The stress on the improvements in conventional forces reflected the Kennedy Administration’s conviction that there was no good way to fight a nuclear war and that this would be as evident in Moscow as it was in Washington. Threats to use nuclear weapons first, even in the most tentative way, therefore lacked credibility. But building up conventional forces was going to take time. The Europeans seemed wedded to massive retaliation and reluctant to increase their defence budgets. In addition, when it came to the most likely flashpoint, Berlin, the conventional options developed by military planners generated little enthusiasm as they seemed to carry more risks than benefits. The quality of the available nuclear options therefore had to be addressed. When they were addressed they appeared alarming. There was nothing tentative about them. With the available SIOP, once a nuclear war began the Soviet Union, and incidentally China, would be battered to virtual extinction with huge numbers of thermonuclear detonations.
Robert McNamara in particular could not see how a credible deterrent could be fashioned out of an incredible nuclear posture and he ‘didn’t have any plan to use nuclear weapons that I would want to implement.’1 Nor could he just relax on the basis that the worst would never happen. The Americans had the responsibility to come up with the plans for they had the weapons and they wanted to ease as much as possible the decisions the President would have to face should the terrible moment come. McNamara later reported that Lord Louis Mountbatten, British Chief of Defence Staff, had remarked to him that ‘under no circumstances, even with the great superiority in nuclear weapons that NATO had at that time … should we consider the use of nuclear weapons’. McNamara added that he had agreed and ‘subsequently recommended such a policy to President Kennedy’.2
This was one reason why the Kennedy Administration was anxious about SAC’s approach to nuclear strategy. The other was its apparent indifference to the risks of being caught out by a Soviet first strike.3 To deal with the second issue a new criterion was set for planning: the ‘ability to strike back after a direct Soviet attack designed to destroy our retaliatory forces’. This required reducing the vulnerability of individual systems, and also action to avoid war coming about ‘in an irrational or premeditated fashion.’ There was a list of possible causes: ‘mistaken triggering of alert forces, by miscalculation by one side of the opponent’s intentions, by irrational or pathological actions by individuals, by spread and escalation of local wars, or by nuclear attack by a minor power.’ At the same time there was a demand for the development of options to influence the conduct and outcome of a nuclear war, so as to exercise some control over its impact.4
As we have noted, when the Kennedy Administration came to power it was still widely assumed that the United States would have to catch up with a Soviet programme that was more advanced. Yet it soon became apparent that the actual missile gap now favoured the United States. Ellsberg noted the impact on RAND and the vulnerability studies which had made its name. It now had to come to terms with the possibility that it had ‘in all good faith, been working obsessively and with a sense of frantic urgency on a wrong set of problems’.5 This suddenly raised the question as to whether the US was now setting its force levels too high.6 In late 1961 numbers for the Minuteman ICBM force were put at 1200 (they were cut in 1963 to 1000). The Air Force was still after twice as many. McNamara and Kennedy suspected half that number would be sufficient but this was a level that could be sustained politically.7 They were more likely to be criticised for undershooting than overshooting.
Evidence of the slow pace of the Soviet missile programme, combined with new Minuteman ICBMs in hardened silos and new nuclear-powered Polaris submarines on patrol, eased what had expected to be the most urgent challenge, which was to protect US systems against a Soviet first strike. This allowed for more focus on preparations for nuclear exchanges. The first requirement was for more civilian control. This required improving and protecting the national command and control system.
The issues raised by the rigid and devastating SIOP as developed by SAC were exposed to senior civilian figures in the Administration during its first months. Ellsberg developed the basis for a new SIOP that allowed for more choice and not only the single scenario of a massive attack including Soviet targets as well as Chinese and East European ones.8 Although Ellsberg’s draft was circulated in May 1961 by Roswell Gilpatric, McNamara’s deputy, asking for views on the advisability of avoiding initial attacks against industrial, population, and governmental centres, the Joint Chiefs reaffirmed their view that it was best to keep any response simple. Kennedy and McNamara were not convinced by the military line but nor were they yet ready to adopt any new guidance.9
They did take the view that any use of strategic forces required the expressed authority of the President. These moves should be as controlled, thoughtful and unhurried as possible and not just a reactive spasm. As McNamara explained early on, he wanted a strategic force ‘to be of a character which will permit its use, in event of attack, in a cool and deliberate fashion and always under the complete control of the constituted authority’.10 In February 1962 he spoke of approaching nuclear exchanges in terms of bargaining, along Schelling lines: ‘We may seek to terminate a war on favourable terms by using our forces as a bargaining weapon—by threatening further attack’.11 This approach was consistent with the concept of populations as hostages; they would be spared great harm while the means by which they could be destroyed were demonstrated.
