During the mid-1980s, a new craze seized Paris and much of France. Every day, hundreds of thousands of French telephone customers, in homes and offices, sat down at a chic computer terminal called Le Minitel and went "on-line." Typing various combinations of letters and numbers, they summoned to the terminal's nine-inch black-and-white screen a stream of text and images representing an astonishing array of services and information. Through the Minitel, for example, they could check a bank credit-card balance, buy a suit or dress, scan the morning newspaper, read the latest weather report or leave a message in a friend's electronic mailbox.
More popular than any of these activities was one called Dialog. Though the words were typed rather than spoken into the machines, Dialog often resembled a telecommunications free-for-all, in which anyone who wished to could join a conversation. Flirtatious participants adopted pseudonyms that freed them to spice their repartee with sexual innuendo. Dialog and other services expanded so rapidly that several times during the summer of 1985 the volume of subscribers attempting to use their terminals overloaded the system's main computer. Like a power grid sapped by too many air conditioners turned on in steamy weather, the computer shut down, leaving Minitel addicts to sweat it out until service could be restored. (Top)
This remarkable system of computer communications is called Teletel. Launched in 1981 by the government-owned telephone and postal monopoly, Postes, Té1écommunications et Té1édiffusion (P.T.T.), Teletel was conceived mainly as an electronic telephone directory. P.T.T. officials expected the service to pay for itself by reducing both the number of telephone books printed and the need for directory-assistance operators. Private companies soon began to offer banking, entertainment and information services on the system, drawn by the promise of revenues amounting to 60 percent of the fees paid by Teletel customers to P.T.T. By 1986, the number of on-line services exceeded 2,000, having increased tenfold in just one year.
Teletel is not inexpensive. Home subscribers to the service pay nothing for the terminal, but they are charged for connect timethe period they are dialed into Teletelat the rate of about 13 cents per minute. At such prices, it is easy to run up a substantial Teletel bill, as one Parisian marketing executive discovered. In a single month, his mildly insomniac 17-year-old daughter spent 500 francssome $65just playing games available on the network. (Top)
Teletel has proved to be much more than a break-even proposition, earning about $70 million a year for P.T.T. According to Daniel P. Resnick, a history professor at Pittsburgh's Carnegie-Mellon University and former executive director of a French government agency established to encourage the use of computers, Teletel "is an enormous public success, the most promising public experiment using telephone lines in the world, and the audience is unlimited."
Granted, the future of an infant such as Teletel is always difficult to predict. And telecomputing, a word coined to describe the computer's impact on telecommunications, has had its share of unfulfilled promises. But there is little doubt that a revolution is under way, leading to the day when -as John F. Akers, the president of IBM, has put it- "everything is connected to everything." (Top)
Communications between computers have become extraordinarily fast. When the first long-distance communications link between a computing machine and a remote terminal was established almost five decades ago, information in the form of binary digits, or bits, was transmitted at the rate of fewer than 50 bits per second. With modern satellites and fiber-optics technology-the art of communicating with light pulsed along fine filaments of glass-data may fly between distant computers at 1.5 million bits per second -a speed that would allow Herman Melville's 222,000-word masterpiece Moby Dick to be sent from one computer to another in less than 30 seconds.
Since the beginning of computer communications, the vast majority of computer data has traveled along ordinary telephone circuits, the ubiquitous network that makes it possible for anyone with a telephone to speak with virtually anyone else on earth who also possesses such an instrument. But the union of computers and the phone network was an unnatural one. Telephone circuits were designed to handle the electrical equivalent, or analog, of the human voice. The telephone converts a speaker's words into an electrical voltage whose frequency varies according to the pitch of the voice and whose amplitude increases or decreases according to its loudness. Computers, in contrast, express information in digital form, as a stream of discrete electrical pulses. (Top)
This incompatibility between computer data and the telephone network was resolved by a clever translating device called a modem (from modulator-demodulator).
Inserted between a computer and the phone system at the origin of a transmission, it converts, or modulates, the computer's digital output into analog signals, which can be handled like a telephone call. At the receiving end, the modem demodulates the incoming telephone signal, restoring the information to the digital form acceptable to the computer.
For many, the modem has been associated exclusively with computers, especially the desktop variety. Yet in the fast-paced context of electrical communications, the modem is practically as ancient as the wheel. Its origins can be traced to the merging of devices invented for similar purposes but fundamentally different in their technology -the telegraph, the telephone and the teletypewriter.
