From a Clay Shirky comment on his post on xMOOCs:
The thing that seems to me to differentiate MOOCs from iTunes U and other ‘Access to the lecture’ platforms is the linking of the idea of a course that takes place at a certain time (including ending at a certain time), certification of the results, and, of course, massive scale.
I don’t think we’ve thought through the implications of this, namely that xMOOCs are relatively easy to make out of existing OER materials.
We continue to treat these platforms as if there is some super secret technology at work in them. In reality , it’s OCW in an LMS wrapper, serialized and made available to a large class. There’s some peer-grading involved, yes, but we were doing this stuff anyway.
You can go and reuse existing OCW and put together an xMOOC in a week, in a platform like Canvas.net or Class2Go. If I understand the Open Yale Courses license correctly, you can go get a bunch of OYC videos and string them together with quizzes and a short peer-reviewed paper and you are done. It’s the cohort that makes the difference, not the twiddly bits and the machine that goes “ping”.
Hapgood 
We can also set an open enrollment and timed release in Moodle, and we could have been doing it in WebCT for the past 10 years. The hype surrounding the Cous/dacity style platforms provide a marketing advantage for the moment, but at an institutional level the platforms themselves don’t seem to offer an advantage over what we already have.
I’d go even further and suggest that the LMS wrapper is unnecessary and constraining. With the Mechanical MOOC (http://mechanicalmooc.org), we are using an e-mail list to manage the cohorting and pointing learners directly to the (O)ER in situ. The platforms of many MOOCs limit the community artificially, and force producers to adapt assessments to whatever the local system can handle. With the MechMOOC, we don’t have to fold Codecademy’s exercises into our platform or recreate them. We just use them as is. The challenge in this model is certification, an issue we can sidestep in this first iteration by focusing on a skill like Python programming.