Also from Caterina, a description of the semantics of flickr tagging.
Dave Weinberger has written about it also. Here are two interesting thoughts:
t will be fascinating to watch the social effects as people adjust their tag sets in order to get aggregated either into the most popular tags or to be segmented into smaller groupings. That is, if you want to be found when people are searching for blogs about America, you will learn to tag it with (say) “USA” and not “U.S.A.”, “US,” or “America.” And if you want to have your posts be found by people searching for posts written by members of your Dungeons & Dragon’s group, your group will make up a tag that no one else would use. How this sort of stuff occurs at Technorati depends to a large degree ? but not entirely ? on how Technorati chooses to enhance the system. Little changes will have rippling effects.
Third, this represents the externalization of tagging. That is, Technorati is a broker of tags, not a place where you create tags. There are other important functions that could be handled externally, including the creation of thesauruses so that items tagged as “USA” get clustered with ones tagged “America” and “Etats-Unis.” The particular apps where you tag stuff can, of course, compile their own thesaursi. And, they’re likely to be compiled automatically by noticing the different tags that are applied to the same item. But having a thesaurus compiled from a superset would help smaller-scale apps cluster tagged items well and would provide additional useful information to all clustering apps. Local thesauri are always going to contain the most valuable information, but info from the aggregated thesaurus can also help. But, there will be social effects from having external thesauri. I don’t know what those effects will be, but I suspect that they’ll be significant since thesauri are about meaning across groups differentiated by meaning.