Content Management and Information Architecture
By Tony Byrne at 2003-05-23 00:00:00 |
Lou Rosenfeld is an independent information architecture consultant. He has been instrumental in helping establish the field of information architecture, serving as president of Argus Associates consulting firm from 1994-2001. With Peter Morville, Lou co-authored the best-selling book, Information Architecture for the World Wide Web (O'Reilly), often known as the "Polar Bear" book.
CMSWATCH: How do you define Content Management?
ROSENFELD: I think it's important to note that CM is a field more than just a definition -- an emerging field that's concerned with making information easier to manage and access inside corporate environments. Within a CMS, the information challenge is first and foremost one of management as opposed to making the information findable, which is what we're concerned with in information architecture (IA).
The CMS audience is specialized as well. The audience really are editorial users of Content Management systems or participants in workflows, and that's different than whom we think of as users in the information architecture world. At CMS conferences I have to make sure to keep using the term "end users" when I was talking about content consumers in the IA sense, as opposed to just "users," which CM people often think are content contributors.
CMSWATCH: So, where does IA end and CM begin?
ROSENFELD: I think it's really hard to separate the two because with both types of users -- people working with the content versus people using the content -- you can't really talk about the needs of one without the others' considerations being involved. While both IA and CM specialists typically look at much the same content, we come at it from different directions. In the IA world we've focused largely on the unstructured stuff that typically lives on websites, whereas I think in content management the material has maybe been around a bit longer in other guises. Document management systems for example have predated the advent of the web.
One of the places where we are really similar is that we are taking holistic views of information systems. But I see IA as sort of a snapshot, a spatial representation of an information system. Content management is a more temporal representation of flow, in and out of that snapshot. That's why I see them as two sides of the same coin.
The content management world, however, is different in that it has a suite of technologies that drives the development of the CMS field. In IA, we don't have an industry -- we have a field and I think the field comes first for us. I think the industry maybe comes first for you. That's good because you have a lot more money behind you and a lot more companies that have to justify the field as a whole because they have to come up with justifications for the products.
On the other hand, you have a lot of companies putting tons of money into marketing and often misleading people with those messages. In our world, there's really nothing out there really that would be considered IA software. So, we are kind of technology agnostics, which is really good but we have a certain poverty in the field. We can't get sponsors for our conferences like you folks can. We don't have a bunch of white papers out there from companies or analysts who look at the whole industry and serve as resources to decision-makers.
I think the time has come for a closer discussion and at least mutual awareness between the two fields of IA and CM. We have a lot to offer each other and I'm concerned and a little surprised that more hasn't happened at this point. I'm curious as to why that is. It may be that because of how the fields have developed -- one with a software industry and one not -- that we somehow just missed each other. We end up being in different business units inside the same companies and we end up at different conferences. But it seems like a natural pairing and I hope that people in the content management community will look at us and reach out to us and help us see them.
CMSWATCH: Would you consider autocategorization and autoclassification software as maybe the first automated IA tools out there or do you consider them something different?
ROSENFELD: Very clearly they are part of the suite of tools that Information Architects need to be aware of and conversant with. I wouldn't say though by any means that they are the first. Technologies that support directories, catalogues, and portals have been around a bit longer. Actually I think the most common IA tool is "Search", which has obviously been around for quite a long time.
CMSWATCH: Let's assume I'm going to lead a Web Content Management project at a major corporation and I want to bring in an outside IA consultant. What are the top three things I should look for in that consultant?
ROSENFELD: There are a lot of different types of information architects, with different types of skills.
First, if you're coming at it from a content management angle, you need an information architect who has skills dealing with finely-grained content and can look for patterns to help you determine things like content objects, metadata, and rules for linking objects to support contextual navigation. However, that information architect will do this from a slightly different perspective, a more user-centered perspective. So, whereas perhaps the metadata attributes and values that are important in content management might be more of the administrative variety -- supporting content management -- you need to balance that with what the information architect is going to bring to the table, namely, descriptive metadata that supports searching and browsing by end users.
So, an ideal information architect in your setting is someone who's going to be focused and detail-oriented enough to be able to conceive of the small granular chunks of content and how they fit together, as well as have the metadata skills that can determine the descriptive vocabularies and other means of getting people to those little chunks of content. Implied in what I'm saying -- and I hope this is going to be true of any information architect -- they should have skills in doing research and user testing, because if they are going to bring this user-oriented sensibility to the table, they have to possess some experience with methods to actually come up with the data that's going to inform key content decisions.
