Looks at the development of various classification systems leading up to tagging, or user created metadata. Argues that tagging more closely mirrors the nature of web information.

Argues that ontologies are a bad ideal for organizing the world online. Points out that library classification systems are designed to optimize space on the shelves, not to describe the essences of identities. Also, that library classification systems are fundamentally about organizing books, not about organizing the enormity of human knowledge. The same flaws exists in a hierarchical file system. That it is designed with the assumption that a thing can only be in one place at one time -- it makes some attempt to have the organizional structure of ideas match the physical world, where in fact a pointer, or an idea, or a metaphorical path can be in countless places at the same time, and can have many equally important and useful relationships which describe it.

That ontologies are useful where there are expert users, clear categories and a limited domain. But, much less useful for non-expert users or large domains, and fuzzy categories. Links are the universal pointers on the web, and the addition of tags is simple, and provides a much more useful finding system than an ontology. With a system like delicious, you get to know who's doing the tagging, not just what the tags are, so you get to limit searches by people and time, limiting the size of your group [penntags tie-in].

Very clear pros and cons of folksonomies versus more traditional classification systems. Looks at when and for what each kind of classification is most useful.

A Columbia Librarian posted a long article about tagging systems and their use in libraries. Interesting reading.

A beautiful brief essay about tags.