The telegraph preceded the telephone by more than 30 years and the teletypewriter by half a century, and in some ways it was the forerunner of telecomputing. To begin with, the telegraph, like the modern computer, was a digital medium. Depressing the sending key sent a discrete pulse of electricity along an iron wire to a distant destination. The transmission had three states: off, on for a short time (a dot) and on for a longer time (a dash). (Top)
Speed of transmission was an important consideration from the outset. Adopting the dot-and-dash system made it possible for telegraphy's inventor, Samuel F. B. Morse, and his assistant, Alfred Vail, to represent the alphabet and numbers with far fewer keystrokes than a two-state system would have permitted. To reduce to the minimum the time it would take to transmit a message, Morse and Vail decided that the most frequently used letters of the alphabet would be indicated by the shortest possible sequences of dots and dashes. They visited a printer and counted the pieces of type in each letter case; e's and t's outnumbered all other letters, reflecting the frequent appearance of the pair in the English language. So Morse and Vail represented the letter e with a single dot and the letter t with a single dash. This rational approach resulted in a telegraphic code that could send messages at the rate of about two characters-equivalent to the now-modest rate of 15 bits per second. (Top)
In 1844, with the aid of a congressional appropriation of $30,000, Morse inaugurated long-distance telegraph service in the United States. From Washington, D.C., he tapped out the dots and dashes signifying the Biblical text, "What hath God wrought!" The electric pulses sped along 37 miles of wire to Baltimore, Maryland, where Vail received the staccato stream of pulses and repeated it back to Washington. Seventeen years later, the telegraph spanned the continent, replacing the fabled Pony Express, which had previously relayed messages through 1 57 stations along the 1,840-mile route between St. Joseph, Missouri, and San Francisco.
One of the earliest teletypewriters was invented during the late 1890s by Donald Murray, a New Zealander. Murray was working at a newspaper in Sydney, Australia, at the time, and he originally intended his machine as an automatic typesetter. When little came of that endeavor, he applied the same principle to telegraphy. The heart of Murray's teletypewriter was a rotating drum. When the operator pressed a key, a complex chain of mechanical and electrical events began, culminating in the transmission of a letter.
In its early form, Murray's teletypewriter was so crudely built that Scientific American magazine likened it to "a sort of cross between a sewing machine and a barrel organ." Because the drum had to be turned manually by a wooden handle, skeptics called it "Murray's coffee mill" or "the Australian sausage machine." But Murray improved the machine-first in the U.S., and then in England with the backing of the British post office-by adding, among other features, an electric motor to turn the drum; he put his device into service in 1903. Teletypes, as they came to be called, required far less training and skill than the Morse telegraph. Anyone who could type now qualified as a telegrapher. (Top)
SPEED OF PAPER TAPE
Clattering away in their five-bit Baudot teletypes blazed many trails later followed by the computer. Paper-tape readers and punches, for example, were essential to many early computers, and telegrams were actually the firste lectronic mail. Teletype networks also handled airline reservations and served as information services, providing up-to-the-moment weather data and stock-market quotations. They acted as a communications tool for big corporations and, during World War II, for U.S. President Franklin D. Roosevelt and British Prime Minister Winston Churchill, who wished to"chat"across the Atlantic and have a written record of their conversations. Teletypes allowed department stores to swap credit information and banks to transfer funds. They furnished a kind of electronic bulletin board for police agencies, linking them from coast to coast and printing information on stolen vehicles and fugitive criminals. They transmitted news from around the world to local newspapers and even fulfilled Donald Murray's turn-of-the-century dream by perforating tape that, when fed to Linotype machines, automatically set the news in type. (Top)
The solution, of course, was the telephone. When Alexander Graham Bell constructed the device in 1876, he was actually attempting to create a harmonic telegraph. This apparatus was to permit several telegraph messages to be transmitted at the same time over one wire, saving the expense of stringing multiple telegraph lines to accommodate traffic on busy routes. Bell's idea was to have many sending keys in a telegraph office connected to a single circuit; the keys would send dots and dashes at different frequencies, or pitches, instead of identical ones that, if mixed together, would be impossible to resolve into the original messages. Receivers at the destination, each responding to a different frequency, would untangle the jumble of messages. In practice, the system was unreliable, but the principle worked well enough to convince Bell that many frequencies-even as many as produced by the human voice-could be sent simultaneously over a single wire. (Top)
The telephone soon gained popularity, spreading through communities large and small, and ultimately extending across the nation. Western Union, principal provider of telegraph service in the U.S., benefited enormously from the expansion of the telephone network. By 1888, when the first intercity telephone circuits were established, it was possible to send telegraph messages and telephone conversations at the same time over the same wires by employing an early version of a process known as frequency-division multiplexing). Because voice messages contain little sound below about 250 hertz, or cycles per second, the band of frequencies from that point down to zero hertz went unused in the phone system. Inasmuch as a telegrapher of superior skills could transmit only about 30 words per minute- the equivalent of 20 hertz- there was a gap of more than 200 hertz between the top of the telegraph's bandwidth, or possible range of frequencies, and the bottom of the voice bandwidth.
