'I think there is a fair chance that much of the semantic web dream (i.e. "the internet organizing itself") will arrive pretty suddenly, with exponential-like growth.'

I'm skeptical. It seems like you'd have to either solve some pretty tricky AI problems, or get people writing for the internet to do all this extra work. It might still to feature in some larger, professionally developed sites, but I'd be surprised if it took off.