So, you are unhappy with a five percent completion rate. Hire tutors (lots of them, if it is massive). Try to get machines to tutor. Use learners as tutors (never mind the pedagogical affordances, you only care about scale and completion). Set up automated phone calls to remind people to turn in their homework. Ring the (behaviorist) bell.
Or not.
Google’s Coursebuilder team has an interesting take on completion rates. Let’s start by asking learners what they want to achieve. Then examine their behavior against their own expectations, rather than against fixed criteria. Surprise, surprise: take learner agency into consideration, and it turns out that most folks finish… what they wanted to.
Bill Cope has an interesting take on scale. He says: there is no scale. It is not only that face-to-face/online is a false dichotomy. The intimacy of learning can be recreated, irregardless of how many people are learning. Public schools break down an entire population of children into classes of twenty-five. The Red Cross and Red Crescent train 17 million people each year to do first aid, one workshop at a time. That makes the best aspects of those experiences ‘personable’. But depersonalization is not  a function of scale. It is a function of learning environments that limit the affordances of learning and assessment.
In the United States, 26 million already have Type II diabetes. That is already massive problem on a national scale, part of the very wicked problem that makes non-communicable diseases the world’s bigger killer, responsible for over 36 million deaths every year. Prevent is a start-up that just raised 28 million U.S. dollars to deliver personalized health education on the very intimate issue of pre-diabetes, where a positive outcome equals a change in real-world behavior. In its model, each person is matched to a small (read: personable) group of no more than a dozen peers, and then works as part of this small group. The first published clinical study (apparently sponsored by the start-up, but due for publication in a scientific journal) indicates that the approach helps people lose weight in clinically-significant, long-term ways. The scale is in the opportunity, not in the experience of Prevent participants.
There is no scale.Â