Challenges for Computing Research
Advances by computer science and engineering (CS&E) researchers have, over the past forty years, changed the world. Similar opportunities still exist, but excitement is tempered by challenges beyond our control. We explore issues facing our field and describe efforts by NSF’s CISE (Computer and Information Science and Engineering) to better understand future opportunities and also to maximize the impact of current resources.
During the past ten years there has been unprecedented growth in our field. The community of researchers has grown dramatically both in size and in the scope of research undertaken. Our undergraduate and graduate programs are the envy of the world and attract many of the best students from around the globe. Important discoveries have led to major advances in a wide variety of areas. There has been a blossoming of cooperation between computer scientists and colleagues in other science and engineering fields. Our research has continued to spawn new industries that account for a significant fraction of the growth in the US economy. So overall, for computer science and related disciplines, the recent past has been a time of scientific vitality and significant impact on society as a whole.
With this incredible vitality and growth have also come problems that potentially threaten our ability to maintain US leadership over future decades. While our domestic community has grown and has welcomed a wide variety of “non-traditional” colleagues, the resources needed to support the wide gamut of potentially high-impact and world-class research have not kept pace. This is due to the increase in the size of the community and also to factors beyond our control, including US government budget constraints, retrenchment of industrial research support, and re-orientation of research away from universities by various agencies.
Another issue beyond our control is the increasing competition for international students. With the upsurge in the economies of countries such as India and China and the attendant professional opportunities for students who remain at home, increasing numbers have opted for local universities. In addition, the difficulties facing foreign students trying to enter the United States since September 11, 2001, have sometimes made universities elsewhere (e.g., Australia) more attractive. If current trends continue, our cutting-edge industries may not be able to acquire sufficient numbers of first-rate graduates, government may not have the people and technology it needs, and, worst of all, the most innovative ideas, products, and services may spring from organizations elsewhere.
There are also problems that we have created ourselves. Proposal submissions have soared in recent years. This has led to lower success rates and to a tendency by researchers to submit even more proposals, leading to even lower success rates. As part of this process, CISE Program Directors have, at times, reacted by cutting budgets so larger numbers of projects can be funded. Ideally, CISE could fund all truly outstanding research and education projects at levels that are justified by the scientific needs of the project. Unfortunately, budget constraints have made this difficult to achieve.
Some of the growth in submissions is due to the increasing numbers of researchers, the broadening of the scope of CISE-funded activities, and the introduction of cross-cutting programs such as ITR and Cybertrust. However, much of the problem has resulted from the currently common strategy of submitting multiple proposals in a year. Given that resources have not expanded to match this increased submission rate, multiple submissions have played a major role in the lowering of success rates.
We also need to address the incredible workloads imposed on CISE investigators and on NSF personnel as a result of the large number of proposals submitted. Researchers today spend a significant fraction of their time on proposal- writing. NSF personnel are overloaded by the number of proposals, including the difficulty in identifying top-notch panelists who do not have conflicts and the demands of providing a high-quality review process. Indeed, this topic and the related issue of low success rates seem to be a major concern of many in the community.
Given the above analysis, it appears that constraints on multiple submissions by individual researchers will go a long way in addressing both the overload issue and the perceived drop in funding rates. For example, we might limit researchers to, at most, one research proposal per year per solicitation. Another option might be to disallow submissions to a solicitation by those who have been successful in the prior year’s competition. Still another might be to limit the number of research proposals submitted to CISE in a given year. These, together with the change to annual solicitations, have the potential to cut the number of submissions and increase success rates. An associated benefit will be the extra time that researchers have to work on research instead of proposal-writing. It also will make it possible for CISE Program Directors to play a more active role in managing the scientific programs for which they are responsible. Some combination of these mechanisms is under consideration.
One further change that we have made in recent solicitations is the introduction of clusters. Each cluster is a collection of related areas. In the past, each program had a small budget and it was difficult to move funds between individual programs. A principal motivation of the cluster approach is to make it possible to fund the best projects submitted to the cluster, regardless of area. This may entail, on a year-to-year basis, more or less funding for a particular area. Limiting submissions to clusters also has the potential to deal with the workload/success-rate issue.
One positive factor is that, despite the FY2005 budget cut experienced by NSF, CISE received a (small) increase in its budget. In addition, while the ITR program has ended, the associated funds remain in the CISE account. As ITR commitments end, some of these funds will be channeled into CISE divisions to enable growth in core CISE programs. In addition, some will be used to introduce new cross-cutting emphases such as the current CyberTrust, Information Integration, Science of Design, and Broadening Participation in Computing.
We also need to consider what proactive steps we can take to ensure that the United States maintains its research leadership in computing. To address this, we have initiated a study of the scientific vitality of research areas. The first phase of the study involved developing a taxonomy of CISE areas. This was done cooperatively by the CISE Program Directors. The second phase involves asking researchers to assess the state-of-the-art in their areas, including past achievements and future opportunities. Many of you have agreed to participate in this study, and for this you have both our thanks and those of your colleagues. The third phase of the study will be to convene panels of prominent individuals from academia, industry, and government. These panels will be given the information obtained in the second phase. They will be asked to provide evaluations of areas for consideration by CISE. It is hoped that, in the aggregate, the above information will be helpful in a variety of ways.
While much of the tone here may sound negative, there are positives, including those cited in the first paragraph. At the present time, the glass is certainly more than half full. The challenge for all of us going forward is to maintain our global leadership in computer science and engineering research and education, while adjusting to the external constraints over which we have no control.
Let us close our final CRN article for this academic year by thanking all of you for your help during the past year. Your willingness to serve as panel members and advisors, as well as your feedback on the full array of issues affecting CISE, is greatly appreciated. As is always the case, we welcome your comments on this article. Have a productive and pleasant summer!
Peter A. Freeman (pfreeman [at] nsf.gov) is Assistant Director and Lawrence H. Landweber (llandweb [at] nsf.gov) is a Senior Advisor, both in NSF’s Computer and Information Science and Engineering (CISE) directorate.