A LearnWorld Text

Race to Oblivion

A Participant's View of the Arms Race

Herbert F. York




Table of Contents57Missile-Gap Mania125
INTRODUCTION78The McNamara Era147
Prologue: Eisenhower's Other Warning9PART TWO: UNBALANCING THE BALANCE OF TERROR171
1The Arms Race and I159MIRV: The Multiple Menace173
2The Race Begins: Nuclear Weapons and Overkill2711Other Lessons from the ABM Debate213
3The Bomber Bonanza4912The Ultimate Absurdity228
4The Elusive Nuclear Airplane60A Glossary of Acroyms241
5Rockets and Missiles75Index245








As the decade of the seventies begins, four technological advances threaten to upset the strategic stability, or balance of terror, that I described in the last chapter. These are ( I ) improvements in reliability, (2) further improvements in guidance accuracy, (3) MIRV, and (4) ABM.

Now, as in the recent past, accuracy is the factor that has the strongest effect on the probability of destroying a hard target. But the accuracies now in prospect are so good that missile reliability may become most important. In many cases, the kill probability, if everything works as it should, will be so close to one hundred percent that the probability of everything working as it should can become the dominant factor in determining the net kill probability. The measure of whether or not everything works as it should is the "reliability,, of the system. This reliability is not known for operational missiles under operational conditions, it is known only for test missiles under test conditions. Dr. Daniel Fink, who had been Deputy Director of Defense Research and Engineering under bath Harold Brown and his successor, John S. Foster, Jr., estimated it at forty percent to sixty-five percent. Improving this figure is largely a matter of procedures such as quality-control methods. If basic changes in the system design are held to a minimum, thus avoiding new "bugs," it is reasonable to suppose that


ultimately this figure will improve somewhat. Also, those failures that occur early in missile flight can in principle be observed early enough so that the failed missile can be immediately replaced by another. This method of early replacement works in theory if the over-all reliability is not too low and if nearly all failures do indeed occur early in flight.

However, doing this in practice during a huge massive attack with perhaps a thousand missiles and many thousand warheads in the air simultaneously is quite another matter. For one thing, an experimental check of the technique at full scale is impossible. However, during the 1969 ABM debates, Foster and Albert Wohlstetter, one of the sounder ABM supporters from outside the government, both implied that the Soviets would (or could) have a combination of intrinsic reliability plus an early-replacement capability such that the equivalent reliability would be practically one hundred percent. I very much doubt that this will be true in the real world in the mid-seventies, but if Foster and Wohlstetter can think so, then others with authority and influence can think so, too. And on this question of the stability of the deterrent, what such people think has as much effect as does reality.

The second technological advance that threatens to upset the current strategic stability is further improvement in guidance technology. Accuracy is measured in terms of CEP, the circular error probable. This CEP is the distance from the aiming point within which one half of a large number of otherwise similar shots would fall. To get a feeling for the significance of accuracy, we may note that a one-megaton weapon with a CEP of half a nautical mile fired against a target hardened to survive a blast overpressure of three hundred pounds per square inch would have a kill probability of thirty-two percent. If the CEP were reduced to a quarter of a mile, the kill probability would be about seventy-eight percent, and for an eighth of a mile it would be essentially one hundred percent.

The following table shows how these "kill probabilities" vary


with accuracy for warheads of other strengths. We can see that for accuracies of a quarter of a mile or worse (the present situation) only "big,' bombs (one megaton and up) have high kill probabilities when used against missiles stored in very hard silos. But for accuracies of an eighth of a mile (coming) all sizes of bombs considered, even the so-called "small" MIRV warheads carried by the Poseidon, have high kill probabilities.

(300-p.s.i.* hard taget and 100% reliability are assumed)
Missile Warhead
Yield CEP:
1/2 mile
1/4 mile
1/8 mile
MIRV 50 KT 5 % 19 % 56 %
MIRV 200 KT 12 % 40 % 87 %
single 1 MT 32 % 78 % 100 %
single 1 MT 32 % 78 % 100 %
MIRV 5 MT 68 % 99 % 100 %
* Per square inch.

The third item on a list of technological advances that threatens to destabilize things is MIRV.

MIRV was one of the few new weapons systems started in the early sixties. It has had a much more profound effect on


the strategic situation and on the arms race than was anticipated at its beginning. It has been so much discussed and debated bath in public and in private that the letters "MIRV" have come to be accepted as a word that can be declined as a noun or conjugated as a verb.

