I'd elaborate on your 3rd point: some administrative units that are coordinating MOOCs underestimate the amount of time needed to plan, construct, and iteratively develop a truly excellent experience. The outcome is more or less predictable. I think a MOOC might probably be more like an iterative project (with bug fixes, updates to reflect new information, changes to reflect better explanations, etc.) -- it's not really "done" so much as "evolved". Designing a course is hard enough; designing one that can rest alone without direct feedback of the professor who gives the lecture series is more like a book project: those can take months and months of preparation and refinement. If you don't get credit for this effort, how could you dedicate time to it?
Once universities gives on-line MOOC career kudos similar to a book (or at least a peer-reviewed articles), the incentives to make the lectures truly great will emerge. I think this could transform community colleges and even some state colleges into institutions that provide one-on-one tutoring to those who really need/want the guided tour.
As a side note: Art History can be fascinating -- it's an excellent way to teach people to slow down, look closely, and form well-written, coherent logical arguments supported by evidence. It has a low barrier to entry. Unfortunately, I think much college eduction today is trying to make up for a completely broken primary and secondary school experience. Perhaps that should be our focus? Perhaps MOOCs will help in this regard -- by providing excellent eduction accessible to young children.