Turn Your Backs on the Origins of Computing!

 BOB TAYLOR

Bob Taylor’s job at the University of Utah, where he had gone after ARPA, had lasted only a year. The same disregard for rules and bureaucracy that had led him to barge into superiors’ offices at the Pentagon left him battering heads with university administrators.¹ They did not appreciate his attempts to sell time on the graphics group’s computer to students elsewhere in the university.^XII

So at about the same time that Mike Markkula began his job at Intel, Taylor had again packed up the family station wagon and continued west, going nearly as far as he could this time, almost all the way to the Pacific. Half an hour shy of the ocean, he stopped in Palo Alto, California, the site of what would be his professional home for the next thirteen years: Xerox’s Palo Alto Research Center. Nearly every breakthrough for which the research center, known as Xerox PARC, is famous—from the personal computer to networking to printing—can be traced in some way back to the Computer Science Laboratory that Taylor launched in 1970.

By the time PARC was two years old, the computer science lab was as unconventional as Taylor himself. The conference room, for example, was furnished with beanbag chairs. Taylor knew that most conference rooms at Xerox, which was headquartered in Rochester, New York, featured standard-issue rolling chairs and large wooden tables polished to a shine that reflected the clean-cut faces, neat collars, and conservative neckties of the company’s managers, salesmen, engineers, and executives. Taylor also knew that other Xerox offices did not have bicycles leaned up against the interior walls or the work-when-you-want policies that kept his lab busy through the night.

Taylor knew it, but he didn’t care. His suspicion of bureaucracy had grown to encompass many aspects of American business, which he believed was too concerned about appearing to do something rather than actually doing it.

So when someone in the lab suggested the beanbag chairs, other people liked the idea, and the chairs proved cheap, Taylor was happy to have his unconventional meeting room in PARC’s low-slung, glass-faced building in the Stanford Industrial Park, an easy walk from the spot where students had overturned a school bus during the anti-SRI protests three years earlier.² Now he had some twenty-odd corduroy-covered chairs in the colors of the 1970s rainbow: burnt orange, goldenrod yellow, avocado green, midnight blue.^XI

In the spring of 1972, a young reporter named Stewart Brand came out to write about the lab for a magazine article on the “youthful fervor and firm dis-Establishmentarianism of the freaks who design computer science.”³ The article would be published in Rolling Stone, established five years before with support from Arthur Rock, the venture capitalist behind Fairchild, Intel, Scientific Data Systems, and soon Apple. The magazine sent a young photographer named Annie Leibovitz to accompany Brand to PARC. She took many shots of the beanbags, along with candids of the researchers in their plush beards and sideburns measured in inches. Taylor loved Leibovitz’s portrait of him: clean-shaven, his hair carefully parted at one side and just grazing the tops of his ears, the bottom of his face almost obscured by the pluming smoke from the deep-bowled pipe in his left hand.⁴

Liebovitz also shot the mocked-up model of the computer that one researcher, Alan Kay, had been urging the group to build: a notebook-sized machine, mostly screen and keyboard, that would be easy enough for a child to use. Kay, handsome and luxuriously mustachioed, was a former professional musician who built a pipe organ in his home and would go on to win a Turing Award, the highest honor for a computer scientist. Kay worked in the systems science lab, but Taylor had recruited him to PARC, and he collaborated closely with Taylor’s group.

Kay’s idealized computer—he called it the Dynabook—was as different from the roughly 150,000 computers humming away in the world’s back offices, banks, and universities as the beanbag room was from its executive counterpart.⁵ In 1972, computers no longer needed to be room-size mainframe behemoths that cost millions of dollars and ran batches of punch-card programs. But the new minicomputers, the type of computer for which Sandy Kurtzig’s ASK was writing software for HP, could still fill multiple cabinets. Minicomputers ran the scoreboard at Red Sox games and the lights for Broadway’s A Chorus Line, but at $50,000 or more, the machines remained a specialized—and costly—business expense.⁶

Meanwhile, the notion of a computer small and easy enough for a child to use was preposterous. In 1969, the Neiman-Marcus Christmas catalog had offered a $10,600 Honeywell “kitchen computer” aimed at the homemaker whose “soufflés are supreme, her meal planning a challenge.” The catalog also featured a baby elephant, and perhaps the animal seemed no more foreign, intimidating, or useless than the computer. Not a single machine was sold, despite the apron and cookbook included in the purchase price.⁷

