acm - an acm publication

Need 'Therapy' for your 'information pain'?

Ubiquity, Volume 2001 Issue March, March 1 - March 31, 2001 | BY John Gehl , Suzanne Douglas , Peter J. Denning , Robin Perry 


Full citation in the ACM Digital Library

Louis Rosenfeld is an authority on information architecture, and is co-author (with Peter Morville) of Information Architecture for the World Wide Web (O'Reilly), which established itself as a best-seller among Internet books and will soon be republished in a revised and expanded edition.

UBIQUITY: Let's start out by getting our bearings. What exactly is "information architecture"? What relation does it bear to information science? And to Web design and management?

LOUIS ROSENFELD: Information architecture the thing is the structure of an information system and the ways it's organized and labeled. Every information system, be it book or Web site, has an architecture. For a book, it's the stuff we all know and count on-chapter and section headings, tables of contents, back of the book indices, sequential pagination, consistent information on the cover and spine, and so on. But for a Web site, your guess is as good as mine-the architecture might consist of many different kinds of hierarchies for navigating, sets of terms used for labeling content, search interfaces, site maps, indices, data models . . . . Unlike books, sites have no conventional information architecture, but that's another story.

UBIQUITY: What about information architecture as a field of study?

ROSENFELD: Information architecture the field is an applied field, just as business was a century ago. Information architecture draws from any established discipline that concerns itself with information as valuable stuff. Information science, for example, contributes many guiding principles of information organization, but so do journalism and data modeling. Human factors techniques help us get inside system users' heads and understand what kinds of information they seek, but so can ethnography and librarianship. Software design, technical communications, process engineering, graphic design, and other disciplines all offer principles or techniques that can help us design better information architectures.

UBIQUITY: Why do we need information architecture to be its own field?

ROSENFELD: Because none of the disciplines mentioned above can, on its own, offer sufficient expertise to address the complex design challenges presented by new media. There are too many questions to answer about users, content and context in a world of cheap, easy-to-use information technologies that allow anyone to publish anything globally. To answer those questions, we need a multi-disciplinary melting pot to bring together all the appropriate principles and techniques from established disciplines and put them together into a meaningful whole.

UBIQUITY: How do you rate the skills of multi-disciplinary chefs and the stews they're serving up?

ROSENFELD: The role of Webmaster has been incredibly important for the commercial Web's development. Webmasters took responsibility for making a suite of new and unknown technologies work in practical ways for their employers. In doing so, Webmasters provided their organizations' Web education. But at this point, I rate Webmasters "M" for moot. Most sites are too large and complex for a generalist to manage. There are too many skills, as I've mentioned, for one person to provide in any reasonable fashion. That's why you don't see many people going by that label any longer. Webmasters have been replaced by hordes of specialists, such as graphic designers, information architects, programmers and editors, not too mention project managers who tie their teams' efforts together.

UBIQUITY: What is the percentage of Web sites that are (to use a high-tech term) awesome? And what is the percentage of ones that are just awful? Do people other than specialists know the difference?

ROSENFELD: Like "Webmaster," I think "awesome" and "awful" are somewhat dated terms. Let's face it, the best sites out there today are far better than the cream of 1995's crop, or perhaps even last year's. Many of us just don't get too excited by the Web anymore. Like the telephone, it's become ubiquitous and ho-hum. Can't live without it, for sure; but "usable" and "profitable" have become much more important terms for describing sites than "awesome" and "cool." I think that's true of most Web users, not just jaded old farts like me.

UBIQUITY: Aren't these odd comments coming from a person who is a principal of a consulting firm that specializes in information architecture design? You don't want to be cool anymore? What's the world coming to? Doesn't cool mean money? Does Peter Morville, your partner, know you're saying these heresies?

ROSENFELD: Look, my background is in librarianship -- do you think I ever was cool in the first place? I'm not sure I'd know what cool was even if it fell on my head like a five-volume stack of Library of Congress Subject Headings. Our clients don't care about cool; they are in business to make money. Most have learned the hard way that their Web sites can't serve as a form of creative self-expression for a Webmaster or a VP. Access to appropriate information is key to all their organizational functions, from a sales rep providing her leads with the right product information to an employee finding how to reconfigure his 401K investment portfolio. Heresy? What could be more strategic than the unsexy task of matching users with information? Besides, it's hard for something that's mostly invisible to be cool. An architecture is a form of infrastructure that, if successful, users won't notice it anyway.

UBIQUITY: Somebody -- not end-users, maybe, but somebody -- gets chills thinking about infrastructure. What does an information architect looking at infrastructure get enthused about or perturbed about? Is it not possible to say, just looking at sites X and Y, that site X is much better (not "cooler" but better) than Site Y?

ROSENFELD: It sounds like you're bringing up the issue of metrics. "Why is this architecture better than that one?" First, anyone interested in this intriguing topic should attend a panel that one of my Argus colleagues, Keith Instone, and I are co-moderating for CHI 2001 this April in Seattle. It's called "Measuring Information Architecture Quality: Prove It (or Not)!" ( l-category .html#ia). We've convinced an excellent group of panelists to argue different positions on this very issue. As a moderator, I'll leave my biases at the door. But I'll tip my hand here and tell you that I think it's poppycock and moot to boot.

First, give us the poppycock angle.

You can test aspects or components of an information architecture. But you can't quantitatively evaluate the whole beast, save for some extremely narrow and focused situations. For example, you can test the relative speeds of both Land's End's and REI's search engines by looking up the same blue parka. But you can't measure the performance of the architecture itself. It's made of too many individual components that make up each site's architecture (including various browsable taxonomies, search functions, labeling schemes, navigational approaches and interface widgets). Users interact with many or all of these components as they look for information in a site, and it's impossible to ascertain their collective performance.

