Chemical & Engineering News,
March 27, 1995

Copyright © 1995 by the American Chemical Society.

NSF's computer guru, Paul Young, discusses information revolution

For the past eight months, Paul Young has headed the National Science Foundation's Computer & Information Science & Engineering Directorate. The programs within that directorate are aimed at keeping computer, information, and communications research at the cutting edge of discovery. NSF's activities are part of the larger High-Performance Computing & Communications Initiative, which consists of several agencies that perform and support research in high-speed technology. From his position, Young is able to oversee the totality of computer-communications activities in both research and applications. Before coming to NSF, Young, whose field is computational complexity and its connections with mathematical logic, was dean of the college of engineering at the University of Washington, Seattle. In a recent interview with C&EN, he spoke about his activities and concerns.

What is the state of coordination of the field of high-performance computing and communications among the different agencies?

All the agencies in the High-Performance Computing & Communications Initiative are represented on the HPCCI Coordinating Committee. They talk about common programs to make sure the programs are lined up properly with each other. That process, while time-consuming, actually works pretty well. It's also true that at the other level, the Information Infrastructure Task Force, which is [run] out of the Commerce Department, has a lot of people from research programs sitting on its committees.

What sorts of activities are involved?

The Computers & Information Committee is involved in planning for research programs across the field. So it is responsible for interagency communication and coordination, research agendas - quite a bit. I think the whole program is working well, but the proof of the pudding won't be clear for a couple of years. Strategic research plans are being developed.

How does all of that interplay with all the other bodies within NII?

They all want to discuss the importance of developing information technologies in terms of helping them with their more basic charge. Data is often a big issue. There's a lot of it out there. They want to manipulate it in different ways. They want to understand how to integrate these different modalities. We often don't know how to represent the semantic content of things very well. Talking just about text is not so bad. But in talking about things like understanding and integrating pictures and text, the methods of doing that are not fully understood.

HPCCI has changed its orientation a few times within the past couple of years. It had consisted solely of the so-called grand challenges involving the high-speed technology. But now the emphasis seems to be on the broader national challenges that extend the field to the more general problems of society.

The national challenges are programs that correspond to the grand challenges of science and engineering. But the grand challenges for science and engineering tend to be more traditional science and engineering with very high computational requirements. The national challenge problems are those broad programs like the digital libraries program or telemedicine, which are probably going to rely on communications more than computation and which at the same time are going to directly involve many more people. I want to be careful how I say that because a lot of the programs in high-performance computing may have very broad impact even though few people are working in them.

Policies do shift. As communications get better, it is possible to shift to programs that have broader impact, as the NII sorts of programs have. I think HPCCI and its predecessor have transformed the way science and engineering gets done. We used to speak of theory and experimentation as fundamental ways of doing science and engineering. Now there's a third mode, computation. It's different. It's new. And that has pervaded a lot of scientific areas.

How do you follow, in your busy life, the technical developments in the field, their social and policy implications, and manage at NSF the many important technical areas as well?

I serve on the High-Performance Coordinating Committee of the National Science & Technology Council [the Administration's interagency coordinating body for science and technology]. NSTC has nine subcommittees. One of them is the Committee on Information & Communications [CIC], which oversees the high-performance program through that coordinating committee. CIC has a broader agenda than HPCCI because it includes everything of relevance to computer information and communications. And, of course, I read a lot.

Would you go so far as to say that this whole field of computers, communications, and information underlies the next social and economic infrastructure of the country?

I would. I think that there are predictions of how rapidly that would occur. I think that a lot of those predictions are much too short term. But it's already transforming the way we do major portions of science and engineering. We often do computations that in some sense replace physical experiments. I tend to call those efforts experiments, although somebody a little purer in terms of the traditional approach would say that experiments would have to actually touch the physical basis of science and engineering. In a good sense, though, it's experimental. It's not theoretical in the sense of mathematical analysis.

Would you also say that the information infrastructure the scientific and engineering communities enjoy is a prototype of what the general information infrastructure will be like for citizens and communities?

Yes, I would. You go to a college campus and you see that most of the students have access to e-mail. I think what you're going to see is a lot of things involving visualization and huge amounts of computation. But we are driving the amount of computation done on supercomputers and workstations beyond the ability currently to handle those things gracefully. So you haven't seen anything yet.

If you think about the sorts of things you can do with visualization, it would be nice to get the communications and network infrastructure into place to enable that sort of thing to move over large geographic distances. But I think we'll see that.

How would computer-communication developments affect future person-to-person interactions, such as the one we're having here?

I can only speculate, but if we had videoconferencing down to a good enough level, you wouldn't feel the need to come here and talk to me personally, because you'd feel that you're getting the same interaction with the videoconferencing system. One doesn't feel that, and correctly so, with systems in use today. They don't replace traditional interaction, but they will get better.

