Plenary Panel II:
An In-Depth View of Computer Science
Marye Anne Fox (Moderator)
Chancellor, North Carolina State University
The next panel will focus on computer science and in particular on women and information technology from both an academic and industrial perspective.
Dr. William Wulf is the President of the National Academy of Engineering and Vice Chair of the National Research Council. He is on leave from the University of Virginia, Charlottesville where he is the AT& T Professor of Engineering and Applied Sciences. Among his activities at the university were a complete revision of the undergraduate computer science curriculum, research on computer architecture and computer security, and an effort to assist humanities scholars to exploit information technology.
Dr. Wulf's distinguished professional career has found him serving as Assistant Director for the National Science Foundation, as Chair and Chief Executive Officer of Tartan Laboratories, Inc. in Pittsburgh, and as professor of computer science at Carnegie-Mellon University. He is the
author of more than 80 papers and technical reports, has written three books, and holds a patent.
Dr. Lilian Wu is Consultant to Corporate Technical Strategy Development at IBM and a member of the President's Committee of Advisers on Science and Technology. Dr. Wu received her Ph.D. in applied mathematics from Cornell University and her bachelor's degree from the University of Maryland at College Park. Her major research interests are in mathematical modeling and risk analysis in business, particularly in the electric power industry, in women in science and engineering, and in energy and ecosystems.
She serves as the Director of the International Institute of Forecasters and is on the Advisory Boards of the National Institute for Science Education and the Douglas Project for Women in Math, Science and Engineering at Rutgers University.
THE DECLINING PERCENTAGE OF WOMEN IN COMPUTER SCIENCE: AN ACADEMIC VIEW
William A. Wulf, President
National Academy of Engineering
An alternative title for this paper might have been “An Exercise in Self-Flagellation.” It is not a happy story I am about to tell you. It is not happy because the ending isn't very happy, and it is not happy because we don't have any explanation for it.
I will present some data on the numbers of degrees in computer science (CS), and then I will present findings from one particular study.
The total number of baccalaureate degrees in computer science awarded peaked in the mid-1980s at about 40,000 (Figure 4). Almost coincident with that, preceding it by a couple of years, the percentage of undergraduate women in computer science peaked at 38 percent of total computer science degrees.
Unfortunately, nobody knows why either of these phenomena occurred. We don't know the reason for the increase to the large number of baccalaureate degrees, accompanied by the increase in percentage of women, in the years leading to the 1980s. Nor do we know the
reasons for the subsequent declines in both the total number of baccalaureate degrees and in the percentage of women.
There are many anecdotal explanations. One thing to understand is that at the time, the field was overwhelmed with undergraduate students. In a visit to the University of Maryland during this period, the computer science department had between 20 or 25 faculty members and 1,500 undergraduates. They were just inundated with students.
When the numbers started to decline, nobody thought to ask why. There was just a sense of profound relief.
In particular, no one noticed that the number of women was declining. We were all very proud of the fact that we were approaching 40 percent women undergraduates, and I think we thought that was the natural order of things. So, as things started to drop, surely they were going to go back up again. It didn't happen, and nobody studied why.
Unfortunately, data from the last few years just isn't available yet. In particular, data on incoming students as opposed to graduating ones would be interesting, because the anecdotal evidence is that we are once again experiencing an explosion of undergraduates.
At my own university, the University of Virginia, the number of undergraduates in computer science has roughly quadrupled in the last six or seven years. The same phenomenon is occurring at other colleges and universities. I don't have even anecdotal data on
what is happening to the percentage of women, although I am going to tell you about a study that is not encouraging.
Some of the anecdotal reasons given for both the increase and the decline in total number of baccalaureate students note that this is more or less coincident with the release of the personal computer. This explanation posits that the hype associated with IBM's release of the personal computer in 1980, or the availability of machines to a larger number of people, piqued interest. If this is the case, then the current increase in interest may be due to the Internet.
