Are Taxonomies Dead?

By Deane Barker on January 9, 2004

The taxonomy was always supposed to be the be-all and end-all of information architecture. A good, solid category structure was how all the information in an enterprise was supposed to fit together.

But they’re harder to build than you think. There are shades of gray and complications. You need related categories so people can jump from branch to branch; you can slice information so many different ways; who can agree where something fits, etc. I’ve tried to build a half-dozen, but I can’t point to any major successes.

Is the ideal of taxonomy possible? Or is it just better to invest in a good search engine? Think about it, when you visit a site, do you ever browse a taxonomy, or do you just go right to search? If you’re looking for something you’ve seen on this site, do you wade through the category list, or just hit the search engine?

When was the last time you actually browsed Yahoo! or DMOZ? I know they’re there, but I haven’t visited them in ages. Last time I did visit, what was the first thing I did? That’s right — typed something into the search box.

Search is a lazy man’s taxonomy. It’s not as organized or structured as a taxonomy, but human beings — imperfect creatures than we are — tend to settle to what’s easier. So, as an information architect, do you stand on principle, or do you cater to the lazy way your users are going to look for information?

This comes from my current infatuation with wikis. There is no categorizing of pages in wikis (even after my railing against all their shortcomings a few months ago), there’s just search and linking between pages. But the search is good, and it always seems to work. Same with the search on this site — when I’m looking for a previous post, it just always seems to work, and that search is nothing but a SQL “LIKE” query, the dumbest search of all.

So, are taxonomies an ideal that just don’t survive the reality test?

Gadgetopia
What Links Here

Comments

  1. You’re making a pretty convincing argument against them, and I wind up thinking that you’re right. (Mark your calendars – I’m replying to Deane and I’m not taking the opposite position) However, at some point the rubber has to meet the road, and it becomes a question of how do you store your documents. Most people still want folders and subfolders, which implicitly implies at least a basic taxonomy. Should I instead name all my filed with GUIDs and keep enough meta on them to sort them out? It would really suck then if you lost your database. I don’t know if I could ever implement an information system this way. I find myself clinging to structure.

    (The rambling starts here, FYI) There’s an interesting parallel here with libraries. A library at a major university implemented an automated filing system. All the books were kept in bins and moved around by automated lifters and conveyers. The librarians hated the idea. Why? It made NO ATTEMPT whatsoever to store the books in any sort of order. The books were all just in random bins, and the computer just knew which books were in which bins. Blasphemy against Dewey and his system.

    The thing is, students were then able to find books faster, and the filing was easier, because the system took care of it all.

    I guess we’re approaching that with large digital information stores now too. The computers are happier if the meatsacks don’t try to impose their view of the world on things.

  2. I’ve been doing some work with It’s the Content (www.itsthecontent.com) and, when you get down to it, that’s what you can do with that system.

    You can just dump everything in a big bin and then create a “dynamic” folder structure, where the folders aren’t real folders (what is a “real” folder, anyway?), but are just searches based on certain criteria.

    So you could have a “Users” folder with a “Joe” folder that just returns a search for everything in the bin owned by Joe. You could have a “Proposals” folder that returns all Word documents based on the template “Proposal.dot”.

    It’s weird — too weird for me to make the leap yet. You have to stand your thinking on its head.

    From what I hear, however, this is what WinFS in Longhorn is going to bring us closer to.

  3. I am struggling with a taxonomy design to represent all the Content Management Systems. As editor of the DMOZ/Google category on CMS, I am really frustrated by their policy of on website in just one place in the hierarchy.

    We have subcats for XML, for PHP, for Open Source. What if a CMS is all three of these?

    My efforts are in the Directory link in the banner at CMSReview.com and in a new website forum CMS-Forum.org.

    Your infatuation with wikis is shared by many. cms-listers asked for a wiki instead of a forum, so we created CMSWiki.com.

    But so far, the CMS Forum has 10 times the hits of CMS Wiki.

    It has been mentioned in a lot of places.

    Comments and criticisms of these works in progress most welcome…

  4. “We have subcats for XML, for PHP, for Open Source. What if a CMS is all three of these?”

    Then perhaps your category structure is too deep. If you have problems deciding where to put an item from several possible nodes in the hierarchy, your hierarchy is most likely more complex than the information demands.

  5. On some e-commerce sites, it seems like search and taxonomy can fit together quite nicely. I may search for a particular product and once I’ve found something the site may have a section informing me, “You may also be interested in these items.” In some instances, I’ve found that information very useful. In fact, sometimes, the suggested item is more what I wanted than the item I chose in the first place.

    I’m not sure how this would relate to a content based (rather than product based) site. Except for maybe a “related articles” section. Or like the “See Also:” links you put on articles here.

  6. You’re on the right track here. I’ve consistently found that “Related Items” links at the bottom of articles are very helpful and really contribute to browsing. If I get some time, I’m going to implement them on this site.

    I’ve been looking through a Sim City 3000 site lately, and they have “See Also” links at the bottom of these articles, like here:

    http://tinyurl.com/3ao3x

    I almost always follow them because they’re always relevant. The trick is being able to identify the relevant articles and be able to “back reference,” meaning when you add an article, it becomes a “See Also” reference for previous articles.

  7. Taxonomies are good as an idea, but people are often not good in implementing it good enough. Yes, I’ve used “categories” on this site to find only the post I’ve been interested. Unfortunately the category tree isn’t deep enough to sort content more granullary. The same is for many other sites. They often have only one “true” hierarchy which fits the needsonly some visitors, but not all. But well planned taxonomy can be very usefull. If I find an article interesting for me, I often click tags to see other articles from the same category.

Comments are closed. If you have something you really want to say, email editors@gadgetopia.com and we‘ll get it added for you.