As things would turn out, even Taylor’s lab—which was stocked with so many bright computer scientists that the president of MIT blamed PARC for causing faculty shortages at the top universities—could not build anything like the notebook-sized Dynabook in 1972.⁸ What PARC built instead was a machine that would be recognizable today as a personal computer. It had a large monitor, a mouse, menus, a word processing program, and multiple windows. It could compose and edit documents and send them to a printer (also developed at PARC)—and the printout would look like the document laid out on the screen. The PARC personal computer could store files, documents, and images. It could connect through an Ethernet network, also developed at PARC, not only to printers but also to other computers to send emails and files.

Taylor named the revolutionary computer “Alto.” To him it was “a continuation of the work I was doing at ARPA,” the logical progression in the vision that he and Licklider had laid out years before, the vision that they had separately funded at ARPA: the notion that the computer could be a communication device connecting people and ideas.⁹

At ARPA, Taylor had funded the vision in two ways. He had supported efforts to develop time-sharing computers that were more responsive to their users than ever before, and he had launched the Arpanet, which connected computers to one another (and, by extension, connected the people using those computers to one another).

At Xerox PARC, Taylor’s team, working with the systems science lab, would weave those two strands together into a single Alto system that combined the world’s then easiest-to-use computers with a network that connected them to one another. It would take a decade for anyone to catch up to them.

Taylor first learned of Xerox PARC in the summer of 1970, shortly after Xerox’s chief scientist, Jack Goldman, convinced the company to launch a research facility supported by a $6 million annual operating budget and staffed with “the best in computer scientists; electrical engineers; systems analysts; operations researchers, mathematicians and statisticians; and biophysicists and biochemists.” The goal was nothing less than the invention of “the Xerox of the future.”¹⁰ The original name of the research center, possibly rejected for its unfortunate acronym, was the “Xerox Advanced Scientific and Systems Laboratory.”

Goldman proposed that Xerox locate the research center in a university town with an “intellectual night life” that would attract and benefit researchers. Within the center, he envisioned three laboratories to target basic sciences, systems science, and computer science. He also had a few specific ideas for future products that might emerge from the labs—“a machine which is half xerographic printer and half computing machine,” for example. But the essence of his plan could be distilled to seven words: hire great researchers and leave them alone. He told Xerox’s top management to expect nothing useful from PARC for at least five years.¹¹

Taylor arrived at PARC in typical damn-the-torpedoes fashion. His first visit came within months of the lab’s founding, when the physicist George Pake, the new director of PARC, invited him to Palo Alto. Pake was a former Stanford physics professor and provost at Washington University in St. Louis. A member of the National Academy of Sciences, Pake, an expert on magnetic resonance, would go on to receive the National Medal of Science. Pake’s goal was not to recruit Taylor to PARC but to see who Taylor thought should be recruited for the computer science lab. “That would depend on what the lab is going to do,” Taylor said. When Pake told him that the lab would support Scientific Data Systems, a computer company that Xerox had just bought for a breathtaking $918 million, Taylor informed him that Xerox had “bought the wrong computer company, an incompetent organization.” He went on to call the latest offering from Scientific Data Systems, a Sigma computer, “an abomination.” He told Pake that no one worthwhile would want to join a lab in support of such a second-rate company.^X

Taylor left his meeting with Pake expecting never to hear from him or PARC. Soon, however, Pake invited him for a second visit, this time to talk about a job. As the two men sat together in Pake’s PARC office, Pake told Taylor that he admired his ARPA connections and would like to have him at PARC’s Computer Science Laboratory—but not as its director. Taylor did not have a PhD, Pake reminded him, and a lab aspiring to world-class status needed a man with a doctorate at its head. Moreover, Pake believed that only someone who had conducted high-level research could manage an advanced lab.¹²