The origin of the concept of multiple independently targetable reentry vehicles can be usefully traced back to 1958. By that time we had already spent many millions for research on and development of various concepts for defense against missiles. One ABM, our anti-ballistic-missile, the Army's Nike-Zeus, had reached such a stage in its development that it began to appear possible to intercept simple large warheads of the type then intended for our offensive missiles. Also, about the same time, our intelligence agencies confirmed the existence of an already large and still growing ABM development program in the Soviet Union. And Khrushchev in his own inimitable way had boasted of Soviet capabilities- to-come in this type of weaponry. With all this in the background, the designers and promoters of ballistic-missiles systems began to take the challenge of the anti-ballistic-missile seriously. In a relatively short time they considered a great many ideas, inventions, and tactics designed to enable ballistic-missile warheads to penetrate any ABM defense then known. Collectively, these are known as penetration aids. The most important insofar as the then unforeseen consequences were concerned was the notion of putting more than one warhead on each launch vehicle. At first this last idea simply involved a "shotgun" technique in which the several warheads landed in a tight pattern around the aiming point. But of course almost immediately it was proposed that each of these be separately aimed and targeted.

It is this last refinement that we now call MIRV.

Several programs to explore these ideas and then develop the better ones further were initiated almost right away. One of these had as its objective the development of the threefold


Claw warhead for the A-3 version of the Polaris. It seemed especially important to provide Polaris with this multiple-reentry-vehicle capability, since at that time there didn't seem to be any other good way of gaining the type of advantage for the offense that is provided by a barrage of simultaneously arriving warheads. The Claw was to be of the "shotgun" type, however, since the techniques for independently targeting the individual warhead had not yet been worked out adequately. As time went on, the designers of the defense began to invent methods for discriminating decoys from warheads and for coping with jammers and chaff. In my mind there has always been and still remains very considerable doubt about the true efficacy of such discrimination techniques. However, the possibility of their coming into being ultimately pushed the designers of the offense into placing greatest emphasis on multiple warheads as the one sure way of making the defending ABM fire at every incoming object. Thus the defense would be overpowered by simple exhaustion even if no other penetration aids or tactics worked.

A favorite MIRV design nowadays involves launching an object called a "bus" onto a normal missile trajectory. This bus carries on board all the various individual warheads. It is equipped with its own inertial-guidance system and it is provided with a set of small motors that can be used to change its orientation, direction, and velocity. When the bus is properly aimed at the first of a series of programmed targets, it very gently drops off its first warhead. This warhead then continues on the same course toward target number one. Next, the guidance system and the rocket motors modify the velocity and direction of the bus so that it is aimed at a second target. It then gently drops off its second warhead to proceed inexorably to target number two. The same maneuver is repeated again and again until it has aimed each of its passengers toward a separate preselected target.

We have already seen that without MIRV, and with only


approximately numerically equal forces on each side, neither side can hope to "wipe out" the other in a surprise attack. Because of missiles that fail and warheads that miss, a significant number of missiles will always survive the attack. Even if the two forces are somewhat out of numerical balance, uncertainties in the estimates of how effective an attack would be will still very powerfully inhibit even the side that possesses the numerically superior force. Further, if one side does start to get too far ahead in numbers (say, one side has more than double what the other has) the other side can in principle bring things back onto an even keel by proliferating his forces, and other drastic but effective means, such as the dreadful launch-on-warning technique which I will discuss later, can be employed if necessary during the time it takes to readjust the numerical balance.

MIRV completely eliminates even the possibility of this sort of balance. MIRV makes it easily possible for each side to have many more warheads than the other side has missiles. With presently planned warhead multiplicities, and with foreseeable accuracies and reliabilities, each side will eventually be able to wipe out better than ninety-five percent of the other side's silo-based missile force in a surprise attack. The possibility of the Soviets' achieving such a capability against us in the mid-seventies was one of the reasons given by the Nixon Administration in 1969 for proposing deployment of the Safeguard ABM system. Indeed, the question is when, not whether, this can happen. We will come back to this matter later; for now, suffice it to say that I believe this time will not come before 1980, and that in any event the Safeguard ABM is an inappropriate solution to the genuinely serious problem that will be created when one or bath of us achieves this capability in his operational forces.

The fourth item on the list of technological developments that threatens to destabilize the present balance of terror is the development of the ABM. We will discuss the technical aspects


and possibilities of that type of system in a separate chapter. Here we need note only the arguments usually presented by strategic analysts about how the ABM can influence stability. According to such analyses, if an ABM system is deployed so as to protect the population generally, it will add to the instability brought about by MIRV. The reason is that even after a surprise attack by MIRVed forces (as described just above) there will still be some residual forces left for retaliation. But these will be numerically small, and a population defense, even though inevitably imperfect, might be able to intercept enough of them to reduce the expected casualties to what these theorists refer to as an "acceptable" level. In this view, MIRV plus population defense make preemption lock more promising than MIRV alone. Hence their combination is destabilizing.

