AS THE SOVIET NUCLEAR DOCTRINE evolved to preempt an American nuclear surprise, the Americans realized in the late 1970s that their plans for continuity of government were far too vulnerable to a Soviet attack. The leader of the Free World could not endure or survive a nuclear first strike.
On July 7, 1977, at 7:34 a.m., a simulated alert pulsed along secure White House Communications Agency circuits to the Anacostia alert facility for the Marine One helicopter, and to other locations. Seven minutes later, the designated “gold” squadron pilot landed on the South Lawn. Nine minutes later, at 7:50 a.m., the helo, having scooped up the president, a few key staff members, and military aides, rendezvoused at Andrews Air Force Base with the National Emergency Airborne Command Post jet, (known as NEACP, pronounced “Kneecap”). The airplane then sluiced down the runway and took off. Time was called. In just 22 minutes, the president could be evacuated, in the air, and safe from attack.1 President Jimmy Carter knew that the exercises did not realistically simulate the threat that existed. Even before the Soviets developed ICBMs capable of avoiding the eyes of American imagery and measure-and-signature satellites, the USSR’s deep-water navy was capable of launching a submarine-based ballistic missile at Washington, DC, that could cruise undetected entirely until conventional air defense radar along the coasts of the United States picked it up. Since these missiles didn’t arc into the atmosphere, and since their launch did not give off much of a flash, this strategic capability of the USSR became one basis for stealth attack rehearsals that almost never succeeded in evacuating key leaders in time. The second: a submarine-launched ballistic missile (SLBM) fired anywhere off the coasts would explode during the upward portion of its trajectory—about 3 minutes after launch—and generate an electromagnetic pulse that shorted every electrical circuit on the East Coast.
The nuclear war plan Carter was given and the defense budget he inherited had been revised to reflect these new fears. The updated nuclear war menu had four major attack options:
1. destroy all threatening military targets in the Soviet Union and Eastern Europe,
2. destroy Soviet Union and Eastern European military targets plus all economic and industrial targets that would allow the Soviets to reconstitute a nuclear capacity,
3. destroy Chinese military and leadership threats, and
4. destroy Chinese military and leadership threats and their reconstitution capability.
In addition, it included eleven selective attack options and six categories of withholds—the “destroy these but not those” options to preserve the appearance of an ability to terminate a war.2 The strikes he could launch were smaller in mega tonnage. The communication networks he would use to launch them would be resilient, according to the Pentagon. Their idea was that the president could limit the spreading contagion of nuclear war by devastating the world a bit less, or by controlling its escalation in the aftermath of an attack. Carter’s National Security Advisor, Zbigniew Brzezinski, didn’t buy it. But no one had a better idea.
The Pentagon’s guestimate was that the president would have, at best, three minutes between notification and detonation.3 One reason was that Soviet SSBNs often parked themselves just outside of US territorial waters, perhaps 12 miles off the coast. They didn’t yet carry the more accurate nukes, but one day soon they would. Evacuating the president was a necessary, albeit wholly insufficient, part of preserving continuity. If the attack seemed limited, the White House Emergency Plan called on the Secret Service and White House Military Office to fly or drive the president to one of eight underground bunkers within 100 miles of Washington, DC. Each offered line-of-sight microwave communications with the Pentagon and the Alternate National Military Command Center, located beneath granite rock on the border of Pennsylvania and Maryland.4 The Soviets knew where these failsafe locations were, though; this safety net was very loose. Even if the president were to find his way to a plane in the air, or to a safe house, he would be a figurehead without a way to talk to the country unless several hundred others, including a thirty-two-person emergency operations team, had successfully made it to their own pre-prepared sites. The president could speak to the nation only if the Emergency Broadcast System was operational; this required the activation of the circuit within the enormous bunker at Mount Weather, in Virginia, a circuit that had no reliable backup until the end of the Reagan presidency.5 And if, in the air, the president needed to transmit emergency war orders, it was not certain that he could so on a secure voice circuit, particularly if he was overseas.