Taylor never got over the insult that Pake saw him as unqualified. Even after Pake hired him and let him hand-select his own PhD-bearing boss (Jerry Elkind, a former student of Licklider’s who had been a senior manager at Bolt Beranek and Newman, the company that built the routers for the Arpanet); even after Taylor had decided that he was happy for Elkind to handle the administrative side of the lab while Taylor led the research team; even after Pake supported Taylor’s bid to pay PARC computer scientists 15 to 20 percent more than scientists with comparable degrees at Xerox’s research lab in Webster, New York;¹³ even after Pake left the job as head of PARC only to be replaced by someone Taylor thought was worse—even forty years after Pake told Taylor he was not qualified to run PARC’s computer science lab, the sting of the initial insult burned.^IX

Despite his reservations about Pake, Taylor accepted the job of principal scientist at PARC with the understanding that he would be named assistant manager of the Computer Science Lab as soon as a proper manager was found.¹⁴ Taylor liked the climate and “aliveness” of the San Francisco Bay Area, which he had been visiting since his earliest days of supporting Doug Engelbart at SRI, and he liked Stanford, which he thought produced good “computerists.” The heart of the sprawling campus was only a mile from the PARC lab, the spire of the university’s Hoover Tower jutting high above any other structure in the area. Taylor moved his family into a large Craftsman-style house in Old Palo Alto, where the streets are named for Romantic poets. The neighborhood teemed with young families. Taylor had three sons.

The general mandate from Xerox corporate headquarters was to focus on “the architecture of information.”¹⁵ For Taylor, it was a matter of simple semantics to repackage the vision of a network of interactive computers to fit this mandate. Whatever the effort was called—the computer science lab’s focus on building “the office of the future” came later—Taylor believed that “it was time to get rid of centralized machines . . . and give everyone their own machine.”

“And I knew who to hire to do it,” Taylor adds. “I knew, personally, who the strongest young computer scientists in the country were.”¹⁶

Taylor, who lacked formal technical training, knew where he wanted to end up but not how to get there. Severo Ornstein, a member of the inner circle of PARC researchers that Taylor called his “graybeards” and a man who respects Taylor, calls him “a concert pianist without fingers.”¹⁷ Taylor could hear a faint melody in the distance, but he could not play it himself. He knew whether to move up or down the scale to approximate the sound, he could recognize when a note was wrong, but he needed someone else to make the music.

There have been many great technical visionaries whose ideas never reached full expression under the visionary’s guidance. In Taylor’s own day, indeed within a few miles of the Xerox PARC office, there were two. In the 1970s, Ted Nelson, who coined the word “hypertext,” wrote about a complex information architecture called Project Xanadu that never came to fruition, despite anticipating and in some ways exceeding the World Wide Web. Likewise, many of Douglas Engelbart’s ideas were not realized until they were refined at PARC, in the computer science and systems science labs.

Taylor was different. He could recruit to PARC an outstanding group of researchers—selected based on his belief that “a very good researcher was worth two dozen good researchers”—and keep them working together for years.¹⁸ He also had the support of his boss, Jerry Elkind, who handled much of the lab’s administrative work. As one researcher, who respected both men, put it, “Bob was my leader. Jerry was my manager.”¹⁹

Taylor’s work at ARPA had pointed him to the researchers he desired, and many of the young graduate students he knew from those days fondly remembered him—and the generous ARPA funding he offered. As a result, he could handpick his team.^VIII “All it took was a few phone calls from Taylor and the chosen researchers signed up,” recalled one Xerox executive.²⁰ Taylor began recruiting at the University of Utah. He swept in several employees from Bolt Beranek and Newman. At Harvard, he pursued the networking expert Bob Metcalfe, even after Harvard balked at giving Metcalfe a PhD.^VII

Closer to home, Taylor bypassed the young microchip and electronics operations to pursue clusters of talent with academic, not business, leanings. From Berkeley, he plucked a group that had worked together on an ARPA project at the University of California and then launched a commercial operation near the campus. The Berkeley Computer Corporation was about to go bankrupt, and many of its young computer scientists were excited to leave business concerns behind to join a research lab. As Butler Lampson, who would go on to gain recognition as one of the world’s greatest computer scientists, put it, “We started BCC because it was the only way we could think of to carry out this particular research program.”²¹ Joining Lampson were Chuck Thacker, who would, like Lampson and Kay, win a Turing Award. The team also included the expert system designer Peter Deutsch; the top design engineer Richard Shoup, who would go on to win an Emmy for his work in color graphics; the elite programmer Jim Mitchell; and Charles Simonyi, who is best known for writing Microsoft Word after he left PARC. Taylor would call the Berkeley contingent “the cadre of my computer science lab” at PARC.²²

