Regarding Stephen's post on "Why the Semantic Web Will Fail." First off, since I've been slagging on predictions lately, I'd point out that I agree that Stephen has "a pretty good record" for public predictions. I mostly agree with his arguments about commercial vendors not being able to play nicely together in something as complex as the Semantic Web.
For schools, I'm just hoping for a semantic intranet, where data will at least be easily shared between applications licensed by the school and hosted on their own computers. Even that seems a way off, not only because of vendor greed, but because the fundamental process of moving to Semantic Web technologies is to take intelligence out of applications and apply it to data. If that distinction makes no sense to you, consider that one reason this hasn't taken off.
The problem isn't that we have to write different types of programs, but that we have to do less programming and more, um, ontology-making. Right now we at least have a sizable population of programmers in the world, but I'm not even sure what the title is of the people we need to write ontologies. "Information architects?" Wikipedia's description doesn't seem to fit.
The problem is, whenever I read someone's vision for school computing or computing in general for 10 to 15 years in the future, as each feature is described I usually think to myself, "You need the Semantic Web for that..." I guess that's the next level of predicion here. Before I die, I imagine I'll be able to say to my computer, "I want to see Bjork on her next tour." And it will take care of it. The question is, by what method will my agent perform this? Using techniques that are recognizably the "Semantic Web" or some entirely different method? Or will this not happen at all in the next 40 years?