Giving up on discovery

A recent day-long taskforce exercise that I attended that was set up by the Ontario Council of University Libraries (OCUL) explored the collaborative potential that a consortial URM purchase could unleash. In a general sense, this concerns the intersection of collections and technology. Thankfully, we’re taking a “what could be possible” rather than a “which tool is best” approach.

At any rate, it was a well structured and vibrant day, with a lot of provocative yet grounded thinking. As an AUL with a large IT portfolio, what we are going to do next with respect to the ILS/SFX/ERM/… technology matrix is more or less constantly on my mind. Moreover, lately I’ve been veering ever more strongly toward questioning some of the core functions and services we provide in libraries, and am finding myself more ready to dispense with some sacred cows.

This taskforce meeting raised two issues that I’d like to tease out here and then suggest some fairly radical conclusions. One of the key themes of the day was discovery, i.e.- what role do we play in helping our users locate information? We all know from myriad studies that a vanishingly small percentage of users prefers our interfaces to those they find on the open Web. We’ve wrung our hands about this for years, and the emergence of federated searching part deux, aka the discovery layer, seemed to promise salvation. One box! All the content! Problem is, it hasn’t quite worked out that way.

Taking a step back from the discovery layer, one can see that it is really a response to years of decline with regard to the library catalog as a place to start research. We tried sexifying the catalog with various features over the previous decade: RSS feeds, call number texting, LibraryThing recommendations, etc. None of those could really halt the decline. So now we’ve dropped the discovery layer on top of that, and while behind that interface now sits an infinitely larger pool of content, it would seem that most users still see it for what is is, i.e.- a librarianly window into a world that we define. And so they have not returned in droves from Google and Co.; the discovery layer solves a ‘problem’ that we’ve defined, not our users. Given the level of financial and organizational investment in these tools, that’s a less than ideal outcome.

Now we, and presumably others, are talking about creating what are essentially consortial-level discovery layers. Surely, the reasoning goes, if we all join forces and collections, then we’ll hit that critical mass that will bring users back to us. Why bother, really? Google won the search discovery wars years ago, and nothing we can do is going to change that.

My radical suggestion here is that we–finally–acknowledge that our inventory control systems are of little interest to our users, whether an OPAC or a discovery layer. As such, we should invest as little as possible in the public interface while making it as useful as possible for the few who do find it critical to their needs. At McMaster we use VuFind on top of our catalog, and it’s an attractive and useful interface that no one seems to dislike. Done. Next.

So what could one do with time saved not implementing and maintaining a discovery layer? There are a couple of ideas here that I took away from the taskforce meeting, both of which come from the oft-repeated notion–for example, Robert Darnton’s remarks at the 15:00 mark of this interview–that one of the most valuable assets we have in the library moving forward is our store of unique content, i.e.- our special collections. Many libraries have done a fair bit of work to digitize portions of their special collections, but collectively we are decades away from being caught up with that job. Along the way, however, we have neglected their discoverability, perhaps ironically because we’ve been too invested in creating discovery layers that we thought users would use as gateways to this content. Recent work done by Kenning Arlitsch and Patrick O’Brien has laid out in detail how poorly our digital collections are indexed. So while they may be present in our local tools, they’re not visible via Google and other search engines, rendering them essentially invisible. At the very least, we should be looking at Arlitsch and O’Brien’s research and making sure that we do everything we can to surface our content in the tools people actually demonstrably use.

To that end, some brave soul mentioned that we should be doing more with linked data. I’ll be the first to admit that while I understand linked data conceptually, I do not actually know how to realize a statement like that operationally or organizationally. That said, at last year’s Access conference, I saw a talk that began to address that issue (text; video), demonstrating a tool that actually enables staff–who may not be able to rattle off a tidy definition of linked data–to create useful linked data and publish it.

One could suggest at this point, and correctly, that this is advocating for us to invest in an unproven technology and hope it pays off in the future when and if the major players on the Web create tools that leverage it. I would argue we need to take such risks. We know what’s not going to be a major driver of our future: library-developed search interfaces, print collections, etc. In fact, our whole gatekeeper role between users and licensed content is eroding as scientific literature becomes more generally available as open-access mandates mature and generate impact and as commercial information providers bypass us to market directly to users (Oyster is just the latest), a far more lucrative market for their wares.

Do we spend too much effort on our interfaces? Are we doing enough to develop the future?

