The Idea Factory: Bell Labs and the Great Age of American Innovation

by Jon Gertner

  Systems engineers—the ones who looked at new ideas and decided whether they could improve the system—lived by Kelly’s rule: Better, or cheaper, or both. In the years immediately following the war, one idea that met with their approval involved a project whereby the Labs, working in conjunction with AT&T’s Long Lines Department, could ease the congestion on the long-distance phone network. The plan was to create a new, long-distance corridor that would forsake cables altogether—in effect, it would be wireless. At the time, all long-distance calls, along with some coast-to-coast TV transmissions, were carried on thick underground coaxial cables that crisscrossed the country. But it was believed that a nationwide chain of microwave relay antennas, linked to one another in a relatively straight line, could efficiently move calls and programs great distances. Indeed, the microwave links would liberate the phone company from the onus of buying and burying or stringing more of its expensive cables. To test the idea, “an eight-hop route,” as the Bell Laboratories Record described it—with seven antenna stations along a 220-mile corridor—was built in the late 1940s. The test route linked New York with Boston, via eight microwave towers, some built from concrete blocks and others from steel girders, most located on hillsides (or in some cases on tall urban buildings), topped by special horn-shaped antennas, vaguely resembling megaphones, which had been invented at Bell Labs largely under the direction of Harald Friis, the head of Bell Labs’ Holmdel, New Jersey, research office.22 Usually, two horn-shaped antennas on the towers would receive calls; a repeater apparatus inside the tower would amplify them; and then two other horn-shaped antennas, facing the opposite way, would instantly relay them to the next tower in the phone link. The height of these towers was crucial: Transmissions by microwaves traveled in straight lines and required a clear line of sight. Any interruptions—buildings, trees, mountains—would affect the signal.

  When the local test proved successful, AT&T and Western Electric built a national system of microwave links, requiring the construction of 107 towers across the United States, or about one every thirty miles. A phone call could now be handed off automatically—that is, received and immediately resent—between towers in some of the most remote locations imaginable. For instance, a long-distance call made from a Manhattan advertising agency could be switched through the network to a microwave tower atop a New York skyscraper; from there it could move from east to west at nearly the speed of light, passing from station to station—through a 406-foot steel tower in Des Moines, for instance, and another installation high atop Mount Rose, in Nevada—before ultimately arriving at a receiving station perched high in the Oakland hills east of San Francisco, where it would be routed to a switching center and tied into the local phone exchange.

  “Mark, how are you?” an AT&T vice president in New York asked Mark Sullivan, president of Pacific Bell, in San Francisco, during a call on the system’s opening day in August 1951.

  “It’s nice to hear your voice,” Sullivan said. “I’m fine, thank you.”23

  This exchange of dull pleasantries seemed fitting for the occasion. In contrast to the opening of the first transcontinental line some thirty-five years before, the relay system produced only a slight riffle of excitement, suggesting that the public had come to take for granted easy coast-to-coast communications. Microwave towers would shape the future of telecommunications, as well as the fate of Bell Laboratories. But at this point, nobody could see how.

  Eleven

  EMPIRE

  One project under Kelly’s direct supervision captured the public’s imagination, and expanded the system’s reach, in a way that microwave towers could not. The project was given the name TAT-1; it was the first transatlantic phone cable, a joint project of AT&T and the British Post Office, that was intended to carry thirty-six phone conversations at any given time from the tiny village of Clarenville in Newfoundland, Canada, to the city of Oban, Scotland. Actually, TAT-1 was two cables that would be laid side by side. One cable would carry voices to Europe, the other would carry responses back.

