interesting. wouldn’t it make more sense for tags to emerge as from natural language used in the content itself?

that way the content–content relations are made explicit by the nature of their inclusion within various tags. i.e., i’d expect both the content of “fuel combustion engines – how they work” and the content of “gasoline – how it works” to be related with at least the one similar tag of automobiles.

to me it seems like tags should be intersection points that accrue greater “density” as it were as more content-content lines cross those points. then tags wouldn’t have to be user-defined, either, though they could be as well.

does that make any sense?