On the other hand, an ABM system in which the defender could have very high confidence, and which was used to defend his deterrent (the ostensible purpose of Safeguard), should be stabilizing by virtue of increasing the anticipated retaliation that would follow a surprise attack.

I believe these analyses are correct in an abstract sense; however, I heavily discount their relevance in the real world. It's not all that easy to distinguish between population defense and deterrent defense, especially at an early phase of deployment. Also, this kind of highly convoluted argument, even if correct, does not carry nearly as much weight among ruling politicians as it does among theoreticians. In my view, the main reason an ABM program (any ABM program I have heard of) is dangerous and destabilizing is that it is simply another step in the arms race. It presents a technical challenge (again, not a political provocation!) to the technologists who design the offense. In designing around the ABM, these latter will usually come up with a more complex, more expensive, more deadly and more volatile offense. Our ultimate response to the Soviet ABM was MIRV. If current plans continue, fewer than one hundred Soviet ABM missiles new deployed around


Moscow will have resulted before very long in an increase in the number of U. S. warheads by more than five thousand. Thus, as a matter of historical fact, the Soviet ABM program has been fantastically counterproductive and has had anything but a stabilizing effect.

This sequence of steps did not come about as a result of careful analyses of the political significance of each preceding step and considerations of such matter as stability. The technologists on each side took what the other side was doing as a challenge; they designed new systems to counter the challenge, they sold these to the officials and politicians in charge, and then sometimes they performed careful after-the-fact analyses that rationalized the whole thing when it was already too late to do anything about it. The technological side of the arms race has a life of its own, almost independent of policy and politics. McNamara himself described this situation in his famous San Francisco speech (1967) by saying, "There is a kind of mad momentum intrinsic to the development of all nuclear weaponry. If a system works--and works well there is a strong pressure from all directions to procure and deploy the weapon out of all proportion to the prudent level required."

Let us return to a more detailed look at MIRV and its political meaning. Some American analysts have said that the U. S. MIRV program is very different from the Soviet MIRV program insofar as the threat to the stability of the present strategic-arms balance is concerned. I believe this supposed asymmetry exists only in the eye of the beholder. U.S. analysts knew about all the little nitty-gritty problems that have beset our programs, and in the interest of "erring on the side of military safety" they downgrade (often unconsciously) the capability of our missiles. These same people, when they look at the Soviet system, are largely unaware of the myriad small problems that add up to big difficulties in the real Soviet Union. They emphasize "the worst plausible case," which in this situation means making almost mystic assumptions about


Soviet technology, and they (again often unconsciously) upgrade the capability of Soviet missiles.

Let us consider the American MIRV as it might be viewed by a Soviet analyst. Suppose the entire Minuteman force is converted to the Minuteman III with a MIRV warhead consisting of three 200-kiloton individual warheads. If each of these has a one-quarter-mile CEP, then each one of them, if successfully launched, would have a forty percent chance of destroying a Soviet missile in its silo. If all three are launched at the same silo, the kill probability goes up to almost eighty percent.

Real reliabilities reduce the net "kill probability" further, and the resulting figure is not enough to make a preemptive attack "interesting" in the case of a one-quarter-mile CEP. But what about a one-eighth-mile CEP? Under such conditions, only two of the three Minuteman MIRV warheads are needed to achieve well over ninety-eight percent kill probability, leaving the third warhead for use either as a replacement for other missiles that failed at or near launch or for other purposes. And is a one-eighth-mile CEP so improbable? We have improved our CEPs almost twentyfold in the sixteen years since we started our ICBM development program. Another factor of two does not lock very large by comparison. And the accuracy requirement set for our first ICBM, the Atlas, was just twenty times the accuracy actually achieved by the V-2 only ten years before that. The net improvement in twenty-five years has been four- hundredfold! Furthermore, there are no fundamental laws forbidding still further improvements, and eventually we will no doubt go well below one eighth of a mile. Even the so-called "small" fifty-kiloton warhead postulated for the Poseidon MIRV will be quite effective at destroying silo-based missiles when this system achieves a one-eighth-mile CEP. A single shot with such a missile would result in a kill probability of about fifty-six percent. This means that only four shots at the same target would result in about a ninety-


six percent net probability of destruction, and each Poseidon MIRV will have ten such warheads!