With such ample separation between the signals, even the comparatively crude, frequency-sensitive filters that were available around the turn of the century could easily separate a telegraph message from a voice message traveling the same wire. Thus the analog voice waves of a telephone conversation could travel in the bandwidth between 250 hertz and 3,400 hertz, while the digital pulses of telegraphy streamed along in the lower band. As a result, Western Union was able to establish telegraph connections anywhere that a telephone line went. Though fees were paid to the Bell Company for use of the lines, Western Union was spared the expense of stringing its own wires. (Top)
Frequency-division multiplexing was greatly enhanced during the 1920s when the telephone system "went electronic," in the words of one historian. The newly perfected vacuum tube permitted Bell engineers to devise a way to transmit multiple telephone conversations on a single circuit. They accomplished this by boosting the 250- to 3,400-hertz frequency band of a voice on the telephone up to higher frequency ranges—4,250 to 7,650 hertz, 8,250 to 1 1,650 hertz, and so on—before transmission between telephone exchanges. At the telephone exchange near the destination, the signals were restored to the original frequency band and then routed to the recipient. Improved filters and a gap of 600 hertz between adjacent voice frequency bands prevented transmissions from contaminating one another. By this process, up to a dozen conversations could travel simultaneously on one circuit.
Analogous techniques were then applied to both telegraph and teletype messages, making those transmissions indistinguishable from a voice message as far as the telephone network was concerned. This piggybacking on the telephone system was particularly significant to the growth of teletyping because telephone lines went into the offices of business and government, where teletypes soon found widespread application.
At first the equipment that modulated and demodulated the teletype signals consisted of bulky, vacuum-tube circuitry located at the telephone company's central offices. By the late 1 950s, as the demand for computer-to-computer communication began to increase, the electronics had acquired the name "modem" and were built into the terminals-which bore an external resemblance to the teletypewriter -used with computers of the day. (Top)
But Stibitz' penchant for tinkering eventually led to several notable landmarks in computing. In 1937, Stibitz put together in his spare time the first machine developed in the U.S. to do binary arithmetic. Stibitz built the prototype at home, wiring together telephone-system components, batteries and other parts into the device, which he called the Model K -because he assembled it on the kitchen table. Then, in collaboration with Samuel B. Williams, a veteran Bell engineer who specialized in the switching equipment and techniques used in telephone networking, Stibitz developed a more ambitious machinea digital calculator designed to handle complex numbers, a class of mathematical entities encountered in the design of long-distance telephone networks and elsewhere. (Top)
The resulting device, which Stibitz called the Complex Number Calculator, was one of the earliest digital computers. It became operational early in 1940 at Bell Labs' Manhattan headquarters. Input and output for the machine were accommodated by a keyboard and printer adapted from standard teletype equipment. The calculator, which was locked in a closet, was soon linked to three of the teletype terminals, each located on a different floor of the building.
The Complex Number Calculator drew the attention of the American Mathematical Society, which invited Stibitz to present a paper on the machine to a meeting at Dartmouth College in Hanover, New Hampshire. To show off the computer's prowess, Dr. T. C. Fry, Stibitz' boss, suggested that he take numerical problems from the audience and telephone them to a keyboard operator in New York for solution. But Stibitz had a better idea. "With my usual genius for making things more difficult for myself and others," he wrote later, "I suggested direct telegraph operation from Hanover, and this was decided upon." (Top)
In New York, the three terminals communicated with the Stibitz computer over several wires-one for each number, mathematical sign and control character-bundled inside a cable. Running a cable the 250 miles from Hanover to New York was out of the question but, Stibitz suggested, an ordinary teletype circuit would serve the purpose.
