THE CODEBREAKERS

by DAVID KAHN

  Noise might be added to the TV signal. Synchronization problems would again require this to be of a fairly simple type. Black or white bars might stripe the picture. A problem here is that using the ends of the video spectrum for enciphering reduces the amount of spectrum available for the picture, resulting in less contrast and a washed-out image. The transmitter might scan the successive horizontal lines at varying rates instead of the standard uniform speed. “This would pose a great problem in all TV receiver designs,” Kirk wrote, “because much attention has gone into cutting as much of the cost out of the sweep circuits of the receiver as possible. This has resulted in a sweep circuit of fixed design. To either speed up or slow down this sweep circuit by an appreciable amount would require major modifications, and, therefore, costly additions to the average TV receiver.”

  TV normally fixes the horizontal and vertical position of the picture on the screen by sending the horizontal position pulses and the vertical position pulses at regular intervals. If a transmitter electronically manipulated these, the image would become displaced on the screen. Varying the manipulations would shred the picture, so that parts normally appearing in juxtaposition would become separated. A nose might appear to the right of an eye, an ear below a mouth. Other manipulations would “jitter” the image like a movie film that has jumped its sprockets. These arrangements, Kirk wrote, would “require only inexpensive changes in the position-circuits of the receiver.”

  Various pay-TV companies proposed different forms of the keys that they would sell to subscribers to enable them to unscramble the program. Skiatron Electronics and Television Corporation would send out each month an I.B.M. card on which electric circuits would be imprinted. The subscriber would insert this into his decoder, which could then reverse that month’s enciphering key to reverse the two basic video enciphering processes that Skiatron would use—shifting the horizontal lines to one of three different positions and inverting the video signal. It would also decipher the audio scramble, a band-shift. Zenith Radio Corporation’s decoder had five key knobs, each of which had seven positions. To unscramble a particular program, a customer would have to set his decoder knobs to one of the 16,000 possible positions. And to prevent one person from buying numbers and furnishing them to his neighbors, Zenith would wire its decoders differently so that different decoders would require different numbers for the same program. Zenith would encipher its video in essentially the same way as Skiatron. A third system, that of the International Telemeter Corporation, employed a decoding card within a coin-operated device. Holes in this card, which was to be changed monthly at the same time coins were collected from the decoder, permitted electrical contacts to make circuits that unscramble the programs. More recently, Blonder-Tongue Laboratories and Teleglobe Pay-Television System have proposed other methods.

  Kirk demonstrated how fee-vee bootleggers could steal the keys of the three systems. In each case, the bootlegger would have to subscribe legitimately to the system to get the unscrambling information, which he would then use to break the system and peddle the results to nonsubscribers at a rate lower than subscribers would be paying. For Skiatron, he could sell a card that would fit into the decoder and would contact all of the circuits; switches attached to it would permit shutting the circuits on or off. Then he would simply translate his knowledge of the current key into settings for the switches and sell the settings. For Zenith, all the decoders would have to be able to unscramble the same signal, even though the knob settings would be different; the bootlegger could quickly correlate the setting for a knob on one decoder with the setting for a knob on another decoder and obtain equivalences between the two decoders. To sell to a particular subscriber, the bootlegger would merely have to compare a few of that man’s keys with his own keys for the same period. Telemeter’s system would require breaking into the decoder to copy the card.

  But how would he break the system in the first place? Kirk observed that “for successful operation of a scrambled television system, long-term security—in fact, indefinite security—is mandatory.” Yet, “even if one used a very complex coding arrangement and delivered the coding devices only into hands of those people who made adequate payment for them, this would represent only short-term security. With decoders located in living rooms and thousands of technicians (not sworn to secrecy as in the military case) receiving training, it is unlikely that a bootlegger could long be prevented from obtaining a decoder or an instruction book containing circuit information. In fact, these technicians, trained at the expense of the television scramblers, might well become the first bootleggers.” Since any such system would be used extensively, it would have to be relatively simple, and once installed on millions of TV sets it naturally could not be changed very easily. Moreover, television transmits at a very high rate, providing an enormous volume of traffic for solution. TV’s approximately 500 lines per picture and 300 spots per line make about 150,000 spots of light per picture; Kirk equated the 25 levels of brightness with the 26 letters of the alphabet, and noted that 30 pictures are sent per second. “This corresponds to sending 4½ million letters per second, or in terms with which we are more familiar, some 900,000 words per second. This is the equivalent of five or ten books per second in terms of information transferred by letters.” Finally, the TV signals are very highly redundant. The image does not radically change every one-thirtieth of a second; usually it stays essentially the same. The only television picture that would have low redundancy would be one that shows grains of sand of various shades of black and white blowing across the screen: that picture would change from image to image.

