Computer Science Enrollments: The Real News

July 11th, 2008 by Peter Lee Post a comment »

I regularly am contacted by reporters who read the CRA “Taulbee Survey” and inquire about the current state of computer science undergraduate enrollments. Here’s what I said last night to the most recent reporter who inquired:

The Taulbee Survey “headline” this year was (roughly) “computer science bachelors degrees drop again.” In my view, this is not news — it was entirely predictable from the legitimate headline four years ago: (roughly) “freshman interest and new enrollments drop again.” The actual news right now in the CRA data is that freshman interest and new enrollments seem to be stabilizing and turning the corner — starting to trend upward. “Degrees granted” is a lagging indicator — it lags freshman interest by 4 years. The fact that the number of bachelors degrees granted this past year decreased is not news — anyone could have looked at the freshman interest data from 4 years ago and told you it was going to happen.

The natural question is “What’s responsible for these oscillations?” There are a number of factors. I want to be really clear that I am not in denial about various substantive areas in which we need to continue to work to improve our field and its attractiveness. But by far the most important factors are (a) the job market (or people’s sense of the job market), and (b) the level of “buzz” associated with the field.

Let’s start by considering graduate enrollment, rather than undergraduate enrollment. For the past 15 years, the number of Ph.D.s granted annually in computer science has been in the 900-1100 range. Suddenly, though, in the past 2 years, it has climbed to 1800. Why is this? The answer is totally obvious:

  • In 2001, lots of startup companies went bust.
  • This dumped onto the job market a number of the best bachelors graduates from a few years before, who now had two or three years of experience under their belts.
  • This made it hard for some excellent new bachelors graduates of 2001 and 2002 to get the super-exciting jobs they had anticipated — they were competing with people whose academic records were every bit as good as theirs, but who also had 2 or 3 years of experience working at a hot startup.
  • Because these great new bachelors graduates couldn’t get exciting jobs, they went to graduate school instead.
  • And, mirabile dictu, 6 years later, they’re emerging with Ph.D.s.

This is not a news flash — it didn’t take a genius to predict, a few years ago, that it was going to happen, and it doesn’t take a genius to explain it, either.

Similarly for bachelors degrees. Starting in about 2002, there was lots of news about the tech bust. Tech was no longer sexy. Jobs were no longer plentiful. Subsequently, there was a lot of misleading information about the impact of offshoring. And the newspapers never bothered to report that by late 2004, US IT employment was back to the 2000-2001 level — we had fully recovered from the bust — somehow that wasn’t considered newsworthy. So it’s not surprising that interest in bachelors programs decreased sharply, and that 4 and 5 years later, the number of degrees granted precisely mirrored this decline.

Also, it’s not surprising that things are turning around. Google is hot. Tech in general is hot. There are startups everywhere. It’s clear to anyone that there are plenty of jobs. (By the way, given the incredible state of today’s bachelors job market, it doesn’t take a genius to predict that the number of Ph.D. graduates in 2014 will show a decline. When you read the scary headlines 6 years from now, remember that you heard it here first!)

So, what about the enrollment at my university — the University of Washington — and enrollments vs. projected need nationally?

At UW, our enrollments are stable, because they are limited. We take 160 new bachelors students every year. This is far lower than the demand. So changes in the level of interest result in changes in the proportion of applicants that we must reluctantly turn away.

One place where we can easily measure changes in student interest is in the enrollment in our first introductory course, which serves the entire university. Between 2003-04 and 2007-08 (a 4-year period), enrollment in this course is up by 27%. Enrollment by women is up by 45%. (Annual enrollment of women into the major is up by 64% over that same interval.)

Our national peers see this same sort of upturn in interest — not surprisingly, as explained above.

Here’s one more comment about enrollment. There is lots of discussion about computer science enrollment, because it actually matters to the nation! But take a look at this NSF data for bachelors degrees granted in 2004 by all US institutions:

  • Computer science: 57,405
  • Mathematics: 13,755
  • All of the physical sciences combined (astronomy, physics, chemistry, …): 14,240
  • All of the earth, atmospheric, and ocean sciences combined: 3,903
  • All of engineering combined (aero/astro, chemical, civil, electrical, industrial, materials, mechanical, …): 64,675

Now, we know that computer science degrees have decreased since then, but they’re heading back up, and it’s important to keep things in perspective relative to other fields. Nobody is crying in their beer about the death of the physical sciences, yet computer science produced 4x as many degrees as all of the physical sciences together!