Stibitz' collaborator, Sam Williams, went to work. Williams, whom Stibitz remembered later as "a quiet, very hard worker who seldom got very excited," built a special terminal that would encode each keystroke for transmission over the Bell system's Teletypewriter Exchange (TWX) service. The system he devised was a modification of the teletype's standard Baudot code. A decoder at the receiving end in New York restored the signals to the form required to operate the calculator. Furthermore, to avoid overloading the calculator, Williams designed the terminal to lock the keyboard momentarily after each keystroke. A fraction of a second later, after the calculator in New York signaled that the instruction had arrived, the keyboard unlocked to accept the next part of the problem. (Top)
On September 11, 1940, Stibitz staged his demonstration of long-distance computing as part of his presentation at Dartmouth. In the audience were three men in particular who would exert a profound influence on computer science: mathematicians John von Neumann and Norbert Wiener, and John W. Mauchly, who a few years later helped invent ENIAC, the world's first large-scale electronic digital computer. Stibitz, Wiener and others typed in complex-number problems on the keyboard, and within a minute the correct answer to each problem came racing back across the wire from the computer in New York.
QUEST FOR SPEED BEGINS
A major push for greater speed came during the I 950s with the construction of the SAGE air-defense system, the first of the large-scale, computer-centered communications networks. SAGE—for Semi-Automatic Ground Environment—linked hundreds of radar stations in the U.S. and Canada to 27 regional command and control centers. The centers were built around computers modeled after M.I.T.'s Whirlwind, the world's first computer capable of responding in real time—that is, without noticeable delay to the user.
SAGE eventually included approximately 1.5 million miles of communications line. To feed information from the radar stations to the command centers for computer analysis, this huge network needed both new techniques for coding the data in digital form and modems that could transmit it rapidly over telephone lines. In the event that enemy aircraft were detected by the radar network, these modems had to effect the alert virtually instantaneously.
The first high-speed modems were developed by scientists at M.I.T.'s Lincoln Laboratory, which supervised SAGE development for the U.S. Air Force. "At first the telephone company was dubious about what we were doing," recalled Robert R. Everett, a key member of the staff at Lincoln. "When the first telephone line for radar data came into the Whirlwind building to be wired into one of the modems, the telephone installer insisted on wiring it into a handset. We told him we didn't want the handset, but he said it was regulations and that was that. When he left, we connected it to the modem." (Top)
INTO HIGH GEAR
The rigors of meeting SAGE requirements helped prepare Bell researchers for the rising demand for modems and other computer communications technology. Bell's Dataphone system, introduced to the public on a limited basis in 1958, provided the first commercial modems expressly designed for transmitting computer data over ordinary telephone lines. Initially, Dataphone modems could transmit at only about two thirds the rate achieved on the SAGE network. This limitation was imposed largely by the tendency of digital signals to become distorted during transmission. Distortion results in part from different frequencies of a signal traveling over a wire at slightly different speeds, causing a phenomenon that communications engineers call echoes. Over the years, the telephone company had succeeded in reducing the effect of these echoes to the extent that they no longer hindered voice conversations, but the residuum, which made it difficult for a modem to distinguish between ones and zeros in the data stream, was more than enough to ruin computer data. The faster the pace of the data, the shorter the duration of each bit and the more likely a zero will be mistaken for a one or vice versa. (Top)
In the SAGE system, engineers were better able to deal with distortion -and thus maintain high-speed transmission -because certain telephone lines were reserved for use exclusively by the system. On dedicated lines such as those, the physical path that the data followed never changed; engineers could analyze distortion in the line and correct it by installing devices called equalizers. The ordinary, switched telephone lines for which the Dataphone modems were intended posed a different problem. On these lines, the physical link between a sender and a receiver had to be established each time a call was made; depending on which circuits were f ree, the route could be different every time. Though it was theoretically possible to equalize such a path once it had been established, doing so proved impractical at first. Thus the only way to keep errors within to lerable limits (about one bit in a million) in the Dataphone system was to transmit fewer, longer pulses per second.
Inspiration for the solution came to Lucky about a month later as he was driving home from work. "I was sitting at this red light," he said, "and it just came to me, the whole thing. And I went home and couldn't sleep. I raced into work with the first light of the sun, and I put it on the computer and naturally it worked perfectly. Sometimes I stop at that light now and wait for an inspiration to hit me. But the light never worked for me again."
Lucky's breakthrough approach to automatic equalization required that the circuits he designed undergo what he called a "training period" of perhaps a second or two after a telephone connection was established. During that time, the originator of the transmission sent out sample pulses at a slow rate. This step was essential, said Lucky, because "in a normal data transmission, you couldn't tell what was an echo and what wasn't; it's all jumbled together." Lucky's equalizer examined the echoes from the widely separated pulses, then generated artificial echoes to counteract the ones responsible for the distortion. Lucky later improved the device so that it could compensate for echoes without a training period and adapt continuously to a telephone circuit's changing electrical characteristics. (Top)
Lucky's adaptive equalizer, together with improved techniques for modulating digital signals, opened the way for faster modems. By the early 1970s, Bell Telephone modems were able to transmit data over ordinary switched telephone lines at rates of up to 4,800 bits per second. As it turned out, the adaptive equalizer improved data-transmission speeds over dedicated lines as well. Though such lines feature permanent connections, various electronic factors cause noticeable fluctuations in distortion. Lucky's invention automatically compensated for such variations; earlier techniques could not. Consequently, it became possible to send as many as 9,600 bits per second over dedicated lines, six times the initial speed achieved for SAGE.
