Well, it's not that thorough, and does raise the question of why this information is not made available by purveyors of MOOCs, but there are data here, and that's a start. One of the reasons for non-completion is answered in the text - that a focus on completion assumes this was among the goals of the learners who participate. In many cases, perhaps the majority, it is clear that this is very much not the case.

This highlights a fundamental difference between open courses (massive or not) and closed courses and one of the reasons that MOOCs are disruptive innovations, not just a big and free version of mediocre examples of what we already have. Traditional courses are focused on accreditation, for which it should be fairly vital to complete everything in the course, so people tend to do so a little more often than when left to their own devices. For those participating in MOOCs, the goal is typically quite different: it is to learn something. There is no reason on earth that a whole course is always needed for that, any more than that we should have to read an entire edited book to learn something useful from it. Nor is there any reason to finish a book we have started once we find that it doesn't interest us or is too difficult for us right now. Or we might find a different book, or article, or blog, or Wikipedia page that does it sufficiently or better. This is not news.

Give learners control, and they can choose when, how and what they learn. It is not up to us, the teachers. This is why I suspect that the future lies less with the xMOOC or even with the cMOOC (though that is far more interesting), but with the kMOOC - with the kind of things provided by the Khan Academy and a million how-to videos, help forums, Q&A sites and wikis, where learning comes in chunks appropriate to the needs of the content, not to filling a number of weeks or credit hours: where learning is on demand, not on command. Some things take five minutes to learn, some take five years, some are a never-ending process.

But, let's assume that completion rates actually do mean something. It seems to me that what this most likely shows is not that MOOCs are problematic as a matter of principle (if they were, no one would reach the end), but that those created so far are insufficiently compelling to be of value to more than a few. If they really offered value, neither hell nor high water would stop people from finishing them (OK - both might be a strong disincentive, but completion rates would be a great deal higher). This means that the content is perceived as insufficiently compelling and/or they are boringly or confusingly enacted. If we look at what people learn in appropriately sized chunks, there are things that people look at and gain value from, and things that they don't. The bigger you make the chunks, the more likely it is that there will be things within them that are of lesser value. So people lose interest. That's not much of a story though. We love to generalize from the particular to the general and we like news that simplifies complexity, especially if it demonizes something strange to us.