by Tom Wheeler
During his absence the telephone landscape of the United States had changed dramatically. Bell’s patent had expired in 1894, and independent telephone companies were springing up everywhere. By the time Vail returned to enjoy his riches, the independents’ 3 million users exceeded the Bell System’s 2.5 million. Strategically, the independents, which controlled most of the West and the nation’s rural areas, had the wherewithal to threaten Bell’s dominance in the major cities. The Wall Street bankers who had wrested control from the Bostonians who had overseen the Bell Company since the days of Alexander Graham Bell persuaded Vail to return to the company (now renamed American Telephone & Telegraph) in 1902 as a member of the board of directors. Five years later, in May 1907, a delegation of fellow directors appeared at his Vermont estate to ask Vail to return to the helm of the Bell System.
Twenty years after a dispute over corporate vision forced Theodore Vail out of Bell, he was back as the boss.30 He was sixty-two. More than any other individual, Theodore Vail is responsible for the existence of the ubiquitous national network that was necessary for the launch of the information age. At the same time, Vail also bears responsibility for policies that stymied the ultimate rollout of digital service.
As the new president of AT&T, Vail imposed policies that had been rejected twenty years earlier. At the heart of his vision was the concept of “one policy, one system, one universal service.” Vail set about to make his company the sole provider of that three-pronged offering.
In the company’s annual report for the year of his return, Vail laid out his philosophy. “The strength of the Bell System lies in its universality,” he explained. Then he declared war on competition. “Two exchange systems in the same community, each serving the same members, cannot be conceived of as a permancy [sic], nor can service in either be furnished at any material reduction because of the competition, if return on investment and proper maintenance be taken into account. Duplication of charges is a waste to the user.”31
A quarter century earlier Theodore Vail had been a competitive crusader fighting to keep Western Union from expanding its market dominance. At the helm of AT&T, he became the principal advocate for the concept of a “natural monopoly”—the notion that economic efficiency would be undermined by the existence of multiple firms and enhanced by the scale achievable by a single firm. It was the pre-antitrust era of integrated market dominance throughout the economy. Theodore Vail intended to become the integrator and dominant force in the electronic communications business.
Vail immediately began buying the competition. The Bell System swallowed independent telephone companies. Those who chose continued independence found it difficult to connect with AT&T’s long-distance network. When independents were allowed to connect, the Bell System dictated their operations as a condition of that interconnection.
Then Vail bought his old nemesis, Western Union. In 1879, Vail had warned of the effect of a monopoly controlling both telegraph and telephone. By 1909 he was that monopoly.
To achieve his “one policy, one system, one universal service” vision, Theodore Vail enlisted an unlikely partner: the government. As concern about monopolies—even “natural monopolies”—expanded, Vail made government both a watchdog and a co-conspirator. “Private management and ownership, subordinate to public interest and under rational control and regulation by national, state, and municipal bodies is the best possible system,” he expounded.32 In 1913, Vail agreed with the federal government to divest the newly acquired Western Union, cease from acquiring competitors, and to interconnect his long-distance services with other telephone companies to provide long-distance service. This agreement, named for Nathan Kingsbury, the AT&T general counsel who negotiated it (the “Kingsbury Commitment”), convinced the government to back off its antitrust inquiry into the company’s activities. It was the embodiment of Vail’s vision of a symbiotic relationship between his monopoly and government. There would be only one provider of telephone service per market and only one long-distance company to connect them. With the imprimatur of the government in return for regulatory oversight, the Vail vision was complete. The company flourished as service expanded.
The Vail-built troika of universal service, based on a government-regulated monopoly and driven by massive revenue with small margins, would define telecommunications for the twentieth century.
Competition and Innovation
In his 1910 report to stockholders Theodore Vail expounded on his vision: a “universal wire system” for the “electronic transmission of intelligence.”33 He was prescient in the synergy between universality and the electronic transmission of intelligence; the analog phone network would become the universal gateway to early digital activity. He was less insightful about the impediment to such innovation that his monopoly created.
Vail’s success in transferring AT&T’s focus away from a competitive marketplace had the effect of creating a sclerosis that hardened the network against change. While competition traditionally drives innovation, the company Vail created had succeeded in eliminating competition and thus could innovate (or not) at its own pace. The universality of the network Vail created may have become the threshold enabler of the information age. Yet that innovation would not occur without a fight, as the monopoly’s antibodies fought against the threat of disruptive innovation.
Under Vail, innovation was to be embraced but controlled. Vail, who had watched as Jay Gould used Edison’s technology to manipulate the activities of Western Union, appreciated how new technology could negatively affect an established business. He saw a means of mitigating such a threat if the company became a technology leader itself. He began a process of expansive technology development that looked into the needs of the future, but on Bell’s terms.34
In 1925, AT&T created Bell Laboratories, which ultimately became home to arguably more IQ per square foot than any other place on the planet. Bell Labs engineers originated ideas as important as Claude Shannon’s in “A Mathematical Theory of Communication,” which envisioned information as a physical quantity to be manipulated. Scientific American called it “the Magna Carta of the information age.”35 Bell Labs was the progenitor of the essential components to fulfill Shannon’s thesis, including the transistor, magnetic storage, and early computer languages.