The trend toward portable modems began in the 1960s. At that time, a modem was built into the computer terminal and then permanently connected to the telephone lines. When users wanted to move the machine to a new location in the office, they had to call in the phone company to make expensive new connections. (Top)
Confronted with this inconvenience in his own office, John Van Geen did something about it. In 1966, Van Geen was a 37-year-old engineer at California's Stanford Research Institute (now SRI International). He and other researchers were connected through Bell's TWX teletype network to the pioneering computer time-sharing service a continent away at Dartmouth College. Every time Van Geen or his colleagues wanted to hook up with the computer to make mathematical computations, they had to walk to the terminal's permanent installation at the end of a hallway. The machine was too expensive for everybody to have one.
One way around the problem was to build a modem that could use a telephone as the link to the network. With such a device, the permanent wiring between the terminal and the TWX network would no longer be necessary; a terminal could be rolled to any office, where it could be connected to an ordinary phone line. When Van Geen began his work on this idea, there were a number of such modems on the market. Typically, the connection to the terminal was made by wires fastened to the same attachment points that the built-in modem had used; the connection to the telephone was made by a device known as an acoustic coupler, which had sound-deadening receptacles for the mouthpiece and earpiece of a standard telephone handset. The modem converted digital signals from the terminal into analog signals for the telephone network and sent them out through the mouthpiece to Dartmouth. Results from the computer, arriving through the earpiece, were demodulated and then—in an era when television-like computer screens were rare—printed like a telegram. (Top)
Van Geen, however, found the existing modems lacking; they worked well for links to nearby computers but functioned unpredictably over long-distance telephone lines, such as those connecting Van Geen and his colleagues to the computer in New Hampshire. The modems could not distinguish reliably between the weak data signal that arrived in California after a 3,000-mile journey and the noise that accompanied the information. Consequently, data received from the computer contained an unacceptable number of errors.
In his own device, Van Geen retained in principle the acoustic coupler and transmitter from current modems and then went on to solve the reception problem by incorporating a different type of receiver in his modem, one that tolerated wide variations in signal strength and noise. it was so successful that he was able to receive virtually error-free data over distances as great as 6,000 miles.
Even so, some errors could be introduced into the modem by loud sounds nearby, which were able to penetrate the sound-deadening material of the acoustic coupler. Such errors might have been eliminated had Van Geen wired his modem directly to the telephone system. But in 1966, such a connection would have been illegal. At that time, telephone-company rules called tariffs, in addition to establishing rates, prohibited connecting to the telephone network any devices not furnished by the phone company. The argument against so-called foreign equipment was that it might damage the network; the penalty for ignoring this rule could be termination of telephone service. (Top)
Nonetheless, Van Geen and his colleagues worked closely with the telephone company to ensure that his modem's transmitter would exactly mimic the signals sent out by Bell's modem, thus defusing the potential charge that Van Geen's device would generate signals incompatible with the network. Moreover, because the new modem caused no harm—and because the phone company saw the possibility of increased revenues from the growth of data transmission—no action was taken to halt its use.
In 1968, two years after Van Geen's acoustic-coupled modem appeared, the Federal Communications Commission (FCC) issued a ruling that ended the tariffs' strangle hold on the use of foreign equipment. In this far-reaching decision, the manufacturers of a device called the Carterphone, which linked two-way radios and telephones, prevailed over the telephone company. As a result, AT&T could no longer ban equipment from other suppliers, although it retained the right to insert protective circuitry between such products and the AT&T network. By 1977, even that restriction was dropped as the FCC inaugurated a program of certification for telephone equipment. Any device that conformed to certain published standards could now be plugged directly into the network. (Top)
The Carterphone decision spawned an entire industry devoted to the manufacture of telephone-related equipment, from ordinary phones to sophisticated office communications systems. As for modems, competition soon made them smaller, faster and cheaper. In the early 1980s, as personal computers proliferated, sales of modems soared. Ultimately, the modem, etched onto a single, integrated-circuit chip, may become a standard feature of computers. (Top)
Copyright Time-Life Books, 1986