The common vision: Nimble Internet startups were destined to sweep through academia, the last walled garden, just as they had in the worldwide media markets years earlier.
In retrospect, the hyperbole and excitement around MOOCs was understandable. While "online education" -- broadly defined as a class delivered in part or whole across a computer network -- goes back decades, the scale of MOOCs, in which a single course can attract a quarter-million students, was new and breathtaking.
Most histories of MOOCs start with the 2008 course "Connectivism and Connective Knowledge," created by George Siemens, then an associate director for research and development with the Learning Technologies Center at the University of Manitoba, and Stephen Downes, an online learning and new media designer and commentator.
The course content was available through RSS feeds, and students could participate via threaded discussions in Moodle, blog posts, Second Life or synchronous online meetings. The course was taken by 25 tuition-paying students at the University of Manitoba; another 2,300 members of the general public registered for the course online, free of charge.
[ Incremental revolution: MOOCs Lead Duke To Reinvent On-Campus Courses. ]
The current surge in interest in MOOCs began in 2011 after Sebastian Thrun and Peter Norvig attracted 160,000 students to a free online course on artificial intelligence, based on a course they had taught together at Stanford. Norvig is Google's director of research, and Thrun until recently ran GoogleX, the innovation lab behind the self-driving car and the Google Glass wearable computer. Thrun went on to found Udacity, which along with rivals Coursera and EdX began offering large-scale courses in early 2012. Udacity produces its own courses, primarily on math and technology topics, while Coursera and EdX work with university partners and cover a broader range of subjects.
How the massive amounts of data pulled from these platforms can be used to iterate courses or inform pedagogy per se is an area of intense work by MOOC vendors and academic researchers alike.
As Coursera co-founder Andrew Ng is fond of pointing out, the MOOC has collected more educational data in one year than all the universities in the history of mankind. Coursera claims more than 4 million registered students.
Speaking at the KDD-2013 Conference on Knowledge Discovery and Data Mining in Chicago this August, Ng described an online machine learning course, in which 2,000 students submitted exactly the same wrong answer because they reversed the order of two steps.
Such errors can be found and addressed, Ng said, adding this wouldn't be feasible for a class with 100 students.
Even harsh critics of MOOCs like Kenneth C. Green, founding director of the Campus Computing Project, think the MOOC vendors are onto something here, if they can use their scale to garner important data about student activities and outcomes.
"They're doing the best research in technology-aided education," he said. "If you have 150 students, and 2% miss something, it's easy to say it's a student issue. But if you have 10,000 and 2% miss something, that can be a catalyst for discussion."
A second defining characteristic of MOOCs has been their "open" aspect.
Traditional online instruction has been aimed at enrolled, paying students and offered for credit. Take George Washington University's online MBA healthcare degree, which costs the same as the standard, on-campus program.
Likewise, Howard University announced plans this summer to dramatically expand its online education offerings in a partnership with Pearson. But Howard University Online (HU-Online) will be degree-bearing and paid. Howard has no interest in MOOCs.
MOOCs, by definition, are free and open to all comers. A single course may include high schoolers, retirees, faculty, enrolled students and working professionals. Designing a course for this diverse student population is a challenge, as is making these courses "sticky" to the end. It comes as no surprise that while initial registrations for a MOOC can be enormous -- tens of thousands of sign-ups are common -- they tend to have much lower completion rates (typically in the high single digits) than credit-bearing, paid courses for enrolled college students.