AT&T became a frenetic propagator of innovation, but a sluggish adopter. While its leaders understood that the nature of networks must ultimately evolve, the first purpose of the corporation was to preserve the market position of the present network. One example of this policy at work was the development of magnetic storage. In the early 1930s a Bell Labs engineer, Clarence Hickman, developed the first answering machine by developing magnetic tape that would record sounds. It was a discovery about the magnetic storage of information with implications far beyond recording a missed phone message. AT&T ordered Bell Labs to cancel Hickman’s research because management worried that if the public had the ability to leave a message people would place fewer phone calls.36
The ultimate tool to exploit the symbiotic relationship between the government and the phone company was a federal rule (actually an FCC-approved tariff describing rates and services) that gave the AT&T monopoly total control over anything that attached to the network. The company used the full force of the federal government to prohibit the use of any “equipment, apparatus, circuit or device not furnished by the telephone company.”37 In a fit of public relations inspiration, such non-Bell devices were labeled “foreign attachments,” or sometimes “alien attachments.” The specter that a non-Bell “alien” device could cause physical harm to the network, perhaps even disabling the national defense system, was a carefully cultivated myth.38
The telephone company may have successfully built a universal pathway, but its absolute control of that pathway meant that AT&T, and through it the United States, would enter the digital age when AT&T damn well pleased. An example of this chilling control was the digital modem, the tool that ultimately opened the door to the intern
et.
Whereas the telephone delivered sounds via a smoothly undulating signal (called a waveform), a digital signal is choppy with square, discrete components. In order to carry a computer’s digital output over an analog phone line it is necessary to remodulate the digital signal into sounds that made it similar to a voice signal. The term “modem” is shorthand for how a device modulates digital information into analog signals (the screeching you hear when two fax machines connect) for transmission as if it were the sounds of a voice, and then demodulates it at the other end into digital format for the computer.
Since a modem connected to the telephone network, AT&T controlled both the unit’s capabilities and its availability. Using the power to dictate standards for connecting to the network—a ploy that Theodore Vail first used to impose his will on independent telephone companies—AT&T dictated the availability of modems, their cost, and their design. When the federal courts finally saw through the charade and began to chip away at the absurdity of the “foreign attachment” claims, the door to innovation in modem development and design began to open.39
Gradually, as the rules that allowed AT&T to unilaterally determine who and what attached to the network disappeared, the nation began to utilize the universality of the telephone network to deliver computer-to-computer information on a widespread, generally available basis. In the early 1980s, following the government’s decision to eliminate the restrictive rules, a new generation of inexpensive non-Bell modems came on to the market and the information age became manifest for anyone with a personal computer and a phone line.
From Data Transmission to Digital Networks
Our story of network evolution now requires a look back to the birth of digital packet switching discussed in chapter 1. Paul Baran’s 1964 paper, “On Distributed Communications,” identified how the hub-and-spoke analog telephone network could be made less vulnerable to attack by adopting a fishnet-like topology and digital packet routing.40 In the process he had also identified a potentially more efficient, lower-cost network technology than that used by AT&T.41
When the Department of Defense went to AT&T to request that the phone company adopt Baran’s ideas, it was turned down cold. “The Air Force said to AT&T, ‘Look, we’ll give you the money. Just do it,’ ” Baran would later recall. “AT&T replied, ‘It’s not going to work. And furthermore, we’re not going into competition with ourselves.’ ”42
The “it won’t work” response is not illogical. Baran’s concept meant trashing network truths that had been drilled into the brains of generations of engineers. The advocates for change described these monodimensional individuals as having “Bell-shaped heads.”
Ever since “Mr. Watson—Come here,” communications networks had been about establishing an open circuit between two points and maintaining that circuit for the duration of the transmission. It was a highly inefficient technology for which circuit capacity had to be built in anticipation of peak demand. Such costly inefficiency, however, was the economic backbone of Theodore Vail’s “natural monopoly.”
The connection between network economic inefficiency and monopoly underpinned Baran’s other observation about the phone company’s unwillingness to embrace a new, potentially competitive technology. The executives entrusted with Theodore Vail’s monument, the world’s greatest communications network and the mainstay of the stock holdings of widows and orphans, were not about to change their business model and the guarantee of a fixed rate of return on every dollar invested, even if that meant using inefficient infrastructure.