The bad news is that psychology and the social sciences generated 220,067 degrees. And the biological and agricultural sciences generated 80,933 degrees. So the physical and mathematical and earth sciences and engineering are not faring well. But compared to the other physical and mathematical and earth sciences and engineering, computer science is not looking like the problem child! (Skill testing question: What was the fastest-growing bachelors major in America the last time I checked? Answer here.)

What about workforce demand? There are several things to say about this:

  • Computer science, increasingly, is great preparation for all sorts of careers. That is, lots of people get computer science degrees to go to law school, business school, medical school, biotech labs, etc.
  • And even “Information Technology” is much broader than the software industry. 70% of all IT jobs are with “IT consumers” (companies that use it) rather than with “IT producers” (companies that invent it).
  • And the software industry is really hot right now, and also it is really cool — it’s creative, interactive, vibrant.

Here is a spreadsheet with charts showing Bureau of Labor Statistics projections for employment between 2006 and 2016 for all fields in the sciences and engineering (including the social sciences). What it shows is that of all of these fields, between now and 2016:

  • 70% of all newly-created jobs will be in computer science.
  • 62% of all job openings (both newly-created jobs and jobs available due to retirements) will be in computer science.

(The latter is a smaller % than the former because other fields are “older” and thus will have a greater number of retirements.)

There is a huge gap between “people” and “jobs” in computer science — there is plenty of opportunity!

What do we do at UW to attract students? Many many things. As one example, starting tomorrow at UW we’re running an annual 3-day workshop for high school teachers of math and science, sponsored by Google. The goal is to show these teachers that computer science is important to their fields, and is a great field to send their smartest students into. Information is available at http://cs4hs.cs.washington.edu/. (We do this jointly with Carnegie Mellon and UCLA.)

We have a set of terrific videos that illustrate several important points:

  1. People enter the field of computer science for all sorts of aspirational reasons.
  2. People do all sorts of things with their computer science degrees in addition to working in the software industry.
  3. Working in the software industry is highly exciting and creative and interactive.

You can take a look at these videos at http://www.cs.washington.edu/WhyCSE/.

Most importantly, we really invest in our students. Word gets out. At the University of Washington, we have the strongest undergraduates, because students know they can get a great education here.

How do we “calibrate” our program — make sure our students are ready for careers? Here is a Word document I prepared recently for another purpose. Every year we are a top-5 supplier to Microsoft, Google, and Amazon.com — our students are fantastic.

You asked what US companies can do. They can demand greater federal investment in K-12 science and math education, and greater federal investment in research in the physical sciences and engineering (which drives education at research-intensive universities, who are the producers of most of the students that high-tech employers seek). And they can invest themselves — they can be good corporate citizens. Obviously they can run internship programs and things like that — there’s lots of mutual benefit in that.

But most importantly, companies and individual citizens can insist on federal policies that support education and research. Stop pissing money away in Iraq. Stop pissing money away on “security theater” in airports and along our borders. Stop being hostile to talented immigrants. Stop destroying K-12 by obsessing over high-stakes tests that can be gamed. Stop denigrating science. Stop substituting ideology for objective analysis in designing federal policies. Stop all of the disastrous policies of the past 8 years. Return sanity and accountability to our national leadership. Institute policies such as those outlined in the Executive Summary of the National Academies’ Rising Above the Gathering Storm.

Please note that these are not partisan political statements. Historically, support for education, support for research, and belief in a reasonable degree of intellectual integrity have not been partisan issues.

I do want to emphasize again that I am not in denial about various substantive areas in which we need to continue to work to improve our field and its attractiveness — and I don’t think our community is, either. But it’s important to keep things in perspective! Public lamentations are not likely to improve the situation.

Ed Lazowska

  • Pingback: Conspirama

  • http://www.cs.rice.edu/~vardi Moshe Vardi

    I agree with Ed that the decline in CS enrollment seems to have bottom out. What is not clear to me, however, is what the trend over the next few years will be. I don’t think we should expect enrollment to go back to the peak of 2000-2, which, IMHO, reflected the “irrational exuberance” of the dot.com era.

    The real question wrt enrollments is whether the country is educating enough IT professionals. The numbers we see from BLS seem to indicate a huge educational shortfall, but not all IT jobs require IT degrees, so it is not clear that there is a shortfall. We need to get used to the idea that CS enrollments fluctuate widely. This has been the case for the last 30 years. If we had solid idea what the ideal level of enrollment should be, we’d be able to look at the enrollment situation in a more objective way. Moshe

  • http://www.it.usyd.edu.au Sanjay Chawla

    The trend in Australia is similar. From 2002 to 2007 undergraduate enrollments in IT plummeted. 2008 is the first year when UG enrollments are up. During the same period PhD enrollments went through the roof. 2008 is the first year when PhD enrollments have flattened as the first post dot com bust batch graduates.