The U. S. missile force with the planned MIRV modification and a perfectly plausible extrapolation of our present accuracy is every bit as suitable for a preemptive attack as is the Soviets' missile force after MIRVing and with what we take to be a plausible extrapolation of their present accuracy (which, generally, has been behind ours). It is easy to imagine that in times of tension such a situation would be fraught with danger. Each side, believing the other could make a preemptive attack, would certainly be stimulated to consider doing so first.

Of course, the situation described above differs from reality in one essential respect. It ignores the fact that there are other components to the deterrent besides the silo-based missiles. The MIRV-accuracy-yield- numbers combination assumed above does not similarly threaten either our missile submarine force or our bomber force, since these are protected against surprise attack by other means. The submarine force is protected simply by being hidden in the murky seas, and the bombers are protected by being normally placed on an alert status such that a substantial fraction of them could become airborne in the warning time that can be anticipated.

However, even having just one component of our deterrent become so vulnerable to a preemptive attack is bound to cause concern. It is always possible to imagine that somehow something parallel may happen to the other components and then we really might be in a fix. I regard this simultaneous vulnerability of all components of our strategic forces as being completely incredible, but, even so, it is prudent to think about what should and could be done to avoid a situation in which even one major component of our strategic forces becomes vulnerable to a preemptive strike.

Many ideas for protecting our land-based missile forces against a reliable and accurate MIRV attack have been sug-


gested and studied. In my judgment, all save one are inadequate from a technological point of view, and that one is unacceptable for other reasons. The inadequate solutions include superhardening, the "Safeguard" ABM system, proliferation, a "shell game" in which there are more silos than missiles, and land-mobile missiles. Although I was personally hopeful before the 1969 ABM hearings that at least one of these approaches would maintain its invulnerability, a review of those debates leaves me new with the pessimistic view that none of them holds much promise beyond the next ten years.

Silo-hardening most probably does work now, in the sense that the combination of SS-11 accuracy and yield and Minuteman silo-hardening works out in such a way that one incoming warhead (and hence one SS- 11 missile) has less than a fifty-fifty chance of destroying a Minuteman. If, however, one considers the technological trends in hardening, yield per unit weight, MIRVing, and accuracy, it does seem convincing that this is a game that the offense eventually will win. Albert Wohlstetter, testifying in favor of the Safeguard system before the Senate Armed Services Committee, quoted a paper he wrote with Fred Hoffman in 1954 (long before any ICBMs were actually in place anywhere) predicting that the ability of silo-hardening to protect offensive missiles would run out by the end of the 1960s. This remarkably prescient study is wrong only in numerical detail. If we take the same rosy view of technology that was taken in almost all the pro-ABM arguments, then hardening will not work for more than another five years. My own view of the technological future is clearly much less rosy, but I do believe that the situation in which hardening is no longer the answer could come by, say, 1980 or' more appropriately, 1984.

Protection of a land-based ICBM by an ABM system also appears to me to be quite inadequate as a means for reestablishing a high level of confidence in the assured survival of this component of the deterrent. The Safeguard ABM as originally


proposed would be too weak and thin to provide protection against a determined enemy; an ABM which would provide an effective shield would in my opinion cost vastly more than we would be willing to pay. The same thing applies to the Soviets. I will have more to say about this in the next chapter.

Proliferation of Minuteman would have worked in the absence of MIRV. Now, however, it would seem that the ability to MIRV, which no doubt can eventually be carried much further than the fewfold MIRV we see for the immediate future, clearly makes proliferation a losing game as well as the dangerous one it always was.

The "shell game" has not, in my view, been analyzed in satisfactory detail, but it would appear to have a serious destabilizing effect on the arms race. Schemes have been suggested for verifying that a certain fraction of the missile holes are in fact empty, but one can foresee a growing and persistent belief on each side that the "other missiles" must be hidden somewhere.

Road-mobile and rail-mobile versions of Minuteman have been seriously studied for well over a decade. These ideas have always foundered on two basic difficulties: (1) such systems are inherently soft and hence can be attacked by very large warheads without precise knowledge of where they are, and (2) railroads and highways all pass through population centers, and large political and social problems seem unavoidable. Off-road systems have been considered, too, but they suffer from severe geographic restrictions.

Where does all this leave us? Can we find a technical solution for protecting Minuteman? Is a balance of terror possible with MIRVs in the strategic forces? One and only one technically viable solution seems to have emerged for the long run: launch on warning. Such an idea has been considered seriously by some politicians, some technical men, and some military officers.

In theory, launch on warning could be managed entirely


by automatic devices or the command and control system could be such as to require authorization to launch by some very high human authority.

