THE CODEBREAKERS

by DAVID KAHN

  This withholding of information constitutes the essential element of that which is called “secrecy.” All the manifestations of secrecy—hiding places, disguises, locked doors—share the basic idea of not communicating objects or information. Its extreme form is silence (which conjures up an Orwellian nightmare of the extreme form of eavesdropping—detection and interpretation of brain waves). An exhaustive investigation of the concept of secrecy would require, as Maurits de Vries has pointed out, “a complete examination of the relations between individuals and between groups in our society,” because secrecy is the antithesis of communication, and communication—as that which makes man a social being—encompasses all aspects of cultural behavior. Cryptography combines these antitheses into a single operation; a wag might define it as “noncommunicating communication.”

  The relation between cryptography and cryptanalysis is not logically necessary; it is contingent. One can envision men communicating by secret means with others not even thinking of prying. But in the real world, the cryptanalyst—or more accurately the potential crypt analyst—comes first. What need for cryptography if no one would eavesdrop? Why build forts if no one would attack? Thus the assumption that someone will attempt a cryptanalysis, no matter how tentatively or incompetently, engenders cryptography.

  Experience of the interreaction between cryptography and cryptanalysis has precipitated out certain practical principles. They all refer to time, because all practical matters involving mortal men connect eventually with that one inexorable, irreversible, irretrievable factor.

  Time, for the cryptographer, controls a variable relationship. The most general of the cryptographer’s principles deals with the sliding ratio between speed and security; as the need for speed in communications increases, the need for security decreases. Early in the planning of a major operation, messages demand great security because the enemy, if he could read them, would have time to prepare countermoves. But in the heat of battle, commanders may use plain language because the enemy, though he intercepts the messages, may not have time to react effectively. This principle arranges a nation’s cryptosystems in a hierarchy in which front-line systems are simple and diplomatic systems secure and more complex. “Of each such system,” Friedman wrote, “the best that can be expected is that the degree of security be great enough to delay solution by the enemy for such a length of time that when the solution is finally reached the information thus obtained has lost all its ‘short term,’ immediate, or operational value, and much of its ‘long term,’ research, or historical value.”

  The paramount requirement for all cryptosystems is reliability. This means that cryptograms must be decipherable without ambiguity, without delay, and without error. It implies, for example, that cipher machines will be sturdy enough to withstand ordinary abuse so that they will be ready to operate properly when a message comes in. Usually the simpler the system, the more reliable. The requirement excludes from the combat zone ciphers of more than two steps. Any encipherer’s errors or garbles should be correctable without having to call for a retransmission. This bans systems in which a single error garbles the message from the point of error on, as in autokey ciphers (such systems are said to have an undesirable error-propagation characteristic). Obviously, if a general cannot rely upon the validity of messages that come out of his cipher machines, the cryptosystem is worse than useless.

  Secondary requirements for a cryptosystem are security and rapidity. Which one comes first depends upon the needs of the users. Further down the scale of importance stands the requirement of economy. This rules out any system that requires several men to encipher, makes the ciphertext more than twice as long as the plaintext, or is too complicated or expensive to manufacture or distribute.

  In addition to these general requirements, military and diplomatic crypto-systems must meet two specific ones—both first enunciated by Kerckhoffs. The first rests upon the almost universal employment of telegraphy or radio-telegraphy for military and diplomatic communications. No system is acceptable whose cryptogram characters cannot be sent in Morse code; excluded are squares, angles, crosses, or other designs. The second rests upon a working assumption of military cryptography: that the enemy knows in general how a cipher works. Secrecy must depend upon the keys used. No method is acceptable that does not accede to this requirement, that does not provide for both a general system and specific keys.