Another explanation is that students thought that if they wanted to have anything to do with computing, they had to be a computer science major. According to this theory, they really were more interested in applying computers to physics or business, but they thought they had to be a computer scientist to do that. The corresponding explanation for the decline is that they realized that it is not true.
Some of my colleagues would like to promulgate the idea that the decline is because students figured out computer science was hard, and that it wasn't an easy major.
The data on Ph.D. students is somewhat different. The total number of Ph.D.s graduated per year has climbed to about 1,000 (Figure 5). As with the baccalaureate degrees, cause and
effect are not known, but a report issued about 15 years ago stated that 1,000 computer science Ph.D. students per year was about the right number, and something in the system seems to have converged on that number.
The percentage of women continues to climb a little—a very little. We are seeing an effect here of the law of small numbers. We have lots of fluctuations, and it may actually be flat.
Interestingly, the number of women taking CS courses in high school is almost identical to the number of young men taking CS courses in high school (Figure 6). Broken down further into Advanced Physics (AP) or various kinds of advanced CS courses, again, the numbers are nearly equal. Yet there is this continuing drop from high school to baccalaureate, to master's, to Ph.D., and to faculty. The numbers for associate and full professors may be related more to the population of women that were available in the cohort. Although it's difficult to discern, we should be concerned that those small numbers may be the result of discrimination in promotion and tenure decisions.
Another issue is the distribution of women in baccalaureate programs in computer science. Currently, the percentage of women receiving baccalaureate degrees in computer science is about 28 percent. That percentage is not uniformly distributed among departments.
An interesting phenomenon about the discipline is that some CS departments are in engineering schools, some are in colleges of arts and sciences, and some are even their own college.
Of the CS departments that are in engineering schools, some are separate departments and some are combined with electrical engineering. For the years 1991-1993, there is a substantial difference between the number of degrees awarded to women from departments that are in engineering schools (23 percent) and the percentage awarded from departments in colleges of arts and sciences (over 28 percent; Camp, 1998). There is no definitive explanation for this finding.
Now, I will discuss the findings of one particular study. Because at least anecdotally the number of bachelor's degrees is again on the rise, this time perhaps we will collect some data to understand what is actually occurring. The study in question, at Carnegie-Mellon University, attempts to do that (Fisher, Margolis, & Miller, 1997).
The researchers are not looking at the number of women who enter computer science, but rather at the number who entered, and then for one reason or another transferred out. Since the researchers don't know who didn't apply, they are using the transfer-out rate as a surrogate for why students didn't enroll in the first place. The answers they come up with are a little bit different than the usual explanations. One of the things that they can verify is that the nature of the interest of women in computing is quite different from that of men.
Men seem to be interested in computers per se. They are fascinated by the device, the programming, and by the mathematics involved.
Women seem to be much more interested in the application of computers to other things. Their real interest is in some other area.
Second, there is the issue of confidence. Although the authors don 't have data on this yet, they are studying this issue and may be able to document that there is a real crisis of confidence that happens with women in the first 2 years of their undergraduate programs. In some sense, that is a positive statement, because this is something we can do something about!
Third is this notion that the undergraduates have that you must do all of your work between midnight and 5 a.m., and live on Twinkies and Coke. That doesn't seem to be particularly attractive to women.
Why do men and women get interested in computing? There is a marked gender difference in the number who have been programming since they were very young and expressed an interest in computing per se (Figure 7). These data are reversed when they express an interest in using computers to do something else.
Men really seem to enjoy computing per se (Figure 8). Women, on the other hand, see this as a much more safe and secure way to gain employment. Women have also been much more influenced by other people, fathers, very often. The notion that computing can be used to do other things tends to be more important to women.
The most troubling and bizarre issue in all of this is the issue of women's lack of confidence in their abilities. The objective measures are that women do just as well as men. There is no difference in grade point average and no
• Programming since they were young
• Expressed enthusiasm for computing
• Interest in CS expressed in other areas (e.g., teaching)
Source: Fisher, Margolis & Miller, 1997
FIGURE 7 Interest in computing.
discernible difference in performance. Yet, when interviewed, the women consistently believed that they were not doing as well.