In short, faced with the most likely contingencies, NATO, not the Soviets, would have to make the momentous decision to use nuclear weapons, and we would have to do so in the knowledge that the consequences might be catastrophic for all of us.
As we understand the dynamics of nuclear warfare, we believe that a local nuclear engagement would do grave damage to Europe, be militarily ineffective, and would probably expand very rapidly into general nuclear war.
Not only had the United States increased its own tactical nuclear forces in the area; it had also encouraged the European allies to buy nuclear delivery systems and had committed itself to stockpiling nuclear warheads in Europe for those systems. This was not a situation that defense officials could, or wanted to, change drastically at the time. To attempt to do so would have raised the spectre of an imminent US withdrawal from Europe.14
TNW numbers were actually raised during the 1960s (especially 1963–1966) from 3500 to 7000 warheads.
come to the conclusion that to the extent feasible basic military strategy in a possible general nuclear war should be approached in much the same way that more conventional military operations have been regarded in the past. That is to say, our principal military objectives, in the event of a nuclear war stemming from a major attack on the Alliance, should be the destruction of the enemy’s military forces while attempting to preserve the fabric as well as the integrity of allied society.
In June 1962 McNamara repeated the speech at the University of Michigan at Ann Arbor, this time on an unclassified basis and so giving the new doctrine a higher status. This was unusual in that never before had the principles behind US targeting been discussed in public (they had barely been discussed within government up to this point). There was a considerable debate within his circle as to whether this was a good idea, as this meant a far more open discussion of operational considerations than had ever been offered before.
McNamara was clear that he saw no possibility of a successful first strike—‘destroying enemy forces while preserving our own societies.’ His point was more modest. The outcome would be less awful if both sides refrained from attacking cities. ‘The more discriminating the attacks the less the damage.’ This was a plea for restraint once nuclear hostilities had started, preventing a lurch into the most cataclysmic set of attacks, even though these attacks would be disastrous enough. Looking back, he later explained that his aim was to ‘influence the Soviets’. If there were ever nuclear exchanges the aim was ‘to limit severely the initial launches of nuclear weapons in the hope that we could avoid destruction of our societies.’ It was a ‘damage-limiting’ strategy, but only appropriate, he acknowledged, at a time ‘when they had so few weapons relative to ours’.15
The new plans involved significant changes. Reserves would be maintained. To ensure both sides retained control of their forces, not only would US command and control facilities be protected but those of the Soviet Union would also at first be spared. The options created allowed for attacks ranging from those against Soviet retaliatory forces, through air defence installations distant from cities, to those near cities, to command and control systems, to an all-out ‘spasm’ attack. The fatalities could be varied according to the altitude of the explosions, size of the warhead, and the amount of radioactive fallout released.16 Multiple options would mean that any American action could be tailored to the particular Soviet challenge. Preserving a wide range of options for as long as possible meant that should one need to be chosen it could be geared to the circumstances of the moment. Pentagon estimates suggested that casualties might be reduced by a factor of ten, from 100 million to 10 million in the United States, if the Russians could be persuaded to avoid cities in the event of a nuclear exchange (the comparable figures for Western Europe were 115 million with Soviet counter-city targeting and 15 million with counter-force targeting).17 The difficulty remained that all of the options for nuclear use appeared awful. In principle it was better for the US to lose (to use the numbers of the Athens speech) 25 million than 75 million people, but that also meant that war with this sort of death and destruction could not be considered to be in any sense restrained, let alone controlled.
