Joho the Blog
An Entry from the Archives

Clay talks about how taxonomies always have values built in. Even the periodic table's "noble gases" division reflects an assumption about the "essential" state of elements. He points to the Dewey Decimal System's skewed religion category. [Yikes! I've been doing that, too! I probably heard it from Clay first. I will attribute it from now on. Ack!] Even the Library of Congress puts the Balkan Peninsula and African on equal footing because it's measuring the number of books on the shelves. The categorization reflects not the ideas but the physical storage.

He points out, that even though Yahoo has cross links, it has a concept of which is the real categorization; it only shows you the number of links in a category if you are seeing the category in its "real" environment.

Hierarchical systems, he says, inevitably adds hyperlinks to cut across the taxonomy. (Yahoo only allows three, he says, because they didn't want to be spammed.) If you have enough links, you don't need the hierarchy. "There is no shelf."

When does ontological organization work? When you don't have a lot of stuff, it's stable, things have clear edges, an authoritative source and trained users. I.e., the opposite of the Web.

People have assumed that tags that mean the same thing are actually the same, but (Clay says) "movie" people don't want to hang out with "cinema" people, and "queer" people certainly don't want to hang out with "homosexual" people. There is information in the differences that thesauruses and categorization schemes miss.

He shows graphs that provide evidence that tagging forms power laws — who tags, and the tags that individuals use.

Organic categorization uses market logic, merged from URLs not categories. The categories overlap, not synch. The mergers are probabilistic, not binary. User and time are core attributes; you can do grouping, inclusion/exclusion, and decay. The semantics are in the users, not in the system; this is not a way to get computers to understand one another.