Code Warriors

by Stephen Budiansky

  SAC’s targeting briefing was entitled “To Kill a Nation,” and it proposed to carry out the emergency war plan by destroying seventy Soviet cities with 133 atomic bombs. The precise targets were only vaguely specified. But LeMay grandly argued that the effect of the “Sunday punch” would be to make the Soviet Union lose the “will” to wage war, to “stun the enemy into submission,” even to cause the entire country “to collapse.” Given the huge costs of matching the Soviets’ conventional forces in Europe, U.S. Air Force planners, sounding more like actuaries than generals, argued that the ability to deliver a crushing atomic attack in response to any Soviet aggression “was an opportunity to put warfare on an economical, sensible, reasonable basis.” All of this required making the threat of atomic airstrikes wholly believable, and that meant that SAC’s bomber force had to be “combat ready”—one of LeMay’s favorite phrases—capable of flying thousands of miles, penetrating Soviet airspace, accurately locating their targets, and delivering their atomic payloads. Initially appalled at the poor performance of SAC’s aircrews, LeMay drove them just as he had harried his B-29 squadrons during the war in the Pacific, making them operate under realistic wartime conditions of high altitude, bad weather, and difficult targeting situations. “We had to be ready to go to war not next week, not tomorrow, but this afternoon. We had to operate every day as if we were at war,” LeMay wrote.7

  The Soviet atomic bomb, of course, added a more urgent reason to be sure that American bombers could get to their targets and successfully hit them on a moment’s notice: in 1950, the Joint Chiefs of Staff revised the war plan to give top priority to “the destruction of known targets affecting the Soviet capability to deliver atomic bombs.” So grave a conundrum did the threat of a Soviet atomic strike pose to American war planners that the Joint Chiefs proposed asking Congress to authorize the president to carry out a preemptive atomic strike on any nation engaged in acts of aggression, defined to include “the readying of atomic weapons against us.”8

  Almost nothing was known about Soviet air defenses, however, and if LeMay was right that the Cold War had to be treated as a real war, taking wartime risks to ensure operational success was justified. In the spring of 1950, Truman approved a plan for surveillance flights to assess Soviet air defenses.9 Although the ferrets were to avoid actually crossing into Soviet territory, the aim was to fly close enough to the border to provoke a reaction by radar stations and fighter squadrons so that their communications and electronic emissions could be detected and monitored.

  —

  Three hours and forty minutes into the Turbulent Turtle’s flight early on the afternoon of April 8, 1950, the crew radioed that they had reached the end of their assigned mission path, a point roughly even with the northern tip of Latvia, and were turning back home for the return sweep along the Soviet coast. Half an hour later the plane radioed in again to report its location. That was the last it was ever heard from. At 2330 that night the squadron’s headquarters at Port Lyautey received a dispatch from the U.S. naval base at Bremerhaven reporting that the plane was declared overdue. At first light the next morning, three PB4Y-2s were dispatched from Morocco to head east to look for the missing aircraft. Over the next ten days two dozen more aircraft joined the search. A few days after the search was called off a Swedish fishing boat picked up an empty life raft bearing the missing plane’s serial number.

  An official diplomatic protest by the United States was met by a Soviet statement that “a B-29” had overflown Soviet territory, been challenged by Russian fighters over the Latvian coastal town of Liepāja, and was shot down over the sea when it refused to obey orders to land and opened fire as it tried to flee. As the squadron’s history drily observed, “The credibility of the Soviet report was seriously weakened by the fact that the Privateer’s only armament was a .45-cal. pistol carried by one of the officer crewmen.” The ten-man crew was never found, though years later several prisoners released from Soviet prison camps claimed they had seen or heard about some of the men.10

  The flight path of the U.S. Navy PB4Y-2 electronic surveillance plane shot down by Soviet fighters off the coast of Latvia on April 8, 1950.

