Stanford University got lots of attention for inviting the public to participate in a series of free online computer-science classes. One thing that’s drawn less notice is how some of the technologies that help facilitate those mega-classes are changing the experience for Stanford students learning the same subjects. Now a Stanford student is provoking a debate on those innovations, with a blog post critiquing the rigor and format of the “flipped classroom” teaching method deployed in his machine-learning course.
In one version of that course offered to Stanford students, the traditional teaching format was inverted, with lectures presented through online videos and optional once-a-week class meetings devoted to problem solving with the professor. The videos, plus auto-graded assignments, were also offered to the public in the free online version of the machine-learning class. As of November, a staggering 94,000 people had signed up to take that course.
But Ben Rudolph (pictured above), a junior majoring in computer science, provoked a lively discussion on Hacker News with a blog post attacking the largely online version of that class that was offered to Stanford students. He faulted the course, “CS229A: Applied Machine Learning,” for insufficiently challenging assignments and for swapping live lectures with canned videos.
“Online lectures suck. Sure, they’re great for rainy days or people learning at a distance or people that don’t go to Stanford. However, these new classes are getting rid of in-person lectures completely. I met barely anyone in my CS229a class. Everything was done alone in my room, which is kind of crappy especially when there is such a nice campus right outside.”
“The initiative that Stanford has taken to open up education is great. However, God help me if all my classes become 2 hour weekly online lectures with review questions and auto-graded programming exercises. Stanford can expect a letter from me asking to get a cut in my tuition if the classes begin to go the way of CS229a.”
Mr. Rudolph took particular exception to the programming exercises, in which the computer automatically informed students whether or not they got 100 percent on the task. “It’s so black and white,” he tells Wired Campus. “They have to make it easy enough so everyone can get 100 percent, basically. In the past I’ve turned in programming assignments, and only the really smart kids got stellar scores, because they went above and beyond. This model kind of discourages that.”
In one sense, Mr. Rudolph doesn’t have much to complain about. Stanford also offered a traditional version of machine learning via another class—CS229, taught by the same professor, Andrew Y. Ng—with live lectures twice a week. What’s more interesting is his worry about the spread of technologies that facilitate teaching to a mass audience, as well as his take on the “flipped classroom,” a teaching format getting lots of buzz lately largely through the work of the ed-tech star Salman Khan.
According to Mr. Ng, that format first appealed to Stanford professors as a way to improve education for the university’s own students, not to accommodate the general public, as Mr. Rudolph suggests. Prior to flipping his class, Mr. Ng had already been offering students the option of watching video recordings of his lectures at home rather than attending the lectures in person. The average in-class attendance hovered around 30 percent of students.
“The vast majority of students in my class were not showing up to class, and they were choosing to watch videos of me instead of showing up to the live classroom,” Mr. Ng says. “If students prefer to watch videos online anyway, then maybe we can do much better.”
In particular, Mr. Ng thinks the online experience can be improved through shorter, more topically coherent videos, and interactive quizzes.
Right now, when Mr. Ng asks a question in class, here’s what happens: Half the room is still madly scribbling away, he says. Maybe one-quarter of the students are on Facebook. And there’s sometimes one “smarty-pants” in the first row who blurts out a response. “Essentially, only one student in a class gets to attempt an answer,” he says. “Whereas in contrast, using online videos and interactive quizzes, every student gets to attempt an answer.”
As for auto-graded assignments, Mr. Ng stresses the benefits of instant feedback over the traditional method of returning assignments weeks after students learned the material.
And notwithstanding Mr. Rudolph’s complaints, early responses are encouraging: Teaching evaluations for the “flipped” version of the class (CS229A) were comparable to those for the traditional class (CS229). So, too, were students performances on the final project, which was graded based on the same criteria for both the traditional class and the new “applied” version of it.
“This was surprising, because even though CS229A had fewer prerequisites and thus many students in CS229A had come in to the class with a weaker mathematical background, this shows that they nonetheless learned enough to perform essentially identically to CS229 students in terms of getting machine-learning algorithms to work,” Mr. Ng says. The professor sees this as evidence that “the flipped classroom medium in CS229A is even more effective than the traditional lecture-based CS229 one for teaching students machine learning.”
Another student, Kevin Khoa Nguyen, agrees with Mr. Rudolph that the new class had been one of the easiest computer-science courses in his time at Stanford. Mr. Nguyen tells Wired Campus that he is a big fan of the online lecture format, however.
“I think this is the future of education,” he says. “The videos save a lot of time in going to class and the use of efficiently displayed diagrams takes away the overhead in drawing diagrams on a whiteboard. Obviously, this approach comes at a cost of classroom interactivity, which can be remedied with sites like Piazza,” an online forum where students can ask questions.
What do you think? Is this the future of education? Or a misguided innovation?
