recursion

Jenny and I have been taking some online courses lately. The flowering of college-level online courses has been a real boon to self-learners. I am not sure it will always be the case, but as startups “disrupt” the industry and traditional universities scramble to figure out the new medium, a huge number of free courses from great schools and amazing teachers is available.

I’m currently taking Introduction to Recommender Systems offered through Coursera by the University of Minnesota (my alma mater) and taught by Professor Joe Konstan and Ph.D. candidate Michael Elkstrand. This class appealed to me because I’m interested in recommender systems and I never got a chance to take a class with Professor Konstan. It’s been a pretty big time commitment, but I’m enjoying the class.

Taking this class has exposed me to a number of other classes that I now want to take. There are some terrific classes like Stanford’s machine learning, compilers, and database classes; the University of Texas’s linear algebra class; a public speaking class from the University of Washington; and Princeton’s statistics class. My main problem is figuring out a schedule that allows me to take these classes – since many of them require 10+ hours per week of coursework, it’s difficult to take more than one while working a full-time job.

A great thing about these classes is that you can learn from some of the best teachers in the world. Introduction to Recommender Systems is taught by one of the pioneers in the field. Alex Aiken created the Cool language to teach compilers. And of course, MIT’s legendary 6001 computer science class is available online.

The biggest challenge with MOOCs (besides the terrible name) is the absolutely dismal completion rate. The average completion rate for MOOCs is under 7%. In a paid, for-credit trial San Jose State, Udacity managed a completion rate of 83%, but the failure rate ranged from 56-76%. I have been party to the completion problem. I dropped Udacity’s Intro to Computer Science because it was too basic to be interesting (that’s good news, I guess). Less happily, I bailed on Udacity’s self-driving car class in the last unit because I got busy. In general, MOOCs seem to offer self-starters a good opportunity for guided learning. I would guess the outcomes are better than just buying textbooks and trying to learn from them. But the time commitment is real and the classes can be quite difficult.

That’s why I am a skeptical of entirely replacing college with MOOCs. I think online classes can augment the learning experience, but the traditional classroom still provides a huge benefit for many learners (not to mention the social capital attending a particular university generates). As traditional universities scramble to get into the MOOC game, much free learning is available – which I am happy to take advantage of – but the ultimate goal of Udacity and Coursera is profit. Universities are looking for a way to cut costs, and outsourcing their lecturers to Stanford, Harvard and MIT may seem appealing. Or maybe Podunk State thinks they’ll be exporting their lecturers to Podunker State. Good luck, this is likely to be a winner-take-most game.

Ultimately, some form of recorded instruction might offer a way out of the university cost spiral. Baumol’s cost disease can be blamed for part of the eye-watering increase in college costs:

The bubble analogy does work in one respect: education costs, and student debt, are rising at what seem like unsustainable rates. But this isn’t the result of collective delusion. Instead, it stems from the peculiar economics of education, which have a lot in common with the economics of health care, another industry with a huge cost problem. (Indeed, in recent decades the cost of both college education and health care has risen sharply in most developed countries, not just the U.S.) Both industries suffer from an ailment called Baumol’s cost disease, which was diagnosed by the economist William Baumol, back in the sixties. Baumol recognized that some sectors of the economy, like manufacturing, have rising productivity—they regularly produce more with less, which leads to higher wages and rising living standards. But other sectors, like education, have a harder time increasing productivity. Ford, after all, can make more cars with fewer workers and in less time than it did in 1980. But the average student-teacher ratio in college is sixteen to one, just about what it was thirty years ago. In other words, teachers today aren’t any more productive than they were in 1980. The problem is that colleges can’t pay 1980 salaries, and the only way they can pay 2011 salaries is by raising prices.

But I think we are a long way from figuring out how to integrate these new ways of learning. In the short-term, if we are looking to cut costs, university administrators seem like an attractive target to me. Meanwhile, I will enjoy the free online classes and thank my lucky stars I attended college back when it was cheap.