  For the cryptanalyst, time’s demands remain fixed. Always at his back he hears time’s winged chariot hurrying near. He seeks to get out his solutions as quickly as possible. It is probably true that a message will always have some historical value, but that is small comfort to a commander who does not get a cryptanalysis that would have warned him of an enemy attack until after the attack is under way. The factors that affect the time required to solve cryptograms—aside from external factors, like the speed of sending the intercepts back to the cryptanalyst—are the strength of the system, the soundness of the regulations for its use, how closely the cipher clerks follow those regulations, the volume of text, the size and skill of the cryptanalytic organization, and the amount and character of collateral information.

  Bringing skill into the picture raises the question of whether cryptanalysis is a science or an art. It is both. On the one hand, cryptanalysis—or, more properly in this context, cryptanalytics—is an organized body of knowledge. It studies and controls phenomena. Its whole spirit is scientific, but that of an applied science, like engineering. On the other hand, cryptanalysis—here meaning the steps performed in solution—clearly depends upon personal ability. Some cryptanalysts are better than others. In this sense, cryptanalysis is an art. So, in this sense, is any human activity that demands a certain aptitude for its superior practice. Yardley said that outstanding cryptanalysts were gifted with “cipher brains,” and rather glamorized the faculty, but in fact “cipher brains” are just the cryptologic manifestation of a general characteristic—talent in a given field. Who possesses “cipher brains” and why, however, raise complicated questions.

  Human knowledge not only cannot answer them now, it does not even understand how the mind performs the basic psychological operation of cryptanalysis—pattern recognition. How the brain can supply the missing letters to a fragment of plaintext which it has never before seen resembles such problems as how one can read words in a handwriting one has never seen or recognize a piece of music as Mozart’s even though one has never heard it before. These problems remain among the still unsolved ones of psychology and biochemistry, as convoluted as the cerebral cortex and molecular chains which may hold the answer.

  Nor does anyone know the emotional bases of cryptology. Freud stated that the motivation for learning, for the acquisition of knowledge, derives ultimately from the child’s impulse to see the hidden sexual organs of adults and other children. If curiosity is a sublimation of this, then cryptanalysis may be even more positively a manifestation of voyeurism. This view has won some qualified support. Psychoanalyst Theodor Reik, author of Listening with the Third Ear, The Secret Self, and many other books, replied to a query about it: “I am inclined to assume that there is at the bottom of the wish to break a code a continuation of the infantile desire to find out what is the secret of sexuality which the parents or the adult hide before the little boy…. I think that it is … one of the roots of scientific inquiry,” he added, suggesting the Freudian view that scientists, art critics, and anyone else whose work involves either literal or figurative looking is similarly motivated. Psychologist Erich Fromm, author of Escape From Freedom and May Man Prevail? conceded that the voyeuristic explanation “may sometimes be correct, but by no means as generally as the Freudians believe.” The eminent psychiatrist Karl Menninger assented to the idea, adding that “this is not a new ‘theory.’ ”

  But this view has been challenged, and by a Freudian psychiatrist. Jeptha R. Macfarlane thought that cryptanalysis represents a power drive. “The codebreaker isn’t interested in the content of the message
but in the solution of the code,” he said. “He does not take a sneaking interest in the cryptogram but pride in its mastery. Cryptanalysis is not peeking through a keyhole. It is breaking down a door.” Supporting Macfarlane’s hypothesis are remarks by or about cryptanalysts. Werner Kunze of Pers Z, excusing his lack of knowledge of the results of his work, said that he did not pay much attention to what the messages said, and lost interest after the system had been solved—hardly a voyeuristic impulse! Of John Wallis it was said that he “never concerned himself in the matter, but only in ye art & ingenuity.” Even the experience of the solvers of newspaper cryptograms attests to Macfarlane’s power explanation: they are not curious about the answer, they only want to solve the puzzle.

  Though the evidence seems to support the power hypothesis better, neither it nor the voyeuristic one has ever been put to the test. Perhaps part of the answer to the apparent contradiction lies in the theory that the voyeuristic motivation might lead a man to a general interest in cryptanalysis and that the power drive leads him to success in specific solutions.