Back near the PARC offices, Taylor paid a call to Engelbart’s lab at SRI and tapped Bill English, who had built the prototype of the first mouse and organized the technical production of the 1968 Mother of All Demos. English, in turn, recruited other members of Engelbart’s lab; fifteen would join PARC, most in the systems science lab. Taylor never considered wooing Engelbart because Taylor wanted what he called “hands-on engineers,” and Engelbart was anything but. “He was a visionary, and if there was anything this group didn’t need, that was another visionary,” Taylor says. He thought that Engelbart “could not explain what he wants.”²³ Engelbart, meanwhile, was not comfortable even visiting PARC. “I went over there a couple of times to visit but people were always showing me what they were doing and it was in a sense almost like ‘See this is the real way to go,’ ” he said.²⁴

Every one of his recruits, Taylor knew, could play a different, important role in his campaign to transform the computer from a glorified calculating device into a communication tool. He had hired hardware and software experts, engineers and computer scientists, specialists in programming languages, and authorities in human-machine interactions—all of them blessed with an agility of mind and network of contacts. “Be willing, where necessary, to turn your back on the origins of computing!” Taylor exhorted his twenty-person team.²⁵

The recruits had one other trait in common: Taylor’s vision was theirs, as well. Reminiscing with Butler Lampson, Alan Kay recalled, “We already came in as born-again interactive computerists. Taylor was pretty vigilant about not inviting people who weren’t really into that thing.” (“Didn’t sign up for that, right,” Lampson agreed.)²⁶

Taylor was a draw, but PARC also offered high salaries and excellent resources. PARC researchers could easily interact with others at SRI and Stanford, particularly the Stanford Artificial Intelligence Laboratory (SAIL). They could attend or give talks on campus, or meet with people visiting one of the other institutions. Several of the researchers from Taylor’s lab taught classes at Stanford. A job at PARC offered the intellectual stimulation of an academic career without teaching or publication demands.^VI

There was only one inviolable administrative requirement in Taylor’s lab. Once each week, everyone had to be in the beanbag room for a meeting. Taylor did not care when his researchers got to work or what they wore while there. It did not matter when they went to lunch or whether they shaved. But he wanted them in the beanbag room every Tuesday—and he expected them to stay for hours. The meetings served as the intellectual pulse of the lab and a way for him to keep the team moving in the same direction. The gatherings also reveal the inner workings of what both Thacker and Lampson have described as Taylor’s “magical” leadership.

Taylor opened the sessions with administrative issues such as available billets and announcements of upcoming visitors. He then moved on to solicit project reports and ask for social announcements. Many members of the group, young and single, biked together in the hills or grabbed lunch at the Alpine Inn (fondly called “Zotts”) in the rural town of Portola Valley, a few miles from PARC. Researchers shared long weekend hikes among the coastal trees in the misty coolness above the valley. They barbecued at one another’s houses and worked on one another’s basement workshop projects. Taylor, a competitive tennis player, enjoyed challenging the best players at PARC to hard-core, sweat-streaming matches on the court behind his neighbor’s home. Afterward, the players would adjourn to Taylor’s house. “The Dr Pepper was cold, the doors were open, a breeze was blowing through, and everyone was always welcome to come in,” recalls Bob Metcalfe.²⁷

During the weekly lab meetings, Taylor also spent time fielding complaints. The cafeteria food was bad. Equipment was inadequate. Secretaries were overworked. Surprisingly often, grumblings about balky Xerox copiers led the list of problems.

Taylor listened and took careful notes. He knew that truly creative computer scientists tend to be opinionated individualists. He catered to that tendency. When the group was preparing to move to a new building, he polled every researcher to determine the ideal phone system. All twenty researchers wanted Touch-Tone phones rather than the old-fashioned rotary dial, but beyond that, there was little consistency. One did not want any secretarial backup, but nine wanted the so-called Alan Kay switch that transferred incoming calls immediately to an attendant without interrupting the intended recipient. Five shunned the message waiting light, while seven wanted an intercom line.²⁸ Customized phones may seem trivial, but they are an example of a feeling that Taylor created at PARC—the sense that he was working for the researchers, rather than the other way around.

