After sending a final coded Critic message reporting their capture, the radio technicians in the SOD hut managed to smash the wired rotors of their cipher machines using a chipping hammer on the steel deck, just before the North Koreans boarded. Some of the material that they had been unable to burn had been shoved into lead-weighted canvas destruction bags, mattress covers, and laundry sacks on deck, but Bucher—incorrectly—understood that Navy regulations required that they be in water at least a hundred fathoms deep before jettisoning classified material, and the bags remained on deck when the ship was forced to turn toward land after failing to reach open sea.47
The crew was held for nearly a year by the North Koreans, who beat, tortured, and threatened to kill them to extort a series of highly publicized “confessions.” Insisting that the crew would be released only if the United States issued a formal apology for its “crimes,” the North Koreans finally accepted an absurd charade in which the American military representative, meeting with his counterpart at Panmunjon in the Korean DMZ in December 1968, would sign an acknowledgment that the ship had violated North Korean territory, but first make an oral statement repudiating the concession (“I will sign the document to free the crew and only to free the crew”). The Pueblo remained in North Korea, later becoming a museum and tourist attraction.48
A subsequent damage assessment by NSA upon debriefing the returned crew confirmed its worst fears. The destruction of classified material had been “highly disorganized” and was “accomplished in almost total confusion.” Some 80 percent of the documents on board were compromised, and only about 5 percent of the equipment had been destroyed beyond repair or usefulness. But even that was optimistic, as a large collection of maintenance manuals and spare parts gave away nearly as much as the intact machines would have. The 397 SIGINT documents on board included encyclopedic reference material on North Korean communications and a list of 126 “Specific Intelligence Collection Requirements” detailing the current state of U.S. knowledge of North Korean, Soviet, Chinese, and other target systems and the gaps NSA was seeking to fill. Combined with what their captors had learned from interrogating the crew, the loss of the Pueblo’s material had revealed “the full extent of U.S. SIGINT information on North Korean armed forces communications activities and U.S. successes in the techniques of collection, analysis, exploitation, and reporting applied to this target,” NSA found. Although not definitively linked to the Pueblo’s capture, changes in Soviet, Chinese, and North Korean communications security procedures in the following months may have been a result of the information revealed in the compromised material about U.S. successes. The overall loss to U.S. SIGINT capabilities, NSA’s “worst case” damage assessment concluded, had been “very severe.”49
In a final assessment, NSA deputy director Tordella was able to take solace on one point: the loss of the cipher machines had certainly resulted only in “minimal” damage to U.S. communications security. On the one hand, the KL-47 cipher machine and other U.S. encryption equipment had been a particular focus of the North Koreans in their relentless and at times brutal interrogations of the crewmen they identified as most knowledgeable:
Selected qualified cryptographic technicians of the USS PUEBLO were extensively interrogated by special and apparently highly competent North Korean experts on cryptographic principles, operating procedures, and the relationship of keying materials. The materials were neither displayed to the crew nor used for propaganda purposes. This is not the case for a variety of other significant intelligence materials. It appears that the North Koreans thoroughly understood the significance of these materials and the concealment of the details of the acquisition from the United States.
But, Tordella confidently continued, “All U.S. machines have been specifically designed to withstand attacks” based on recovering an intact machine or detailed information on how it works: only if an adversary also has the specific instructions for setting the machine for each key period would it be possible to actually break and read the traffic. Tordella would have been far less sanguine had he known that a few months before, a U.S. Navy communications technician named John Walker, serving aboard a nuclear submarine, had walked into a Soviet embassy abroad to offer his services, or that the North Koreans had swiftly passed on to the USSR the fruits of the Pueblo capture. For the next eighteen years the Soviets would have both the knowledge of how the KL-47 worked and a regular supply of the key lists needed to decode every month’s traffic.50
—
The U.S. Navy reacted to the first capture of an American commissioned vessel on the high seas since the War of 1812 by cutting its losses. All of the mistakes that had led to the loss of the Pueblo and its classified cargo were eminently correctable: there was no excuse whatever for a ship engaged in the Pueblo’s secret activities even to be carrying hundreds of extremely classified documents; her commander’s failure ever to conduct an emergency destruction drill was incomprehensible; the lack of a contingency plan for air support or a destroyer escort to come to the aid of a ship sent into such a dangerous spot was criminal.
