Jechorek’s warning didn’t fall on deaf ears. After further discussions, high-level officials in the Pentagon and the Justice Department concurred that the government would not submit to Regan’s blackmail. Carr’s team had to do everything it could to find the materials that Regan had hidden. The investigators realized that the key to tracing the documents likely lay in decoding the sheets with trinomes that Regan was carrying at the time of his arrest. So far, nobody had been able to make sense of them.
• • •
The use of codes for secret military and diplomatic communications dates back twenty-five hundred years. The Spartans are believed to have devised a simple cryptographic system known as a scytale as early as 400 BC, in which a strip of papyrus was wrapped around a staff like a ribbon to cover its entire length. The secret message was then written on this cylinder from one end to the other. In its unwrapped state, the text on the papyrus was meaningless, since the letters appeared in a nonsensical sequence. But when the receiver of the message wrapped it around a staff of the exact same diameter as that used by the sender, the message could be read out.
Centuries later, Julius Caesar used what may have been one of the first ciphers in history to correspond with military commanders and trusted aides. It involved shifting each letter in the message a certain number of places forward in the alphabet: if that number were 5, say, the letter A in “ARMY” would be enciphered as F, the next letter, R, as W, and so on.
Starting in the ninth century AD, Islamic scholars began systematically studying code making and code breaking. The philosopher Al-Kindi, born in what is now Iraq, birthed the field of cryptanalysis with his Manuscript on Deciphering Cryptographic Messages, written in the ninth century. In the treatise, he introduced a well-reasoned method for cracking a coded message without access to its key or cipher. In Al-Kindi’s words, translated from the Arabic:
One way to solve an encrypted message, if we know its language, is to find a different plaintext of the same language long enough to fill one sheet or so, and then we count the occurrences of each letter. We call the most frequently occurring letter the “first,” the next most occurring letter the “second,” the following most occurring the “third,” and so on, until we account for all the different letters in the plaintext sample.
Then we look at the cipher text we want to solve and we also classify its symbols. We find the most occurring symbol and change it to the form of the “first” letter of the plaintext sample, the next most common symbol is changed to the form of the “second” letter, and so on, until we account for all symbols of the cryptogram we want to solve.
The method described by Al-Kindi—now known as frequency analysis—went on to become a cornerstone of modern-day code breaking. Given that the making and breaking of codes is essentially a game of linguistic hide-and-seek, it’s not hard to appreciate the principle behind Al-Kindi’s approach. In every language, some letters and words are used more frequently than others, such as E and “the” in English. A coded message can often echo those frequency differences. Analyzing the relative number of occurrences of different characters in the encoded text can help figure out the rules used by the code maker to convert alphabets and numbers in the original message into gibberish. By the same token, a good code is one that provides no such statistical clues to the code breaker.
Starting in the thirteenth century, advances in mathematics helped drive the development of cryptography in Europe, where diplomats, administrative functionaries, and spies were increasingly using coded messages to secure their communications. Through the Renaissance years, as new and more secure ciphers were invented—among them the Vigenère cipher, which Brian Regan would use to encrypt his letter to the Libyans—cryptanalysts came up with techniques to crack them. By the nineteenth century, the use of cryptography had become commonplace in Europe and the United States.
With the birth of the telegraph, followed by the invention of the radio, encryption was no longer restricted to written communications. During Prohibition, from 1920 to 1933, rumrunners used coded radio messages to arrange the transfer of liquor from ships off the American coast to boats that would then smuggle the contraband ashore. The U.S. Coast Guard began intercepting and decoding these messages to determine the locations of the boats and capture them.