Every week, once the social and administrative housekeeping were out of the way, Taylor brought the group to the heart of the meeting. “A time for half-baked ideas,” he called the sessions. A researcher would present work in front of a blackboard and set all the rules of engagement. Could he be interrupted with questions?^V Were people allowed to interject at all? Would he talk for an hour and invite discussion for two more? Or would he toss out only a few ideas before opening the floor? Taylor called the featured presenter the “dealer” because the speaker set the terms much as a dealer sets out the rules of a poker game. Soon enough, the meetings themselves came to be called Dealers.

“Dealer” also had a more antagonistic connotation. Taylor had read the popular 1962 how-to guide on card counting, Edward O. Thorp’s Beat the Dealer. He hoped that meeting participants would use their expertise to poke holes in a presenter’s work in the same way that card counters used continuous computations to get the edge on a blackjack dealer. Nearly every presentation was interrupted by mutters or even shouts of disagreement from people sprawled in the soft chairs on the floor. Taylor believed that the confrontations, which he tried (not always successfully) to keep focused on the work, not the presenter, yielded better results.²⁹ In a phrase that could serve as a personal motto, Taylor once said that “controversy is healthy and should not be inhibited.”³⁰

The dealer sessions owed much to the principal investigator meetings that Taylor had directed when he was at ARPA. Then, as now, he wanted the researchers to ask one another questions he might not know how to ask himself. Then, as now, he believed that tough questions would help clarify thinking and illuminate each participant’s viewpoint and particular expertise.

Throughout his time at PARC, as at the Dealer meetings, Taylor framed the conversation in the lab and then let the technical experts take over, interceding only to correct course. “The theme that assembled [the team] was Bob’s notion of interactive computing in the service of people,” wrote Bob Sproull in 1977, when he left the lab for an academic job at Carnegie Mellon. “Whenever the theme fades among the clamors of daily activities and crises, Bob works to raise all our sights out of the details.”³¹ That, too, had been Taylor’s approach when launching the Arpanet: set the goal, hire the right technical people, and then step in only when necessary.

Allowing the research team to define its own path to the goal did not mean that Taylor abdicated control. He cut off discussions if he felt they had reached the end of their usefulness.³² When he thought the quality of Dealer presentations was slipping, he threatened to move the weekly meetings to a monthly schedule, drawing the objections of his staff.³³ He mediated disputes between researchers, insisting again and again that each person needed to be able to explain his antagonist’s point of view to that person’s satisfaction, even if he did not agree with it.^IV He was also willing to make the least popular decisions in the lab: shutting down research efforts (advanced work on color graphics, for example) that he considered tangential.

At the same time, he deferred to his staff’s technical opinions, particularly those of Butler Lampson. Twenty-nine years old when he came to PARC, Lampson was wire thin and pulsed with energy. Brilliant and kinetic, he had once been scheduled to speak at a computer science conference, and when the previous speaker had gone on too long, he had delivered his entire hourlong talk in a whirlwind but cogent fifteen minutes before rushing through the door to catch his flight.³⁴

Lampson was also as hardheaded as Taylor. He refused to write abstracts for his long memos at PARC; if there were a way to summarize what he needed people to know, he said, he would have written a shorter memo.³⁵ He once declared he would rather sit on the floor of his office than have ugly furniture in a new lab Xerox was building.³⁶

“Taylor listened to him as though God was speaking,” says Severo Ornstein. “It would have been terrible if Butler had had any crazy ideas because they would have been followed. But Butler was pretty much right about things.”³⁷

Several of Taylor’s managerial innovations were radical. The lab had no formal organizational hierarchy, aside from everyone reporting to Taylor. For any new hire, nearly every employee had to raise a hand indicating that he—or she; a few female researchers joined over time—would be “deeply disappointed” if the person did not come to PARC. The group internalized Taylor’s own standards. One unstated criterion for hiring, explains Chuck Thacker, was for “every person we hired to increase the average IQ of the group.”³⁸ The need for near unanimity could at times encourage what Butler Lampson called “groupthink” and lead the lab to miss important hires. (Alvy Ray Smith, for example, who went on to cofound Pixar, worked briefly at PARC but was not offered a permanent job.) But the people who joined the staff at PARC did so with the support of the entire lab.

