Here’s a really smart post by Clay Shirky [via Jeremy] on tradeoffs between controlled vocabularies for meta-data and “folksonomies” (i.e. letting people tag things with whatever words come to mind, a la Flickr and del.icio.us). He votes for folksonomies.
Clay is suspicious about the costs and feasibility of large-scale coordination of meta-data, and I am too, although Clay puts it all much more clearly. But can we learn from the general web here? It’s interesting to me that, while websearch engines are cited as ultimate beneficiaries of controlled metadata schemes on the web, I know of no instances of such a centrally coordinated scheme having really helped a search engine. (The nearest thing to a counter-example I can think of is PICS labels for adult sites, but needless to say those are not reliable.) And it’s also interesting that the one wildly successful and extremely helpful annotation scheme we’ve ever had is completely unstructured. (Here I’m thinking of link anchortext.)
My dubious reaction does not extend to microformats, and definitely not to folksonomy-style tagging. But any multi-part plan for large-scale metadata usage that has as its first two steps:
1) everyone will agree on a vocabulary, and
2) everyone will comply with this vocabulary en masse
has me dozing off at step 3.