Jonathan sent me a link to a blog post called MOOCs as capital-biased technological change. It argues that the effect of MOOCs will be to reduce demand for university teaching, which is what supports creative research, and so society's academic output will decline.

I'd argue that if MOOCs lower the cost of teaching, the money saved can (and should) be spent on research, and so the number of people paid to increase the sum of knowledge would go up.