  Over the next fourteen years, twelve more U.S. Navy and Air Force ferret aircraft would be shot down over the Baltic, the Sea of Japan, the Kamchatka Peninsula, East Germany, and Soviet Armenia; dozens more would be buzzed or fired on but escape more serious harm. In many cases ferret flights approached as close as a few miles to the twelve-nautical-mile territorial limit, often paralleling the coast for long distances, in what seemed almost a direct taunt to Soviet air defenses. Many of the Air Force ferret aircraft were modified B-29s, which by external appearances were no different from atomic-capable bombers. The riskiest missions sent a SAC bomber racing toward the Soviet border to deliberately provoke the Russians to scramble fighters and activate their radar networks while a ferret flew nearby to record the reaction. More than ninety men would lose their lives in these electronic intelligence-gathering missions around the periphery of the Soviet empire. When forced to make a public announcement about the losses, the U.S. military would say only that the planes had been conducting “weather reconnaissance.”11

  Signals intelligence was starting to look a lot less like a clean, austere, intellectual exercise and a lot more like the dangerous and dirty business that spying had always been.

  —

  The gathering of data from radar signals acquired its own acronym, ELINT, which stood for electronic intelligence, the distinction being that these sorts of radio emissions were not communications per se and contained no messages to be deciphered or read. All three military services considered ELINT a separate business entirely from the work of the cryptologic agencies, adding yet another turf battle to the increasingly messy bureaucratic struggle over the postwar intelligence structure. A chief purpose of ELINT collection was to record technical details of each of the Soviet radar systems—the frequencies they operated on, the kind of pulses they employed—so that jammers and other electronic countermeasures could be designed to block or spoof them effectively. The Air Force also wanted to thoroughly map Soviet air bases and their associated radar stations so that they could be targeted as part of SAC’s plan for penetrating Soviet airspace in an all-out attack.

  But ELINT was also a potential source of a great deal of operational, strategic, and tactical intelligence. Once the radar systems associated with a particular kind of unit, such as a tank division or a motorized rifle regiment, had been identified, it was possible to begin constructing an “electronic order of battle” that could reveal volumes about the organization of the Soviet military, and even offer immediate warning of the movement of Soviet ground forces if those characteristic signals started showing up in new locations.12

  With the ongoing blackout of significant cryptanalytic intelligence following Black Friday, ELINT assumed even greater importance, especially when it came to sources of possible strategic warning of Soviet preparations to launch an atomic strike of their own against the United States. Existing radar coverage of the continental United States could offer little better than one hour’s warning of approaching Soviet bombers. The planned joint U.S.-Canadian Distant Early Warning, or DEW, line, a chain of radars across northern Canada, Alaska, and Greenland, would extend that to four to six hours, enough time to shift six hundred to one thousand fighter aircraft to the Air Defense Command from other tactical units.

  But despite all of LeMay’s boasts, a secret investigation in 1957 found that even with a decade of effort to put SAC on a constant war footing, not a single one of its bombers would be ready to get off the ground within six hours on a randomly chosen day.13 If warning of Soviet preparations for an attack could be extended to three to six days, however, the consequences would be “of tremendous magnitude,” a high-level scientific review committee that examined signals intelligence and strategic warning concluded. “In fact, warning of this nature” would permit SAC to fully deploy its forces to be ready for a
n immediate counterattack if the Soviets then went through with their plans. Moreover, SAC’s own preparatory mobilization, “if known to the enemy, might induce him to cease his preparations for an attack.”14

  The scientific committee was chaired by Howard P. Robertson, a physicist at Caltech. The Robertson panel was not very encouraging about the dismal state of efforts to break back into the Soviet high-level code systems. In the absence of being able to read the actual transmitted orders of the Soviet high command, signals intelligence derived from collating plain-language intercepts, low-level tactical code systems, and traffic analysis—a process known as “T/A Fusion”—remained the only hope. In fact, the panel found, “Since late 1948, we have been forced to rely on information obtained by T/A Fusion for almost all COMINT on the Soviet Armed Forces,” including almost all of the direct indicators of movements by long-range Soviet bomber forces or other steps suggesting an impending attack.15

