The development of Digital’s first computer provided both Licklider and McCarthy with the opportunity they needed. In 1961, the two joined forces to adapt a brand-new Digital mini so that it could work as a time-sharing machine. The same year, the MIT computing center took up McCarthy’s challenge to reconfigure a newly delivered IBM so that it, too, could accommodate multiple users at once. Time-sharing had moved from wild idea to reality.
* * *
—
Then, time-sharing blew up to immense scale. For in 1962, J. C. R. Licklider went to Washington, appointed to run a new computing initiative within that little Defense Department agency created in late 1957 to chase those post-Sputnik “will-o-the-wisps”: ARPA. Leveraging his considerable reputation and the full force of his personality, Licklider secured a remarkably generous budget and considerable administrative freedom to spend it where he wished—and he proceeded to direct a considerable chunk of these resources to MIT and to Stanford, where John McCarthy had just joined the faculty.
As money flowed from ARPA to academic computing projects dedicated to advancing “man-computer symbiosis,” the Sputnik-inspired research agency effectively created an entirely new discipline at America’s leading universities: no longer merely a variant of physics or electrical engineering, but the stand-alone intellectual pursuit of “computer science.” Stanford got its computer science department in early 1965, the last new program established by Fred Terman as Provost. John McCarthy became one of six founding faculty members of the small-but-mighty unit.7
With ARPA dollars as the spark, time-sharing spread rapidly. Now, operations that couldn’t afford their own mainframe (or even their own minicomputer) could tap in to the thinking power of an electronic brain in another room, another building, or across town. Policymakers and journalists soon were rhapsodizing about the networks’ ability to bring computer access to the masses, becoming a “computer utility” just like electricity or telephone service. “Barring unforeseen obstacles,” cheered MIT management professor Martin Greenberger in the Atlantic Monthly in 1964, “an on-line interactive computer service, provided commercially by an information utility, may be as commonplace by A.D. 2000 as the telephone service is today.”8
THE TIME-SHARING BOOM
Although time-sharing services were born to serve academic people performing academic work, there now was money to be made in providing them for users hungry to access computing power from their offices and homes. Thus, the second half of the 1960s saw a surge of start-up companies trying to hack into the business of sharing computer time. In Texas, there were operations like H. Ross Perot’s Electronic Data Systems and Sam Wyly’s University Computing Company, which distributed computer power and provided software services to mainframe users. But companies like those were really not networking in the way that McCarthy and Minsky and Licklider imagined it; human programmers, not computers, fed data into the machines and determined each run’s order and length. It was batch processing by telephone wire.
True commercial time-sharing came from a smaller, more technologically sophisticated group of start-ups, who programmed powerful scientific computers to determine the order and run time of jobs, and who built out computer-communication networks atop the phone lines to provide faster, cheaper, and more powerful processing services to their customers. It was in a company that would eventually become a leader in that second category, a rangy little Palo Alto start-up called Tymshare, that Ann Hardy found herself at the beginning of 1966.
Tymshare was a California creation, through and through. It was the idea of two engineers working at the Santa Clara Valley outpost of General Electric, Tom O’Rourke and Dave Schmidt, who looked at the region’s electronics companies and saw a robust customer base for time-sharing just waiting to happen. Tymshare’s first demos happened at Lockheed Missiles and Space, where engineers had been waiting twenty-four hours for their FORTRAN code to process and jumped at the chance to speed things up. It used a powerful scientific computer manufactured by Southern California–based Scientific Data Systems, or SDS, which had been modified into a time-sharing machine by an outfit based in Berkeley.9
Ann Hardy had been living and working at Livermore Lab for five years, but her husband’s job moved over to Palo Alto, and it simply wasn’t feasible to do the cross-Bay commute (and unimaginable that the wife’s work would take precedence). New in town, jobless and aimless, Hardy decided that she needed a home time-sharing terminal to keep her skills sharp, and began to look around for a service to which she could subscribe.
Hearing about Tymshare, she cold-called Dave Schmidt, who was so bewildered by the idea of a local housewife who wanted to “play around” on a time-shared computer that he confessed that Tymshare’s machine wasn’t yet operable. Accustomed to working on a time-sharing machine at GE that already was able to accommodate multiple users, O’Rourke and Schmidt hadn’t realized that the machine they ordered up from Berkeley wouldn’t have a similarly capacious operating system (OS) on it. The machine had been tested to accommodate up to two users; a commercial operation like Tymshare needed one that could share time among at least twenty. They needed to upgrade the system. “I told them they needed to hire me,” remembered Hardy. Schmidt agreed, little realizing how fundamental Hardy’s programming would be to Tymshare’s eventual success. “If I had known what the system meant to this company,” he later admitted to her, “I never would have hired a woman to work on it.”10
Tymshare quickly became a hit. Aerospace engineers at Lockheed and Philco-Ford thirsted for access to computing power, and now at last came a service that let them tap into all the computer power they needed and wanted. But with a user base that was almost entirely engineers, the Tymshare computer quickly began to get hacked. They probably didn’t mean to stir up trouble, Hardy reasoned. “As soon as you let an engineer on a computer, they try everything.”11
The tech world was still decades away from encryption standards, but the Tymshare team almost immediately started to encrypt their passwords to stabilize the system. Hardy buckled down to start building other checks and balances into the OS so that it was harder for users to precipitate a crash. While big computer companies were still figuring out shared protocols to allow different machines to talk to one another, Tymshare had leapt ahead to find ways to make these communications secure.