Carter “wanted to be able to be awakened at three o’clock in the morning and not be confused, and understand what he was going to see, what the voice would sound like on the other end of the line, and that sort of thing,” an aide said.6 After the drill, Brzezinski directed Colonel William Odom, his military assistant, to conduct a soup-to-nuts review of the entire system. He would first examine presidential continuity programs; he would also, upon Brzezinski’s orders, be given full access to the Pentagon’s command and control procedures. Were they sufficient? And if so, what were they sufficient for? Odom found a few positives. One was that there were a number of redundant communication systems that linked the place which would first pick up signs of a nuclear launch—NORAD—with the person who would made the call to retaliate—the president—and the nuclear commands that would execute the orders, primarily, the Strategic Air Command in Omaha, Nebraska.7 This satisfied one basic requirement: “to recognize that we are under attack, to characterize that attack, to get a decision from the president, and to disseminate that decision prior to the first weapon impacting upon the United States,” according to SAC’s commander in chief, Richard Ellis.8 Odom looked at ongoing tests of the emergency conference network that would connect the president to his nuclear commanders.9 The calls could be convened quickly. But the network itself was not secure. The Soviets could easily tap in.
Although AT&T had figured out a way to give these presidential communication systems priority routing on their public telephone exchanges, rerouting the traffic after nuclear blasts would be akin to moving every passenger to a back of an airplane in free-fall.10
Odom found that a system called SACCS (the Strategic Automated Command and Control System) was highly vulnerable and often blinked out in bad weather. SACCS was critical to communication in any nuclear war because it allowed ICBM wing commanders to receive targeting instructions and distribute them to launch silos.
What disturbed Odom, as he learned more about the system, was what he heard from the nuclear commanders. They did not believe that civilians on the White House staff paid any attention to the war plans. The Pentagon had no way to reconstitute nuclear forces after an attack, or to reestablish presidential command and control after it was lost. The military had spent billions since the days where they were clearly superior to the Soviet Union on communications networks (eight of them in tandem operating worldwide), its command centers (more than one hundred), its computer networks (about sixty-five), and its regional communication nodes (eighty-five), all part of a contraption called the Worldwide Military Command and Control System (WWMCCS).
The Cuban Missile Crisis accelerated the fielding of command post aircraft for theatre commanders, a limited Post Attack Command and Control System 135 aircraft that would allow a general on board to transmit launch codes to ICBM silos from the air, and failing that, a purse full of communications rockets designed to transmit nuclear “go” codes without human interventions to the poor souls in surviving ICBM bases.11
The Strategic Air Command had planes that could receive information about damage and detonations and then report it back to whoever was in charge, and they could transmit a limited number of emergency war orders from the air—all improvements. But “things would just cease in their world about 6 to 10 hours after they received the order to execute the SIOP,” Odom later wrote.12 The military relied on studies that assumed that the five primary command centers that the president might use—the White House, the NMCC at the Pentagon, the Alternate National Military Command Center, the secure communications facility under Camp David (codename Orange One), and the underground bunker at Mount Weather (codename Site B)—would be incapacitated by as few as nine ballistic missiles within 30 minutes of any attack.13
Another 35 ICBMs would obliterate the fixed primary and alternate command posts that all SAC bases relied on.14 As one study concluded, “the national political and joint military command structure . . . is highly vulnerable and could not be counted upon to complete its minimum essential retaliatory functions if attacked.”15 The American nuclear war plan was supposed to be used only as a deterrent, for defense. But increasingly, its writers realized it could never be used if the other side struck first.