Thus, when Paul Baran presented AT&T with a network concept that increased efficiency by not holding a circuit open, but rather opening and closing it multiple times a second in order to load it chockablock with digital packets, the Bell System executives rolled their eyes and ran the other way. “I might as well have been speaking Swahili,” Baran once told me.43
Contrary to urban legend, the internet was not built as a means of surviving a Soviet attack. Such survivability had been the impetus for Paul Baran’s work that resulted in the packet-switching technology now used in the internet, but Baran’s second-strike network was never built. In fact, after the responsibility for such a network was transferred to the Defense Communications Agency (DCA), Baran worked actively to scuttle the project. “I told my friends in the Pentagon to abort this entire program—because they wouldn’t get it right,” Baran recalled. Having just fought circuit engineers at AT&T, Baran wasn’t anxious to leave his baby in the hands of yet another group of analog-minded engineers at DCA who, because they didn’t get it, would condemn the concept to failure. “The DCA would screw it up and then no one else would be allowed to try, given the failed attempt on the books,” Baran explained. He would rather wait until “a competent organization came along.”44
The Defense Department did build their own packet-switched network—but it was not for controlling bombers and missiles. In 1969 the department’s Advanced Research Project Agency (ARPA) connected the networks serving the mainframe computers of four U.S. research institutions so that scientists could share data and access each other’s computers.45 The project was known as ARPANET.46
It seems such a simple concept today, but ARPANET was a huge reach, both technically and philosophically. While not the internet, ARPANET was the internet’s opening act. Research institutions had acquired large mainframe computers, each accessible only via the institution’s private network. As in the early days of the telephone, these were purely local networks. But what would be the effect if they could be interconnected, just as “Vail’s Folly” had connected Boston and Providence?
ARPANET was not only a technological feat in connecting private networks to create a giant computer time-sharing network. Perhaps even more daunting, it required a cultural change as well. Competitive institutions signed up knowing full well that they were joining an “open” network that allowed access to each other’s previously sacrosanct computers.47
ARPANET was a hit in the academic world. Two years after its launch connecting four supercomputers, twenty-three universities and government research facilities were connected. By 1984 there were more than a thousand ARPANET hosts.48 The connections between the ARPANET nodes were Theodore Vail’s universal network.
Other data networks were starting up as well. The functionality of packet switching was adopted by commercial network service providers such as Tymnet, Telenet, and CompuServe. Essentially, each of these (and other) networks was using the efficiency of packet switching inside the network, but with its own particular protocols. This made the networks incompatible with each other. Information on one network had to be transformed in order to work on another. Theodore Vail had solved this problem by bludgeoning independent phone networks to adopt AT&T’s standards as a condition of interconnection; now it was necessary to develop similar standards, absent the bludgeon. In 1972 the renamed Defense Advanced Research Projects Agency (DARPA) commissioned a project to tackle this problem. It was called the “Internetting Project” and the term “internet”—a network of interoperating networks—was born.49
At the heart of this internet was a lingua franca known as the TCP/IP protocol suite. For their work in developing this lingua franca, Robert Kahn and Vinton Cerf have appropriately been dubbed the “Fathers of the Internet.”50 On New Year’s Day 1983, all of the host computers on ARPANET adopted the TCP/IP format. It remains today the rules of the road for the internet.51
The first challenge Kahn and Cerf faced was to define a uniform addressing methodology. “Packetizing” the information into small unique units of data begins the process. Then Internet Protocol (IP)—the principal communications protocol for the internet—delivers the digital packet to the right address. As shown in the figure in chapter 1, the digital-packet-switched network looks like a fishnet. At each intersection of this labyrinth is a small computing device known as a router. These routers chat back and forth with nearby routers to learn about each other—whom they connect to, and th
eir availability and status. When a packet arrives, the router looks at the database created by this gossiping for the IP address of its destination to pass the packet off to another router in the general direction of its destination. All of this occurs in a fraction of a second.
But with each packet traveling independently, the packets don’t arrive in the same order at their destination as they were launched. This is when the Transmission Control Protocol (TCP) goes to work. Each packet also contains information about how it fits into the whole of the collected packets. At the destination, TCP reassembles the packets in the correct order.
Many mark the birth of the internet with the implementation of TCP/IP protocols. While ARPANET previously had been about connecting computers, TCP/IP made it one giant network of interacting computers. The “internetworking” that allowed a collection of disparate but connected networks to interoperate through the same language became the internet’s hallmark.
Concurrently, the other determining characteristic of the internet was also in development. As the number of routers in the network expanded, the gossiping between them became inefficient. In the early 1990s, a new Border Gateway Protocol moved that routing function further to the edge of the network. With routing now accomplished at the borders of the network, the fishnet topology that Paul Baran envisioned in 1964 had become a reality.
But these developments were all about computers connecting with each other. What about finding information on these networks? And even if you did find what you were looking for, would it be in a format that was usable? Those issues were solved in 1990 when Tim Berners-Lee developed a means to identify, retrieve, and interrelate networked information.
Berners-Lee called his development the World Wide Web. When the internet became the web, it leapt out of the world of computer science into everyday usability and functionality.