What follows is a thought experiment I've been toying with for the last few days, a real work-in-progress. It may not be interesting until there is real software to play with, but in the meantime I'm posting it as a project log.
Part 1: Tags and Trees
Tags and Searches

"Tags", to put it mildly, are the new hotness on the web. The most popular social web services such use "tags" (or "labels" or "keywords") as a quick way for user categorization of information, and in most cases they work quite nicely. Plug in a tag like "Seattle" into del.icio.us or flickr and you will get a mish-mash of links and pictures related to the city with the world's first Starbucks. More sophisticated systems will let you include multiple tags to find content and suggest related tags that can be incorporated in a search. Tags are most useful when you know what you want and simply need to locate something quickly from a lengthy list. After all -- search engines specialize in automatically creating tags or keywords from existing content.