Dale Askey is the Associate University Librarian for Library & Learning Technologies at McMaster University in Hamilton, Ontario, as well as the Administrative Director of the Lewis & Ruth Sherman Centre for Digital Scholarship. He also blogs about libraries, technology, and media at

Posted in Collections, Technology Tagged with: ,
4 comments on “Giving up on discovery
  1. Peter Murray says:

    I’ve looked at this from another perspective and came to the same conclusion. In an article called Riding the Waves of Content and Change I described three phases of content curation: that which is commecially published physical materials, then commercially published digital materials, then local digital materials. It is that third wave of content — the items that are unique to each library — that is an important part of the value proposition for libraries going forward. I also recommend looking back at David W. Lewis’ “A Strategy for Academic Libraries in the First Quarter of the 21st Century” (College & Research Libraries, 68(5), p. 425 — also available from the IUPUI Digital Archive). He talks about strategies for shifting the focus of an academic library from the commercial to the local.

  2. Anja Smit says:

    In Utrecht University Library in The Netherlands, we have decided NOT to implement a discovery system at all. We think it is the wrong investment if you look at the added value for our users. Sep 1 2013, we turned off our home grown search engine for scientific articles and changed our website to advise our users on how to find materials on the web. This transition went very well. GoogleScholar is a good solution it seems for main stream scientific articles. In time, we plan to also retire our OPAC. Again, without offering a local discovery system. Our ambition is be able to use WorldCat (not local!) as a discovery system for our collection.
    So, our focus is on investing in DELIVERY, rather than on providing discovery systems ourselves.
    Anja Smit
    University Librarian Utrecht University Library

  3. Todd Grappone says:

    Interesting post Dale. I would generally agree with your points, OPACs have a pretty low return on the amount of funding we put into them. Conversely, I’ve always been bewildered by how little libraries put into search as a concept. Who is in charge of search for your library? We focus on minimal MARC records in a world where the semantic web is creating very rich search results. Linked data is not the answer by itself, you need something rich to link to.

  4. Robert Hilliker says:

    Having recently served as an institutional repository manager, I spent a lot of time working on optimizing our open source discovery layer (we used Blacklight, not VuFind): primarily though, that work was aimed as passing our metadata through to Google and Google Scholar rather than trying to customize the look-and-feel of the search experience itself.

    Frankly, the decision was a no-brainer. Roughly two-thirds of our traffic came directly from a Google Search, and, after adding the appropriate meta-headers, Google Scholar became one of our main referrers. In fact, if you include the fact that our institution’s own search box was powered by Google, then you could argue that we owed 80% of our traffic to them.

    My colleagues and I wrote a paper that summarizes some of this work which just came out in the special issue of Journal of Library Metadata: If you don’t have access and don’t feel like paying $37 for it, tweet me: @infomnivore.

4 Pings/Trackbacks for "Giving up on discovery"
  1. […] skrev Rurik Greenall et lite innlegg på UB’s intranett hvor han anbefalte artikkelen Giving up on discovery av Dale Askey fra McMaster University, som bl.a. spør hva vi kan bruke ressursene til i stedet […]

  2. […] the items that have been written in the past few weeks about library discovery by Lorcan Dempsey, Dale Askey, Aaron Tay, and Carl Grant, among others. Library discovery, of course, is the capability to search […]

  3. […] through blog posts and twitter discussions – silencing, diversity within the profession, and our investments in discovery systems, come quickly to mind. While this growing online presence is certainly important for building an […]

  4. […] encompassing many different types of information.  Some, like Dale Askey, say that “Google won the discovery wars years ago“, and conclude that it’s not worth spending a lot of resources on library-specific […]

About “Gentle Disturbances”

The title of our new blog, “Gentle Disturbances”, is a tribute and a reference to the husband and wife artists Christo and Jeanne-Claude. Christo and Jeanne-Claude’s art consists of vast, temporary outdoor installations, such as the wrapping of the Reichstag in Berlin, the 24-mile Running Fence in Sonoma and Marin counties in California, and The Gates in New York City's Central Park. Christo has asserted that their art creates “gentle disturbances," designed to challenge traditional perceptions of the spaces and landscapes they inhabit. By encouraging viewers to see familiar landscapes in new ways, their art disrupts assumptions about permanence, ownership, and categorization.

While we claim none of the artistic or political impact of Christo and Jeanne-Claude, we hope that this blog and its many contributors will challenge us all to look at the landscape of academic libraries and higher education in new ways. We aspire to “gentle disturbances” of the kind that will lead to productive conversations and creative approaches to our common challenges.