Computer science is unique in that interest in the discipline has varied markedly in a short period of time, and that the driving factor for interest appears to be different for men and women. We need to gather data to allow us to understand the reasons underlying both the fluctuations in interest and the different motivating factors that attract students to computer science. With such an understanding,
we may be able to keep this young field growing, one which enjoys the contributions of the best minds, to provide the best solutions that computing can offer.
Camp, T. “CS Programs in Engineering Colleges—Fewer Females,” Journal of Women and Minorities in Science and Engineering, vol. 4, no. 1, pp. 15-25, 1998.
Fisher, A., J. Margolis, and F. Miller. “Undergraduate women in computer science: experience, motivation and culture,” in ACM SIGCSE Technical Symposium, February 1997.
WOMEN IN INFORMATION TECHNOLOGY: A VIEW FROM INDUSTRY
Lilian Shiao-Yen Wu
Consultant to Corporate Technical Strategy Development, IBM
President's Committee of Advisors on Science and Technology
I would like to give my thanks to two people. One is Jong-on Hahm who is the director of the Committee on Women in Science and Engineering; she has been invaluable in bringing together data and research in this area. The other person is my husband, Ralph Gomory. He has been my partner in my work on women. Today's talk in many ways is jointly ours.
My characterization of how women are doing in information technology and computer science in industry is that there is good news and bad news. In the previous presentation describing the declining percentage of women obtaining computer science degrees, Dr. Bill Wulf has given you the bad news that the percent of women receiving computer science B.S. degrees has been falling, and the percentage receiving M.S. and Ph.D. degrees has remained essentially around 25 percent and 15 percent. Also, during the very important K-12 period, a recent study by the American Association of
University Women shows that there is already a technology gap where fewer high school girls take advanced computer programming classes. They enter the classroom with less prior experience with computers than boys, and they are not as confident about their technology skills.
The good news is that the information technology or IT world is a terrific place for women. There are a large number of careers in IT that women both enjoy and are very good at. There are careers that are about going deeply into how the computer works, but there are also careers which involve understanding how computers can be used in innovative ways to touch our lives.
There are a number of reasons for the lower numbers of women. I will touch on two. First is what is taught. Many of the courses that a student would take to get a computer science degree are centered around the computer itself, its hardware, its operating systems, or how a compiler operates. It is not about using a computer to affect people's lives. This lack, for some women, makes computer science less attractive than other careers.
The second reason for the low numbers can be seen if we look at individuals who are successful in computers—they are portrayed in the media as nerds and geeks, people who are obsessive about getting the computer to do intricate things they want. This is not a particularly attractive role model for young girls. As a consequence, fewer girls aspire to be in computing, too many shy away from computers, and they think that because they are not nerds they are not as good at computers as boys. We then have fewer women with computer science degrees.
Now for the good news. There are two parts. The first is, despite their small numbers, women have made many fundamental contributions to the science of computers. I will briefly describe the contributions of four women: Brenda Baker and Margaret Wright from Lucent, and Fran Allen and Diane Pozefsky from IBM.
Brenda Baker has contributed to the science of computing in analysis of algorithms. She has discovered techniques for efficiently computing approximate solutions to NP-complete problems (i.e., problems that are extremely difficult to solve exactly) involving planar graphs, scheduling, and other areas of combinatorial optimization, and she has invented efficient algorithms for use in routing wires on chips, guiding robots, and finding duplication or patterns in text and software.
Margaret Wright has contributed to the science of computing through algorithms and software for numerical optimization. She has invented, theoretically analyzed, and implemented original algorithms for nonlinearly constrained optimization, including widely used sequential quadratic programming methods, and more recently, direct search and interior-point methods.
Fran Allen co-invented the framework, which has been the scientific basis for all optimizing compilers. She perceived the need to determine at compile-time how values flow through a computer program. Her paper with John Cocke, “A Catalogue of Optimizing Transformations,” codified for the first time techniques that could be used to improve the running times of programs written in different computer languages and for different machines. These are still the transformations in use today.