During the Second World War, the Axis and the Allied powers spent an inordinate amount of effort on encryption and code breaking, which proved to be pivotal to the war’s outcome. Working out of a research station called Bletchley Park near London, British mathematicians devoted themselves to cracking German messages that had been enciphered using Enigma machines—a typewriter-like devices with rotors and keys to convert plaintext into code and vice versa. In September 1940, around the same time that the British effort produced the first decryptions of Enigma messages, a team of U.S. Army code breakers cracked a Japanese diplomatic cipher called Purple. The UK’s continued success in cracking Enigma ciphers, to which the Germans added complexity as time went on, and the United States’ decryption of Japanese messages provided key intelligence to the Allies, helping to end the war sooner than would have been the case otherwise.
In the United States, the mission of decrypting foreign intercepts during World War II was led by the Signals Security Agency, which later became the National Security Agency. From 1943 to 1980, NSA cryptanalysts working on a secret program called Venona deciphered more than twenty-two hundred Soviet messages, many of them diplomatic communications. From these intercepts, the United States learned how the Soviet Union had attempted to steal the secrets of the Manhattan Project and identified a number of spies working in the United States on behalf of Moscow.
Through the decades of the Cold War, as cryptology became more advanced, the NSA increasingly relied on computer algorithms to break ciphers. By the 1990s, the focus of the NSA’s cryptanalysis efforts had shifted from traditional code breaking to the creation and cracking of digital encryption systems. In the fall of 2001, however, when investigators requested the NSA’s help in the Regan case, the agency’s cryptanalysts were forced to revisit the basics of cryptography as they sought to make sense of Regan’s trinomes. By the end of the year, with success nowhere in sight, the NSA called off the effort. The FBI assigned the task to one of the bureau’s rising stars—a thirty-one-year-old cryptanalyst named Daniel Olson.
• • •
The son of Swedish immigrants, Olson grew up in a small town in central California, a cherubic kid with a round face, gray eyes, and blond hair. From his childhood up until his early twenties, Olson’s life bore some striking similarities to the one Regan had lived, although he was eight years younger. He was the third of four kids, and the family—like Regan’s—suffered dysfunction and discord because of his father’s alcohol addiction. Olson didn’t do nearly as badly at school as Regan did, but just as Regan struggled with dyslexia, Olson struggled with what he had come to accept as a severe, even if far narrower, handicap: the inability to do math. If he had sought medical help, he would possibly have received a diagnosis for dyscalculia, a dysfunction of the brain that makes it enormously difficult to perform arithmetic calculations and grasp math concepts.
Although Olson did not share Regan’s problem of being perceived as unintelligent by friends and teachers, his difficulties with math were the source of a deep inadequacy that he felt throughout school and beyond. The last C he ever got in the subject was in third grade; from then on, it was consistently D or worse. It wasn’t that Olson was incapable of mathematical reasoning—he did fine when it came to solving word problems. But numerical operations and formulas and equations—especially polynomial expressions—gave him mental paralysis.
The fear of disappointing his father—a tough man to please anyway—made Olson’s math problem that much worse to handle. Among other traumatic moments, he wouldn’t ever be able to forget his last day in eighth grade, right before the family was to go on a trip to Disneyland, when
he came home from school feeling mortified about how his father would react upon seeing that Olson had received yet another D in math. His mother, an artist with a far less exacting attitude toward academics, came to his rescue, by changing the D on his report card to a more respectable C.
Olson’s family didn’t have the money to support a college education, and Olson’s grades weren’t good enough for him to get a scholarship. Like Regan, he decided that the best option was to join the military. By doing so, he hoped also to win the approval of his father, who had served in the U.S. Army.
After enlisting in the National Guard and flunking math in community college, Olson joined active military duty in 1988. He was chosen to serve in Army intelligence, thanks to a demonstrated aptitude for pattern recognition, and ended up being sent to Goodfellow Air Force Base in San Angelo, Texas, to train as a signals analyst. Many of the courses he took there, including one on basic cryptanalysis, were those that Regan had taken years earlier. Olson discovered that he had a knack for code breaking. He later returned to Goodfellow for twelve weeks of advanced cryptanalysis training, during which he learned about an array of different ciphers and code systems, from those used in medieval wars to encryption schemes invented by Russian mathematicians during the 1920s and 1930s.
