MOOCs today are about less data for the teacher
My former student, Jeff Rick, has posted a reflection on MOOCs (on Facebook, so I can’t easily link to it from here), with an important point:
There’s an additional element that strikes me as critically missing from MOOCs: feedback to the instructor. Teaching is not about throwing good information out into the world; if so, Wikipedia (or public libraries a la Goodwill Hunting) would make formal education unnecessary. It is about making sure that the students get something out of it. For me, that requires a feedback cycle: realizing what problems students have, changing your teaching to meet their needs / interests, realizing and correcting your mistakes, etc.
Peter Norvig has said that he did the first AI MOOC with Sebastian Thrun explicitly to get more feedback. He was working on a revision for his AI textbook, and he didn’t want to just build it again and throw it into the world. By offering the book/course as a MOOC, he was able to get fine-grained data from many students on how they were using his book.
Teachers offering courses via Coursera or Udacity today get quite little data. The data is all captured behind corporate walls. I talked to Tucker Balch about the data he was gathering from his Coursera course “Computational Investing.” He said that he had the right to survey his students, but Coursera didn’t share any data that they had on the students. He got data on numbers of unique registrants, percent that took the first homework, percent that completed, etc. But nothing about how students did on particular problems, or how long they spent reviewing any particular video. No data that would help you figure out, “Hmm, I don’t think that’s working for the students.”
Isn’t that surprising, that in era of “Big Data,” MOOCs would be about “little data” getting back to the teacher who can most easily improve the course?