When I spoke at the Enterprise Search Summit on May 23, I was privileged to be able to hear Peter Morville, the father of Information Architecture, deliver the keynote address. Peter, the author of Ambient Findability
Peter discussed how in the mid-90s he wanted to take his background in library and information sciences and apply it to online media—that interest fostered his explorations into what we now call Information Architecture. Peter found the most interesting part of IA was moving beyond librarianship into ethnography and human-computer interaction, and has recently been feeling constrained by IA. So Peter started looking at Findability as a key part of the user experience that crosses disciplines.
Peter talked about how the “the polar bear book”
Peter criticized some IA’s approach of creating a single taxonomy, “People get stuck in the Yahoo model of creating a single taxonomy that will solve all problems.” But he also minced no words about the cult of usability, saying, “I hate the word usability—its meaning has grown and grown until it is almost synonymous with ‘quality’—we want a better Web site.” Peter decided to tease apart the characteristics that a site really needs that go beyond usability: useful, desirable, valuable, accessible, credible, and findable.
To Peter, findability poses several questions to Web site designers: Can users find our Web site? Can they find their way around our Web site? Can they find what we have despite our Web site’s navigation (so can they search and drop into a part of the site and get oriented quickly)?
But every quality that goes into usability triggers questions—that’s why it’s useful to break these ideas apart. We can do a separate credibility audit, for example. We know that your site needs to appear in the top of the Google results to get hits to your site, but users also trust the companies behind the top results more. In that way, findability and credibility are linked. The other qualities have relationships as well.
Peter related a case study for the National Cancer Institute: Most people coming to the site are cancer sufferers or families and friends of sufferers. When they first come to the site, they navigate to their diagnoses. Peter asked the site owners “How do people find your site?” and “I was told not to worry about that because they rank #1 or #2 for ‘cancer’ but a quick keyword check showed that many other ‘cancer’ words are searched far more.” Peter told them, “Your mission is not to design a good web site, but to connect cancer victims with resources that help them.”
It gladdened me to hear Peter decry the old thinking of marketers fighting even harder for people’s attention—with gasoline pumps to talk to us while we fill our cars. “In a world where people have so little attention to give, we must help people find what they want when they want it—when they are interested. We must shift from push to pull so people can pull things when we want it.” I couldn’t agree more.
Peter predicted in the 1990s that the Internet would turn everyone into librarians and now says, “Metadata used to be trapped on little cards in the card catalog and now it has become sexy, if you look at Flickr and del.icio.us.” But he parts ways with folks such as David Weinbeger, who compared taxonomies and folksonomies, saying the old way creates a tree while the new way rakes leaves together. Peter called this “the perfect metaphor—those leaves rot and return to the earth and become food for trees that live for a very long time.”
Peter argues, and I agree, that we don’t have to choose between folksonomies and taxonomies. He mentioned
Stewart Brand’s concept of Pace Layering, where layers are able to evolve and change and different rates and paces. Taxonomies evolve slowly and provide stability while folksonomies adapt very quickly. “Our challenge is taking the lessons from folksonomies and driving them into deeper layers,” Peter said.
Peter had particularly urgent words for those of us responsible for enterprise search facilities, calling them “challenging” but “terrible” in their current state. “We need to start with an analysis of our user’s queries—do they value precision or recall? We need to do more work on the interface so people know what they can search and how they can search.” Peter identified content improvement as the biggest opportunity (which longtime readers know I have been preaching): “The easiest way to improve an information retrieval system is to make the content smaller—we can give our users confidence by letting them shrink the space (based on metadata queries). If I know the year something was published, let me put in the year to shrink my navigation space.” Peter expects more of this kind of search interface in the future: “Ten years from now, I think we will still begin our search process by typing in two or three keywords and pressing a button—I don’t think we’ll be using natural language. Does that mean we don’t need to worry about UE or IA? No. That simple search is the same, but it is only a start. Where does it lead us? To a page at Amazon? Look how complex and dense that site is.”
Peter lauded Endeca for its faceted search capability, which provides the user the “illusion of getting started quickly,” comparing that style with drivers who get in their car and start driving, only consulting a map when lost, just as we enter a few search words and then navigate from the search results page. Peter compared Endeca’s approach in enterprise search with Google’s approach to Internet search, where searching for “IBM” yields a stock price chart (structured information) as well as Web pages about IBM (unstructured information). Endeca helps blur the line between structured and unstructured information within the enterprise.
Peter’s book, Ambient Findability challenges all of us in Web site search to concentrate on designing our search facilities to locate what searchers are looking for. Whether it is improved content, new technology, or a better user interface, increasing user expectations means that none of us can afford to sit still without taking up that challenge.