If Taylor’s team collectively felt something was a good idea, he saw it as his job to push that idea as far as he could, to “give it a lot of air cover,” according to Chuck Geschke, who would go on from PARC to cofound the software giant Adobe. Taylor cared less about his own political position within the wider Xerox hierarchy. In the first months of Taylor’s tenure, his research team’s most urgent need was for a computer to design and test their ideas about the machine they wanted to build. The lab computer would serve as the digital equivalent of a chemistry lab’s collection of beakers, chemicals, and spectrometers: the basic equipment that made research possible. Xerox corporate wanted the lab to use a machine built by the recently acquired Scientific Data Systems. Taylor’s researchers thought the computer—the one Taylor had called “an abomination” in his first interview at PARC—was not capable of supporting their work. They wanted a PDP-10 machine made by Xerox’s competitor DEC.

Taylor backed their request, taking it to Pake, who passed it on to corporate. Eventually a Xerox vice president flew out from the company’s headquarters to see if, after listening to the cases presented by the researchers, he could justify the purchase of a million-dollar PDP-10 system that competed with Xerox’s own product.³⁹

Unpersuaded, he denied the lab’s request for a DEC machine. When the news reached Taylor’s lab, several researchers threatened to leave, rather than use the Scientific Data Systems machine. Taylor stood behind them. Pake brokered a compromise: Taylor’s lab could not buy a DEC PDP-10, but they could build a clone.

Designing and assembling the cloned computer, which Taylor’s researchers named MAXC in mocking homage to Scientific Data Systems founder Max Palevsky, proved an extraordinary team-building exercise. Constructing a computer in 1970 was not a simple matter of welding or screwing together some kit parts and loading in an operating system and a few software programs. Components had to be designed and built. Code had to be written and programmed. People in the lab had to agree on the type of memory to use (they ended up going with a new semiconductor memory, Intel’s 1103, whose marketing the newly hired Mike Markkula was helping with). PARC employees had to review proposals from suppliers. In wrestling with those decisions, people who had never worked together learned how to do so, and a disparate group of what Taylor calls “soloists” became an orchestra. The group also developed a network of trusted local suppliers and fabricators.

But from the corporation’s vantage point, the decision to allow the researchers to build their own machine was a terrible capitulation; Xerox employees were permitted to spend months and hundreds of thousands of dollars copying a competitor’s computer instead of buying a Xerox product they could use immediately. Xerox also lost an opportunity to have an exceptionally qualified group of computer scientists develop software for the Scientific Data Systems machines.

Taylor and his research team say that they could not have done the work they wanted to do on the Scientific Data Systems (soon to be renamed Xerox Data Systems) computer. True.^III But many Xerox corporate executives thought that point was irrelevant.⁴⁰ What Taylor’s researchers wanted to do to further computing progress did not matter. What mattered was what they could do to make money for Xerox. They should use a Xerox computer and build products for it.

The corporate executive who had visited PARC as part of the computer-purchasing debacle later noted, “This small incident set the tone for everything that would be coming in the future. PARC was allowed to operate in complete isolation and to build technologies that were designed to demonstrate the brilliance of PARC researchers without any consider-ation of how this could possibly relate to the future of Xerox.” Taylor’s team, he says, were “people filled with messiah [sic] zeal. Their loyalty was never to Xerox but to the intellectual challenge of taking information technology research out of the stranglehold of mainframe computing.”⁴¹

Taylor did indeed want to break “the stranglehold of mainframe computing,” but he, and nearly everyone in his lab, also felt loyal to Xerox. Again and again over the course of the next decade, Taylor would try to interest the company’s leadership in the technology coming out of the computing labs at PARC. His efforts would meet with little success.

