The Rabbits Hopped Away

 BOB TAYLOR

In May 1983, Bob Taylor received an email from a panicked employee working on the Xerox STAR, the pricey computer system that Xerox had brought to market instead of the Alto.^III

“About a dozen of us went to a lecture/demo of the Apple Lisa at UCLA this afternoon, and WE’RE SCARED TO DEATH!” the employee, Bruce Hamilton, wrote. The Lisa, introduced four months earlier, had a few problems—applications started slowly, and email and networking were allegedly “a few months away”—but these, in Hamilton’s opinion, were of little consequence when stacked against Lisa’s advantages. The Apple machine incorporated the mouse, a graphical user interface, and the interactivity of the Alto and Star. Moreover, Apple was allowing outside programmers to write software for the Lisa. As a result, “There is NO SIGNIFICANT STAR FEATURE WHICH WILL NOT HAVE A LISA COUNTERPART WITHIN 12–18 MONTHS,” Hamilton predicted. “It is a MYTH to think Xerox can identify some ‘safe market sector’ where we can maintain our expertise.” He wrote with alarm about any “dirt ball . . . company of two guys in a garage” that will be able to create and release new products in less time than Xerox needed to write specifications for a new feature.¹

Hamilton’s fears, which he broadcast to two Xerox email lists, were exaggerated but not unfounded. It had taken Xerox more than seven years to move from the first working networked Alto system to a product. Apple had made essentially the same move in half the time. The scenario was what Larry Tesler, the PARC researcher who had left for Apple in 1980, had told Taylor would unfold. Others were refining, improvising on, and popularizing PARC’s innovations . . . and then moving beyond them.

One of Taylor’s employees had forwarded Hamilton’s email with the subject line “For your amusement,” but Taylor was not amused. His team had proved what was possible years earlier. Someone else at Xerox—not Apple—should have determined, in Taylor’s words, “what will work in a constrained consumer environment.” In general, Taylor liked a clean division between research and production. “You do not want a researcher building your roads, your subways, your airlines, your houses,” he wrote.² The Lisa rankled nonetheless. Xerox should have owned personal computing by now, not ceded it to unpedigreed corporate upstarts. Moreover, the success of the IBM PC and its clones had stunned Taylor and many PARC researchers. “We never thought people would buy crap,” Alan Kay explained.³

The past few years—and the last months in particular—had been tough for Taylor. His marriage was falling apart. He would soon leave the Craftsman house in Palo Alto with the neighborhood tennis court and breezes blowing through, to move to an isolated house in Woodside at the end of a winding road overlooking the valley. At work, problems that had been brewing for a long time were now boiling over. He continued to battle with other PARC divisions and PARC administrators. Key employees were frustrated and leaving. Budget freezes had become cuts. At his last performance review, he learned that his annual raise would be less than the cost-of-living increase generally given as a matter of course at Xerox. At $118,440, his annual salary was at the top of his pay grade.⁴ He could not expect more.

Through all the troubles, though, he had found an unlikely ally: his immediate boss for the past five years, an engineer named Bob Spinrad. Spinrad was named director of PARC after George Pake had been promoted to head of corporate research. Spinrad had worked in Xerox corporate, beginning his career at the company with Scientific Data Systems, the computer firm that Xerox’s headquarters had acquired in 1970 and shut down $84 million later.

Spinrad understood Taylor. He wrote of Taylor’s “sharply bimodal” performance: the outstanding leader within his own lab, who was at the same time a source of much of the interlab tension at PARC. One day, Taylor, feeling particularly unappreciated, used a ruler and graph paper to chart his lab’s resources. According to his calculations, the amount of money his lab received per employee had plummeted from $37,000 to $2,000 during his tenure. He sent the graphs to Spinrad with a cover note that asked, “Is this message one of appreciation?” The tone was not challenging but plaintive, a request for help in prying more money out of Pake.⁵ Spinrad appreciated the importance of the computer science lab and its work. “It is a source of distress to me that we cannot adequately support all the excellent and varied work that the CSL staff is capable of,” he wrote in early 1981. Around the same time, he sent a five-year plan to Pake at headquarters proposing that the PARC budget be reallocated to send more money to the computing labs.⁶

Shortly after Spinrad submitted his five-year plan, Pake, in effect, demoted him, splitting PARC in two and making Spinrad head of only the computing side. Rumors in Taylor’s lab held that Pake had removed Spinrad from the directorship out of fear that Spinrad was falling under Taylor’s sway. Spinrad left PARC for a job on the corporate staff the next year.

With Spinrad gone, the buffer between Taylor and Pake had disappeared. A new one appeared in the form of the physicist Bill Spencer. A tall, handsome Kansan, Spencer had come to Xerox in 1981 to run a new integrated circuits lab at PARC. He and Pake got along well. After several months, Pake asked Spencer to serve as PARC’s director.