UBIQUITY: What are some of the other issues involved in evaluating a Web site's infrastructure?

ROSENFELD: Many times user satisfaction isn't very easy to delineate. If it's not clear that users' needs have been satisfied, how do you know when their tasks are complete and therefore subject to measuring? My example above is a fairly straight-forward "known-item search": The user wants a blue parka, the user knows that such a parka will be available from the site, and when the user finds the parka, her task is complete. You can time it, measure clicks, track eye movements, whatever. But when the user doesn't exactly know what he wants, it's a different situation. For example, a user is interested in saving money for retirement. He goes to a number of places to explore and discovers that an IRA may or may not be what he's looking for. He learns something about the difference between standard and Roth IRAs, but still isn't sure of what's right for him. He also learns some more about asset mixes, stumbles on a couple of candidate investments, and maybe, just maybe, finishes the experience by actually investing in an IRA plan with a specific investment configuration. Or not. What's important here is that the user didn't exactly know what he wanted when he started. He learned along the way, picking up both new concepts and words to describe them. He also may have learned a lot about what he didn't know. All of these outcomes are positive, but how do you measure the outcome of this experience quantitatively?

UBIQUITY: But you said that this was all moot?

ROSENFELD: Well, sort of. While information architecture remains a new and strange concept to business people, they will constantly want to know the ROI proposition. IA practitioners will constantly answer with fuzzy math. For instance, we're all always citing Jakob Nielsen's findings from his work on Sun's intranet. "If we save every employee three minutes per day, multiply that by 200 days per year, and by what each of those minutes, the savings will be $X,000,000 per year". Yeah, sure, but who's to say that those extra minutes won't be spent playing Tetris? Over time the people with money are becoming comfortable with the more abstract areas of design (like information architecture). They don't need to be sold on the ROI so much any more because they realize that an improved IA is good business. The analogy is therapy: Along with millions of hours, Americans spend billions of dollars on the couch. And no one asks what the ROI is. There's another aspect to that analogy worth noting: Successful therapists, like successful information architects, draw on both science and art. The artsy stuff is, of course, impossible to quantify, but it's real and it's there. I'm increasingly convinced that we should rename IA "information therapy," as that's what we do. We get companies on the "couch" and help them articulate their information "pain".

UBIQUITY: Give some examples of the kind of information pains companies would tell you about. Or would they be too embarrassed?

ROSENFELD: They're always embarrassed until we tell them that we've already talked with a dozen other Fortune 500s with the same problem. For example, I can't tell you how many times we've encountered companies that don't have an authoritative list of their own products' names. Consider the negative implications for a customer or sales rep who is trying to find the product he's looking for on a company's site. It's really very frightening. How can we bet our futures on e-commerce when we can't even come up with a consistent way to label our products? Information architects help by coming up with authoritative vocabularies that describe a company's products, areas of research, benefits offerings -- whatever vocabulary that's needed that is important for users to find quickly and effectively.

UBIQUITY: What about an organization's internal pains, office politics, and such? Do these show up on your "couch"?

ROSENFELD: An example is the corporate intranet that's organized into "silos" of content that correspond to individual business units. Employees don't exactly have a map of the corporate "org chart" in their heads as they try to navigate such a mess. In any case, they often want information that bridges multiple business units. Many of our projects have centered on coming up with ways to cut across the grain of these content silos that make sense to employees. There are many other examples, too many to cover here. Suffice to say that many of them, including the two mentioned here, are often caused by political and cultural tensions within the enterprise. It's no secret that most organizations' individual business units don't play well together. Hand them something that's very inexpensive and easy to set up, namely the Web's suite of technologies, and suddenly these political and cultural problems are ratcheted up a notch or two. You can actually see an organization's frictions by examining its sites' information architectures. Even though these sites are ultimately intended to serve users, they're the ones who suffer the most from intrigue and infighting.

UBIQUITY: The situation seems bleak. Beyond hope. Impossible -- or is that too strong a word? Is "information therapy" able to cope with a patient organization that's in the suicide stage?

ROSENFELD: It's bleak only if you're stuck in "Internet time" or some other bizarrely speeded up view of the universe. We're all in such a hurry -- let's slow down a bit and remember just how long it can take for major changes to happen. Think about it: Democracy has been around on and off since ancient Athens and yet it's just making its first appearance in many of the world's nations. The impact of the 60's civil rights movement is still rippling through our society. And many corporations just clued into networked computing five or ten years ago. These organizations are dealing with a very new suite of technologies that, by its decentralizing nature, roils up all sorts of issues related corporate to governance, communications and culture. But it gets worse: these organizations' decision-makers then ask their various business units to work together to reign in this technology and subsequent content explosion. That's hard; often these disparate units have never before worked together successfully. These units are then asked to create a single, unified interface and information architecture for their enterprise-wide content. That's really hard; issues of interface and information architecture design are very new to most people. The kicker: all of this is supposed to happen in a three-to-six month time frame. Because when it involves the Internet, everything has to be fast. Right? Well, of course not. Three-to-six years is much more realistic. And most folks in the trenches know this. But corporate decision-makers don't, and until they do, it's an uphill battle.

UBIQUITY: Are they making progress?

ROSENFELD: We're seeing ever brighter rays of hope among many of the companies with whom we talk. But the people most directly affected by information pain must realize that it will take years, if not decades, for many of these problems to abate. They're simply too complex and systemic to change in the course of a year.


Leave this field empty