  What relation, if any, these explanations have to those for an interest in cryptography, in inventing secret codes and unbreakable ciphers, is not wholly clear. Reik thought this interest “might be a suspiciousness that others could find out about ourselves (not only about our sex-life, but our hostility, aggressiveness, and so on) and a wish to prevent them.” Fromm’s view was not dissimilar, though he linked cryptography and cryptanalysis: “I think the interest in deciphering, as well as in secret codes, may have a great deal to do with a person’s relatedness to the world, and specifically with the sense of aloneness and isolation and the hope that he might find the related soul with whom he could communicate … the world is closed to him … and hence he has to decipher what is not meant for him.” Psychologist Harold Greenwald, who was once interested in cryptology, wrote: “The patients I have seen who were interested in this subject seemed to display another motive [than voyeurism]. Predominantly they were the kind of people who attempted to establish feelings of power through either hiding their own behavior (putting things in secret code) or discerning the things that others wish to keep secret (breaking codes).” A psychological study of secrets in general states that secrets have their origin in the anal stage of psychosexual development. This implies that cryptography may come ultimately from the infantile sexual pleasure that Freud says children obtain from the muscle tension of retaining the feces.

  There appear to be as many theories as there are writers. Ernest Jones, the distinguished psychoanalyst and biographer of Freud, put his finger on what may be a significant point. Many youngsters become interested in cryptology when they are 12 or 13, and Jones told in his autobiography how “When I was nearly twelve, the secretiveness that so often heralds the approach of puberty combined with an always strong curiosity to impel me towards a passionate interest in ciphers, about which I still know a good deal. 1 devised one myself which I was satisfied would baffle any opponent; I must admit, however, that it would not have been a convenient code for purposes of rapid communication, since it involved the interplay of so many subciphers that it took the best part of a day to transcribe a sentence into it. The complex, thus allayed, was able to transmute itself into a more useful form … ,” which was shorthand. Unfortunately, Jones never said why he thought secretiveness was related to puberty.

  Novelist Aldous Huxley seemed to have glimpsed the voyeuristic idea for cryptanalysis and inverted it into antivoyeurism as an explanation for cryptography. In Those Barren Leaves, he wrote:

  Did she love me? At any rate she often said so, even in writing. I have all her letters still—a score of scribbled notes sent up by messenger from one wing of the Hotel Cecil to the other and a few longer letters written when she was on her holiday or week-ending somewhere apart from me. Here, I spread out the sheets. It is a competent, well-educated writing; the pen rarely leaves the paper, running on from letter to letter, from word to word. A rapid writing, flowing, clear and legible. Only here and there, generally towards the ends of her brief notes, is the clarity troubled; there are scrawled words made up of formless letters. I pore over them in an attempt to interpret their meaning. “I adore you, my beloved … kiss you a thousand times … long for it to be night … love you madly.” These are the fragmentary meanings I contrive to disengage from the scribbles. We write such things illegibly for the same reason as we clothe our bodies. Modesty does not permit us to walk naked, and the expression of our most intimate thoughts, our most urgent desires and secret memories, must not—even when we have so far done violence to ourselves as to commit the words to paper—be too easily read and understood. Pepys, when he recorded the most scabrous details of his loves, is not content with writing in cipher; he breaks into bad French as well. And I remember, now that I mention Pepys, having done the same sort of thing in my own letters to Barbara; winding up with a “Bellissima, ti voglio un bene enorme,” or a “Je t’embrasse en peu partout.”

  This is illuminating, though no more conclusive than the other theories. But if the psychological roots of cryptology remain obscure, the biological roots are clear. Those roots reach back through the eons to the first protozoa struggling for life in the warm seas of the primordial earth. For cryptography and crypt analysis, though they are highly sophisticated technologies, retain at their inmost cores, like chromosomes that determine their heredity, the most primitive of functions.

  Cryptography is protection. It is to that extension of modern man—communications—what the carapace is to the turtle, ink to the squid, camouflage to the chameleon. Cryptanalysis corresponds to the senses. Like the ear of the bat, the chemical sensitivity of an amoeba, the eye of an eagle, it collects information about the outside world.