Getting word, too, was a problem. The commander of NORAD had one job: to provide the president with tactical warning and attack assessment. The Pentagon had tried for years to update the early-warning sensors that NORAD used to detect incoming missiles and planes. The sensors were sophisticated, but the computers were not. Slight changes in atmospheric density could perturb the entire system. Everything ran on the same set of computers, and software updates to, for example, the air defense algorithms could, and did, migrate to cause glitches in other systems, like those that translated raw data about ballistic missiles into pixels that humans could see on a screen.16
On November 9, 1979, engineers uploaded a computer tape containing a hypothetical version of the Soviet nuclear war plan onto an auxiliary computer. For reasons that remain unknown to this day, that exercise data leaked into the processing systems that fed the main display terminals at NORAD’s command center. Then, the information relayed itself instantaneously to the National Military Command Center in the Pentagon, to the White House Situation Room, to the Strategic Air Command, to the Alternate National Military Command Center. Immediately, the NMCC convened an air threat conference call with watch officers. The alert ended within six minutes when none of the independent sensors used to confirm missile attacks registered any anomalies, but not before the duty officer at the White House had called Brzezinski and informed that 220—check that, 2,400—missiles were on the way.17 NORAD wisely decided to move all its testing outside of Cheyenne Mountain after that.18
Five months later, a data processor at an early-warning radar site at Mount Hebo in California computed an erroneous trajectory for two submarine-launched ballistic missiles that had taken off from the waters off the Kuril Islands in the Pacific. The US had been informed about the tests beforehand, but, as the battle staff watched at NORAD, the projected ground zero for one of them suddenly appeared in the shape of a fan over the Western United States.
Three months after that, an inexpensive transistor malfunctioned three times in three different systems over the course of nine days, and started injecting bits of data into the stream of ones and zeros that NORAD’s central processors were receiving from its sensors. Each time, the Pentagon had to convene an emergency conference call. It happened again a few months after that.19 The Soviets naturally saw these alerts for what they were: a demonstration of incredible vulnerability and a source of danger. During each one, the Strategic Air Command automatically generated some of its alert assets, including, at various points, the emergency presidential backup plane and even nuclear-armed fighter bombers.20 Malfunctions had loosened the safety catch on the nuclear trigger without human intervention or intention.
That was one reason why Brezhnev, in a telegram to President Carter, said that error should cause the leader of any country “extreme anxiety.” “What kind of mechanism is it that allows a possibility of such incidents?” A defensive Pentagon gave the White House fighting language to use in reply. Instead of acknowledging the error, they said that Brezhnev’s information was based on false Soviet propaganda.21
The snotty response by the United States covered for the truth: the early-warning system was not reliable. Forget decision time: how could the president know that the information getting to him was real? Odom at the time worried that the Russians could just as easily have spoofed the system. He might have known that the air force and the National Security Agency were developing elaborate plans to spoof the Soviet’s own sensors, too.
In April 1978, NORAD generated an air sovereignty alert, and the NMCC convened a threat assessment call. The US was testing its new Poseidon missiles that week, and the Soviets were conducting provocative drills with their bombers, some of them flying close to US territory. For unknown reasons, the White House Situation Room duty officer thought the call was a test and did nothing. The National Security Council was out of the loop.22 Human error was rife throughout the supposedly fail-safe nuclear command and control system.
For political leaders, this was scary stuff. The system could not keep pace with the demands placed on it. As Brezhnev said in his first missive to Carter, “It turns out that the world can find itself on the brink of a precipice without the knowledge of its president or of other US leaders.” The Soviets took a hard look at the American false alarms and learned from them, speeding up the fielding of their own systems and emphasizing the idea that warning had to be independent. The KGB directed rezidenturas to interpret them as false-flag efforts to suggest that American nuclear command and control systems weren’t very advanced, and therefore the Soviets had nothing to worry about—“thus providing a cover for possible surprise attack” by the Americans.23 It would turn out that the Soviets had a lot to be worried about, too. Warning was an illusion.
