This isn't the killer post on tags and folksonomies I've been working on. That one has been killed for now, because it points to so many other posts, that I can't make enough sense of it to finish it. (I really need to start an anti-blog for these things that I never post.)

But, I started to write this (related) comment on David Weinberger's blog post about his new Joho essay, Trees vs. Leaves, and it got too long. So, I'm posting it here instead and just tracking back to David.

A few questions / comments about "Trees vs Leaves":

Isn't a dictionary a shape that knowledge itself takes? Isn't a dictionary not a tree?

Let's say it was a dictionary of only nouns and it was unsorted. The "system" would be a big pile of nouns attached to descriptions which include "tags" that are other nouns (and therefore are pointers to other nodes in the dictionary), right?

So, I realize you could look at how these nouns point at each other, and map a tree of relationships (e.g., poodle: a kind of dog; dog: a kind of mammal; mammal: a kind of animal).

But, a dictionary primarily represents knowledge as terms cross-referenced in a (controlled) vocabulary. In other words, even if some of how these nouns point to each other could be a tree, it also isn't in every way constrained by a tree structure. So, for example, consider every noun a tag in this definition of a poodle:

Any of a breed of dogs originally developed in Europe as hunting dogs, having thick curly hair of varying color, and classified by shoulder height into standard, miniature, and toy varieties.

With a nice computer, one could browse every noun whose definition includes the tag Europe, for example.

In my experience, there is a kind-of "tree school" that people tend to believe in. But, for the most part, for a long time, most folks (and even most of the folks who belong to the tree school) have relied upon not-tree structures for knowledge and classification. And, I think dictionaries and thesauri probably represent the most common fully elaborated shapes of these not-tree structures that are so relied on.

On top of that, I would say that the "shape of knowledge" can be said to be the library itself, with all of its dictionaries, thesauri and other works. Whatever bug Dewey had for a tree structure, I'm not so sure he was missing the forrest of the whole library for that tree.

My own experience is that good organization relies on the joining of multiple structures and / or multiple approaches to organization. Libraries may use the Dewey Decimal system, but they also use alphabetical indexes, authority files, and librarians' and patrons' "recommended books of the month"—both to help people find things and to help them know things.

And, among many nice things I would want to say about the concept of folksonomy, I would say that it allows each person to join their own organization to many others'—and I'm certainly not surprised that this might have obvious advantages over an isolated taxonomy in many, if not most, cases.