Body of Secrets: Anatomy of the Ultra-Secret National Security Agency from the Cold War through the Dawn of a New Century
Crypto: How the Code Rebels Beat the GovernmentSaving Privacy in the Digital Age
The four-engine US Navy aircraft which made an emergency landing on China’s Hainan island on April 1 was on a routine ELINT mission, so called for what it collected—electronic intelligence. A crew of four actually flew the plane; the other twenty Americans on board, all Navy personnel but on a mission ultimately sponsored by the National Security Agency, were there to find, identify, collect, and record a range of electronic emissions from routine military chatter on radios to the characteristic signature of Chinese defensive radars.
The NSA, with the help of the Navy and Air Force, has been doing this since the late 1940s, sometimes aggressively, and the target countries detest it. In the early days the Soviet Union shot down as many as forty American aircraft on ELINT missions, some of them deep inside Soviet airspace, killing perhaps two hundred American civilians and military men. The most recent incident, however, occurred over international waters in the South China Sea; reckless shadowing of the slow-moving, propeller-driven American EP-3E by a Chinese fighter aircraft appears to have caused a midair collision. Accident it may have been, but the message was the same as that of the Soviet shootdowns of yesteryear—back off.
But backing off is the last thing the United States is likely to do. Collecting intelligence is what great powers have learned to do instead of going to war, and the risk of war between the United States and China, not great, and at first glance crazy and unthinkable, has nevertheless been growing year by year since the collapse of the Soviet Union in 1991. The big irritant is Taiwan, over which China seeks to reassert political control. American administrations haven’t ruled this out, so long as it isn’t achieved by military force. But the specter of military force is very much part of the strategy used by China, which has been threatening Taiwan in symbolic ways, such as test-firing missiles near the island, and the United States has been demonstrating support in symbolic ways, such as agreeing to major new weapons sales, but not, so far, the sophisticated Aegis defense system.
What happened to the Navy’s EP-3E has its symbolic side, too—the Americans were flying it up and down the Chinese coast partly to show we can’t be pushed around, and the Chinese were shadowing it aggressively to show we’d better be ready for a lot of pushing. The civilian observer watching the drama unfold on CNN probably feels much like an adult watching toddlers squabble in a sandbox—what are they fighting about? Why can’t they just get along?
But the making of symbolic gestures is not why the United States spends uncountable billions on ELINT flights and all the rest of the intelligence-collecting activities of the NSA. So what did the Chinese find so threatening and how did the Americans plan to use what they learned? These questions are addressed, with numerous examples and a wealth of human and technical detail, in the new history of the NSA, Body of Secrets, by James Bamford, who wrote one of the really good books about American intelligence twenty years ago, and has now done it again.
The new book revisits old ground but there is nothing tired about it. Bamford has learned some things that ought to make headlines and ignite serious argument, but the real strengths of the book are to be found in its portrait of the NSA—an institution of staggering size and capacity—and in its firm conviction that every American with enough interest in the world to read a daily newspaper ought to know what the NSA does, how it does it, and why. This may sound like elementary civics but candor about intelligence comes at a cost: the secrets uncovered by intelligence organizations are always inconvenient to somebody, and sometimes the way secrets are obtained, once it has become publicly visible, is ruled out of order. No government chooses candor if it can hide what it’s up to, and without Bamford’s efforts, beginning with his first book, The Puzzle Palace, in 1982, the initials NSA would probably still stand for “no such agency.”
The principal target of the National Security Agency is communication by foreign powers, and especially enciphered communications. During World War II the Allies were so successful in learning to read German and Japanese codes that some historians argue it made the difference between victory and defeat. The German navy’s reliance on the Enigma code machine, cracked by British wizards at Bletchley Park, cost them the battle of the Atlantic, and ultimately the war, just as the Japanese navy never recovered from their crushing defeat at the Battle of Midway, the fruit of American success in reading the Japanese naval code called Purple. At the end of the war, Bamford tells us in Body of Secrets, the Allies discovered that the Germans had also succeeded in cracking enemy codes, especially those of the Soviets transmitted by radio over a machine that broke messages into nine separate channels at one end and reassembled them at the other. The Germans read the Russian messages with a machine of their own, and once we had a copy of that machine we could begin to read Soviet back traffic—messages that had been intercepted and filed away in their coded form in the hope of just such a happy breakthrough.
Brilliant as that success was—Bamford calls it a “once-in-a-lifetime discovery” for the American soldiers who dug up the German files and equipment from beneath a cobblestone street—it was soon matched by US army codebreakers who exploited a “bust,” or procedural error, in Soviet diplomatic cables enciphered on one-time pads, which were normally unbreakable, and managed to read thousands of communications in whole or in part. Among the several hundred people disguised by cryptonyms in those messages, collectively called “Venona,” were the atom spies who had betrayed important design secrets of the first plutonium bomb to the Russians.