In some respects McNamara was following the formal strategists. He recognized that any war would have to be terminated somehow, and that this would be a political process, involving bargaining. He also shared their desire to keep such nuclear exchanges that did take place as limited and controlled as possible. Where he differed was in not viewing limited nuclear war in isolation from conventional war, as something totally distinct with its own unique rules of combat. The relevant concepts were those developed at RAND during the 1950s by analysts such as Andrew Marshall and Albert Wohlstetter. They were focused on a conflict already under way on the ground. Rather than engage in either hyper-destructive but militarily pointless attacks on cities, or formalized exchanges for bargaining purposes, their preference was for nuclear strikes that helped hard-pressed military commanders attempting to hold back the Soviet advance. Nuclear strikes would gain their effect through a combination of the impression of determination created by their actual use, and the military value of the targets attacked. If they did prevent the invading forces making progress then this might create the space for negotiations and a political settlement.
In short, if both sides have well-protected forces, nuclear war, if it comes, may be much slower than we now envisage. It may involve attacks on strategic forces which are vulnerable or whose vulnerability is discovered or develops during the war; or attacks on military targets or on economic targets (such as oil fields) which are not close to major population centers.18
At issue here was what should be done if deterrence failed. It was not about the messages communicated by means of either declaratory policy or actual deployments designed to shore up deterrence. The new policy reflected an assumption that a major war was entirely possible. If that happened the President should not be handicapped by rigid pre-war planning but instead should have access to a range of options.
The core idea was largely about expanding available options to help limit the destructiveness of a future war. By raising the possibility of nuclear attacks on the enemy’s strategic forces, however, McNamara unavoidably opened up two big issues. The first was in normalising the idea of using nuclear weapons when conventional forces were not up to the task at hand. Thus while Schelling pronounced the Ann Arbor speech ‘a sensible way to think about war’, he complained that the attempt to create links with traditional military operations was misleading. To him, the approach he had tried to develop involved concentrating on the capacity to hurt; it was ‘utterly different’ to the battles of the past.19 Mixing the two types of warfare could create difficulties. It could mean a rush into nuclear use, because these weapons were the only ones available to meet certain contingencies arising out of the conventional conflict.
As serious a problem was that stressing the targeting of military forces, including nuclear assets, as in traditional military operations, could raise fears of a disarming first strike and encourage the temptation to pre-empt. Why would attacks on Soviet nuclear forces stop short of a decisive surprise blow? The ‘rear echelon’ targets, well behind the front line, could include Soviet forces relevant both to the immediate battle and strategic attacks (such as Soviet missiles aimed at Western Europe). Their location could mean that an attack upon them would be interpreted as a serious escalation. The Pentagon could insist that the aim was not complete damage prevention, which would require a first strike, but only damage limitation, a more modest ambition. Should there be a general nuclear war one key objective must be to keep the damage down to the minimum. This was the sort of minimax outcome offered up by Game Theory, that is the best of the worst outcomes.
But it was difficult to explain why developing targets for a city-avoiding second strike was different from plans for a first strike, which in terms of serious damage limitation would make more sense, at least in the circumstances of 1962. That Moscow might put this sort of gloss on the new strategy was given added credence by an article in the Saturday Evening Post in March 1962, when Stewart Alsop described Kennedy’s belief that ‘Khrushchev must not be certain that, where its vital interests are threatened, the United States will never strike first … in some circumstances we might have to take the initiative.’20 The article was intended to reassure the allies that Kennedy still adhered to NATO doctrine: the Russians interpreted it in a much more ominous light. A special military alert was ordered.21 Malinovsky argued that Kennedy was prepared to initiate a preventative nuclear war against the Soviet bloc.