In the case of the first alternative, people who think about such things envision a system consisting of probably two type' of detection device that could, in principle, determine if a massive launch had been made and then, somewhat later, determine that such a launch consisted of multiple warheads aimed at our missile-silo fields. This information would be processed by a computer which would then launch the Minutemen so that the incoming missiles would find only empty holes; consequently the Minutemen would be able to carry out their mission of revenge. Thus, the steady advance of arms technology may be leading us not to the ultimate weapon but rather to the ultimate absurdity: a completely automatic system for deciding whether or not doomsday has arrived.

To me, such an approach to the problem is politically and morally unacceptable, and if it really is the only approach, then clearly we have been considering the wrong problem. Instead of asking how Minuteman can be protected, we should be asking what the alternatives to Minuteman arc. And, much more important, instead of blithely moving toward a balance of terror of a still more gruesome and precarious sort, we should be asking what the alternatives to a balance of terror might be.

Evidently most other people also find the launch-on-warning idea unacceptable once they think about it. The Army found it necessary to reassure people repeatedly that ABM missiles would not be launched without the approval of "the highest authorities'" even though this is clearly a far less serious matter in the case of the ABM missiles than in the case of Minuteman.

The theoretical alternative is to require that a human decision-maker at the level of "the highest authorities" be introduced into the decision- making loop. But is this really satis-


factory? We would be asking that a human being make in just a few minutes a decision to utterly destroy another country. (After all, there would be no point in firing at their empty silos.) If, for any reason whatever, he were responding to a false alarm or to some kind of smaller, perhaps "accidental," attack, he would be ensuring that a massive deliberate attack on us would take place moments later. Considering the shortness of the time, the complexity of the information and the awesomeness of the moment, the President would himself have to be properly preprogrammed in order to make such a decision.

Those who argue that the command and control system is perfect or perfectable [sic] forget that human beings are not. If forced to choose, I would prefer a preprogrammed President to a computer when it came to deciding whether or not doomsday had arrived, but again I feel that this solution too is really unacceptable, and that once again, in attempting to defend Minuteman, we are simply dealing with the wrong problem. For the present it would seem that the Polarises and the bombers are not, as systems, subject to the same objections, since there are now enough other approaches to the problem of ensuring their invulnerability to sudden massive destruction.

It is important to realize that this launch-on-warning idea neither is new nor has been casually offered. Back in the fifties when the design requirement for being able to launch the large liquid-fueled ICBMs in fifteen minutes or less was first established, it was precisely this capability of being able to launch the missile on warning that determined this time interval. The warning time available from radars located on the fringes of the continent would be fifteen to twenty minutes. Newer technology over-the-horizon radars and satellites can in principle observe the launching itself and hence provide thirty minutes of warning, or about twice the launch-time requirement established back in the fifties. Also, one of the main reasons for substituting the smaller solid-propellant Minuteman for the


larger liquid-propellant Atlases and Titans as the main component of our deterrent forces was that this missile could be launched in an even shorter time. Its name implies that it can be launched in about one minute after receipt of warning. In 1960 I paid an official visit to the North American Air Defense Command headquarters in Colorado Springs. Among other things, the commander told me that a further expansion in our ballistic-missile early-warning system (BMEWS) was urgently needed so as to reduce the probability of a false alarm resulting in the launching of our ICBMs. It was clear from the conversation that he had in mind minimum human intervention in the electronic network which would connect the warning system to the launch-control circuits, or "the button," as it is often called. I replied to the effect that I felt our warning system was never likely to reach a state of such absolute reliability that we could be willing to risk launch on warning, especially if no outside higher-level human intervention was allowed. The commander and some of the other general officers present were obviously appalled by my lack of confidence in the potentialities of technology, and one of them said to me quite literally, "Well, in that case, we might as well surrender now." Before MIRV, we could fight off such nonsense by simply pointing out that, in a careful analysis, it wasn't necessary. Now, with MIRV, and with further advances in reliability and accuracy in prospect, we are being forced to consider these horrors again. The only possible conclusion is that it is the arms race as a whole, and not just some of its details, that is rapidly and inexorably diminishing our national security.

Nuclear Designs: Great Britain, France, and China in the Global Governance of Nuclear Arms
[Transaction Publishers, 1996]

British SSBNs

French SNLEs
Other Sites
"Comprehensive Test Ban" [28 February 1996] and a 21 June 1996 addendum on China's CTB policy. The Acheson-Lilienthal Report [16 March 1946]: Report on the International Control of Atomic Energy. Re CTB

LearnWorld Explained FAQ Bosnia War Theory Templates

Our Home Page Our Home

Please direct comments and corrections by email to Bruce D. Larkin, who has prepared this page.