In the early eighties, the networking technology then used by the majority of large corporations was designed for networks that rarely changed and were tightly managed. Driven by the introduction of mini-computers and personal computers, and newer connection capabilities that made it easier to add and move computers, Diane Pozefsky's work transformed the static networking technology to one that allowed networks to change with less administrative complexity, and to let network traffic adapt based on current traffic patterns. Her work has allowed networks to grow and provide immediate connectivity when new users and machines were added.
The second part of the good news is, I believe, that information technology in industry is turning into a good place for women. IT is becoming pervasive, more complex and intertwined with the telecommunications, media, and entertainment industries. It is in our homes and where we work.
As a consequence, there are careers both for people who care deeply about the uses of computers in our lives, as well as for those who care passionately about the science of computers themselves, and for every shade of interest in between.
The IT industry is moving extremely fast, e.g., we talk in terms of Web years which is 3 months. To invent, at this speed, requires working closely with customers to understand what is needed so we can find better and original approaches. Many women both enjoy and work well with customers.
A person needs some basic understanding of the capabilities and limitations of the technology but the challenge is often in being very creative in using the technology. This often involves bringing together engineers, design artists, people in IT, people who run networks, and business professionals who know the application.
One example is an IBM project that created the Internet trading operations for Charles Schwab. E-Schwab is now the largest U.S. online and discount broker. This web site regularly gets 30 million hits a day and has gotten as high as 55 million hits in one day. A project like this starts by getting the infrastructure, architecture, and technology right, including areas such as privacy and security. But just as important is to understand the customer's needs so that the application is what is really wanted and is easy to use. This requires working with and understanding business professionals and their needs.
Another example is one I have worked on. In the newly deregulated electricity industry, utilities are changing their way of doing business from a monopoly to an open market where consumers can choose who will supply their power. As monopolies, utilities dealt primarily with the engineering problem of supplying electricity in an area without interruption. But in today's open market, a utility must be concerned with other companies selling cheaper power to its customers. In this world, I have had to work not only with power engineers, but also with Wall Street traders and quants, so that the capabilities of the physical network to provide power and the optimal pricing of that power connect. My observation is that many women both find this type of work interesting and rewarding, and are good at finding inventive solutions.
Often more creative and daring ideas and solutions happen when people with different backgrounds and experiences work together on problems. I would like to illustrate this through a few examples coming from IBM's Research Division and the Women & Technology Institute, which is a nonprofit that works on technology designed for women by women. One IBM design was a small handheld device where an icon would appear if an out-of-town friend was within “being able to get together for dinner” distance. You could then send e-mail or call the friend by poking at the icon. Another design was a handheld device so that a child could add to mom's shopping list. An idea from the Women & Technology Institute was a very large video conferencing screen that could be put on the dining room wall. Then a family could have Thanksgiving dinner with other family members across the country.
All of these are examples of what I would call “person-to-person technology.” They all have a bigger interpersonal component than the technology of today.
Companies appreciate the benefit of this different point of view. This together with women's ability to work with customers and in groups helps produce an atmosphere in industry that is often one of genuine support.
Before I close, I would like to mention that I believe the public rankings of companies on where they stand in areas important to women have also played an important role to advance women in industry. For example, each year companies are ranked on their family and child support policies by publications such as Working Woman, and the nonprofit organization Catalyst publishes the number of women on company boards. These rankings help hold companies accountable and true to their words.
In closing, my message is that a curriculum centered around technical understanding of the computer and computer software pushes away too many girls and women, although it does not push away some women who remain involved and excel. But many careers in information technology in industry are centered around the uses of information technologies. They involve both technical knowledge and the ability to work in groups with people from a variety of professions. Many women enjoy this work and do well.
I believe recognizing this, publicizing these careers, and adding courses which better reflect the broad possibilities in this industry will be interesting to more girls and women, and will better prepare them for a successful career in industry.