Kennedy later clarified the remark, making it clear that he was thinking of first use in the context of an ongoing war provoked by the Soviet Union over the fate of Western Europe rather than some nuclear bolt from the blue. But strategic discourse had concentrated on massive nuclear exchanges, causing unacceptable damage to an aggressor by way of punishment or relieving an enemy of its strategic power in a first strike. Despite his disclaimers, and because of the stress on damage limitation, many of the statements of McNamara could be, and were, interpreted in these terms. The sensitivity of both sides to the possibilities of disarming first strikes and the references to limiting damage to US cities in the event of general nuclear war ensured that the sort of exchanges McNamara was discussing would be viewed in isolation. The most germane scenario of the time—a crisis beginning in Berlin—happened to be one in which early nuclear use would be especially likely. Otherwise one had to delve into concepts of signalling and bargaining, which might make for interesting diplomacy but was not the stuff of traditional military doctrine and unlikely to be trusted in Moscow.
How could a counter-force capability developed with the objective of striking only some military forces, including nuclear capabilities, be distinguished from a first-strike capability? One answer provided by McNamara during Congressional hearings was that even after a Soviet first strike there would still remain ‘many installations which we would wish to strike’.22 These might include aircraft which had returned to an airfield for a second sortie or had yet to take off at all, or else missile launchers yet to be used or those with a reload capacity. He also argued that there was no incentive for the US even to contemplate a first strike. The US would have no need to pre-empt, because its forces would be protected sufficiently to ride out a Soviet attack and still be available for retaliation. The highly vulnerable IRBMs in Europe were to be phased out, while a new long-range bomber would not be authorised precisely because they would be vulnerable to a first strike. He also assumed that the Russians would take steps to protect their own strategic forces in the same way.
[I]n planning our second strike force, we have provided, throughout the period under consideration, a capability to destroy virtually all the ‘soft’ and ‘semihard’ military targets in the Soviet Union and a large number of their fully hardened missile sites, with an additional capability in the form of a protected force to be employed or held in reserve for use against urban and industrial areas.
We have not found it feasible at this time, to provide a capability for insuring the destruction of any very large portion of the fully hard ICBM sites or missile-launching submarines.
Fully hard ICBM sites can be destroyed but only at great cost in terms of the numbers of offensive weapons required to dig them out.
Furthermore, in a second strike situation we would be attacking, for the most part, empty sites from which the missiles had already been fired.
The value of trying to provide a capability to destroy a very high proportion of Soviet hard ICBM sites becomes even more questionable in view of the expected increase in the Soviet missile-launching submarine force.23
Despite all the disclaimers, McNamara failed to dispel the notion that the logic of the strategy was more first than second strike. If it was at all possible to target Soviet missile forces accurately it made more sense to attack these forces in their totality, rather than destroy the residue only after the bulk were en route to their targets. As one Soviet strategist wrote: ‘A strategy which contemplates attaining victory through the destruction of the armed forces cannot stem from the idea of a “retaliatory” blow; it stems from preventive action and the achievement of surprise’.24 The sort of counter-force targeting against Soviet strategic forces that McNamara claimed to have in mind could make sense only in the most unlikely of circumstances.
Many of these missile bases are right close to our cities, right close, so are many of our other bases. So an attack on our major bases would necessarily destroy a great many cities and a great many of our people. When those missiles start coming over you do not know whether the intent of the enemy was to hit or not to hit a city if he hits it. The same thing is true with the Russian military installations.25
The doctrine as McNamara attempted to present it, therefore depended on the Soviet Union initiating a nuclear war by a limited and accurate attack against some of the less decisive US forces, situated if possible well away from cities, while holding some of its own forces in reserve, sufficiently vulnerable and available for the US retaliation. McNamara did not pretend that an attack of this nature was likely. On the contrary he admitted that an attack on US cities was as likely if not more so. All he argued was that it was a sufficiently plausible contingency to make it a worthy object of planning, and that if nuclear war could be fought this way, it would be less of a disaster than an all-out war.
But at the time the doctrine was being enunciated, the Soviet Union had no capability at all for mounting any sort of serious counter-force attack on the United States; it was having trouble with its missile programmes while the US was forging ahead. Soviet ICBMs were few and inaccurate. As a number would have to be used to destroy with certainty each military target it would be impossible to destroy a significant segment of US forces even if the entire stock of intercontinental missiles and bombers were employed. The Soviet position could not be improved by launching such an attack. Instead it would worsen because it would have exhausted itself while the US remained strong. Nor, of course, would the Soviet position be improved by attacking US cities, but at least it might be possible to influence US behaviour by threatening to do so.
