I think the Wikipedia and MOOCs are perfectly good for learning a well-developed body of knowledge. I skipped a lot of my undergraduate classes because the lectures didn't add anything to the textbook. I don't think they're adequate for learning to practice most disciplines yet. If you want to do mathematics you need to learn to write proofs, and you're going to write an awful lot of wrong proofs in the process. You really need someone to tell you when you're doing it wrong. The same is true of science, engineering, philosophy, and writing. It's possible to be an autodidact, but most autodidacts end up eccentric as best and cranks at worst. Universities will still have a place until there's some other way for autodidacts to avoid compounding their misunderstandings until they end up like this guy.

Computing is a special case, and I think that's what leads people with a background in computing to be so enthusiastic about MOOCs. In computing you can easily test yourself against the machine; if you implement your ideas and they work, you're probably not too far off. That doesn't generalize very far. If you're trying to learn math, you might check your proofs with Coq, but that's really the wrong level to be writing proofs at. In the physical sciences you could do experiments yourself if you have the money and can legally buy the materials, but you'll reach a point where you don't or can't eventually.