Alfred E. Buckles enlisted in the air force a few days after finishing high school. It was 1957. From the Bay of Pigs invasion through the beginning of the Reagan administration, Buckles climbed the ranks of “the command and control business,” as he put it. He was a SAC communications officer. He pulled duty as an emergency actions controller at the National Military Command Center at the Pentagon. He operated the consoles that transmitted emergency action messages from commanders to deployed forces in the field. He flew on SAC’s airborne command post. He wrote nuclear war plans—the SIOPs—as part of the Joint Strategic Planning and Targeting Staff, eventually serving as its executive director. His institutional knowledge of strategic command and control convinced his superiors to create special jobs for him, and when he retired from active service, they converted his position to a civilian one, a supremely uncommon vote of confidence in his abilities. Very few members of the US military knew more about the nuclear decision chain than Alfred Buckles. “If the Soviets wanted to know all of our nuclear secrets, all they would have had to do was kidnap Al Buckles and figure out how to make him talk,” a former colleague said of him.24
In 1977, President Carter had nominated General Richard Ellis to be the commander in chief of the Strategic Air Command. Ellis was the youngest pilot to be given command in World War II, a brilliant lawyer with an exacting and discerning mind, and a former head of US Air Forces in Europe. It fell to Buckles, by then the chief master sergeant at SAC, to brief him on nuclear decision making. Ellis had been privy to most of the US Air Force’s most sensitive projects, but nuclear procedures and the SIOP were so heavily compartmented that he came in knowing very little about the relationship between his command, its authorities, and the president.
“We went down and put together the end-to-end story,” Buckles recalls of the briefing. “He could ensure that the forces were equipped and trained and postured, but he was kind of surprised at what he couldn’t do—he couldn’t do anything with the nuclear forces after they were committed.” SAC regularly assessed the hardiness of the US system as it would respond to two scenarios: codenamed SIERRA, a Russian attack when “both countries were fully alert,” and a surprise, spasm attack, or INDIA. It turned out that the level of system arousal and preparation did not matter. “Most fixed, land-based primary and alternate command centers were destroyed within thirty minutes,” which forced all follow-on emergency action messages designed to tell the force how to retaliate to the Minimum Essential Emergency Communications Net, the system of last resort, which used relatively jam-resistant airborne command posts, very-low-frequency antennas, and commandeered satellites. But this system’s performance, too, was “severely hindered,” and connectivity with the president was not assured, according to a top-secret study conducted at the time.25
General Ellis’s top worry, though, was a sea-based threat. By the late 1970s, the US could track Soviet nuclear submarines leaving ports using imagery satellites, direction-finding sonar, and secretly implanted seabed arrays. They’d watch the Soviet subs transit the narrow gap between Greenland, Iceland, and the United Kingdom. But then they’d disappear. The United States was blind to the locations of the Soviet subs as they crossed the Atlantic. The US Navy had spent billions to develop its Sound Ocean Surveillance System (SOSUS). SOSUS arrays measured the sound reflecting and refracting against huge tectonically created trenches under the Atlantic and the Pacific. They could passively detect the noise made by Soviet submarines as they traversed these sound channels, because sound, especially at low frequencies, could travel for hundreds of miles without losing much of its strength.26 Beginning in 1952, the navy began to lay strung-together hydrophones at key chokepoints in the Atlantic. Within three years, SOSUS arrays protected both coasts and Alaska. Within five years, the navy had processing centers devoted to analyzing the frequency chunks, called tonals, that Soviet subs created when they crossed a SOSUS “beam.” With triangulation and some trigonometry, the navy could pinpoint the location of subs in near real time.27 SOSUS worked wonders for the US for more than fifteen years. Soviet subs were always louder than US subs, often by as much as 25 to 30 decibels. Navy intelligence catalogued Soviet efforts to quiet them, but the US, realizing its advantage, continued to refine the SOSUS arrays and their processing, and stayed one step ahead.
This gap disappeared, suddenly, in the late 1970s. The Soviets began to deploy their naturally quieter attack submarines to deeper waters away from the US Atlantic coast, part of a tactical shift to protect their nuclear submarines, which had to be there to maintain the Soviet Navy’s strategic deterrent. The Soviets, thanks to the efforts of their spies and engineers, could now launch missiles from protected “Bastions” near their territory.This, in turn, complicated SOSUS detection, because sound travels less well in shallower waters.28 By investing so heavily in passive acoustical quieting technology, the US had forced the Soviet submarine force to cluster its faster, quiet attack subs—those that didn’t have the capability to launch ballistic nuclear missiles—around its noisier, slower SSBNs—the B stands for “ballistic”—because the SSBNs would inevitably be priority targets for the US during war. At the same time, advances in acoustical detection of submarines ran up against the limits of physics.