In the run-up to Operation Desert Storm in 1991, when Regan was helping to brief Air Force commanders on Iraq’s missile defense, Olson was posted at a control center in charge of a handful of outstations eavesdropping on the Iraqi military. Iraq’s battle units, keenly aware of the United States’ signals intelligence capability, were doing their best to keep their radio communications to a minimum, and all that the outstations had picked up in several days of monitoring were a few stray snippets of conversation. It was impossible to extract any meaningful intelligence from them, but Olson noticed that the snippets were being heard around the same time every day.
Based on the pattern, Olson had all of the listening posts concentrate on one radio frequency during a window of time that he’d identified as the Iraqis’ talking hour. The gambit paid off. Soon the Army was able to intercept a complete message.
Despite a fulfilling four years, Olson wasn’t thrilled about continuing in the military. He left in 1992, got married, and started going to college for a degree in criminal justice. He found it difficult to get by without a monthly paycheck, however, and so a year later, he got a job as an analyst with a Drug Enforcement Agency task force in Savannah, Georgia, working during the day and taking classes in the evenings. Over the next few years, Olson came to specialize in tracking the use of UPS, FedEx, and other couriers by drug dealers to transport narcotics. Then, one day in 1996, a class he was taking as part of a money-laundering course reignited his interest in codes.
The class was being taught by a U.S. postal inspector, who was talking about how criminals use money orders to convert the proceeds from drug dealing into seemingly legitimate income. Covering an example, he asked the class to look at a money-order number that he projected on the screen. “Don’t pay any attention to the last digit,” he said. “It’s coded and you can’t figure it out anyway.”
It was an offhand remark: since the lecture wasn’t about codes, the inspector had no reason to go into them. But Olson, who was feeling bored, took it as a challenge and spent the next few minutes figuring out what the code could be. When he’d worked it out, he shared the answer with the person sitting next to him, who called it to the instructor’s attention: “Hey, this military guy here broke your little code.”
Annoyed at being interrupted, the instructor turned to Olson. “OK, smart guy, let’s check it out,” he said. Olson told him the steps he’d followed, and the instructor went through them on the board. It worked.
After the class ended, another instructor who had been sitting in the back of the room came up to Olson and introduced himself. The man’s name was Eugene Saupp. He was a supervisor in a division of the FBI called the Racketeering Records Analysis Unit.
“Where did you learn to break codes?” he asked Olson.
“In the military,” Olson replied.
Saupp gave Olson his business card and asked him to send him his résumé.
Less than a week later, he called Olson from Washington, D.C. “Hey, I don’t have your résumé yet,” he said.
A week later, Olson was on a plane to Washington, D.C., to interview for an analyst’s job. He got the offer shortly after. The only thing he had to do before he could start was finish his undergraduate degree.
He couldn’t tell Saupp or anybody else at the FBI why he didn’t have the bachelor’s under his belt yet. He’d failed to clear the required math courses, despite numerous attempts. Transferring from one college to another hadn’t helped.
It was only with the help of an educational counselor that Olson managed to complete his undergraduate degree in the three months he had available, earning the math credits from a Savannah college where a particularly kind instructor helped him out with a C. In the spring of 1997, he started at the FBI, proud to report to work with his bachelor’s degree in hand. Nobody asked to even look at it.
Although Olson had been hired for his code-breaking skills, he spent his first two years at the unit making sense of records seized from gambling syndicates and prostitution rings, which often use obscure and cryptic bookkeeping to protect their revenues and associates from law enforcement. In 1999, just as Olson was beginning to feel restless, he was handed a coded message that the Federal Bureau of Prisons had seized from an inmate. Members of prison gangs often use codes to communicate with one another.
