It’s 10:44 am on a Tuesday, and I’m lounging at home in my pajamas, sipping chamomile tea. I am, at the same time, taking a class at Harvard. Professor Gregory Nagy is rhapsodizing about the death of Roy, the cyborg from Blade Runner, and pointing out how certain tropes from his final soliloquy echo important themes from ancient Greek myth. The class is called “The Ancient Greek Hero,” and it’s one of many MOOCs (Massive Open Online Courses) created by HarvardX, the university’s online course production company. It appears, from my limited experience, to be a fabulous class, which comes as no surprise, since it’s based on a well-established in-person Harvard course of the same name.
The videos for this course are remarkably elegant and professional, conveying a certain vividness that lectures at the blackboard sometimes lack. The discourse on Roy’s death, for example, seamlessly cuts to key scenes from the film, as Professor Nagy’s voiceover explains the Greek notion of the “hora” (the “correct moment”). The production values are just as high in a HarvardX course called “Introduction to Neuroscience,” in which filmmakers use high-gloss animation to create a vibrantly wacky clip about cell biology.
HarvardX is one of 29 institutions whose content appears on edX, one of the biggest platforms for MOOCs. The company that is now edX resulted from a partnership between Harvard and MIT in 2012, though each school’s courses are its own.
The main thing that edX provides, beyond hosting space for the videos, is the software necessary to grade—and provide feedback on—student work. This is no small undertaking: MOOCs can have tens or even hundreds of thousands of enrollees; edX alone counts about 2.5 million students since its inception. Typically, however, only about 7 to 9 percent of these students actually finish the course (though, to be fair, many of them don’t ever intend to). Despite the staggering scale, edX’s software provides grades for all of them—not just on multiple-choice quizzes, but also on short-answer items and essay-length responses.
Ever since MOOCs debuted, they’ve been an object of concern for many college professors. In a letter published in The Chronicle of Higher Education, the philosophy faculty of San José State University described the MOOC as a “serious compromise of quality of education.” They feared that MOOCs would come to be seen as replacements for flesh-and-blood teachers. Soon, they warned, there would be no classrooms, only technological simulacra managed by teaching assistants, with all the rewards flowing to a handful of private corporations.
It’s worth noting that these companies do, in fact, present themselves as the future of American education. In the lobby of edX’s headquarters in Kendall Square, near MIT, there are plasma TV screens that repeatedly flash words like “cutting-edge” and “tomorrow” alongside the edX logo. And at the HarvardX office, the overall atmosphere is very much that of a tech start-up: open floorplan; huge iMac screens at every workstation; employees sitting on exercise balls; ideas written in multicolored marker all over the erasable walls. These firms are cutting-edge, and they know it.
They’ve also come along at a time when brick-and-mortar colleges are facing a number of serious challenges. The admissions process has become increasingly competitive; growth in administrative budgets has outpaced raises for faculty; more and more tenure-track lines have been replaced by low-paid adjuncts; and tuition has continually skyrocketed. MOOCs, many seem to fear, will contribute to this trend, ensuring the ultimate obsolescence of the professor.
The vast majority of assignments offered through edX (which you can try here) are graded not by a human being, but by software called Open Response Assessment (ORA, pronounced “aura”). Although edX’s MOOCs do offer some peer-graded assignments—and some even provide students with opportunities to live-chat with a faculty member or teaching assistant—student work is, by and large, evaluated by algorithms that employ techniques called “machine learning.” It works by finding correlations between the grades assigned to a “training set” of responses—which are graded by actual human beings—and certain superficial features of student’s answers.
Piotr Mitros, edX’s chief scientist, offered the example of a chemistry problem in which students are asked which factors determine whether a particular chemical solidifies into a glass or a crystal. (The answers are “the rate of cooling” and “the complexity of the molecule.”) After being trained on the various ways in which students actually phrase these ideas, ORA can assign points according to a rubric and even provide feedback or hints to students who missed one component or the other.
Similar processes can be used for short-answer questions in a variety of disciplines. They can even be adapted to grade essays, which requires the computer to look at things like essay length, sentence length, vocabulary level, and punctuation.