Universities opened courses to millions and counted who finished

The classroom note

In 2012, two platforms opened university courses to anyone with an internet connection. Coursera's own site states it 'was launched in 2012 by Andrew Ng and Daphne Koller with a mission to provide universal access to world-class learning'; edX, per its own about page, was 'founded in 2012 by Harvard and MIT professors' and 'began with Harvard's most popular course, Introduction to Computer Science.' Universities that had spent decades limiting enrolment by campus capacity were suddenly recording enrolment in the tens of thousands per course.

What the evidence says

The first rigorous look at what all that enrolment produced came from the universities themselves. A joint Harvard-MIT report, deposited in Harvard's own repository, covers 'the year from the fall of 2012 to the summer of 2013,' during which HarvardX and MITx 'launched 17 courses on edX' and recorded '841,687 registrations from 597,692 unique users.' Of those registrations, the report states plainly, '43,196 registrants earned certificates of completion. Another 35,937 registrants explored half or more of course content without certification. An additional 469,702 registrants viewed less than half of the content. And 292,852 registrants never engaged with the online content.' On those figures, completion ran to roughly one in twenty registrations — a number the report presents as registrant behaviour across 17 different courses, not a single rate it claims to generalise.

The implementation question

A registration here costs a student nothing but a few clicks, which is precisely why enrolment and completion counts measure different things: mass free access lowers the bar to start a course far more than it lowers the bar to finish one, since the time and motivation a self-paced course at home requires were not part of what either platform was selling. Universities publicising hundreds of thousands enrolled and a completion figure in the low single digits are, on this report's own breakdown, describing the same population from two accurate angles.

What holds and what fails

What holds is the report's own four-tier breakdown of engagement, a genuinely more useful description than a single completion percentage, because it separates never-engaged registrants from those who explored substantially but did not certify. What fails is any assumption, unsupported by this report, that low certificate completion means the courses failed the people who did not finish; the report describes registrant behaviour, not a judgment on educational value delivered short of certification, and treating the two as identical would be an editorial overreach the source does not make.

• Is a completion rate being quoted with its denominator — all registrants, or only those who engaged?
• What does a free, low-commitment registration actually promise, compared with a paid, seated course?
• Is 'enrolled hundreds of thousands' and 'very few certified' being treated as a contradiction, when the same report shows both are true of the same population?

The engagement tiers in the first HarvardX-MITx report are the direct ancestor of today's argument about AI tutor usage numbers: a large registered-user count and a small completion or mastery count can both be accurate descriptions of the same free tool.