Fixing a busted IT research system

James Foley is worried.

As chairman of the Computing Research Association--a group made up of academic departments, research centers and professional societies--his job at CRA is to improve computing research and education. But Foley sees troubling trends in the nation's system for nurturing and training new information technology scientists.

The number of doctorate degrees awarded in the United States has dropped not only in computer science and engineering, but also in noncomputer science and engineering fields in general. And top U.S. undergraduate computer science departments are seeing enrollments fall.

Some industry analysts argue that the country already has a glut of Ph.D.s. But to Foley, also a professor at the Georgia Institute of Technology College of Computing, the educational declines may very well contribute to an economic malaise. He wants to excite youngsters about computers, in part through better-trained teachers. Foley would also pump up federal research funding and give young scholars independent funding.

CNET News.com recently spoke with Foley about computer science education, the flow of programming work offshore and how the computer science profession in America can weather the trend toward offshoring.

Q: The number of science and engineering Ph.D.s awarded in the United States has been falling, from 27,300 in 1998 down to 24,550 in 2002. How big a deal is that decline?
A: One way to look at is in the context of what other countries are doing. The per capita production of engineering students in the United States is a lot lower than in other countries, including Korea and Japan and China and Finland. The per capita data is a little dangerous because we have different bases, and the U.S. has a big population compared to Korea. But on the other hand, compared to China, our population is small. So there is a larger base of smart people to draw from than in the United States. That just says that there is going to be a potential long-term issue.

A major part of our economy has been built on scientific advancement--computers, telecommunications, planes, the whole Internet and increasingly, the bio-world of medicine and life sciences and genetics and genomics and all that. So if other countries are increasingly stronger than we are in the technical base, the economic results of the technical base are going to fall behind after a while. So I think it's a big deal because of the potential threat to our economic strength.

What about the argument that it's hard for a lot of Ph.D.s to get good jobs these days, and also that there is a long period of time in which science Ph.D.s aren't getting independent funding. Isn't that a sign that the United States has a glut?
There is this kind of disconnect in life sciences, physics and chemistry, in which you have to do a post-doc for three or six years before you get a faculty position, which is not the case in computing.

In computer science and related fields, that hasn't been necessary, because there has just been the demand for computer science professors. I frankly cannot make sense of it, because there was a great ramp up in funding for the National Institutes of Health--from seven years ago to two years ago, it doubled--and a lot of that money goes into universities for research. I've got to conclude that the money is supporting lot of post-docs but is not supporting a lot of new faculty positions.

Is that an unwise approach? The argument has also been made that we aren't paying good-enough salaries for some of these federally supported scientists. Do we need to have more fully funded scientists?
We need to provide direct funding to new graduates rather than having them working under the wing of the more senior scientists. There are a lot of benefits to having a senior mentor guide you. But that creative urge and that thinking out of the box is particularly strong early on, and we should be encouraging it by giving direct funding to new and recent Ph.D.s--which does happen in computing, by the way.

The number of Ph.D. degrees awarded in computer science and computer engineering in the United States and Canada, according to CRA's survey for last year, totaled 877--up 3 percent from 2002 but was still the second-lowest since 1989. What does that say to you?
During the boom days of the late 1990s, some students were being distracted by dollar signs in their eyes. I think that we had this phenomenon, in which fewer students were going to graduate school. If you say the average time to get a Ph.D. is in the six to seven-year range, that would go back to then--let's say '96 to '97, when the boom was really starting up.

The good news is that the number of students passing their qualifying exam, which is the first step toward a Ph.D., has been going up in the last couple of years. So that's a positive sign for the future.

Numbers have been down in a time when computing and information technology are more and more central to everything that we do, not just in our everyday lives, but also in research.

In research spanning from computer sciences to life sciences, medicine, physics, chemistry and health care, computers are more and more central.

So here we are with a technology that has been recognized as increasing productivity and therefore national economic competitiveness. And what we are seeing is fewer Ph.D.s. So that's a problem. And we are seeing a smaller or a flattened-out government investment in computing research, and that's also a problem, given what we know about the importance and centrality of computing to economic competitiveness.