And the concept of seaborne collection platforms was still a good one: it offered the best chance to carry out the close-in monitoring that was increasingly necessary given changes in radio frequencies taking place throughout the world and the growing limitations and difficulties of operating fixed land-based intercept sites. But in fact the Navy had never really been more than halfhearted about the mission, as was evident in its haphazard approach to training the ships’ crews and assigning officers. NSA had overall responsibility for “technical direction” of the SIGINT operations but delegated that authority to the Naval Security Group, and in another of those bizarre bureaucratic compromises that do nothing but sow confusion, NSA and the Navy agreed to alternate authority for “tasking” the ships from one mission to the next. The astonishingly poor training of the Pueblo’s SOD hut detachment and the rather contemptuous attitude of the regular Navy captain of the ship toward its “mysterious” doings reflected both the fact that no one was really in charge and that the Navy considered SIGINT missions on sardine-can ships a rather annoying distraction to its much more glamorous job of operating aircraft carriers and nuclear submarines. William O. Baker, who was serving on the President’s Foreign Intelligence Advisory Board (PFIAB), did not hesitate to blame the fiasco on the long-running inability of NSA to assert its statutory authority over the SIGINT operations of the military services. “It appears there was no real evaluation of the mission, certainly no suitable preparation for it, and of course no coordination with the experts and responsible managers in the field,” Baker wrote presidential adviser Clark Clifford, who chaired the PFIAB. “As we have recorded on many prior occasions…there will be a sequence of such episodes (prior ones have been similar aircraft ventures) uncoordinated and unjustified by authentic technical and operational requirements of SIGINT, until appropriate central leadership and authority are asserted.”51
Rather than try to find and correct the mistakes, the Navy simply pulled the plug on the whole shipboard collection program. The secretary of the Navy, declaring that the Pueblo’s crew had “suffered enough,” cut short an inquiry that sought to hold the ship’s officers or higher commanders accountable; facing budget cuts ordered by the new Republican administration, he then proposed eliminating the intelligence ships altogether rather than reduce the Navy’s combatant fleet. By the end of 1969 the last of the seaborne collection platforms was deactivated.52
The cutback was part of a larger retrenchment that all the services were facing owing to competing demands for money and personnel to fight the Vietnam War. NSA still depended heavily on the service cryptologic agencies to operate its field collection sites around the world, but the proliferation of microwave telephone links, radars, telemetry, VHF and UHF tactical voice, and radio teleprinter traffic to be covered was running headlong into the increasing difficulties the military services were facing in training and retaining capable intercept operators
.
In 1965, NSA decided to try manning an entire foreign intercept site with its own civilian employees, many of them former military enlisted men who had previously served with one of the service cryptologic agencies. In July 1966 an advance party from Fort Meade arrived to take over the U.S. Air Force station at Harrogate in northern England.53 Harrogate was a faded Victorian spa town not without its charms but abundantly steeped in the shabby postwar austerity that still hung over the country. Most of the new arrivals ended up in the inaptly named Grand Hotel, a decaying relic from the turn of the century. Its halls reeked of leaking gas and its creaky elevator regularly trapped newcomers who had not yet learned to take the stairs. One Sunday the dining room was hastily evacuated after a twenty-square-foot chunk of plaster came crashing down from the ceiling, missing by a few seconds a family that had started to sit down at the table directly underneath, then chose a nearby table instead.54
To update the dimly lit and badly subdivided operations building on base, thirty volunteers from the NSA team, headed by its chief, Hugh Erskine, spent one Friday night wielding sledgehammers to knock out 130 feet of reinforced concrete and cinderblock walls and 3,000 square feet of ceiling. For security reasons the station staff was required to clean the floors itself, so another do-it-yourself-minded NSA civilian came up with a plan to install a central vacuum system, and with some helpers drilled three-inch holes all over the floor, connected by a network of pipes running below. Only when the vacuum unit arrived after months of delay did they discover that there was not sufficient electric power to run the motor. By then an enterprising golfer on the staff had fashioned little numbered pennants for each hole to create a putting course.55
Despite the initial obstacles, the “civilianization” of Harrogate worked well. The larger problem, however, was the cost and political vulnerability of NSA’s sprawling global collection network, which by 1970 had ninety-one fixed field sites operating in seventeen foreign countries and the United States; in West Germany alone, the services and NSA operated fourteen different facilities. The HF sites were among the most manpower-intensive and expensive to operate; they were also the oldest, and their basic receiving equipment was still of World War II vintage and becoming expensive and difficult to maintain, with vacuum tube receivers that could no longer be accurately calibrated, manual scanners, and rhombic antenna fields that were a generation behind the circular arrays used at the newer elephant-cage installations.56
A 1968 study recommended consolidating and closing many sites on efficiency grounds, but also warned that rising nationalism in the Third World meant that the days were numbered for many of the bases NSA had long counted on. Within two years, U.S. intercept sites in Morocco and Pakistan were pulled out in the face of a wave of anti-American feeling; over the following decade NSA would lose its bases in Ethiopia, Vietnam, Iran, Taiwan, Thailand, and the Philippines as well.57 The cumulative toll that fighting the Cold War had taken on American credibility and standing in the world was for the first time imposing substantial limits on the covert means that from the start had been the chief U.S. way of countering the Soviet threat. The loss of trust in government at home, to which the Vietnam War added immeasurably, was about to intrude even more dramatically on the free hand NSA had always been able to count on to conduct its intelligence operations, free from public scrutiny or even political restraint.
