The issue is that, to a large degree, they mostly solved the easy thing. Making videos of university lectures widely available isn't hard. Neither is making other broadcast aspects of the syllabus available (e.g. in MIT OCW). A lot of good material was made available through Coursera and edX. But, frankly, a Wordpress blog with embedded videos and links would have worked nearly as well for a lot of this material.
Yes, CS somewhat uniquely benefits from the opportunity to reasonably evaluate code written for problem sets. But, for the most part, evaluation is at the multiple choice level.
My understanding is that a lot of schools already videotape their big lectures and many students just watch them online while doing study groups, going to recitations, working on problem sets, etc. more or less like they always have.
But these are neither VC-sized opportunities nor anything especially disruptive.
Leaving aside a few useful tools like online code evaluators, MOOCs largely solved a problem that was long-ago solved. Namely, access to quality broadcast-style lectures and other course content. At the same time, peer interactions and any genuine coaching pretty much didn't work with MOOCs.