected interest to them?
There have been a few pieces written about this issue of library serendipity, but much more could be said and explored. It’s a tough, important question not only for digital libraries but for many other websites with expansive content. When we talk about serendipity, we’re not talking about “more of the same,” as seen (or heard) on services like Netflix or Pandora, which simply provide additional items that are similar to what you already know or like, or items recommended by friends with similar interests. Search engines and library OPACs (online public access catalogs) tend to have text-heavy lists of items with few surprises. Google’s experimental Chrome Bookshelf reveals their awareness that such indices may not be ideal for discovery for items like books.
Although this is by no means a comprehensive list, some ingredients seem critical for serendipity online:
Generous interfaces that take advantage of proximity. It may sound too obvious to state, but humans notice things that are near other things. However, few web interfaces take advantage of this fact, and indeed the penchant for white space militates against effective nearby discoveries. I’ve written elsewhere about generous interfaces, or interfaces that are strongly visual and dense, such as large sets of thumbnails; these can encourage discovery in ways that text lists cannot. DPLA’s map mode or bookshelf view (which has thumbnails of images to the right) makes some headway into this territory.
Quick access and assessment. Being proximate isn’t the whole story. People have to be able to very quickly assess their interest in an item, and that requires full and instantaneous access. The ability to take a book off the shelf and scan the table of contents, for instance, is essential—the spine isn’t enough. On the web, a quick click to view is important, without any transaction costs (registering, logging in, etc.). In this way, open access is a key part of serendipity.
Good data and invisible connections. This element is probably the most critical, and yet the most hidden. Smart cataloguing systems in libraries, archives, and museums has always set the stage, albeit in quiet ways, for serendipity. At DPLA, we feel that one of our key services is the normalization and enhancement of data brought together from more than a thousand institutions. The commingling of that data alone provides some unexpected pairings, but adding geocoding, linked data, and the like will undoubtedly payoff in the future, even if a researcher is unaware of why certain items are placed nearby in an interface.
Of course, we can’t plan completely for serendipity. And that’s where having an application programming interface (API) and open data is extremely helpful—it encourages others to experiment in ways that create new interfaces, views, and tools. Each of those will, at some point and in their own way, connect someone with what they didn’t know they were looking for.
All written content on this blog is made available under a Creative Commons Attribution 4.0 International License. All images found on this blog are available under the specific license(s) attributed to them, unless otherwise noted.