My experience over the past nine months is that there are a lot of platforms with work pretty well with experience facilitators for a modest-sized group. And nothing that is really very satisfactory for interactivity at scale, especially with a heterogeneous audience (except in the most glancing way like poll questions).
I agree that interactivity at scale is super difficult. The problem is that in a course you are teaching more or less solved problems to bring students up to speed to where the field is. So the same questions get asked from class to class and group to group. There's the idea that you can compile all this knowledge and questions and then provide that as a resource to students. But actually the hardship of getting through those questions, even though they are the same, is how they learn. So the resources are nice for instructors but not as useful for students directly. There's definitely a lot of space for innovation but most of the people doing that either lack the technical expertise of developers etc because they come from education departments or they lack the domain knowledge of education research. There's a somewhat large space to fill with people who are good at both.
That's only part of the solution though. Even if EdX, as far as it goes, is an open platform, someone needs to run and support it for schools, you need email and doc sharing, etc. etc. It's fine to say that everyone should just run Linux, OpenOffice, and so forth and just shrug if that's too complicated for a lot of students and teachers. But that's not an actual solution.
In principle, the government at some level could do its own collaborative/learning software and host it but people would perhaps rightly then ask why the government isn't using readily available commercial off the shelf software like most companies do.