Freeman Dyson Interview: Conversations with History; Institute of International Studies, UC Berkeley
Page 2 of 3
Another [shared] element of human nature [among] the scientists who are involved in this work is the puzzle solving, the inquisitiveness, the scepticism. One sees it in your own mind at work as you talk about, for example, the sucker fish that will attach themselves to submerged objects and the development of such a system. So you're saying, in part, that the victims have to learn that there are positive aspects both to the people who work with weapons and to the kinds of technologies they can produce in terms of making a safer world.
Right. Certainly, making a safer world is not only a question of moral rectitude, it is also a great deal of common sense and a great deal of hard-nosed willingness to stand up and fight for one's principles. It's not at all clear in all cases that the warriors and the victims have contrary objectives. In many cases their objectives are the same.
Why do you feel that there is such a parting of ways? Is it the struggle itself that results in this lack of empathy for the people who are making the bombs on the part of the victims, and vice versa?
It has a lot to do with the fact that we have become hooked on nuclear weapons. Nuclear weapons are the symbol of this divergence, and rightly so. A large part of the military has wedded itself to nuclear weapons in a very deep fashion, and the public understands, at least part of the public understands, that nuclear weapons are basically inappropriate for the job, that nuclear weapons are in no way an adequate solution to our security problems. They don't, in fact, do the right thing that weapons ought to do. They are not useable. This has given rise, to a great extent, to the split in our society. Nuclear weapons are morally unacceptable to a great part of us. And as long the military establishment is so wedded to them, there won't be a meeting of the minds.
One of the themes that appears in Weapons and Hope is the possibilities that technology offers us in a long-term solution. You discuss Conant's and Oppenheimer's opposition to the super-bombs, their failure to win in the bureaucratic battles in Washington or to win President Truman's ear. But technology over the longer term may [provide] a solution which they were advocating, which was smaller weapons and more precise weapons. So technology, you seem to want to tell the victims, offers possibilities for hope. Is that fair?
Yes, I think that is fair.
The great advances in technology over the last twenty years have been in information processing. The availability of good technology for finding out what's going on, for sensing what's coming at you and processing the information rapidly and taking it where it's needed -- these technologies favor the defense to a great extent. They allow the country which wants to protect itself from penetration from outside to do that with sensible weapons, with weapons which don't destroy the whole country in the process of defending it. That's the way technology has been going in the last twenty years: away from nuclear weapons toward little computers and little rockets, things that ordinary soldiers can handle. It's not all sweetness and light, of course. Nothing that comes with technology is simple, nothing is an unmixed blessing, but at least it looks better.
You, on the one hand, identify this positive role for technology. On the other hand, you cite numerous cases of what you call technical follies. That is where, because a technology is oversold, overproduced, and misapplied, it leads to a disaster in terms of its application in a military context.
Yes. The most complicated systems don't work. That's what I mean by technical folly. That you build something very beautiful and complicated, but in the real world it just doesn't do the job. That happens time and time again. It's very hard to predict whether any particular gadget is going to work in the real world. In fact, you practically never know until you try it. So one must expect, as technology goes ahead, that most of the things that are invented turn out to be no good. Only one or two of them, in the end, are good. It's very hard ever to guess what's going to work well. But the things that do work well are often cheap and simple.
What are the underlying elements that seem to work toward sensible application of technology? You discuss in great detail a turret gun system that was developed in the second world war in Britain, in the hopes of saving more pilots' lives, but when it was actually put in, it couldn't do the job in the terms that had been promised or hoped for.
Yes, because it would have resulted in shooting down much more of our men than of the enemies'. That is a typical situation, that the apparatus works beautifully in tests, in artificial conditions, in trials. And then when you come to use it operationally, you discover that there's a snag that you hadn't thought of that makes the thing useless. That's conspicuously true of many of the gadgets we have now. But that doesn't mean that technology is always useless. It sometimes can work. The main thing is not to put one's hopes too heavily in one particular thing.