Olson had never broken a code related to a criminal case before. What made the particular message doubly challenging for him was that it was in Spanish, a language that was foreign to him. After wrestling with it for a couple of days, he took it home for the weekend and sat down at the kitchen table with a Spanish dictionary beside him to check if he was on the right track. Late that night, when he thought he had solved it but couldn’t find many of the decoded words in the dictionary, he called a military friend who knew Spanish. The friend recognized the words as slang. The message was a directive from one inmate to another commissioning a murder.
After that, Olson began getting drafted to break codes with increasing regularity. Within a year, he was doing code breaking full-time, working to decipher not just secret messages confiscated from prison gangs but also suicide notes and diaries of murder victims. Each of them was a challenge in some way or another, but as he would find out, none of them rivaled the complexity of the trinomes he was given in January 2002.
• • •
Olson talks fast and moves swiftly, as if his internal rhythm were just a little quicker than everyone else’s. In his office at the FBI Laboratory in Quantico, Virginia, the shelves are lined with videotapes of Hollywood crime thrillers, some of which—like Manhunter—feature code making and code breaking in the plot. A display cabinet showcases relics of cryptography such as an Enigma machine from World War II. On his desk, he keeps a jar of peanuts and a box of graham crackers to sustain himself through stretches when he’s too engrossed with problem solving to break for lunch.
In the first few weeks of 2002, Olson found himself having to replenish that stash of snacks constantly as the Regan papers consumed his every waking hour. Besides the handwritten materials seized from Regan at the time of his arrest, Olson had before him thirty sheets with similarly cryptic writings that jail authorities had recovered from Regan’s cell during a series of routine jailhouse searches in November 2001. Regan had claimed he was preparing these papers in jail for his defense and that they were meant for his attorneys.
On first glance, the writing on some of these documents looked almost identical to the four sheets of trinomes. They were labeled similarly—S234, M341, M123456—and contained rows upon rows of three-digit numbers. In another document, Regan appeared to have created word puzzles, writing a series of jumbl
ed spellings for words like “towel” and “Jesus.” On some of the papers, he had scribbled numerical computations, multiplying, adding, and subtracting various numbers. One document, labeled “European Hunt,” was a letter apparently to his kids providing clues for locating treasures he had buried for them. What was particularly baffling about this supposed treasure hunt was that the clues in it referenced missile sites in China and the Middle East. Regan had even picked the same missile types that he’d looked up on Intelink—CSS and SAM—in setting up the elaborate riddle.
It wasn’t immediately clear to investigators what Regan was up to. Was he writing coded messages for a foreign agent? That seemed unlikely since he’d made no special effort to conceal the papers in his cell. The ease with which the prison staff had found them on four different searches from mid- to late November made Carr and his fellow agents wonder if Regan wanted the new materials to be discovered.
Olson compared the trinomes found in Regan’s folder with those Regan had generated in prison. He noticed a couple of key differences between them. In the originally seized trinomes, the second digit was always a number ranging from 1 through 5, and the third digit was always 1 or 2. The prison trinomes, on the other hand, didn’t follow that rule—the second and third digits could be any number ranging from 0 to 9. Further, in the original set of trinomes, some of the numbers had been crossed out and written over, and some appeared to have been corrected. There were no such errors in the prison trinomes. The difference suggested that the trinomes seized at the airport were not random. They were somehow more meaningful than the ones Regan had produced in his jail cell.
Next, Olson compared the two sets using statistical analysis. For each of the papers with the trinomes—including the four sheets from Regan’s folder, marked “letters,” and a total of six messages from the cell—Olson looked at what’s known as an “index of coincidence,” or IC. A mathematical tool invented by the legendary American cryptologist William Friedman—whose team cracked the Japanese diplomatic cipher Purple—the IC is a measure of the randomness of a text. It indicates the likelihood of finding the same character in the same position when comparing two units of a text side by side. Since such “coincidences” in an enciphered text can reflect an underlying pattern in the plaintext, calculating the IC of a coded message can help cryptanalysts find clues to breaking it.
The Spy Who Couldn't Spell Page 15