  • Phil Levis

    Ed has written an insightful piece here. I was wondering what the PhD situation really is, across the country

  • Pingback: CS Enrollments on the Rise? | CSDiary

  • http://www.harding.edu/it/faculty.html Marguerite (Reet) Cronk

    Ref quote – And even “Information Technology” is much broader than the software industry. 70% of all IT jobs are with “IT consumers”

    Are these jobs also increasing?. The BLS statistics demonstrate a high degree of collectivity in job categorization which doesn’t appear to align with either the diverse range of computer/technology related jobs or with the types of degrees offered. At best I think we see a general increase in the overall IT job market.

  • Aurellius Enzo

    As I read this piece I can’t help but wonder in what world Lazowska is living. A little over a year ago I graduated summa cum laude, double major math and CS, from a flagship state university, with 4.0 GPA and a class rank of 1 out of over 130. I’ve been interviewing for 15 months and I have yet to receive an offer for a permanent position (though I have had short-term contract jobs). Meanwhile, in my hometown, the city’s largest employer has outsourced its entire IT operation of about 250 to an global IT services company. Long-time employees were told that they had a choice: accept a position with the IT services firm with fewer benefits and less career potential, or accept two weeks worth of severance pay. That was three years ago. Now, most of the old employees who accepted the transfer have been replaced with new recruits from India. The Indians are easy to recognize around town, since they usually travel in groups of three or four. On Sunday, they play cricket in the park while their wives and children socialize and call out to each other in Hindi. They are very nice people; polite and respectful. I don’t begrudge them at all.

    What I do find frustrating and exasperating is how incredibly out of touch the computer science elite is to the realities of job market _for the average CS graduate_ or even a top CS graduate _from an average school_. The Ed Lazowskas and David Pattersons of the world apparently don’t have a clue about this because what they see, and really what they care most about, is what happens at the very top of the food chain, where already-bloated software companies do battle to devourer the cream of the crop. Of course, the 150 graduates of UW are going to get nice offers. Ditto for those from CMU, Stanford, MIT, UC Berkley and some others. Is there really anything surprising here? But what about the rest of us? What of the remaining 52,405, after the top 5000 have found jobs at Microsoft, Google and Amazon?

    What we need is a basic reality check. In this regard the 2004 undergraduate degree production stats are actually very revealing, but not in the way Lazowska thinks they are.

    The question I ask is this: do we really need twice as many CS grads as all of math, physical sciences and earth sciences combined? And, do we really need nearly as many CS grads as all other engineering disciplines combined? The answer is: of course we don’t. But don’t just take my word for it. The proof is in the combined intelligence of thousands of individual decisions by people who know that they are making one of the most important decisions of their lives in choosing their college major. These folks are not simpletons being “mislead” by media reports of lack of opportunity in CS. On the contrary, they are being educated by these reports and quite appropriately taking them into consideration.

    If I had it all to do over again would I have picked other majors? No, I wouldn’t have. Even if I never get an offer at a software company, I have always loved math and CS (I read Douglas Hofstadter and Marvin Minsky in middle school). But I know I’m in the minority on this one. Most go into CS expecting and indeed needing a job at the other end. For them, this post is a lot of hype and not much help.

  • Glenn Campbell

    Excellent response Aurellius Enzo!!

    While I don’t think the article is total BS, it does show a level of detachment that is disturbing. As a 1997 CS graduate (with honors) who has never held and has been unable to obtain a total CS job, I find some of the authors coments insulting.

    I’m just tired of taking classes to remain current without a reasonable job as reward.

    I have given up taking classes, I just can’t afford it any more.

    What about all of the 90s graduates who have spent large sums of money educating ourselves only to find out, that we can’t get a job because our skills have fallen behind. I can’t afford to spend the money without some assurance that I will be able to get a job. What about ME.

  • Pingback: UW CSE News » “Computer Science Enrollments: The Real News” (Computing Community Consortium blog)

  • Pingback: Cloud Computing Globalization and the American IT Worker | CloudAve

  • http://professional-suggestion.com/ Uninstall Program

    I can’t agree with your points any more.

  • Pingback: Somebody’s Eating Your Lunch « Machine Learning (Theory)