  Kirk contemptuously dismissed inversion as “not … a coding procedure because a simple inverter and two-position switch is all that is needed” to reinvert it. He ignored the problem of finding the inversion point, but that is a minor problem. Bars across the picture can be removed by examining the voltage of the video wave to find features—such as spurts of amplitude—“which can be eliminated to bring order to the picture.” Kirk did not discuss solving the video equivalent of band-splitting, presumably because no pay-TV firm proposed such a system, and he also did not discuss a one-time system, because synchronization problems render such encipherments practicable in home TV sets only with a wired link from transmitter to receiver to bring in the key pulses. To solve shiftings of horizontal lines involves merely electronically comparing the frequencies of the horizontal sweep signals with the frequencies of the video signals; these will be out of phase by a fixed number corresponding to the shift. “Thus,” Kirk wrote, “it may well be possible for a bootlegger to build a decoding device … which … simply utilizes the fact that the encoding process at the transmitter end will insert in the signal a stop phase shift which can be detected by a phase detector at the receiver end.”

  The most general question involved the displaced or jiggled image. Kirk’s technique resembles the speech-scrambler solution technique of recording the scrambled sounds on a spectrogram and then matching the lines as in a jigsaw puzzle until a smooth flow is established.

  Assume for the moment that the television signal is received, and instead of being displayed on a picture, the lines of the picture are simply recorded on a tape. This means that one will have available on tape the series of lines which contain variations in intensity from black to all shades up to white. These lines will be received in the order that they are supposed to come in the picture. The only thing that is out of order is that the edges of the lines don’t match up. If one matches the ends of these lines, then the picture doesn’t match up.

  If these were given to an individual with the requirement that he sort them out to make a picture, he could certainly do this in very short order. He would simply look for the like looking blobs of black and white on the adjacent lines and then match these up by shoving one first a little to the right and the other a little to the left as necessary. Very shortly he would have the completed picture….

  If now one makes an electronic circuit
which can control the horizontal position of the picture and feeds this circuit from a computer which computes which alignment of picture elements will keep things on adjacent lines most nearly the same, the result will be a circuit which is capable of electronically decoding the television signal.

  Kirk’s exposition made a very strong technical case against the security of fee-vee. Toll-TV proponents in effect conceded this. But they replied on the much broader and, in the case of profit-seeking firms, the controlling ground of economics.

  They denied, first, Kirk’s premise that “long-term security … is mandatory.” “Pay-television systems do not have to have a cryptographic security comparable to the security of military systems,” wrote William C. Rubinstein, vice president of Telemeter. “After all, what is at stake is merely a sports event or a movie.” He then argued that the relatively low security of pay-TV’s own scrambles sufficed.

  The world is full of systems which function satisfactorily in the business world but which have no more security than the most rudimentary secret television system…. Chewing gum and peanuts stand in glass bowls on every street corner exposed to the theft of any little boy with a brick….

  Obviously, a criminal organization can be established to manufacture decoders and to periodically forge code cards. Is the threat of such criminal activity serious? It sounds to me much easier to bootleg liquor instead of paying the high federal tax or engaging in a large variety of other less technically complicated criminal activities….

  What about the genius who uses correlation techniques, builds a device in his garage, etc.? We figure that no substantial percentage of the population can or will do this. This genius probably doesn’t want to watch the programs anyhow. He is probably too busy tinkering in his garage.

  Finally, the pay-TV proponent said, experience backed up these arguments. “In Hartford, Connecticut, a cryptographic pay-television system has now operated successfully for four years. The RKO General Corporation is now expanding the system. The provisions for security are now being minimized because it has become obvious that there is no security problem in pay-television…. I notice that the Jerrold organization, although it has had four years in which it could have gotten rich by building bootleg decoders in Hartford, has done nothing along these lines. Instead it has proceeded along making a fortune minding its own business.”

  Nevertheless, the F.C.C. has not yet approved either televised fee-vee or Jerrold’s wired version. The real reason lies in an economic argument that is even more broadly based than the economic argument against insecure scrambles. This is that the American public prefers free-vee.

  Despite heady flirtations with the dark attractions of secrecy, business chose the dependably beneficial quality of economy for its long-term association with codes. In that association lies the saga of the nonsecret code—a saga that is now all but ended.

  The roots of the nonsecret code reach back to the prearrangements of signals required for the most primitive means of rapid long-distance communication. Tom-toms, smoke signals, beacons of fire at night work only when the recipient knows what the signal-pattern means. These prearrangements constitute a code in the sense that a language is a code; their purpose is the very opposite of secrecy: it is to make communication possible, not impossible. Though the Romans had more than 3,000 towers for fire signals throughout their empire, rapid long-distance communication did not become available for business until 1794, when Claude Chappe installed an “aerial telegraph” from Paris to Lille. This consisted essentially of a semaphore system, with hilltop towers supporting the signal apparatus. A signal was repeated from tower to tower, and in good visibility would cover the 16 stations in the 140 miles from Paris to Lille in two minutes and the 116 stations from Paris to Toulon on the Mediterranean in 20 minutes. This was so much faster than messengers that Chappe’s system spread rapidly, not only in France, where it eventually created a network of 534 stations that served 29 towns, but in other countries of Europe.