Unsurprisingly the Kremlin rejected McNamara’s ideas, concentrating propaganda on the terrible consequences of any war, and treating the notion of ‘Marquis of Queensbury rules for the conduct of nuclear war’ with derision. Should nuclear war break out mass destruction was inevitable. They noted the option of relying on ‘launch on warning’, emphasised that they had a sufficiency of missiles to ensure an adequate retaliatory blow, and criticised the idea that it was possible to separate military from non-military targets. The Soviet leadership, which had failed to get hold of a copy of the Athens speech, read Ann Arbor. According to one account, ‘what Khrushchev heard was that McNamara was somehow trying to make nuclear war less bloody and therefore more acceptable’.26 Later he described the speech as intended ‘to legalize nuclear warfare and thereby the death of millions and millions of people’.27 As US bases were close to population centres the message was deceptive. To get any real value out of a counter-force attack it was either necessary to destroy more weapons than were used up in making the attack (and this was not possible at that time), or else accept the expenditure of extra missiles because of the possession of a much larger stock than the other side. In other words, to extract military benefit from counter-force capabilities it was necessary to have superiority in the quantity and quality of weapons. This made it a much more plausible aspiration for the US than the USSR. McNamara explained to a Congressional committee the need for an invulnerable US force and also one ‘larger than would otherwise be the case. Because since no force can be completely invulnerable, we will lose a portion of it under those circumstances [Soviet first strike] and we must buy more than we otherwise would buy.’28 What were wise precautions to McNamara could well appear as a drive to superiority by the Soviets.
McNamara soon became aware of the problems with the new strategy. It offered no sure-fire way to bring a war to a successful conclusion and contained potential for the dreaded escalation. As seriously, it provided no answer to the question of sufficiency. Rather it was a recipe for an arms race, for as one side added targets so the other must add extra weapons to hit them. At first these concerns were put to one side as McNamara sought to introduce a modicum of control and flexibility into preparations for nuclear war. Also few political points were going to be lost by ending up with a nuclear surplus.
while a potential aggressor might have cause to question the credibility of a strategy that depends on a ‘city-busting’ retaliatory response, he should have no doubts about the credibility of a strategy based on the ability to destroy his remaining military forces and eventually prevail. The loss of these forces is one risk he cannot afford to take, for once deprived of his military strength he is compelled to accept peace on our terms.29
The Air Force therefore embraced the new doctrine with enthusiasm, to an extent that dismayed McNamara and his aides. When Ellsberg was drawing up new guidance in 1961 he was told that the USAF was prepared to go along with his approach, because it helped make the case for the new B-70 long-range bomber, and also required an accuracy that the Minuteman ICBMs could provide but the Navy’s Polaris SLBMs could not, emphasising once again the importance of the USAF’s contribution. Expensive new programmes were forwarded with a ‘counter-force’ label attached. The accuracy and discrimination of bombers as against missiles was underlined to justify the new B-70 bomber. As alarming was the ease with which Air Force officers spoke and wrote about nuclear war, as a normal military operation rather than as a hideous contingency. While he defended the new doctrine publicly through 1963 for the sake of consistency, by the start of that year McNamara had told the Air Force that it was not to take it ‘as a criterion for strategic force proposals’.30 In February complaints appeared in Air Force Magazine of a ‘deliberate intention to replace the necessary strategic superiority with strategic stalemate’.31
Another fact encouraging McNamara to play down counter-force was that, to the extent it was believed by Europeans to add credibility to the nuclear deterrent, it provided them with an excuse for not exerting themselves in the development of conventional options. In the end McNamara’s team killed off the B-70. It would only be useful for a first-strike because otherwise ‘parked and concentrated on some airfield’ they could ‘be knocked out by enemy ballistic missiles’.32
There was also a change in the political and intellectual climate. The turning point was the Cuban Missile Crisis of October 1962. The first consequence of this episode for a counter-force strategy was that, in this textbook crisis of the nuclear age, the strategy had proved to be irrelevant. In presenting his warning to the Russians, Kennedy threatened a ‘full retaliatory strike’; meanwhile, by deploying US B-47 aircraft in civilian airfields, the US could be seen to be denying the Soviet Union a counter-force option.33 A second consequence of the missile crisis was a move towards a degree of détente between the two superpowers, with the signing of the partial test ban treaty in 1963, thereby relieving the tension that had surrounded their relationship for the previous few years.
