Why Not a MOOC?

One of the fundamental principles of engineering is, "Change one thing at a time." We're already trying to revolutionize the way scientists build software and use the web; if we try to change the educational system at the same time, we'll probably fail to do either. (Our instructor training does introduce people to modern evidence-based teaching techniques, but that's only revolutionary because most universities do such a poor job of teaching researchers how to teach.)
MOOCs are expensive. It took more than a thousand hours over the course of a year to record the videos we currently have, and I believe it would take a thousand more (i.e., six months of full-time work) to create robo-graded exercises for just our core curriculum. They might reach more people, but there's no point discussing how cost-effective they would be if we can't afford to create them. (And no, I don't believe we can crowdsource it.)
MOOCs don't work well yet. Many studies have found that user engagement and completion rates are low, and while 2% of 100,000 people is 2,000 people, it's still only 2%. It also seems that most of those who complete have done the material before and are using the MOOC as a refresher, which isn't a good fit for us.

The web will revolutionize education some day—how could it not? We're keen to see how this space evolves, as there are definite opportunities in online, distributed learning, but while online education and MOOCs are still working out the kinks, we're going to continue to watch, listen, and learn.