In 1978, the Soviets deployed their first Victor III-class submarine from the Komsomolsk shipyards. The 350-foot ship bore an impressive array of anti-submarine warfare and intelligence-gathering features, but what worried US intelligence the most was that it was significantly quieter than any Soviet sub that had come before it.29 SOSUS arrays were primed to pick up the tonals, the frequency spasms that were created by the popping generated by the bubbles from a sub’s propellers. The Soviets had somehow figured out how to reduce the noise signatures from these cavitations. Soviet engineers also managed to mask the sounds that emanated from the cooling system in their nuclear reactors, another source of sound for SOSUS microphones. That they did so quickly suggested to the navy that espionage had been involved, because the Victor III seemed to adopt some of the US’s technological advantages in quieting while simultaneously exploiting SOSUS’s inherent vulnerabilities. (Indeed, it would later turn out that an American working for the National Security Agency, John Edward Walker, had been passing top-secret codeword documents on these precise capabilities to the Soviets for years.)30
By 1979, though, the Soviet Navy advances had frustrated SOSUS collectors to the point where the subs would disappear off the coast of the Soviet Union and then reappear, near the coast of Washington, giving the president and his advisors a mere six minutes to decide what to do if a nuclear cruise missile were launched from one of them. The SSNs still possessed enough conventional and nuclear firepower to decapitate the US government. Given this reality, virtually everything that SAC had assumed about nuclear decision planning was rendered moot.
Odom’s attention, though, had turned to larger questions. He focused on the presidential nuclear decision handbook and found its language mind-bogglingly obtuse. The military didn’t really know what sort of language would be useful and comprehensible to civilians executing the decisions. Then he discovered that the new LNOs (limited nuclear options)—which had been added to the SIOP at the direction President Nixon, through his Secretary of Defense James Schlesinger—bore no resemblance to any policy option the real president would actually consider.31 There was no allowance in these war plans for real-world, real-time changes to the situation.32
He wondered, Why couldn’t a policy be developed that figured out how to use nuclear weapons in a way that would advance US national and strategic goals?33
Figuring out how to answer that question, and then to turn it into a war plan revision, became Odom’s holy grail. First, he had to convince everyone that although nuclear war might mean the end of the sitting government, it might not mean that the world had to end. This violated a taboo, and, of course, the tenets of assured destruction. And if the government could be fortified along with civil defense, then survivability automatically implied that the president would have more options.34
Intelligence satellites could fairly accurately assess damage to a degree of granularity that would allow for retargeting, but the SIOP assumed that that process would rely upon satellites in existence, not the satellites that were planned. The big difference: the intelligence agency in charge of satellites, the National Reconnaissance Office, was slowly but surely moving its systems to near-real-time radar and electro-optical imagery platforms and moving away from its legacy Hexagon satellites with film canisters that had to be recovered at sea and processed.
Why couldn’t those satellites be used to give the president maximum flexibility as commander in chief? If the questions Odom and his colleagues asked were provocative, the solutions he proposed did not seem radical. Boost civil defense. Fix defense communications. Redraw the SIOP and its associated systems to give the president maximum flexibility before, during, and after a nuclear attack, and establish a federal requirement for enduring constitutional government.35
The changes to the SIOP were added late—in May 1980. A draft of the final presidential directive was sent to Carter days later, but the National Security Council didn’t discuss it in full until months later, owing (or so the NSC claimed), to the particularities of the president’s schedule.36 Presidential Directive 59, signed on July 25, 1980, committed the US to endure a nuclear war that lasted beyond an exchange of first strikes:
so as to preserve the possibility of bargaining effectively to terminate the war on acceptable terms that are as favorable as practical, if deterrence fails initially, we must be capable of fighting successfully so that the adversary would not achieve his war aims and would suffer costs that are unacceptable, or in any event greater than his gains, from having initiated an attack.