(a) a well-nigh limitless concern with saving face, regardless of risk, (b) a great deal of ground-in automaticity of response and counter-response- resulting in a swiftly accelerating ascent in the scale of violence.
Those responsible would not necessarily succumb to the ‘bundle of psychological factors’ summed up by loss of face and the tendencies to yield to feelings of hatred and rage that were said to explain escalation. In order to assess the genuine risks of escalation it was necessary to predict ‘the behavior of their leaders (as well as our own) under various kinds of crisis situation’. Brodie assumed that the arrival of strategic stability reduced the general risk of escalation, even as a result of tactical nuclear use. Early use of a few nuclear weapons was less likely to push things out of control than intensive and prolonged non-nuclear warfare.36
Each side may take certain positive steps either to bring the other to the bargaining table or to apply pressure during the negotiations. Sometimes these pressures tend to decrease with time or with a temporary solution to the problem at hand. At other times there is a tendency for each side to counter the other pressure with a somewhat stronger one of its own. This increasing pressure step by step is called ‘escalation’.39
The term ‘escalation’ focuses attention on the fact that each side can increase the stakes, and that bargaining is almost always crucial: that deterrence continues to function during and after an attack as well as preattack, and is likely to be as important as denial in preserving a country.
The sort of uncontrolled process that had once been described as ‘escalation’ Kahn now renamed as ‘eruption’.40
It is undoubtedly in the interest of limiting war that some obvious firebreaks and thresholds occur. There are subclasses or patterns or conventional boundaries to help find a stopping place. It is hard to stop without an obvious stopping place … some thresholds have made the claim to being the “ultimate limit”, the last stopping place before all-out war.41
The most durable idea to emerge from Kahn’s discussion was that of ‘escalation dominance’. This he defined as ‘a capacity, other things being equal, to enable the side possessing it to enjoy marked advantages in a given region of the escalation ladder.’ Dominating in this particular region would put the onus on the other side to take the risks of escalating to the next stage, which might be more deadly and dangerous. This only worked with a somewhat mechanistic view of the likely course of a conflict. The logic of escalation could be followed until one side found a point at which it was dominant. Yet in most conflicts there would be a variety of actions being taken at the same time, with each side looking for favourable opportunities and with no obligation to follow the enemy’s moves exactly. The implication of the theory was that a potential enemy must not be allowed to develop a superior capacity to operate at a particular region of the escalation ladder. This assumed that there would be no options to counter this presumed superiority other than high-risk escalation, and that it would be reasonable for the enemy to gamble on a form of surrender being preferable to taking this risk. In this way it suggested a possible form of victory other than a decisive first strike, based on one side being confronted with the consequences of its inferiority in a particular set of capabilities. This suggestion became a persistent feature of discussions of nuclear strategy in subsequent decades, providing an argument for boosting certain types of capabilities lest otherwise a potential enemy might look forward to a moment of dominance.
I feel very strongly that any belief in a limited quick action is an illusion and would lead us into a full war with Cuba on a step-by-step basis which would greatly increase the possibility of general war.42
In addition, the Cuba experience could also be taken as an example of the importance of offering inducements as well as making threats when trying to reach a settlement. Cuba was not a full test of controlled escalation because it did not extend into combat. Yet despite these reasons to question the argument, a consequence of Cuba was confidence in the possibility of effective and proportionate force. ‘Limited coercion’ offered an ‘unmistakable and undeniable politico-military grammar’ comprehensible to ‘pragmatic minds who were alert to war’s horrors, yet not inured to them’.43