Then came the network itself. Tymshare’s founders had started their company with only a local market in mind, but the success of the service soon had them pushing out to new locations. But since the only feasible way for its customers to send data over the wires was via local telephone lines, going national meant creating full-bore computer centers in every city they wanted to serve—an expensive and hardly scalable proposition.
As Tymshare’s executives tussled with the dilemma of building out a national data network in an era when no infrastructure for the thing existed, Hardy reached out to a programmer she had known at Livermore Labs whose name (LaRoy Tymes) and expertise (computer networking) made him seem fated to work at Tymshare. When first hired at Livermore, Tymes had been a twenty-year-old without a college degree; he had intended to become an apprentice electrician. Suddenly, thanks to the government’s great technical hiring spree, he was operating computers in one of the military’s most important, top secret facilities. “It was way beyond anything I’d ever seen in my life or even heard of,” he remembered. “I was . . . very impressed by the fact that they would entrust these multimillion-dollar machines to somebody like me.”12
Like Ann Hardy, LaRoy Tymes had stumbled into programming, and it proved to be his great calling. He came to Tymshare in early 1968 and set about building a brilliant work-around to the networking problem: a national network of minicomputers that could effectively act as information middlemen between the customer and the all-powerful SDS 940, tied together with multiplexed lines that allowed many messages to go across the wires at the same time. Tymes’s work-around—branded Tymnet—tur
ned out to be a pace-setting innovation of its own. From its 1971 launch until the advent of the commercial Internet in the early 1990s, there was no service for business customers that was close to comparable to it.13
Tymshare boomed, but it still felt like a start-up: small, convivial, and technically exciting. After finishing a seventy-hour workweek, staff would retreat to a steakhouse around the corner for beers and shop talk. Hardy had small children, but she stayed on, adapting her family life around the demands of work. Perhaps it was the suburban bedroom-community culture, or the overwhelmingly male demographics of the engineering profession, but it was rare to find a working mother like her in the Valley’s tech ranks, much less one who was a technical executive. The feminist revolution of the 1960s and 1970s was in full flame, but among Palo Alto’s playgrounds and cul-de-sacs it still felt like the Eisenhower era.
As the market grew, Tymshare expanded to become a leading performer in an industry coming into commercial maturity right as Wall Street entered several consecutive years of a bull market. In September 1970, Tymshare went public, outpacing its issue price in its first day of trading and creating a windfall for its founders and early employees. Ann Hardy, however, wasn’t one of them. “They didn’t give me stock options like they gave all the men,” she said matter-of-factly. No early retirement for her, and she didn’t want to stop working anyway. There were far more interesting things ahead.14
DEREGULATED
Although commercial time-sharing was taking off in the late 1960s, the future of the networked computer still depended on politics and policy. Across the Atlantic, Britain and France had kept telecoms government-run operations. In America, the networks were privately owned, but the government exercised a great deal of control through regulation designed to keep service universal and prices low. In return, telecommunication companies like AT&T and Western Union had the market to themselves. AT&T’s control of the telephone wires was one reason it was so difficult for companies like Tymshare to build out nationally, and it extended to control of the machines that connected into them. Clunky Bell telephone handsets, leased monthly from the phone company, became a near-universal feature in midcentury American homes.
In the 1960s and early 1970s, however, several things happened to ensure that America’s regulatory environment for computing would be very different from that of telephony. Little noticed at the time, these differences ultimately proved hugely consequential to the development of the tech economy to come.
First, there was the case of the Carterfone. Texan Thomas F. Carter was a man of the prairie and the oilfield, who in 1959 patented a device that hooked a regular phone line into a two-way radio, giving ranchers and oilfield workers a way to communicate across distances that a regular walkie-talkie couldn’t muster. It didn’t take very long for AT&T to pounce. Only Ma Bell’s equipment could be used on the Bell network, the telecommunications giant informed Carter. The Carterfone would have to go. Thomas Carter’s company may have been tiny, but his outrage was Texas-sized. He decided to fight. In 1965, Carter filed a private antitrust suit against all of the Goliaths of the industry: AT&T, its twenty-two regional subsidiaries, and the number two telephone service provider, General Telephone & Electronics (or GTE).
Three years later, in July 1968, the FCC ended the case with a decisive ruling: AT&T did not have the right to ban third-party equipment from being used on its lines. “We pinned their ears back,” Carter crowed. “We made fools of them.” It was, one Washington lawyer declared, “the most important communications case in the last decade.” The Carterfone decision helped accelerate the entry of new manufacturers into the phone market as well as forcing AT&T to develop more-appealing consumer products. A generation of American children would grow up with Mickey Mouse phones on their bedside tables. More important for the online future ahead, a market opened for a new set of high-tech entrants to build handsets, couplers, routers, and network infrastructure that allowed computer communication.15
A second David-vs.-Goliath battle pushed the door open further. This time the setting wasn’t Texas oilfields, but the pinstriped canyons of Wall Street.