  The objective is self-preservation. This is the first law of life, as imperative for a body politic as for an individual organism. And if biological evolution demonstrates anything, it is that intelligence best secures that goal. Knowledge is power. In an atmosphere of competition, it may exist in two modes: mine and mine enemy’s. All organisms attempt to maximize the former and minimize the latter. Cryptography and cryptanalysis exemplify the two modes. Cryptography seeks to conserve in exclusivity a nation’s store of knowledge, cryptanalysis to increase that store.

  But knowledge alone is not power. To have any effect it must be linked to physical force. Cryptology, like the services of supply and transportation and administration, aids the fighting troops that constitute a main element of national power. Nations use that power to advance their political and social goals. Cryptography and cryptanalysis are means to those ends. And that is their position in the ultimate scheme of things.

  Even when the ends that they serve are purely defensive in regard to other nations, there exists a difference in morality between the means of cryptanalysis and such means as armies and navies. The latter are honest and above-board, open deterrents to aggression; they are like strong men armed. Cryptanalysis is itself an aggression—often a preventive one, to be sure—but still an aggression, a trespass. Moreover, it is surreptitious, snooping, sneaking; it makes its government hypocritical. It is the very opposite of all that is best in mankind. It shatters the highest ethical precept: to do unto others as we would have others do unto us.

  Is it, then, ever morally justified? It is. A single act can be both moral and immoral, depending on circumstances. Killing is permissible in self-defense. So is cryptanalysis. In war, of course, cryptanalysis can look like a positive good, especially when it saves lives. Even in peace, cryptanalysis may be a form of self-defense. It can warn of hostile intent and enable the government to preserve life and liberty, without which there is no doing to others of any kind. But when a nation is not threatened, it is wrong for it to violate another’s dignity by clandestine pryings into its messages, just as it is wrong to indiscriminately tap telephone lines or invade the privacy of a man’s castle. That is why it is indefensible for the United States to read the messages of friendl
y nations like Norway, Britain, or Peru.

  Even when justified, cryptanalysis remains an evil, and it goes against the American grain. Ever since July 4, 1776, the United States has stood for morality and integrity, in international affairs as in domestic, in the Fourteen Points as in the Emancipation Proclamation. It is this stand that, in large measure, makes America great. Cryptanalysis therefore poses a much greater problem for the United States than for other nations. It perhaps reflects this concern that the United States places her national cryptanalytic agency within the Defense Department, where it belongs in ethical terms, while Great Britain puts hers in the Foreign Office, where it belongs in a practical way.

  Only once has cryptanalysis been treated as the sin against morality that it is: in 1929, before Hitler and the Japanese militarists, with no nations potentially dangerous to the United States and self-preservation not at issue, Henry L. Stimson closed Yardley’s Black Chamber. Even though it was done at a time when the United States could afford it, the decision was a profoundly moral one, and it marched in the center rank of American belief. Was it softheaded, unrealistic? No. Idealism is the ultimate realism. Ideas of truth and justice always eventually triumph. Mankind can learn. America’s whole history shows this, as does humanity’s ascent from barbarism. The growth of wisdom and morality—urged on in these present times by the very real danger of total annihilation—may some day lead mankind to beat its swords into plowshares. When it does, it will no longer need cryptanalysis, and will dismantle organizations like N.S.A. and the Spets-Otdel. Their nonexistence then will testify to a true peace on earth. And may such be their glorious destiny!

  Note 1: Variations in Letter Frequency

  The problems of why certain letters should be more frequent than others within a language, and why one language should prefer sounds that another abhors, remain unsolved. Why should e be the most frequent letter in English? (This question is admittedly complicated in English by the discontinuity between speech and spelling, but even languages whose orthography fits their phonetics have high- and low-frequency letters.) Why should o be the most common in Russian, a in Serbo-Croatian, and e again in German, French, and Spanish?