  Still, it was not as forlorn a hope as might have first appeared. Traffic analysis had repeatedly proved its value going back to World War I, when cryptologists at the British Admiralty noticed that sorties of the German fleet from the Heligoland Bight into the North Sea were invariably preceded by a flurry of radioed messages to minesweepers, air patrols, and units in charge of the booms blocking the estuary entrance.16

  In World War II, Bletchley’s traffic analysts were able to map out the radio networks and call procedures used to assemble U-boats into position for “wolf pack” attacks, and even during periods when the naval Enigma messages could not be broken they were frequently able to forecast impending attacks hours or even days in advance just from the pattern of communications and direction-finding fixes on transmitting U-boats. Call signs, transmission schedules, operator “chatter” requesting frequency changes or repetitions of messages or setting up future transmission times, even careful analysis of the distinctive idiosyncrasies of a Morse code operator’s “fist” and the unique radio fingerprint of stray noise produced by a specific transmitter, could identify and locate a particular ship or unit even when the messages themselves could not be read. Rapid sending without any request from the receiving party for service was usually a sign that only dummy traffic was being sent; its sudden replacement with real traffic often indicated imminent plans for a military operation.17

  In view of “our present heavy dependence on T/A” for strategic warning of Soviet military action, the Robertson panel accordingly urged that considerably more attention be given to improving the equipment needed to monitor radar and other signals, perform rapid and accurate direction fixes, and identify transmitters by radio fingerprinting; that trained analysts be stationed at field intercept sites to help notice any suspicious indications in Soviet radio traffic patterns; and that the current system for transmitting “Flash” messages be improved so that warning could reach Washington in minutes if necessary.

  If there was anyone predisposed to value the role of top-level scientific expertise in intelligence, it was Howard Robertson. He had served as scientific intelligence adviser at Eisenhower’s headquarters during the war and helped to interrogate captured German rocket scientists on the V-2 project. He was also one of the world’s leading mathematical physicists, one of the originators of the cosmological concept of an expanding universe. But his panel took pains to emphasize that the jobs the cryptologic agencies still tended to regard as mere “support” to the mathematically adept cryptanalysts had to be given far more prestige, and in particular the disparity in pay and promotion between cryptanalysts and traffic analysts needed to be corrected.18 In an era when cryptanalysis was no longer king, and might in fact never be again, the heretofore unquestioned place of the cryptanalyst, at the top of Arlington Hall’s pecking order urgently needed rethinking.

  —

  At the end of World War II, Soviet and American troops had occupied the Korean peninsula almost as an afterthought. Part of Japan’s Asian empire since 1910, Korea was suddenly freed of its foreign yoke with the surrender of Japan, and the Allied troops that had been preparing to invade Japanese-held Manchuria and the home islands took swift possession of the Korean peninsula, with the United States and the USSR agreeing on the 38th parallel, which roughly divided the country in half, as the north-south demarcation line of their occupation zones. Elections under UN auspices were to follow to select a new government for the entire country. Stalin, however, refused to allow the Soviet-occupied zone to participate in the vote, instead establishing a Communist government under Kim Il-sung, a Korean who had been trained as an infantry officer in the Soviet Red Army during the war and who had promptly begun erecting a Stalinist cult of personality around himself. The election went ahead in the South in May 1948, choosing as the Republic of Korea’s first president Syngman Rhee, a seventy-three-year-old leader of the movement for Korean independence who had spent decades in exile in the United States, where he earned a PhD in political science from Princeton. With the subsequent withdrawal of U.S. and Soviet troops, the Korean peninsula was left split between two rival governments, each claiming to be the sole legitimate representative of the entire nation.

