I disagree with the statement that "nobody should be able to decide what is a category and what is not", at least in its explicit interpretation. If you mean that having a rigid set of categories in a tree structure is not a good idea, then I agree, but the fact that everyone is stuck on trees is rather absurd. The simple fact is that many categories overlap -- I mean, articles can clearly lie in more than one category. While it made no sense in library schemes, where you can't have links and pointers, it makes perfect sense in a file system or database or other logical system that is not space constrained.
Once you abandon the tree structure of categories for a DAG, which is implied by your article, you are free to introduce multiple categories containing the same article, and the rigid hierarchy vanishes. Any discussion of categories really needs to include some insights from set theory and graph theory to have any kind of theoretical foundation. Brent Gulanowski 00:03, 17 Dec 2003 (UTC)