In my mind the crucial thing about this new approach classification is that it gets around the achilles heel of the traditional approach of meta tagging or controlled vocabularies, which is, that most people just don't do it.
There are some special circumstances when a fixed list (or hierarchy) of categories or topics or tags will work for most people. But in the majority of cases, a fixed list of categories to apply to some item of information just causes frustration ("the category I am looking for isn't there") or confusion ("What does tag X even mean?")
The end result is that information just gets mis-tagged or is left un-tagged. This is where so called folksonomies are so superior. Yes, often information will get misclassified. Whether this is so often as to render the whole system useless is an open question, which Technorati Tags, Del.icio.us and others will help us discover.
p.s: One of the dangers to commenting on this discussion is that it is basically impossible to read all the essays and follow all the links and so you risk repeating what was already said. I believe that Clay Shirky has made essentially the same point. At least I am in good company.