In early 1972, the lab was at a critical point. The MAXC computer was almost complete. Soon Bob Metcalfe would connect it to the Arpanet, the computer network that Taylor had kick-started in 1966 when he was at ARPA. Researchers were beginning to complain that the final-stage refinements were “depressing” and too far removed from basic research.⁴²

Around this time, Jerry Elkind, Taylor’s PhD-bearing boss, asked whether Xerox should “buy the ARPANET.” ARPA was looking for an outside institution to run the network as a public service akin to the national phone network.⁴³

After some discussion with Taylor, Elkind, along with Xerox’s head of research, Jack Goldman, put together a group “to analyze the opportunity for buying ARPANET and to recommend the action Xerox should take.”⁴⁴ In the end, Xerox, like AT&T (which had also considered a purchase), declined to bid for ownership of the Arpanet, and in 1975, the Defense Communications Agency took over operational responsibility.

What would have happened if Xerox or AT&T had bought the Arpanet in the early 1970s? The network almost certainly would have been much less independent and freewheeling than it—and the Internet that followed in its wake—later proved to be. A corporate owner would likely have exerted more control over who could join and might also have adopted a more rigid definition of acceptable behavior.⁴⁵

While Xerox contemplated buying the Arpanet, Taylor’s researchers were clamoring for a new project. He needed to figure out what would come next. A talk by Lampson at a February 1972 Dealer meeting provided a spark. “We should turn our attention to building a simple machine. They are not expensive, and they are not hard to build, given what we have now,” Lampson said. “Simple machines will give us most of what we want,” and if a number of them were networked together, “almost all jobs that people will be wanting to do in the next ten years could be taken care of.”⁴⁶

At the next week’s Dealer meeting, Lampson and Alan Kay, who saw Lampson’s proposed computer as an “interim Dynabook”—a first step toward the portable kid-friendly machine he imagined—gave a more detailed talk about “Alan Kay and Butler’s $500 Machine.” The two men laid out a few technical thoughts and invited interested people to a meeting to be held several days later.⁴⁷

Taylor was thrilled. He had wanted to build a small, easy-to-use machine when the lab had launched eighteen months earlier. At the time, he could not describe his ideas about what would come to be called a personal computer in a way that Lampson or Chuck Thacker, whom he also approached, thought reasonable.⁴⁸ “He was waving his arms, talking about interactive mumbo-jumbo,” recalls Lampson. “We interpreted this to mean he was describing something completely impractical.” Researchers in other places, many of them supported by ARPA funds, had built hugely expensive one-off prototypes of personalized interactive computers. That was not the agenda for PARC. From the beginning the lab aimed to build machines that could be used by large groups of real people in real environments. Lampson was interested only in building systems for one hundred or more users.

By the end of 1972, a team headed by Lampson, Kay, and Thacker was planning an entire system built around “ten to thirty” Alto computers. In December, Lampson circulated a seminal memo titled “Why Alto?” That memo moved beyond Taylor and Licklider’s vision to a blueprint for reality, laying out hardware and software specifications for a machine that would fit under a desk and use a graphical display, a keyboard, and a “mouse or other pointing device.” The memo also placed the machine within a network that included other computers as well as a printer.⁴⁹ It was not enough for an individual to possess significant computing power in a dedicated, easy-to-use, affordable machine, Lampson’s memo made clear.^II

“If our theories about the utility of cheap, powerful personal computers are correct, we should be able to demonstrate them convincingly on Alto,” Lampson wrote, using the word to refer not only to the computer but also to what he was now calling “an Alto system.” He continued, “If [the ideas] are wrong, we can find out why.”⁵⁰

Over the next few months, the computer science lab and Alan Kay’s group in the systems science lab worked together at a frenzied pace to build the computer described in Lampson’s memo and in a companion hardware note from Chuck Thacker. Larry Tesler, who worked in Kay’s group, recalls tag-teaming with Tim Mott on the graphical user interface. They shared a computer, one coding all night, the other coding all day, with an hour of overlap scheduled at the shift change so they could update each other.⁵¹ “It had the intensity of a startup,” says Chuck Thacker. “I feel very sad that I didn’t get to know my first daughter except at the two o’clock [a.m.] feeding.”⁵²

The group was grappling toward their shared vision of what it meant, in silicon and wire, in operating systems and microcode and routers and zeroes and ones, to have a truly interactive, networked machine. The memos they produced reveal the shared sense of mission and intensity: “These problems seem worth solving.” “Here’s my take.” “Comments welcome.” “This is just a draft.” The subject lines of two memos are noteworthy: “Vented Frustrations” and “What Am I Doing Here.”⁵³

Discipline was tight. Every memo was stamped “Read and understood,” signed, and dated, in accordance with standard protocol in scientific labs around the world.

