The folksonomy meme is well underway, with a well timed announcement from Technorati feeding the frenzy. I think that one thing that needs to be addressed, though, is the sense of triumphalism—folksonomy over all organized taxonomies—that I hear in some of the posts.
Clay Shirky starts to address this issue in his excellent post about the economic costs of controlled vocabularies. He points out correctly that for systems at the scale of the Internet there is just no way to control and manage tagging information in a centralized way that is even remotely economically feasible.
This is fine, but it’s not the whole story. There are plenty of systems at smaller scales than the Internet where some combination of controlled and uncontrolled tagging is necessary. Put it another way, you don’t really want your users generating all your metadata in an uncontrolled fashion. Examples: accounting codes for general ledger systems; country codes; languages; lists of employees; and so on. On the flip side, as Scott Rosenberg points out, often user-generated metadata is a lot more tractable and ultimately more useful than trying to cook up “official” lists in a clean room.
There is also discussion about issues of synonym control, browseability, and so forth. Yep. Actually, I’m not convinced about synonym control. If the system offers a way to browse by frequency, it’s likely that users will find the tag that the majority of users are using and want to be a part of it—this happened on Orkut with a number of groups, including the Mac and Macintosh groups. Of course, one of the issues there is that changing groups on Orkut was fairly frictionless, whereas changing tags (categories) on one’s blog is quite a bit stickier as a problem. Where in the infrastructure might one want to see synonyms established?
Incidentally, this is one online topic where the discussion at Slashdot, even at the Score:4 level, was completely unhelpful.