If the New York Stock Exchange was the Cadillac of the American stock market, the over-the-counter, or OTC, market was its slightly dented Model T. These were companies too tiny for a listing on the big board or even regional exchanges, their buying and selling controlled since the 1930s by a self-regulating group for brokers called the National Securities Dealers (NASD). The market was opaque and archaic in its methods; its printed stock tables were highly unreliable, and buyers could usually only learn a stock’s price by calling a broker. As the market began to surge, the broker-dealers of the NASD realized they needed to move into the twentieth century, not to mention smarten up their slightly seedy reputation. Computerizing the whole system would be the quickest and splashiest way to do it.
In 1968, the dealers put out the job of “automating” their system for bid, and the firm that won was a four-year-old Southern California start-up named Bunker Ramo. The firm had defense industry roots: founded by Martin Marietta president George Bunker and TRW vice president Simon Ramo, the firm was dedicated to what the two founders termed “a national need in the application of electronics to information handling.” An early client was NASA, for which Bunker Ramo built one of the world’s first computerized information retrieval systems, using the networked computer to classify and categorize large data sets a la Vannevar Bush’s memex.16
At first, the system Bunker Ramo designed for the dealers was simply another digital database that put paper stock tables on line. But when the firm added a feature that allowed brokers to buy and sell over the network, AT&T again cried foul. This wasn’t time-sharing anymore, AT&T lawyers argued; it was two-way telecommunication, and Ma Bell would no longer lease Bunker Ramo its lines.17
Just like Thomas Carter, Bunker and Ramo pushed back. The complaint the firm filed with the FCC culminated in another landmark ruling by the commissioners, this time in 1971. Called “Computer I,” the FCC decision declared a compromise: a new “hybrid services” category of communication into which most two-way communication via computer would fall. It didn’t end the debate; “Computer I” would be joined by “Computer II” and “Computer III” before the 1970s were out. By the time the FCC got around to properly muddling through some rules for how to categorize communication by computer, the pace of technological change had galloped far ahead.
The FCC’s indecision essentially settled the question. Gone was the strong regulatory approach that had characterized the previous four decades of rulemaking and enabled the all-powerful Ma Bell universe. Instead, a range of commercial networked services inhabited the wires, all delivering more and more content and, over time, enabling e-mail and chat and all kinds of two-way communication: CompuServe, founded in Ohio in 1969; The Source, coming out of Virginia in 1979; Prodigy, a joint venture between CBS, IBM, and Sears, in 1984. LaRoy Tymes’s Tymnet never became a familiar brand name like these others, but the first true time-sharing network was the most durable of them all, running without a crash from 1971 to 2003. “He never had a bug in his network code, ever,” marveled Ann Hardy. “Amazing work.”18
Carterfone and Bunker Ramo, and the FCC rulings that came in their wake, meant that digital communications portals became the realm of many private companies, not just one or two. And, while hugely popular, none of these services became a publicly governed “computer utility,” a computerized analogue to AT&T. Instead, Congress and the FCC actively blocked AT&T from delivering online content and deregulated the entirety of the system, forcing Ma Bell’s breakup in the early 1980s.
In that vacuum, the only network powerful and widespread enough to eventually become the backbone for the world’s electronic commerce and communication was something quite different—a nonhierarchical, noncommercial, wild-and-woolly network also made possible by America’s race to space. It was called the ARPANET.19
INTERGALACTIC
By the time the time-sharing business began to take off, J. C. R. Licklider already had moved on to other, far bigger ideas. Nearly from the moment he landed at the Pentagon, Lick started to issue a flurry of memos that took time-sharing networks to the next level—proposing to interconnect time-shared computers to create a powerful national web of interactive supercomputing. With a flamboyant and intentionally humorous flourish that underscored the wildness of the idea, he called it the “Intergalactic Computer Network.”20
Academics embraced the concept, despite the wicked technical problem presented by trying to get different makes and models of mainframes to communicate with one another. Teams of computer scientists on different campuses were pursuing similar research questions, but they worked on entirely separate systems that couldn’t share data or findings. It was nearly as inefficient as waiting thirty-six hours to process your punch cards. Licklider’s eventual successor at ARPA, a pipe-smoking Texan named Bob Taylor, labored over the remainder of the 1960s to fix this problem and make a “Resource Sharing Computer Network” possible.21
The Pentagon had by then entered the age of Robert McNamara’s businesslike budget-cutting, and the brass proved receptive to Taylor’s argument that networked computing would eliminate inefficiencies and free up contractors to obtain the best kind of computer at the lowest price. A second, more apocalyptic argument for a national computer network came out of RAND, whose researchers pointed out that this decentralized network could serve as a military communication lifeline in case nuclear attack wiped out regular telephone service and the command centers of the Pentagon.
The Code Page 8