When all these factors were taken into account, it appeared that the Soviets could have built 500 or so intercontinental bombers by the early 1960s.
Yet there was another assumption that entered into these calculations, something less tangible but, in the eyes of intelligence analysts of the day, far more real and certain. The Soviet Union’s primary goal was to attack a large number of strategic and urban-industrial targets inside the United States. U.S. targeting studies had revealed that the Soviets would need something like 500 bombers of intercontinental range to accomplish the goals that intelligence had imputed to them. Therefore, any evidence that seemed to confirm the assumption about Soviet aims—regardless of evidence that might point to other conclusions—was viewed as truth.
At that time, the Central Intelligence Agency had no charter to do military analysis; that job was assigned to the intelligence staffs of the individual services. However, there was an oversight board, the Office of National Estimates, ONE, headed by a veteran intelligence analyst, Sherman Kent. ONE was responsible for producing the annual National Intelligence Estimate, and so had the authority to look into intelligence in all fields, including military. While technically independent of the CIA, it was housed in the Agency’s headquarters. Still, ONE had to rely almost entirely on Air Force Intelligence for analysis and estimates of Soviet long-range missiles and bombers.
However, there was a division of the CIA in charge of economic intelligence, headed by a young analyst named Ed Proctor, who managed to grab one slice of military intelligence: trying to calculate how much money the Soviets were spending on their armed forces. This task allowed Proctor and his staff to obtain as much data as they wanted on Soviet bombers, which allowed them to get heavily involved with the whole question of bomber production and production rates. In short, through a cleverly roundabout route, the CIA’s economic division got into the business of analyzing all the technical and arcane issues that lay at the very heart of the Strategic Forces section of the NIE—formally the exclusive province of the military services, especially the Air Force.
From their studies of other economic sectors of the U.S.S.R., the analysts in Proctor’s shop knew practically everything there was to know about Soviet factory markings—things like how serial numbers on manufactured goods can reveal what year and month a particular item was produced. In this sense, airplanes were just like any other manufactured goods, and the serial number, so to speak, was the tail number on each plane, which the CIA could detect on a few of the many photographs taken with long-range lenses by the American air attachés observing the goings-on at the Fili Plant in Moscow.
Not long after they began amassing this sort of data, Proctor and his team began to conclude that the Air Force estimate—the official National Intelligence Estimate—could not be right. One of the assumptions behind that estimate was that the Bison bombers were produced in batches of ten. The assumption was integral to all the other assumptions and, thus, to the overall estimate. Ten was the logical number, given the Air Force estimates of the Fili Plant’s floor space, of the plant’s “learning curve,” of the number of labor shifts working on production. And ten was the absolutely necessary number, given the more basic assumption that the Soviets wanted to be able to attack a whole variety of American targets as soon as possible. If the number were significantly less than ten, then all the other assumptions were wrong, including the basic one concerning Soviet aims and intentions.
And yet the CIA was coming to believe that the Soviets were producing the Bisons in batches of only five. The analysts would see Bisons marked with tail numbers ending with 10, 11, 12, 13, 14 or 20, 21, 22, 23, 24 or 30, 31, 32, 33, 34—but not a single plane ending with 15, 16, 17, 18, 19 or 25, 26, 27, 28, 29…. Moreover, there was another set of numbers on the plane indicating when it was manufactured. As it turned out, about as much time elapsed between the plane with numbers ending in, say, 22 and 24 as between planes ending in 24 and 31. In other words, there probably would not have been enough time to produce 25, 26, 27, 28, 29 and 30 in the interim.
Thus, if no airplanes with higher end numbers existed, then that clearly meant they were being produced in batches of five, not ten; and that clearly meant that the estimate was all wrong, that the Soviets were producing only about half as many Bison bombers as the NIE projected, and that meant that all the other assumptions behind the NIE—from the efficiency of a Soviet aircraft plant to the objectives of the Kremlin—were also wrong.
By 1957, not only was Proctor’s shop convinced that the NIE was wrong, but so were a number of analysts on the staff of the Office of National Estimates.
Allen Dulles was in a spot. As manager of the entire intelligence community, he was reluctant to abandon the estimate of the Air Force, the source of all the data that the community was receiving on Soviet bomber production. He was also reluctant to accept immediately the critique made by Proctor’s shop. That division might know a lot about economics and factory markings, but could he really believe that the men who worked there knew as much about airplanes as the Air Force did? They had never even looked much at airplanes before they got involved in this exercise. Furthermore, if the NIEs that the Agency had been supporting the past few years were based on totally faulty data and assumptions, how would they come up with a new estimate—who could produce one—and what would that say, politically, about the wisdom of the CIA?
So, a fight broke out between Air Force Intelligence and the CIA. The Air Force had stakes beyond merely protecting its reputation as a reliable intelligence agency: a large Soviet strategic air force meant guaranteed support for a large American strategic air force, which meant more prestige and a greater share of the defense budget than for the Army and the Navy. Not surprisingly, in this internal clash, Army Intelligence sided with the CIA’s economic division.
The Air Force response to the CIA critique sounded entirely reasonable. The Air Force, its intelligence officers pointed out, made a logical estimate. It accorded with everybody’s perception of Soviet intentions; it accorded with their estimates of the Fili Plant’s floor space, with the July 1955 air display over Moscow and with their judgment that the Soviets produced ten planes per batch. All that the CIA analysts had was the absence of any data that proved conclusively the estimate of ten per batch. The Air Force didn’t think that the sample size was large enough to conclude that the Soviets produced only five per batch, at least not in the face of all the other conflicting information.
The CIA economic division’s response was equally logical, but in precisely the opposite direction. The five-per-batch number that they had come up with was absolutely solid, they said, and the confidence levels were very high. This meant that the Fili Plant was not producing to what the Air Force thought was full capacity, that the plant did not work in two shifts, that its learning curve had not yet peaked, that the Soviets were not planning to build a second Bison plant. The air show of July 1955 was a bit tougher to challenge, but all the other bits of data suggested either that it had been a fluke or that the Soviets must have been flying every single bomber in their inventory, not, as the Air Force had assumed, half of them.[2]
2
Years later, some intelligence analysts would speculate that the same planes might have been flying around the display area twice, but the theory has never been confirmed.