I can't predict what copyright laws will do, but I think we'll surely get different forms of information dissemination. I still read books the usual way, but I print things out from my screen. I write there. I'd feel lost if I had to use a paper and pencil to write. But to read, I still want paper output to scribble out corrections. I think technologies will change. We'll have a virtual book. You can imagine devices that will be like a pair of glasses which might project directly into the retina and give you a clear illusion of a book. We're not there yet, though.

This technology has been practiced by a lot of computationists and scientists. It hasn't made them less "human," as it were, has it?

I don't think so. My experience is that I really don't like interacting with somebody by e-mail a lot unless it's somebody I know very well. Somehow you need to establish a certain rapport before the technology will enable you to interact informally. But for working with groups of people, it's transformed the way we work. It hasn't totally replaced meetings, but it has almost totally replaced telephone calls. I still place telephone calls to somebody I never met before, because I sort of want to sound out the nuances. But once I know someone, dropping them e-mail messages is much faster than a phone conversation.

What do you see as the future federal role in the development of these technologies?

As far as NSF is concerned, we support research. One form of support is for network infrastructure, which, in turn, supports all science and engineering research generally. The other form of support is the supercomputer centers, which also support all of science and engineering generally. The support here is for the long-term goals for some other science and engineering objective because the developing technology will be very important.

Information in a sense is the policy of policies, then. But who is minding the store in terms of ensuring the quality of information for those many uses?

I'd be very careful about that. We are in the funny business of selling the research we do in terms of the applications to society as a whole, and we're also developing the underlying technology that enables all of that. So there's this huge hidden underlying technology that is often the heart and soul of what we do.

On the other hand, society would have no reason to support that research other than the basic intellectual interest in it if it didn't have this payoff. And so at some level, the basic research in strategic payoff is in terms of applications that are useful to society and science and engineering generally.

Yes, but news about this research and technology isn't as sexy as, say, the discovery of a new mechanism for cancer.

I agree. Part of the difficulty we have is explaining this technology. But if you look at, say, time sharing, which goes way back, it is something that really has enabled computers to be widely used. Time sharing helped drive that revolution. In another area, the ability to do graphics displays certainly is going to help with the way we do lots of things in science and engineering, including medicine.

How we use networking, as well as all the individual things we do to make these things happen, isn't sexy. But the sheer computational power that's now become available in computing is growing enormously, although we still have very significant problems with developing the software that enables that to happen in graceful fashion.

As far as parallel computation goes, we really intellectually don't yet fully understand it down to the theoretical level. It's a fundamental scientific question of what the true nature of parallel computation should be. We don't have adequate processes for modeling it. But at the same time, we're driving applications with it that are very successful.

Of what interest to the technical community are changes in regulatory law in the field?

If the question is whether the technologists have an interest in the policy directions, the answer is probably no. But the impact will be fairly large. So it's important to get the policy right. Policies determine the sort of interaction that will occur among the pieces of the technology that will be put in place and that will influence, in the long run, the technology development.

How do you feel about all the current budget uncertainty in research and technology?

We don't know what's impending. There are cuts for science and engineering generally. So far, there are no discussions in cuts in high-performance computing and NII. That doesn't mean I'm not optimistic or pessimistic. It just means that I can't deal with them until I know what they are. So I haven't dealt with that question. Of course, the out-year budgets don't look good, and we're still working on development of priorities regarding the foundation as a whole.

Are you worried?

Sure. I don't know anybody in science who isn't worried. I am particularly worried for computer science and engineering and computational science generally because there is a misperception that the work we do might be picked up by industry. The recent report by the National Research Council on High-Performance Computing & Communication was pretty clear in that regard. The report spoke of a $250 billion industry, and it's probably double that if you include portions of industries like finance and entertainment that are increasingly dependent on the technology. It quoted the total identified industrial R&D budgets at about $20 billion, but there's a very clear indication that a major portion of that is development and very short-term applied product-development work.

Are you fearful that the speed of this information age will lead to more stress in the people who use it and are affected by it?

For many of us, this technology increases the pace at which we work. It gives us many more people we're in contact with. I typically deal with 50 to 100 e-mail messages a day. And we operate on very short time frames because we can do collaborative writing. I think it makes us very much more productive, but there's a stress factor involved, and I imagine that depends on the particular nature of one's job. But I think that's true of American society in general today. Most people seem to be working harder, whether they're in jobs connected with information or not.

What are the hardest things about your job?

The total number of things that need to get done and the lack of time to get them done in. This is a time of enormous strain and enormous change and consequent stress on the system. It comes from trying to find new ways of coordinating programs across agencies - for example, different funding patterns for programs whose budgets are under stress. The coordination effort causes work of many different forms.

[ACS Home Page] [ACS Publications Division Page]