Taylor remained a warm and welcoming presence in the midst of the excitement. “You could go into Bob’s office with a problem, any problem,” recalls Larry Tesler, “and he would shut the door behind you and help you.”⁵⁴ Bob Sproull wrote, “It was Bob who always had more faith in me than I had, and who nourished every shoot of confidence I demonstrated.”⁵⁵

Tesler recalls that Taylor would occasionally hand him one of the cartoons that illustrated “The Computer as a Communication Device” article Taylor had coauthored with Licklider in 1968: “He’d say, ‘This hasn’t happened yet.’ ” But Taylor emphasizes that he did not lay out a comprehensive vision for the Alto in advance. “As we finished one piece, it would be clear, I thought, what would be next pieces,” he says. Since interactive computing mandated that a computer be optimized for communication above calculation, the Alto was built around a display rather than an arithmetic unit. The focus on the display required a new type of operating system. A new OS required new applications, and since one goal was communication, a word processor was an obvious program. And so it went. “As you posited one element, the next would become apparent,” Taylor says. “You knew you wanted to send characters over a network (by the way, you need a network), to a printer (so now you need a printer). And these characters you were sending, you might want to save them, so you needed a file system. Another thing you could do with these characters: you could compose them into an email, so you needed an email system. These were all separate components in a way, but in another way, they were part of a system. One of the things I did was guide this progression.”^56I

Step by step and piece by piece, Taylor and his team were laying the groundwork for the modern personal computer industry.

---

I. Both Thacker and Lampson agreed with Taylor’s characterization of his work, though Thacker amended it to “help guide this progression.”

II. This systems insight, that the network was as important as the computers in it, recalls Taylor’s original vision of the Arpanet as a means of connecting remote computers. As Taylor told Chuck Geschke, “The Arpanet had opened the door to communications, but there were a lot of things that were undone if you really wanted a computer system that was interactive and focused on communication. So we set about doing that in PARC.”

III. The group that came from Berkeley Computer Corporation had intimate knowledge of the SDS computer line, since before coming to PARC they had modified an SDS computer to serve as a time-sharing machine. The first customer for the modified SDS940 computer was Tymshare, the company for which Sandy Kurtzig wrote MANMAN.

IV. Taylor distinguished between what he called “Class 1 and Class 2 disputes.” In the first, the two sides are so estranged that they cannot even hear, much less understand, what the other is saying. In Class 2 disputes, the two sides disagree but understand each other. Taylor’s goal was to move all Class 1 disagreements to Class 2, even if resolution was not possible.

V. Later, women would be hired into the computer science lab, but early on, the researchers were all men.

VI. The first publication out of the lab came three years after its founding—and it was a doctoral thesis. Another option for researchers, to pursue ARPA-funded research, required them to explain how their work supported the country’s defense and to accept money from a Department of Defense still engaged in the unpopular Vietnam War.

VII. Metcalfe would resubmit his dissertation and be granted the degree while at PARC.

VIII. One early hire that Taylor wanted but did not get—Larry Tesler—fell through because Taylor was not directly consulted in negotiating his offer. Tesler later joined PARC’s Systems Science Laboratory.

IX. Taylor mentioned his dislike of Pake in four separate interviews.

X. One possible source of Taylor’s animosity: years earlier, Scientific Data System’s founder, Max Palevsky, had told Taylor that the time-sharing efforts Taylor was supporting with millions of ARPA dollars were bound to fail—a comment that had inspired the irate Taylor to throw Palevsky out of his office. “It was as though an idiot comes into your house and starts telling you what’s wrong with the way you’re living. You know, after a point you can’t tolerate it any longer,” Taylor says. His reservations about SDS, which were shared by Chief Scientist Goldman, proved reasonable. In 1975, Xerox would write off the Scientific Data Systems purchase for $484 million.

XI. By the end of 1972, the computer science lab had a staff of thirty-one, most of whom were conducting research, with a support staff of roughly a half dozen.

XII. Taylor was trying both to earn a bit more revenue for the graphics group (officially the Information Research Lab in the College of Engineering) and to find a way to make the university’s machines and computer expertise available to students throughout the university.