But that, apparently, was it. Despite the immense importance attached to reading Soviet messages throughout the cold war, and the huge effort devoted to the task, the NSA never again achieved a similar breakthrough in reading an important Soviet code on a routine basis. “During the 1960s,” Bamford writes, “NSA’s inability to break high-level Soviet codes was becoming its biggest secret.” Whole divisions of the NSA with platoons of mathematicians and acres of computers evidently tried and failed at a job which is now growing even harder. The NSA was long the silent driver in the development of computers and it still actively supports cutting-edge research—on computers, for example, that make use of living protoplasm in the manner of the human brain. But computers not only make it easier to crack codes; they make it easier to encrypt messages as well, and in the war of the codemakers and codebreakers the makers seem to be pulling ahead.
“Public encryption,” as it is called—the ability of private citizens to have and use strong codes defying sophisticated attack—is something the NSA fought against tenaciously for nearly thirty years in a clandestine campaign recounted in lively detail in Steven Levy’s new book, Crypto: How the Code Rebels Beat the Government—Saving Privacy in the Digital Age. Early in the 1970s a handful of young computer wizards, distrustful of government after the hard lessons of Vietnam and Watergate, began to think of ways to preserve “privacy” in the computer age. What they meant was the ability of people to communicate without fear of the government, and what they wanted, once they started to think hard about the problem, was a means of encrypting private communication.
Levy’s book follows the genesis and development of the idea of “public key encryption,” brainchild of two bright and stiff-necked young mathematicians—Whitfield Diffie and Marty Hellman. Codes traditionally substitute letters according to a formula more or less complex, sometimes with the aid of machines. To read a coded message one needs the key—the formula for unscrambling the substitutions. From ancient times until the 1970s the key was always secret, held only by the sender and receiver of an encoded message. For obvious security reasons keys were frequently changed, creating a logistical problem of nightmarish difficulty for the government of a great power sending scores of thousands of coded messages to diplomats and military units all over the world. The Diffie-Hellman stroke of genius was the public key—a very large number, derived from a so-called one-way mathematical function which could be openly distributed. The first Diffie-Hellman keys were the result of multiplying two large prime numbers, a function extremely difficult to reverse. Over the following decades the original Diffie-Hellman approach was developed and refined as mathematicians created modern encryption systems which can be used to protect cell-phone conversations, e-mail messages, and commerce on the World Wide Web.
How these systems actually work is complicated but not dauntingly so, and I urge interested readers to consult Levy’s book. What matters here is that Diffie and Hellman began working on sophisticated codes outside the Triple Fence—Levy’s term of choice for the heavily guarded, supersecret NSA. Diffie and Hellman asked no one’s permission and believed they needed none. The NSA’s position was, first, that code work, like certain principles of atomic physics, was born secret—classified as soon as conceived. This was such an egregious intrusion on academic freedom and the First Amendment to the Constitution that the NSA retreated to a back-up claim that codes were in effect “munitions” under the law and could be denied export on grounds of national security. Because large computer software companies resisted the complexity of issuing two versions of software—a domestic program with high-level encryp-tion, and an export version easy to crack—and because they correctly imagined foreign customers would avoid programs expressly designed to help American spies read them, public encryption was slow to develop and catch on. This Pyrrhic victory of the NSA is one reason few Americans can protect their communications and Web sites with strong encryption, and are thus too often open to invasion by hackers—or foreign information warfare experts, about which more in a moment. At the end of the millennium the NSA could still read just about all private American communications—although it was enjoined by law from doing so—while it had lost the ability to read high-level codes used by other countries, like China, Russia, and its predecessor, the Soviet Union.
The one really big and central thing the NSA was conceived and funded to do apparently never got done, but that does not mean the agency was idle or the money wasted. If we couldn’t read Russian secret communications enciphered at the highest level, we could nevertheless read those of just about every other country, friend or foe, and written messages, in the electronic age, were only part of what proved interesting. Modern military forces are like vast nervous systems, linked from the loftiest general down to the lowliest private by a connective web of warning and reporting systems sometimes called C3I—pronounced “see-cubed-eye”—command, control, communications, and intelligence. What the generals are reporting to the Kremlin or the Pentagon is important, but so is what test missiles are telling ground crews about trajectories, for example; or what sergeants in Siberia, chatting on the phone, might reveal about the local inventory of nuclear weapons; or how many divisions of Chinese assault troops have been established along the Formosa Strait, and whether assault craft are already there or on the way. Maybe we can’t read the instructions of China to its ambassadors but all the stuff we can “read,” analyzed, insofar as possible, as a whole, brings us pretty close to what the Chinese leaders have on their minds. The important thing to grasp about the National Security Agency is its core belief that the best way to collect the important bits is to collect all the bits, and to understand further that it not only tries to do this—it does. All of it.