10
Brute Force and Legerdemain
Though NSA may not have equaled General Motors (695,000 employees), AT&T (773,000), ITT (392,000), or America’s other largest employers in the year 1970, with a total workforce of 93,067 and a budget of more than $1 billion the agency was quite large enough to be suffering from all the recognized signs of organizational sclerosis. By its third decade of existence NSA had developed its own insular culture and accustomed ways of doing things and a vast multilayered bureaucracy made all the more resistant to change by its hermetically sealed secrecy. In the name of security, the agency had obtained exemptions from many government contracting rules and civil service hiring procedures, but that did not make NSA’s business and personnel offices more efficient or streamlined; it merely freed the most bureaucratic spies in the world to generate endless streams of their own red tape, unchecked by any outside scrutiny. An internal investigation a few years later found that it took NSA’s supply office ten days to deliver a box of paper clips to someone who had ordered one, which sounded like a joke but was not; it took thirteen days to send a letter from the time it was written, due to all the reviews and approvals required as it made its way through multiple layers of management; it took six months to issue a purchase order after the agency had received bids from all interested contractors.1
The NSA in-house newsletter of the 1950s, which read like a church bulletin or a Rotary Club events calendar, was superseded in 1974 by a staff-written Top Secret publication called Cryptolog. Visually it resembled a high school newspaper, with amateurish line drawings and clip art and a cacophony of typefaces, and its pages were filled with poems, recipes for spanakopita, humorous articles, and word games. But what was most striking were all of the insider allusions to the various cliques of the SIGINT world, each with their own fixed and dug-in attitudes, and to the manifest inability of management to make decisions, implement new technologies, or even get the 19,290 workers at NSA headquarters to talk to one another very much. Along with lots of cybernetic, linguistic, and cryptanalytic geekiness on display were endless careerist laments (“Let’s Give Linguists a Bigger Piece of Pie!” “Let’s Not Lose Our TA Skills”); a series of almost desperate pleas to readers from the agency’s reference and support branches to make use of their services (“Have you ever visited the NSA Cryptologic Collection? Have you ever heard of it?” “Next time, instead of 15 phone calls, make just one—to the Central Reference Service, x3258s. Central Information maintains an extensive collection of specialized dictionaries, reference books, journals, and documents. We have people trained in the research techniques necessary to make effective use of these sources”); and accounts of hilariously bungled attempts by management to “improve” working conditions. An article titled “The 2,000-Year-Old Transcriber” offered an NSA insider’s version of the Mel Brooks “2,000 Year Old Man” routine, describing the primitive tape recorders and painfully uncomfortable headphones they struggled with in the early years (not designed for repeated replaying of passages, the rewind function on the machines would almost always give out, forcing the transcribers to use a pencil stuck into the spindle to manually turn the tape back), and then the agency’s elaborately mismanaged program to develop custom-made “Transcribers Consoles” specifically tailored to NSA’s needs. They “looked like somebody’s idea of a Cape Canaveral space console,” but, as the stereotypical product of a committee, got almost everything wrong, with insufficient drawer space, tape recorders placed where they could not be reached while sitting, and metallic hollow spaces that resonated like a struck bell at every clack of the foot pedals or keystroke of the typewriter.2
More serious and revealing was a long and thoughtful series of articles by “Anne Exinterne” about the “unhealthy” atmosphere resulting from the large number of new hires who had been “wooed” to make NSA their career with early promotions; by the time they realized that they were in fact ill-suited to work that required not just the language or math aptitude that led to their being recruited but also an extraordinary tolerance for sheer drudgery, such as listening to tapes eight hours a day, many found themselves trapped by a high salary that they could not equal on the outside and “seven blank years on a resume form” that did little to interest a prospective employer in any case. With scant opportunity for recognition in their field and compartmentalization that often prevented employees from having a sense of how their work even served the agency’s mission, “morale is likely to plunge” and promotions—the only tangible measure of accomplishment in this closed world—“can become an obsession.” A remarkable number of the bright new
college graduates accepted into the agency’s fast-track intern program, initiated in 1965 to “assure that high quality pre-professionals” were given a broad training in the agency’s mission, ended up asking to be transferred to the personnel office within a few years after becoming “disenchanted with [their] assigned field.”3
Louis Tordella, who retired in 1974 after fifteen years as deputy director, was widely admired within the ranks for his technical competence and for bringing much-needed stability to the organization. But the flip side of stability was the near-total power he exercised, a situation that NSA’s extreme secrecy abetted. Even before officially assuming the deputy directorship in 1958, Tordella was already regarded as the most powerful person at NSA: his nickname was “Dr. No” for the decision-making authority he wielded to approve—or more often disapprove—new projects. “Everyone was kind of afraid of him,” recalled one long-term NSA official. Throughout his tenure, Tordella was the only executive aware of all of NSA’s most important programs and the single official who managed the relationship with CIA; his detractors within the agency found him an “ultrabureaucratic conspiratorialist” who kept information to himself and often used it to deflect outside criticism of, and pressure on, the agency.4
(Underscoring his unique position, a week after Tordella’s death in 1996, NSA historian Tom Johnson was delivered sixteen boxes of material from a safe that NSA employees found in Tordella’s office as they were cleaning it out. Tordella had kept the office while he continued to work as a consultant to the agency after his retirement, and the documents turned out to be a compendium of every single one of NSA’s most highly classified, compartmented programs of the post–World War II era. “Everyone knew he must have some such safe, but no one knew where it was,” Johnson remarked.)5
Code Warriors Page 35
