[Chevalier]

Wow, I had no idea. I assumed, as pbhjpbhj wrote, that MOOCs' high incompletion rate was due to the complete absence of barriers to entry. I've enrolled in dozens of MOOC courses to have access to their course archives in the future, contributing to the "failure" ratio.

Completion is already an extremely suspect metric by which to grade MOOCs, but it's particularly dishonest when you depend on expulsion-level populations. As most anti-MOOC arguments depend crucially on the San Jose study as evidence that lectures must be live and artisanal to work, I expect to refer back to your post quite often in the near future. Thanks for the heads up!