  To speed up transmission, Chappe’s cousin, Leon Delaunay, made up a code representing 10,000 expressions by one to four figures. After using this for a while, Chappe devised a more efficient code in which 92 of the semaphore’s 196 positions were set aside as code positions. Three vocabularies, each of 92 pages with 92 expressions on each page, provided a lexicon of more than 25,000 elements. When similar lines were set up in other countries, similar codes sprang up and, by 1825, had become common. In 1830, Chappe expanded his code by assigning 184 semaphore positions to code use, giving a code of almost 34,000 elements. A Russian-language code appeared in St. Petersburg in 1839 for the Chappe network between that capital and Kronstadt and Warsaw that helped bridge the immense distances of the Russian empire. In 1845 a rather extensive Telegraphic Vocabulary for the Line of Semaphoric Telegraphs between Liverpool and Holyhead was published in London.

  At about the same time, England was improving maritime signals. During the American Revolution, Admiral Richard Kempenfeldt issued the first scientific naval signal book, which, after a struggle, finally established itself in the Royal Navy. In 1817, Captain Frederick Marryat published the first international code of signals, in which colored flags represented the numbers of words listed in a 9,000-item signal book. In 1857, the British Board of Trade published a draft code of more than 70,000 signals, which was adopted by many seafaring nations. Hoists of colored flags, of which there were 18, standing for all consonants but x and z, represented codewords in the book, which contained words and expressions used by sailors.

  Part of a page of one of the vocabularies for Claude Chappe’s “aerial telegraph” showing the three-armed semaphore’s two-position signals assigned to ships’ names

  In 1843, the first public electric telegraph line had been laid in England, and in 1844 Morse established the first public telegraph line in the United States. The electric telegraph, much faster than the Chappe semaphore, and usable in night and rain and fog, quickly supplanted the older system and spread very quickly through Europe and America. In 1845, former Maine Congressman Francis O. J. Smith, then 39, whom Morse had taken into partnership in the hope of finding in Smith the business acumen that artist Morse himself so lacked, published his The Secret Corresponding Vocabulary: Adapted for Use to Morse’s Electro-Magnetic Telegraph, the first code intended for the electric telegraph. Smith, a rather unscrupulous lawyer, may have gotten the idea for his code from one that Morse himself had compiled but discarded. The 1835 model of Morse’s telegraph transmitted ten symbols corresponding to the ten digits, and to convey words by means of them, Morse spent considerable time numbering words to form a special vocabulary for use with his apparatus. He used it in his first public demonstration of his telegraph, in 1837. But the invention of what is now the Morse code, which permitted the direct transmission of words by dots and dashes without the extra step of encoding, supervened and Morse jettisoned his vocabulary.

  Smith’s Vocabulary and one compiled at about the same time by Henry Rogers, entitled The Telegraph Dictionary and Seaman’s Signal Book, Adapted to Signals by Flags or Other Semaphores; and Arranged for Secret Correspondence, Through Morse’s Electro-magnetic Telegraph, both emphasized secrecy. But although businessmen desire secrecy in communication, they demand speed, accuracy, and economy before it. These motives soon became paramount in the public telegraph codes that followed Smith’s and Rogers’, such as John Wills’ Telegraphic Congressional Reporter of 1847, and in the private codes that American firms began to improvise as early as 1848. Their lexicons expanded and grew richer in phrases. And since groups of figures were more expensive to send than words and much more liable to error—the change of a single digit, which in the Morse code could result from the simple dropping of a dot, could mean an entirely different word—the codes shifted to the use of regular words as codewords. Thus CAT might mean sell and DOG, buy. By 1854, one eighth of the telegrams between New York and New Orleans passed in code.

  In 1866, the laying of the Atlantic cable gave an
immense impetus to commercial codes, as these nonsecret codes came to be called. Cable messages cost so much that the reduction in length made possible by code afforded enormous economies. Within eight years there appeared the first edition of the first public code destined to have a wide sale and a long life, The ABC Code, compiled by William Clausen-Thue, 40 years old, a shipping manager later elected a Fellow of the Royal Geographical Society. The ABC Code, which went through many editions, probably owed its success to its enormous vocabulary, which represented many business expressions of several words by a single dictionary codeword. The cable companies charged for codewords as if they were plain language, limiting both plain and codewords to a maximum of seven syllables.

  Codes could save so much money for telegraph and cable users that, it seemed, everyone who ever sent a telegram had one. In 1874, the Hebrew Orphan Asylum Printing Establishment issued M. Abenheim’s Telegraph-Code for Exclusive Use With His Correspondents. Detwiller & Street, a fireworks firm, had its own 20-page Telegraphic Chart, with some appropriate codewords (mammoth torpedoes, 3 case = FESTIVAL). India’s Department of Revenue and Agriculture had its 325-page Weather and Famine Telegraphic Word-Code, a two-part code (ENVELOPPE = Great swarms of locusts have appeared and ravaged the crops). The mackerel industry had a 5-page cable code of its own (ABDIC = extra quality, very fat and white), and so did the sausage industry. There were codes for tourists and the press. The big companies naturally had their own codes, the Erie Railroad Company’s running to 214 pages, Swift & Company’s to 554 pages (not counting the 364 pages of the separate code for its provision department), Lehman Brothers’ to two volumes. Wells, Fargo & Company prudently did not supply printed a codeword for the plaintext robbed, evidently preferring to fill this in by hand to afford some secrecy.