  Sending a message over land had always been easier than doing so over or under the water. Engineers had first tried to connect North America to Europe in the 1850s, when successive attempts were made to lay down a telegraph cable on the floor of the North Atlantic. “It was a mathematical impossibility to submerge the cable successfully at so great a depth, and if it were possible, no signals could be transmitted through so great a length,” the British royal astronomer predicted at the time.1 Indeed, the first two tries had ended in failure; the cable layings, done by way of a sailing ship outfitted with a giant spool of copper wire in its hold, were deemed perilous, multimillion-dollar disasters. Cables would snap, snag, kink, and leak; ocean storms would batter the crews and equipment; and in the end, transmissions on the line might work for a couple of weeks before going dead for no apparent reason. But in 1866, a cable—made of better materials, and laid down with more care and expertise—finally succeeded in carrying dots and dashes back and forth between Canada and Ireland. And in the decades after, engineers figured ingenious ways to increase the speed and capacity of other submarine cables. By the early 1900s, overseas telegraph communications had become a lucrative business.

  Human voices were different than telegraph signals. Carried by copper wire, the telephone signals were more complex and delicate; moreover, they would attenuate—that is, fade out—after a hundred miles or so. Years before, in the early part of the century, the same challenge had vexed Frank Jewett as he contemplated the idea of a telephone line that connected New York and San Francisco. “The crux of the problem,” Jewett wrote at the time, “was a satisfactory telephone repeater or amplifier.” Harold Arnold, Jewett’s research chief, had answered his prayers by developing a vacuum tube that could amplify signals and relay them forward. But as difficult as that task had been, stringing a cross-country line meant working on hard ground and in dry air. To put tubes and repeaters in a cable located miles under the sea, and then provide a constant electrical current for them, and then make sure the cable covering never leaked, that sea worms never gnawed through the covering, and that the components never failed? “If such a complex and technically difficult system could be made operational,” Kelly would later reflect, that would not automatically make it viable. The cable would still have to prove that it was economically justified. “In the event of a failure in midocean,” Kelly pointed out, “a cable ship would have to lift the cable at that point and make the necessary repairs. In winter, this might well remove the cable from service for two or so months and be a costly operation.”2 By Kelly’s calculations, the cable would have to work for at least twenty years without a single problem in order to pay for itself.

  For several decades, the safe answer had been to forsake the idea of an ocean cable altogether. Some of the early pioneers of radio transmission, including the Italian inventor Guglielmo Marconi, had proved that it was possible to send radio signals across the ocean by bouncing them off the earth’s outer atmosphere. In the late 1920s, as it happened, Kelly had worked with a Bell Labs team stationed at the end of Long Island that had built a transatlantic transmitting station to create regular radio-telephone service between the United States and Europe. By the 1950s there were sixteen radio channels operating between the continents. These were relatively cheap to operate (and certainly less expensive than a 2,200-mile undersea cable, which was projected to cost about $42 million, or roughly $340 million in today’s dollars). But overseas radio had one ineradicable fault: Weather and atmospheric conditions could wreak havoc on the transmissions. “When conditions were good,” Arthur C. Clarke wrote in his history of overseas communications, “transatlantic speech was of excellent quality, with little distortion or interference. But all too often the radio beams picked up most peculiar noises, like the sounds of cosmic frying-pans. These were usually no more than annoying, but sometimes they could obliterate the signal.”3 In fact, there could be periods of days when weather conditions made an overseas radio
call impossible. In the early 1950s, Kelly noted, “No way has been found to provide day-to-day continuity and reliability comparable to that of good wire lines.”4

  If you were building the best communications system in the world, as Kelly liked to imagine he was, this was not acceptable. At the same time, he noted, there was an obvious and unique answer: Figure out a way to make an undersea cable with repeaters spaced every forty miles or so, and then figure out a way to make it work for twenty years without leaks or interruption. To a large degree, then, the cable was a challenge of engineering rather than science, but even Kelly seemed humbled by “the magnitude of the task, its tremendous scope, the necessity for establishing that the designs of each and every item meet not only functional requirements but also the twenty-year no-failure requirement.”5 Fortunately, by the time Kelly began planning an undersea cable with British phone engineers in 1953, a small library already existed about what would and would not work. In the years following World War II, the Labs had tested various designs for undersea repeaters on shorter routes—notably a submarine cable that successfully connected Key West with Havana, Cuba. In the course of this project, the Labs had created a “flexible” repeater containing three vacuum tubes that could be inserted within the cable at forty-mile intervals and could be wound around a horizontal spool on a cable ship. In the case of the TAT-1, the planners would use a British vessel called the Monarch, the world’s largest cable ship.

