I’ve been thinking a lot this week about two seemingly unrelated news items. The first, the research by David Shermis and Ben Hamner that found that automated essay grading software performs comparably to human graders. (See the Inside Higher Ed story.) The second, the official unveiling of Coursera, the latest online learning startup to spin out of Stanford, that promises to offer a full course catalog, including many classes in the humanities. (Here’s my write-up of the news). The connection: scaling how we assess student writing.
The tech industry talks a lot about “scaling,” and much of the ed-tech industry now seems to follow in its wake. When the tech industry speaks of “scaling,” it typically means “more tech.” Adding servers. Beefing up the processing power. Distributing computing across multiple servers. Automating as much as possible.
So no surprise perhaps, when the engineering department at a university like Stanford or MIT offers a MOOC, enrolling tens if not hundreds of thousands of students, much of the focus remains, again, on the technology. It isn’t just a question of technically “How do you deliver lectures?” or “How do you facilitate discussion?” – it’s “how do you scale the assessment?” And in the case of these engineering-focused classes, the assessment has been entirely automated. You submit your quizzes and your code, and it’s scored by a machine, not by a human. That’s what happens, I guess, when 160,000 students sign up for your AI class.
Since the phenomenal enrollment numbers of Stanford’s AI classes, I’ve wondered how or if we’d see this latest manifestion* of the MOOC work in other disciplines – those that, in my mind at least, are ill-suited to robot-graders. Of course, the Shermis and Hamner research might be used to argue that in fact we can automate the assessment in those classes too. AI isn’t just for AI class if robots can grade essays as well as humans. (I don’t actually believe that, but there you go).
What’s interesting then, now that Coursera says it’s going to offer just these sorts of classes I had in mind – science fiction and modern American poetry, for example – it’s also announced that it will not use machines to grade student work. Instead, it’ll use people. Coursera has built a technology solution (the startup’s founders are engineering professors, after all), but it’s a crowdsourced form of peer review, and as part of the Coursera classes students will be trained to assess one another’s work. It’s an interesting approach, and I’m looking forward to seeing how it works in practice.
It’s also a different way, of course, to view “scaling” with technology. In this case, it means leveraging the human network. It means leveraging the community.
That community piece is really important, I’d argue, and one of the things that frankly I struggled with with the Udacity course I took (and dropped out of). Community – or perhaps a better word is “audience” – is also one of the things that makes me balk at robot essay graders. I like writing for people, not automatons. (Writing for robots reminds me of filling out captchas – alternately frustrating, inane, and hilarious). And when I taught college composition, community – “discourse community,” specifically – was something that I talked to my students a lot about as I helped them consider rhetoric and argumentation.
So what will it look like when/if we scale college composition? What would a comp MOOC entail? The obstacle to scaling comp classes has long been the question of providing adequate feedback and assessment, but now it appears as though there are several technology solutions in the wings that promise to do just that: robot essay graders and a crowdsourced (human) peer feedback mechanism, for starters. If colleges opt to scale their writing courses, what’s it going to be? More like the Web, with students writing for and to their online peers? Or will it be more mechanical, with students writing for automated essay grading software?
* When I say “latest manifestation of the MOOC,” I mean those being offered by Stanford, MITx, and Udacity. I realize, of course, that the MOOC has its origins outside the engineering department.