Increasingly, however, McNamara’s rhetorical backpedaling from counterforce reflected more than just budgetary bouts with the JCS. He was also beginning genuinely to doubt whether the strategy was feasible. In 1962, the JCS had told him that the Soviets were beginning to deploy SLBMs, like the Polaris, and that by mid-1967 they would have 186 of them, as well as 156 sub-launched cruise missiles—none of which could be easily located, much less suddenly destroyed all at once.
Second, photoreconnaissance satellites had revealed that the Soviets were starting to follow McNamara’s prudent example of encasing their ICBMs in concrete underground silos. Finally, the Whiz Kids in McNamara’s Office of Systems Analysis had calculated that under very favorable circumstances, even the number of bombers and missiles proposed by the Air Force would, by 1967, “at best be able to reduce Soviet strategic forces to roughly 100 ICBMs.” These, along with the 100 or so SLBMs that the Soviets would have on alert at sea, could kill fifty million Americans, assuming a modest fallout-shelter program. McNamara wrote to President Kennedy, “I do not consider this an ‘acceptable’ level of damage.” In the December 1963 DPM, he wrote to President Johnson, “The prospects for ‘Damage Limiting’ by counterforce attacks may not hold great promise in the latter part of the 1960s if the Soviets harden and disperse their ICBM force and build up their missile submarine forces as we now expect them to do.”
In January 1964, McNamara’s skepticism was nourished by an analysis produced, interestingly enough, by an Air Force general working out of Harold Brown’s Directorate of Defense Research & Engineering, or DDR&E. The study, by General Glenn Kent, was entitled Damage Limiting: A Rationale for the Allocation of Resources by the U.S. and the U.S.S.R. Kent was an unusual Air Force officer, a mathematician who had been director of plans in the late 1950s, took leave from 1961 to 1962 to study defense policy at Harvard, and returned to the Pentagon as an assistant to Harold Brown. Many were fooled at first by his mild manners and southern drawl, but he soon became every systems analyst’s favorite generaclass="underline" he knew what they were talking about, he could do the calculations.
When Kent first started working in DDR&E, he noticed Harold Brown looking at the results of a JCS computer war game that attempted to examine the effectiveness of antiballistic missiles, ABMs, in a “nuclear exchange” between the U.S. and the U.S.S.R. The methodology of the JCS game was to assume a Soviet attack of x warheads, and then to figure out how many ABM interceptors, with what probability of kill, would be needed to shoot down a high percentage of the incoming warheads. Kent pointed out to Brown that the calculations did not take into account the likely interactions between the Soviet and the American buildup; that the construction of an American defensive system would almost certainly prompt the Soviets to adjust and expand their own offensive arsenal, in which case he would have to do the calculations all over again. A more realistic analysis, Kent continued, would display curves on a graph indicating what was necessary—in terms of interceptors, dollars, whatever—to shoot down enemy missiles, given a wide range of attack tactics that the Soviets might choose.
Kent saw the three issues of ABM, counterforce and civil defense as intertwining facets of a single strategy called “damage-limiting.” Counterforce limited damage by destroying Soviet weapons before they were launched, ABMs by destroying them before they landed, shelters by protecting people from the effects of the weapons that made it through.
Brown told Kent to do a comprehensive study that would calculate what mix of these three programs would limit damage most cost-effectively and that would systematically account for the interaction between the buildup of American defense and Soviet offense, American damage-limiting and Soviet damage-inflicting.
Over the next several months, Kent worked on the calculations and drew twenty-nine graphs, plotting the relationship between the percent of surviving U.S. population and industry versus the number of attacking Soviet missiles. The graphs covered a wide variety of U.S. damage-limiting strategies and Soviet attack tactics.
The calculations revealed that damage-limiting was a fairly hopeless strategy. Kent did show that, under certain conditions, damage-limiting could make a difference of up to 55 or 60 percent in the amount of U.S. industry surviving a Soviet attack. However, if the Soviets reacted by expanding their own offensive forces, they could completely nullify the damage-limiting measures—and, moreover, do so far more cheaply than it would take for the United States to limit damage yet again. In the race between offense and defense, offense would win, and at lower cost.
When Robert McNamara read the Kent study in January 1964, he picked up on this theme immediately. Two graphs in particular riveted his attention. One plotted the cost in billions of dollars of a full fallout-shelter program, an ABM system and an anti-bomber air-defense system against the number of reliable Soviet missiles attacking the U.S. Different curves showed the relationship between cost and the number of missiles for varying percents of U.S. industry surviving the attack. The curves were hardly even curves; they looked more like straight vertical lines. If the U.S. wanted to protect 60 percent of its industry against an attack by 200 reliable Soviet missiles, it would require an $18 billion investment in damage-limiting. However, the curves showed that if the Soviets attacked with 300 missiles, a 50 percent increase in the offensive forces, the U.S. would have to double its damage-limiting budget to $35 billion. For each extra missile that the Soviets added to the attack, the U.S. would have to spend $140 million to offset its effects. If the U.S. wanted to protect 70 to 80 percent of industry (and roughly the same percentage of population), it would cost $260 million to $500 million to offset a single additional Soviet missile. If the tolerance of damage was much lower and one could stand only 50 percent surviving, the U.S. would have to spend $80 million on defenses for each missile that the Soviets added to the attack. As Kent calculated it, adding one missile to their arsenal would cost the Soviets the equivalent of only about $25 million. In other words, to limit damage so that half the industry (and about 60 percent of the population) survived an attack, the U.S. would have to spend $3.20 for each $1 paid by the U.S.S.R.—a losing proposition.
The other graph that impressed McNamara showed the cost in billions of dollars of the most economical mix of all damage-limiting measures—fallout shelters, ABMs, bomber defense and counterforce strikes—versus the cost to the Soviets of an attack against the U.S. Again, the curves measured this relationship for 70, 60, 50 and 40 percent of U.S. industry surviving the attack. The curves were only slightly less steep. For each extra dollar that the Soviets added to the attack forces, the U.S. would have to spend $3 to protect 70 percent of its industry, $2 to save 60 percent, $1.80 to defend half, and the same $1 to defend a mere 40 percent. Again, it seemed a losing battle.
Kent also warned Brown, who passed along the message to McNamara, that all of these calculations assumed extremely effective ABM systems. Specifically, they assumed that each ABM interceptor had a “kill probability” of .8—that for each ten incoming Soviet warheads, the ABMs could shoot down eight. This figure was based on official data provided by the Army, which had high stakes in deploying vast ABM systems. Kent, Brown and McNamara doubted whether the actual kill probability would be even half that high.