  Even to the untrained eye, the repeater in the transatlantic line was not hard to spot. The final Bell Labs designs specified a cable 2,250 miles long and about one and a half inches thick—“such a precise piece of construction,” as Reader’s Digest put it, “that communications engineers speak of it almost with awe.” Unpeeling the ten-layer cable (an impossible task without power tools, actually, owing to its formidable strength), one might see that at its center were copper wires, surrounded by polyethylene insulation, and then another layer of copper conducting tape. These copper wires and tape were the coaxial cable used to carry the calls. Then there were six outer layers, designed with the knowledge gained from the late 1800s onward, for protection and strength. First came a copper layer to protect against undersea teredo worms, then successive layers of cotton, jute, steel, and more jute. At forty-mile intervals, where a repeater would be inserted, the thickness of the cable gently bulged to about three inches in diameter for a length of about twenty-eight feet before graduating back to one and a half inches again—making the cable look like a snake that had just had a modest meal, as Arthur Clarke put it.

  The engineers at the Labs had spent years drawing up an exact architecture for the repeater system so it could, as Kelly explained, “withstand the shocks of laying and recovery and also the pressure of water encountered.” The floor of the North Atlantic wasn’t smooth. Between Newfoundland and Scotland there existed a series of huge underwater peaks and troughs, meaning the cable would descend in places to well over two miles before climbing mile-long underwater inclines to less punishing depths. The engineers also designed the repeater sections with the cable ship in mind—so that it could “without damage to itself or to the cable be passed over the drum and sheaves of the cable ship and paid out without slowing up or stopping the vessel.”6 Clarke described the laying of the cable as a spider spinning out its thread.

  Kelly had spent years considering how the cable’s repeaters could operate flawlessly. It was the kind of question that suited the Labs statistical experts, the same ones who had invented the disciplines of quality control and quality assurance for the Western Electric factories. For something to be “better, or cheaper, or both,” as Kelly insisted, it had to last and last. It was not unusual at Bell Labs to run an experiment on, say, telephone poles—burying them in a swamp, or exposing them to harsh temperatures7—lasting twenty-five years and keeping a meticulous record of the results. In Kelly’s negotiations with the British engineers, there had been a fair number of disagreements over which country’s technology to use for the cable. “It was realized that compromise was essential but innovation and relatively untried methods were too risky,” an internal history of the Labs would put it.8 In fact, by the time the first cable was ready for laying down in 1955 (the return cable would go down in 1956), Kelly had made sure that the undersea project, in sum a great innovation, used the least innovative components. A transistor for the repeater system was out of the question; the technology was too new, and there was no telling how long it would last underwater. Only simple vacuum tubes that had been designed in the late 1930s, back when Kelly was still the Labs’ head of research, were to be considered—and these would be assembled from the best possible materials in a specially designed clean room at a new factory, built expressly for the purpose of the undersea project, in Hillside, New Jersey. The Labs’ engineers had reams of data on how well these tubes would work. Some had been running continually in a testing lab for sixteen years.

  A hurricane blew through the North Atlantic while the Monarch was laying down the first cable; otherwise, the operation went smoothly. The opening of the cable, on September 25, 1956, was a gala event at three main sites of the Labs—West Street, Murray Hill, and Whippany. On the occasion, AT&T chairman Cleo Craig called Charles Hill, Britain’s postmaster general. Technically speaking, Craig’s first words—“This is Cleo Craig in New York calling Dr. Hill in London”—were carried by a phone cable from New York to Portland, Maine; a radio relay system in Portland broadcast it by antenna to Sydney Mines, Nova Scotia. From Sydney Mines, the voice signals then entered a shallow-water cable running to Clarenville, Newfoundland. At Clarenville, the signals entered the deepwater portion of the newly-completed transatlantic cable. All along the bottom of the ocean, the message flashed through fifty-two repeaters over 2,250 miles, before emerging in Oban, Scotland. At Oban the transmission was directed by cable to London, where Hill was poised to respond to Craig. It all took less than a tenth of a second.