CMSWATCH: Should that person have a library science background?
ROSENFELD: That's certainly something that can be helpful and that's my background, but I'm not necessarily someone who thinks that it's only people with a library science background that can really get at descriptive metadata.
It depends more on the person. I'm an LIS-background information architect who doesn't really have those kinds of detail-oriented skills. I'm more of a strategic, big-picture information architect. By contrast, there are people with data modeling skills or technical communication skills that do a really great job on coming up with descriptive metadata without LIS training.
So background disciplines are useful, but they're not enough to go on if you're looking at resumes. You really have to talk to people and get a sense of what kind of work they enjoy and have really done. I've seen some pretty terrible classification work done by librarians and I would probably do no better.
CMSWATCH: What are the key deliverables that I should expect out of that IA consultant before they finish, so I can then continue with the CMS project?
ROSENFELD: Well, one thing is going to be a metadata table that includes descriptive metadata attributes. A metadata table is essentially a list of metadata attributes and values that should go with those attributes.
Those values suggest another deliverable, namely, controlled vocabularies. Controlled vocabularies are used to populate those metadata attributes; for example, a "subject" attribute may have a limited, pre-determined set of possible entries.
An IA person can also be very important contributor to the team working on "Search" and its relationship to the content management system, or they may work with the person who serves as the bridge to the IT people who are working on Search, even if they're not doing that work themselves.
And as I mentioned before, information architects can assist in developing the content model, its content objects, metadata, and linking logic. I'm not sure -- based on my experience with content management people -- that this whole issue of contextual navigation based on rules that link content objects really gets looked at as seriously as it should. The deliverable here is a set of rules. For example, on a music site, one rule might dictate that if someone has retrieved this particular content object -- let's say it's a review of a new compact disc -- and there happens to be a profile of the band, there's probably some shared metadata that can be used to automatically link those two objects, resulting in richer navigation from page to page. It's simply a question of coming up with rules that can then be implemented in an automated way.
And finally, top-down information architecture issues should be considered as well. Obviously the bottom-up work that we've been talking about has to be integrated with the architectural components that help users navigate down from the top levels of the site.
CMSWATCH: If you were going to write the Polar Bear book all over again, what would you include this time that isn't in the current edition?
ROSENFELD: It's funny you should mention that because we just rewrote the Polar Bear book. We doubled it in size and ninety percent of the material is new compared to the first edition, so frankly I want to barf when you bring it up. I can't imagine something I'd want to do less.
One of the problems with doing a third edition and with writing the second edition was the pressure to be comprehensive. The field's scope has simply gotten too big. I hated writing that book; I never felt like I was covering all the topics and I never felt like the topics that we did cover were in the right amount of depth. And it was still a five hundred-page book.
It's amazing to me: even though it's a new field, I'm starting to see people talking about "textbook IA". You can indeed do textbook IA, following basic practices and procedures for a fairly small site in a fairly controlled environment. But as a community we really haven't started to take on how you practice IA in the decentralized, hodge-podge environment of the enterprise. These settings are highly political; cultural and languages issues and different models of resource allocation and sharing need to be resolved when designing an information architecture.
That's why I'm seriously considering doing a book on enterprise IA, which would be something of a non-comprehensive handbook addressing many of the topics I'm covering in my seminars right now. It would complement the more comprehensive Polar Bear book -- which serves as something of a reference. I'm increasingly interested in how you practice IA in the enterprise environment, and I believe we're ready for a book on the topic.
I think you are going to start seeing other IA books that are more topical, which shows that the field is maturing.
CMSWATCH: Are there things on which you and [Polar Bear co-author] Peter Morville fundamentally disagree?
ROSENFELD: We are complete opposites. However, we almost never disagree.
When we were running our company and writing our books and even today when we work together, I think we've managed to come up with the same ideas from our completely different perspectives. When we managed [IA consulting firm] Argus and were writing project proposals and determining project fees, Peter would go sit down and crank out numbers -- you know: this many hours, this many people, etc. He would come up with a number. I would write down something based completely on a hunch. When we compared, we were almost always within a thousand dollars of each other, even with six digit projects.
CMSWATCH: So what is it about Ann Arbor, Michigan that it's kind of a fish tank for IA folks?
ROSENFELD: Two things. There has traditionally been a strong program in library science here at the University of Michigan and that program produced Peter and me. In turn, we decided to build the company [Argus] around IA here in Ann Arbor.
That and the fact that there must be something in the water.