The directive enshrined what Odom had long argued: the president should be able to choose to use nuclear weapons and conventional weapons in pursuit of “specific policy objectives,” and not just the preprogrammed SIOP options. Intelligence would help the nuclear forces instantly retarget after first strikes to “control escalation.”37 The US would not “rely on launching nuclear weapons on warning that an attack has begun,” but it would institute “pre-planning” to make sure that the nuclear force was survivable, so as “to provide the president the option of so launching.”38 Such “pre-planned options, capable of relatively prolonged withhold or of prompt execution, should be provided for attacks on the political control system and on general industrial capacity,” according to the directive. Of the 50,000 or so targets in the National Strategic Targeting Database, only 2,500 were designated as “strategic” or nuclear.39
And exercises would be emphasized to practice: at least two per year had to involve the “National Command Authorities”—the president or his duly designated successor.40
The substance of Presidential Directive 59 leaked almost immediately, and its major tenets were splashed across front pages around the country. This was deliberate. In public, Harold “Hap” Brown, Jimmy Carter’s secretary of defense, was grilled, over and over: wouldn’t the capacity to limit attacks actually make both sides more likely to engage in them?41 The Joint Chiefs leaked, too: they warned that their own studies suggested that, as currently constituted, the US ICBM force would be largely annihilated if absolutely nothing was done before the first missiles arrived.42
The Soviet Union’s Marshal Ogarkov found PD-59 illogical. It was a feint, he believed at first, to justify the expense of a new, more accurate ICBM missile, the MX.43 Then a more grim interpretation set in. The new doctrine was an attempt at psychological conditioning, designed to instill fear in the public and the US Congress about Soviet capabilities and their intention to strike first. In turn, that would raise a clamor for the US to strike preventatively.44
The ambiguous language made for grim interpretations, exactly what the Soviets had feared: the US had committed to a policy of actually fighting a nuclear war and of absorbing a first strike.45 The Soviets understood the US nuclear doctrine in the context of the new missiles that were due to be placed in Europe by 1983. The US could, therefore, fight a war that would be limited to the European continent.46 The Soviets felt confined.47 The sharp bull’s-eye was now on their leadership and rendered them more vulnerable to a first decapitating strike by the United States.
Squishing a new nuclear doctrine in to the end of an administration was bad governing. Not anticipating it would leak—and that the leaked interpretation would come to dominate political discussions—was malpractice, a “case study in how not to make national security policy,” noted Marshall Shulman, a State Department advisor and graybeard on Soviet thinking.48 In a memo, he pointed out a logical fallacy that became glaring later on: if you want to somehow limit war and increase the chances that both sides can communicate after an exchange of first strikes, why on earth would your preferred targets be the very mechanisms that would allow Russian leaders to contact you and then agree to stop the war—and then see that the military carries out orders to stop it?49
As Carter’s presidency drew to a close, nuclear commander Ellis requested an urgent meeting with the president. It was a difficult time. Fifty-two American hostages were still being held in Tehran. The incoming and outgoing administrations were openly sniping at each other. Ellis and Buckles traveled to Washington in early December 1980, hoping that President Carter would not want to leave too much broken glass on the table. The president could make some immediate decisions that would increase the decision time. They did not want to wait for a new administration, focused on myriad other things, to come in and face more delays. Buckles recalls the essence of Ellis’s message to Carter. At best, given the current technology, the president might have only a few minutes to make decisions, and he might have very little information to go on. He might not know how any weapons were en route—just the fact that a bunch were. He might know what the targets would be—just broad geographic area estimates. He might not even be told whether the missiles came from Russia, or China, or somewhere else—it was 50/50 that the satellites could see enough to determine that, if they saw the missile launches at all.
“Mr. President,” Ellis told Carter, “we are spending a lot of money and time to train and equip this force, and actually, we’re killing ourselves to make sure that the nuclear force is ready to go to defend this nation, but until we fix the presidential decision-making and presidential support component and until we assure connectivity, we are kind of wasting our money.”
Carter simply responded: “I understand this, General. I understand it very well.” He drafted a message to Brown, ordering him to create the architecture of a program that would focus solely on strategic connectivity. Carter did not want to leave any more glass on the table for his successor. Brown tasked Ellis to come up with a series of recommendations—near term and long term—on the decision problem. Buckles became the point person on the project. He worked through Christmas that year, and four days before Ronald Reagan’s inauguration, he had his first meeting with the president-elect.