Let me ask you about a comment I heard from Peter Lee, associate dean at Carnegie Mellon University. He said one of the problems in computer science is that the field has been, in some ways, a victim of its success. Computers have become so practical to daily life that the big questions surrounding them have kind of taken a backseat. The field hasn't promoted the idea that these machines can help us improve our intelligence or move us ahead. Do you think the field has been imaginative enough?
I think that we have, over the last five to 10 years, been too worried about short-term things. And I am going to lay that back at the doorstep of the review process of proposals at National Science Foundation and also on the tenure process, which puts a lot of emphasis on publications.

This fall, there are just less than 200 undergraduate majors in the electrical engineering and computer science departments at the Massachusetts Institute of Technology. That's down from about 240 last year and roughly 385 three years ago. Does that concern you? Could you comment on what's happening at Georgia Tech along those lines?
It does concern me. At Georgia Tech, enrollments are up a few percentage points, after in the past having had the same kinds of declines. And I believe that at other schools, there is kind of a mixed story. It's somewhere between beginning to see the end of the decline and actually seeing the end of the decline.

So you expect an uptick in the future in undergraduate enrollment?
I think we will have an uptick.
Why?
We have had kind of a perfect storm in computing in a negative sense over the last 3 years. We had the dot-com crash, we had 9/11 and we had the big offshoring hullabaloo, all of which in one way or another have had negative effects on enrollment in computing.

Is the 9/11 issue related to foreign students?
Yeah, that's mostly the foreign-student issue--students not being able to get visas or choosing to go to other countries where they know there is less of a hassle.

At the graduate level, you have mentioned that you hadn't been seeing the quality of American student candidates that is needed to get into these programs and succeed. Is that something related to not attracting the best and the brightest of the American students, or is it something about how Americans don't have the same proclivity or skill in computer science as, say, people from India or China or Taiwan?

I don't think it's all about proclivity and skill. I think computing is seen as a hard discipline, and I do believe that we do not get into computing or into technology as many of the best and brightest as we need. One reason is that technology has historically been, to some extent, an upward-mobility path.

That's interesting. So it is tending to attract immigrants?
That's right. And, we don't have as many economically disadvantaged folks in the United States as we once did. So that's one element, though I don't think that's the whole story. Law, medicine and business, in some ways, are seen as more interesting.

The other element of it goes back to education and high school--this is a pretty well-documented problem--with having low-qualified science and math teachers in high school. So if math and science are being taught by individuals who are well-meaning but don't have enough of a background, then they are not going to make it be as interesting and exciting as it can be and as it is. The intellectual and emotional excitement that helps kids decide to go to school in science and technology won't be there. So that goes back to how we spend our money with high-school education, incentives for teachers, pay for teachers and all that.

Let me then ask about the offshoring issue, because a lot of people would look at it and say, "It doesn't make sense to get into computer science and then to become a programmer."
Right. It does not make sense to become a programmer. But there is a lot more to computer science and computing than programming, and that's part of our challenge. There is this stereotypical image that computer science education leads to heads-down programming jobs, and it's those heads-down, isolated-from-the-problem jobs that are going to some extent offshore. I think the trend of pure programming jobs will continue to go offshore, because in many ways, our computers and communications technologies enable that to happen.

What are the kinds of computer science jobs that make sense for the future--that are going to be, to some extent, offshore-proof?
The key is big-picture design--what you would call system architecture or system design. It's understanding end-user needs and translating them into the detailed specifications, designs and architectures that can and will be shipped offshore.

It's what we call the user-facing, or customer-facing, aspects of computing, which is sometimes characterized as "computing plus X." We are emphasizing a lot more with our students that they need to understand something besides computing--like business, biology, chemistry, mapping, geography, information retrieval or history. Like anything in addition to computing, because the big win with computing is that you use computers to do things. And to be a creative computer architect or computing systems architect, you need to understand "X" as well as computing.

Taking the temperature of computer science and computer science research, are you optimistic or are you pessimistic?
I am very optimistic. Wearing my hat as chair of CRA, I am seeing a lot of universities understanding this needed change in emphasis, which has been going on at some schools for quite a while.