Guest Post: Eric Roberts on the Dangers of Escalating Enrollments

Eric Roberts just sent out a note of concern to the ACM Education Council, and he kindly allowed me to share it here as a guest post.  When I read his note, I thought about my May 2009 CACM column on “Teaching Computing to Everyone” where I argued that, once non-CS majors figure out the value of what we have to offer, our problem won’t be declining enrollments, but dealing with capacity issues. Two years later…

Just before heading off to the CSTA Advisory Council meeting at the end of last week, I sent out a few thoughts to the other Advisory Council members about what’s been happening this year at Stanford in terms of computer science enrollments. Enrollment trends have been a major topic of concern for CSTA in recent years, just as they have for the Education Board and Council. For most of the past decade, enrollments have been falling, and they remain low at many institutions, some of which have been cutting staff or even closing departments in response.

In many ways, that decline in enrollments has served as the primary narrative for the computer science education community for several years now. In most of the meetings I attend, one still hears a lot about the decline in student interest, the vanishing diversity of the field, and the urgency of developing well-trained people to fill the jobs that will go begging unless we can repair this broken pipeline. I believe those problems are absolutely real, and I myself sometimes talk about them on the road. Doing so has become much harder because I spend my day job worrying about the opposite problem: enrollments that are rising so fast that it is nearly impossible to stay on top of them.

As some of you may have heard from Mehran Sahami’s presentations, Stanford has seen its computer science enrollments recover at about 20% per year since 2007-08, after turning the corner the year before that. A 20% rate of increase is healthy and manageable. What we’re seeing this year — and particularly this quarter — is at a level that I find downright scary, having lived through a period in the early 1980s in which enrollment growth outpaced the ability of academic institutions to respond.

Here are a couple of data points to give you a better sense of our situation. In 2009-10, Stanford fell just short of its all-time record enrollment in our CS1 course, which we call CS106A. That record of 762 for the three regular-term quarters was set back in 1999-2000 at the height of the dot-com bubble. I know that I have already told several of you that I was confident we would pass that mark this year. What I didn’t know was that we would leave that old record in the dust. The numbers are now in, and the enrollment in CS106A in those three quarters is 1087, which represents a year-on-year growth of 51%. More frightening still, the enrollment this quarter in CS106A is running ahead of last spring by 120%, suggesting that the trend is accelerating rapidly. That conclusion is supported by enrollment data from other courses. Compared to last spring, enrollments this quarter are up by 74% in CS107 (Computer Organization and Systems), by 78% in CS109 (Introduction to Probability for Computer Scientists), and by 111% in the course I’m teaching, which is CS181 (Computers, Ethics, and Public Policy).

We have some ideas as to why this growth is occurring at Stanford. In his paper at SIGCSE, Mehran offered data to suggest that about half of growth we’ve seen in majors is attributable to adding greater flexibility to our requirements and that the rest has more to do with the bad economy driving students into different career tracks. The scale of the phenomenon and the fact that the effect seems to have gathered tremendous steam this year are as yet beyond our powers to explain. Several of us here have also observed a change in the way students are perceiving our courses that may help understand the basis of this resurgence of interest. Although I as yet no evidence to support such a claim, I’m convinced that the growth spurt we’re seeing here at Stanford comes from subtly but distinctly different sources than the one we saw during the dot-com boom of the late 1990s. In those years, one of the things that drove me up the wall was the number students who were choosing to major in computer science for reasons that were essentially wealth-seeking. I had many students — probably never more than ten percent of my classes, but nonetheless enough to poison the classroom atmosphere — who would freely admit that they hated the material but knew that if they got their Stanford CS degree, they could soon make enough money to put all that unpleasantness behind them.

What my colleagues and I are seeing today is entirely different. The students who are now inflating the ranks of CS106A are, it seems, deciding to take a computer science course as a way of bolstering their credentials before they emerge into a weak economy. Most have majors in other areas but recognize, probably correctly, that having programming skills will likely increase their chances of gaining employment in their own field. A surprising number of those students, however, once they get into our introductory courses fall completely in love with the material and continue on to double the size of the downstream courses in the curriculum.

My most poignant story of the year is of a student in my CS106B class (the reasonably traditional CS2 course that follows after CS106A) who came to my office hours last quarter pretty much in tears. The problem for her was not that she was having any trouble with the material. Indeed, she was one of a handful of students who got a perfect score on the midterm and an even smaller group that ended up with one of our jealously guarded A+ grades in the class. Her issue was that she now wanted more than anything else to spend her life doing computer science, but couldn’t figure out how to make that happen given that she is currently a fourth-year Ph.D. student in geology. We’re working together on that problem, and I’m sure there will be lots of opportunities for her to combine her skills. What’s amazing is how many students there are who seem to be in the same position. I’ve swapped stories with several of my colleagues here, and there are literally dozens of people with much the same tale.

Stories such as this one are dramatically at odds with the narrative that most of us have been following over the last several years. At least at Stanford, both numbers and enthusiasm have rebounded — with a vengeance — to reach their highest levels ever. What I’m less sure about is whether Stanford is an anomaly or whether we represent the leading edge of a more general trend that other places can expect to see within a few years. If it’s the former, those of us at Stanford need to understand better what we’re doing right; if it’s the latter, I think that many organizations — certainly including both the ACM Education Board and Council — will need to do some thinking about what’s going to happen if this phenomenon spreads.

There is some evidence that our experience is happening at other places as well.  At most of the large institutions, enrollments hit bottom a couple of years ago.  The Taulbee survey this year confirms a rise of about 10% in the last year.  That level of increase is significant, but nothing close to what we’ve seen.  At the CSTA meeting last Friday, representatives from Harvey Mudd and the University of Texas at Austin both reported annual increases in the 20-30% range.  If those numbers are representative, we’re going to need to shift our thinking quickly from how to build demand over how to build capacity.

Even at a well-endowed institution like Stanford, growth rates of the sort we’ve seen this year are significantly stretching our resources in terms of faculty and teaching assistants. Although there will be growing pains, I’m sure we’ll be able to cope. The more vexing question is what happen if this resurgence of interest spreads to institutions that are facing shrinking budgets, many of which have recently been downsizing their computer science faculty and closing departments?

I don’t have an answer for this question, but I do have a nagging fear that grows out of my early years as a professor in the 1980. The enrollment bubble of the late 1990s created some problems for computer science departments, but I think in retrospect that they weren’t so serious. If nothing else, the enrollment pressure subsided of its own accord. After the tech crash in 2000, students moved away from the field, relieving the pressure before it had time to build to a dangerous level. Those students whose primary motivation was wealth-seeking found other pursuits; many managed to get in on the rise of yet another bubble, this time in the financial sector.

Things were quite so easy in the earlier growth spurt that occurred at the beginning of the 1980s. In 1999, I published an essay in SIGCSE Inroads (http://cs.stanford.edu/~eroberts/papers/ConservingSeedCorn.pdf) in which I looked back on that crisis, which reached its peak when I was chairing a brand-new computer science department at Wellesley College. As I indicate in that essay, with support from various NSF studies conducted at the time, academic institutions were completely incapable of responding to the pressure of rapidly growing enrollments. The collapse in student interest that occurred in the late 1980s stemmed not from a failure of demand, but rather a failure of capacity. This boom feels more like that one, and I fear the fall may do so as well.

It seems to me that we need to determine if this growth does represent yet another sea-change in the altogether-too-cyclical history of our discipline and, if so, to find some way to respond before we find ourselves drowning in an ocean of students that we have no capacity to support.