I take the example of this so-called Star Wars program, which is very much talked about. That's a very good case in point, the idea that one should be able to develop some marvelous technology in space which would enable us to defend ourselves against a nuclear attack. That's almost certainly nonsense. There's very little chance that this would ever happen. However, if you look at the basic technology of information processing, of sensors, of the technology that might enable the defense system to work and do it in a modest fashion, and look at it in terms of mostly ground-based defense rather than space-based, then maybe parts of that actually would work.
I have an open mind about it. The way it's being promoted at the moment is mainly total nonsense. So, most of what my scientific colleagues are saying about it is correct, that Star Wars technology, as it is at the moment being sold, is going to be a technical folly of the first magnitude. But if you look at the components of it, it's still possible that some of them might be useful.
What, then, is the political problem? How do we avoid a technical folly, but at the same time explore those possibilities that might be worth pursuing? Is that a problem with political management?
It's partly political. It's mostly a question of disentangling ends from means, of having a firm grasp of what you're trying to do and then allowing that to determine the means rather than the other way around. In the case of strategic defense, one ought to say clearly at the start that it's not going to work unless it's combined with arms control, and a fairly drastic arms control. Only in an environment of drastic arms control limitation on offensive forces can strategic defense possibly be helpful. But if that's granted, then you might be able to do something very good. Trying to do the defensive job without arms control is utter insanity.
Because it won't work, and because there's a lack of clarity about your intentions to the other side.
Yes, because you won't succeed in what is often imagined in outrunning the Soviet Union. That isn't in the cards. The Soviet Union will certainly keep up and will spend whatever it takes. So a move to a defensive world, which is what the world needs, requires as an essential part, an agreed limitation on the offense.
There was an ambivalence in President Reagan's speech as to whether this defense would be coupled with an offense. If you're going upward with defense and you're going up with offense, there is a danger to the other side that you may be thinking about a first strike.
Right. One should make a sharp distinction between what Mr. Reagan actually said in this famous speech last year and all the other propaganda that has come up since. In Reagan's speech you find there is actually very little said about real deployment, about actually building defensive systems. What Reagan was asking for was an exploration, that we should explore what is technically feasible. And that, I'm in favor of. But when it comes to all the rest that comes later, a great deal of talk about building big systems in the next few years, that's moonshine. We wouldn't know what to build if we wanted to.
You mentioned just a moment ago the overall ends that the means were serving, and that that confusion of ends and means was part of the problem. In your book you talk about the problem of verification at the time of the partial test-ban in the sixties and how getting too good at verification technically raised all kinds of political problems. Could we discuss that a little? You were trying to develop verification systems that would test when the Soviets were cheating, and then you ran into a set of problems. What were those problems?
The problem is the problem of false alarms. And that is, of course, absolutely basic to all verification systems. You must consider both the efficiency of the verification, that is to say, the chance of detecting some forbidden activity, and the probability of getting false alarms, that is, of thinking you've detected something when it's really only a glitch on the apparatus.
Those two things are equally damaging, but in many ways the false alarm problem is worse. If you have a system that's giving too many false alarms it overloads the intelligence apparatus. It's no longer possible for the analysts to keep track of it. We had a good example of that a couple of years ago when we had a famous so-called event in the South Atlantic where it was possible that somebody was carrying out a nuclear test in the South Atlantic. The flash of light was seen by one of our detection satellites. That single event tied up the whole of the analytical community for about two months. So if you had such events coming in at the rate of once a week, you wouldn't be able to cope, and as soon as the verification system is overloaded, then it becomes worse than useless.
You should have enough verification, but not too much. Very often, too much is worse than too little.
The real danger is that verification becomes an end in itself, instead of a means toward good political relations which should be existing in and of themselves.
Yes. In most cases, the question whether or not the other side is cheating is not the most important question from the point of view of security. The important question is that we know what the other side is doing. That's true whether or not you have a treaty. The requirement of checking up on the observance of the treaty is, in a way, secondary. You want in any case to know what the other side has, whether you have a treaty or not. So the verification apparatus for the treaty should be considered as an addition to your intelligence sources, not as something that stands by itself.
Next page: Nuclear Weapons and the Future
© Copyright 2005, Regents of the University of California