  In the spring of 1950, North Korea was literally at the bottom of the list of problems that concerned U.S. intelligence analysts. The U.S. Communications Intelligence Board had tried to offer some guidance to the cryptologic agencies on where to focus their attention, drawing up three lists of priorities. The only item involving Korea to make it to the “A” list (topics “of greatest concern to U.S. policy or security”) was “Soviet activities in North Korea,” but even that was dropped when the list was revised on May 15. The “B” list, which was for items of “high importance” to be given “expeditious” treatment if possible, included at the very end, “North Korean–Chinese Communist relations” and “North Korean–South Korean relations, including actions of armed units in border areas.” But even when a “Watch Committee” set up by CIA to monitor Soviet moves became increasingly worried about North Korean intentions in April, that concern was never communicated to Arlington Hall through the USCIB mechanism.19

  A U.S. Army intercept station in Japan the previous year had stumbled on some signals of an unidentified network using Soviet communications procedures, and through direction-finding fixes concluded they might be North Korean. After Army G-2 requested a more deliberate search for North Korean signals on April 21, 1950, the intercept stations in Japan were able to copy 220 enciphered messages and sent them back to Arlington Hall for analysis. It was indicative of the all-consuming attention that the Soviet Union held that the staff available to work on this North Korean material consisted of one part-time traffic analyst, one part-time cryptanalyst, and one Japanese linguist who had been trying to teach himself Korean in his spare time over the previous year; the section had no Korean typewriters, Korean-English dictionaries, or reference books.20

  The situation in the field was even worse, exacerbated by a halfhearted but completely chaotic reorganization instituted at the Pentagon to try to bring the cryptologic agencies of the three military services under centralized control. In May 1949 the Joint Chiefs of Staff had established the Armed Forces Security Agency (AFSA) to consolidate communications intelligence and security activities. But it was the usual bureaucratic compromise that papered over all of the real problems and left the effective power of the Army Security Agency, the Naval Security Group (as Op-20-G was now called), and the newly created U.S. Air Force Security Service (USAFSS) virtually untouched. In the Pacific, ASA remained completely in charge of the three field intercept sites in Japan plus one in the Philippines that the Army operated. ASA also maintained its own regional command headquarters in Tokyo, called ASAPAC—Army Security Agency, Pacific. With a staff of 47 officers and 192 enlisted men, housed in a former Imperial Japanese Army arsenal, ASAPAC not only directed local field operations but even acted as a mini Arlington Hall, processing intercepts and attempting cryptanalytic attacks with its own IBM equipment. Rather than solving the problems of dupl
ication and lack of central direction, the creation of AFSA seemed just to have added a fourth agency to fight for the same spoils, and it was utterly unclear where the dividing line of responsibilities between AFSA and ASAPAC lay when it came to working on the traffic being intercepted in the Far East. Most of ASA’s best officers and equipment had been transferred to AFSA in the reorganization, however, which meant that even when it came to activities that fundamentally were the job of field offices, the work suffered from inexperienced leadership, overstretched staffs, and plunging morale.21

  Reflecting both overall intelligence priorities and their parochial service interests, all three of the services’ intercept units in Japan were focused almost exclusively on the Soviets. ASAPAC was collecting Soviet military and plain-language commercial traffic; the Air Force had a small unit, the 1st Radio Squadron Mobile (RSM), at Johnson Air Base outside Tokyo, which monitored Soviet air and air defense signals; and the Naval Communication Unit at Yokosuka was targeting the Soviet Far East Fleet’s communications. Although ASAPAC did break a few low-level Soviet military codes, the pickings were slim, which added to the frustrations and low morale. The men who operated the radios at the field sites were mostly enlisted personnel, and the stations in Japan suffered constantly from low reenlistment rates and high turnover. At times their most notable distinction seemed to be the records they set for disciplinary infractions, drinking problems, and the number of VD cases reported.

  As an Air Force report would later acknowledge, the radio intercept capability at the outbreak of the Korean War was “pitifully small and concentrated in the wrong places.”22