  Occasionally, fishing trawlers near the shore would cause breaks in phone service via the first transatlantic cable. But for twenty-two years after it was first activated, its technology never failed once.

  BY THE TIME the transatlantic cable came online, only two of the four men most closely associated with the transistor—Mervin Kelly and Walter Brattain—were still working at the Labs. John Bardeen had settled in as a physics professor at the University of Illinios; Bill Shockley had left to form a transistor company—Shockley Semiconductor—in Palo Alto, California. He had received some help from Kelly, who had introduced him to some wealthy investors. The assistance was almost certainly not selfless. Judging Shockley as unsuited for upper management, Kelly had refused to promote him—and as if to prove Kelly’s judgment correct, Shockley had in 1953 and 1954 sent his boss peevish memos written as “pinks,” the nickname for informal notes, complaining about the Labs and Kelly himself.9 Perhaps it was merely evidence of Shockley’s broader frustration. According to Ian Ross, a colleague of Shockley’s who would later become president of the Labs, Shockley simply felt he was not getting the rewards he deserved at Bell Laboratories. By the early 1950s, he was a department head overseeing a small group of scientists. He thought he should be higher in the organization and paid much more. Thwarted by Kelly, Shockley appealed to people even more powerful. In one instance, he even went to the president of AT&T. When he got nowhere, Ross recalls, “he said, the hell with that, I’ll go set up my own business, I’ll make a million dollars that way. And by the way, I’ll do it out in California.”10

  As yet there wasn’t much in the way of technology out in Palo Alto. Mostly it was apricot orchards and undeveloped land, but it had been Shockley’s hometown for most of his childhood. Also, there was Stanford University, where he had a booster named Frederick Terman, the school’s provost. Come to the Valley, Terman had told Shockley, and he could help him find an office in a new industrial park for young, innovative companies.11

  Shockley tried to lure several of his Labs colleagues out west
with him. His notebooks of the era are filled with potential candidates for jobs. For those he liked, he promised grand adventures in a new industry and, at least to the silicon transistor inventor Morris Tanenbaum, twice his Bell Labs salary. Shockley only managed to woo one person from Bell Labs. Mostly, he located and hired some promising young scientists from other companies—most notably Gordon Moore, Robert Noyce, Jean Hoerni, and Eugene Kleiner, all four of whom would do much to put Silicon Valley on the map. None of the recruits seemed particularly aware of Shockley’s shortcomings as a manager. And even so, the attraction of working for his new company was obvious. Robert Noyce famously described what it was like for a young solid-state physicist, toiling in obscurity, to discover that Bill Shockley was calling him: “It was like picking up the phone and talking to God.”12

  Even as Shockley was preparing to leave the Labs, Kelly, as a foreign member of the Swedish Academy of Sciences, had lobbied for the transistor team, though now disbanded, to be awarded the Nobel Prize.13 For years there had been rumors they were in the running, and Brattain, Bardeen, and Shockley heard on November 2, 1956, that they would indeed share a Nobel Prize. Brattain refused to attribute the honor to his own genius. “It was really only a stroke of luck,” he later explained, that he and Bardeen, rather than someone else, had gotten there first. Everyone at Bell Labs knew the prize reflected well on Kelly. Even Shockley was moved, in his own way, to share some credit. “It seems to me that this is a suitable occasion for me to repeat what I told you over the telephone when resigning from Bell Telephone Laboratories,” he wrote to Kelly from California upon the receipt of the Nobel. “It is hard for me to see how a research director and vice president in your position could have proceeded more effectively to get a transistor out of a solid state physicist like myself.”14