What's the business model? I can see why Odersky might want to promote his own language but more generally, why would someone of his caliber devote the time? Philanthropy? Or do the universities pay?
For example, the Princeton Algorithms course says:
.. Coursera will maintain limited data regarding student progress and performance in this course and, with your permission, provide authorized third parties with access to such data.
Given what recruiting fees are, I'd imagine this could work quite well. It is unclear how well it will scale, but if they can keep their costs low enough it may not need to.
So I think letting universities pay for the data or students paying for certificates.
I took a CS intro class at Texas A&M there and he came in at the end of the semester to talk more about C++11x. So he ended up going off about support for threading and things that were way over most people's heads and how the language was getting all of these new features.
From a university standpoint, every intro CS class is C++ for the most part. There is some Java thrown in, and maybe some Python, but all of the algorithms/data structures, really the meat of the basic CS education, is all in C++ because of him. Other professors do some of their research in C++ mainly because they can just walk down the hall and ask him questions about the language if need be.
I think for grad students he would be an awesome resource to have. For the undergraduates, he has cast a strong and complex shadow on the department. The environment has become you learn C++ and then go into the Oil/Engineering/Accounting industry.
No major point here, just reflecting on the topic of "Stroustrop at TAMU"
Not commenting on Odersky in particular, since I'm not sufficiently familiar with him, besides the white Scala book. I will say though that course description claiming Scala "provides the core infrastructure" for my employer is a small overstatement. We definitely use it for some important services, and continue to use it for new projects, but it isn't accurate to imply it's the dominant language in our codebase.
I think that is why lots of people are attracted to working at Google (Pike, Thompson, van Rossum, Bloch, etc)
Edit: Though I don't think he's affiliated with a university. Oh well, there's still lots of good Clojure-oriented stuff available: http://alexott.net/en/clojure/video.html