Spencer had a willful streak that rivaled Taylor’s own. At Sandia Laboratories, where he had worked before Xerox, he had enjoyed provoking a boss he detested. Forty years later, he chuckled as he recalled a time when he had made the man “so angry that we had to cancel a meeting.”⁷ At another job, he received a gift of a giant hypodermic needle—because he was always “needling” others.⁸

Spencer was so similar to Taylor that Taylor would either enjoy or despise him. He did both. For the first months after Spencer arrived, when he was running the integrated circuits lab, the men played tennis on Saturday mornings. Their wives were friendly, and the Taylors attended the fiftieth birthday celebration for Spencer’s wife. All in all, the two families shared “a lot of beer and Dr Pepper nights,” according to Spencer.⁹

But after Pake promoted Spencer to head PARC, making him Taylor’s boss, the friendship faded. Taylor was soon pointing out that Spencer’s undergraduate degree was in physical education. It did not help that Spencer was a physicist with expertise in semiconductor electronics. Taylor’s disdain for physicists was well known at PARC, as were his feelings about semiconductors, which he thought made for boring research, since Moore’s Law had already mapped the expected rate at which the density of components on a chip would increase. He also thought microchips received too much credit for the digital revolution. For Taylor, semiconductors were a necessary but pedestrian base upon which the elegant and complex dance of software could perform. “Without software, hardware is just a piece of hot iron,” he liked to say.¹⁰

Spencer’s first move as head of PARC was to schedule a meeting of the lab leaders. Taylor offered his beanbag room for the meeting. Spencer was shocked when a representative from another lab said that none of the researchers from her lab would attend a meeting held in Taylor’s territory. “They were afraid of being physically attacked,” Spencer says, though he acknowledges that he knew of no physical violence at PARC. “It was worse than I had ever seen in my twenty-five years of being in places like Bell Labs and Xerox.”¹¹ The meeting never happened.

Spencer soon decided that Taylor was standing in the way of interlab cooperation at PARC. Moreover, thanks to the clean division between research and development that Taylor championed, he was impeding Xerox’s ability to commercialize PARC technology. “Getting technology picked up by the rest of Xerox was my problem,” Spencer recalls. “That requires that people who are experts in [the technology] have to spend time with people in other parts of Xerox. [Taylor] was totally unwilling to participate in that at all.”¹²

As Taylor saw it, a number of his researchers were already moving technology into other parts of Xerox. People from his lab had worked on the Star, and he had encouraged one key researcher (Geschke) to launch a new imaging lab within Xerox. But Spencer had a different model in mind: he envisioned Xerox researchers helping an outside company develop products based on PARC technology in exchange for an ownership stake for Xerox. Spencer would later successfully promote this model with Spectra Diode Laboratories and Synoptics.¹³

Taylor thought that Spencer’s real objective was neither interlab cooperation nor product development. His true goal, in Taylor’s opinion, was to seize control of Taylor’s lab.

Looking back four decades later, Spencer says with a sigh, “Had I not been so anxious to leave [Sandia], I probably would have done a better job looking into PARC—and gone to Hewlett-Packard.”¹⁴

Taylor’s defensiveness and Spencer’s frustration peaked at the end of August 1983, when Spencer handed Taylor a typed memo after a heated meeting in Spencer’s office. “Your management and your personal attitude are counterproductive,” Spencer wrote. If things did not change, he said, Taylor would face “disciplinary action that may include your termination.”¹⁵

Spencer gave Taylor three weeks to reorganize the computer science lab to look like the other labs at PARC. No longer would everyone report to Taylor. Instead, Taylor should divide the lab into several groups, each with a dedicated manager. Spencer felt that the move would “help develop future Xerox managers and to improve our contacts with other Xerox organizations.” He directed Taylor “to make no derogatory remarks about Xerox programs or organizations” and “not to discuss the formation of an outside organization with Xerox employees . . . [or] induce them to leave Xerox.” Finally, Taylor needed to meet with Spencer every Monday morning at 9:00 “to go over your compliance and progress on these actions.”^II

No one, including the conflict-averse Pake, had ever treated Taylor so imperiously. Not even Jerry Elkind, the nominal manager of the computer science lab in the early years, had ever told Taylor how to run his lab.¹⁶ Spencer had designed his letter to put Taylor in his place.

“As I read and read again your memo,” Taylor wrote to Spencer six days later, “I was increasingly appalled and shocked by its approach, its tone, and its content. . . . When I began to recover from [the memo’s] effect, my reaction was simply to resign immediately. However, I have worked for a long time, and over one of the most productive periods of my life, to bring the very best people and technology to Xerox. I owe to this investment much more than just walking away from it without responding to your dismaying memo in more detail.”¹⁷ In the following five single-spaced pages, Taylor defended his lab and his management.

He dismissed the allegation that he had encouraged people to leave Xerox or to join him in a new venture. (“It is clear from the tone of your memo that I should have been interviewing for employment elsewhere, but I have not,” he said.) He wrote that Spencer had always been welcome at lab meetings and disagreed that reorganizing the lab would foster new managers. PARC researchers were hired for their research, not their management skills and interests, Taylor noted. Taylor also offered a list of ten technologies that the lab had made “work on a daily basis for ourselves and for thousands of others inside Xerox.” His list included local area networks, an internetwork serving more than 4,400 electronic mailboxes, integrated circuit design aids, and the laser printer.^I Any responsibility for the failure of these and others of his lab’s accomplishments to reach the general public lay beyond Taylor’s responsibilities.

As for derogatory comments: “I have a fundamentally positive, optimistic spirit; I am not interested in being derogatory,” Taylor wrote. He admitted that he could be critical, but “truth and open expression should be the coin of the realm” in a research environment. PARC, by contrast, had become “more politicized than I would ever imagined was possible.”

In closing, he requested that he be allowed to share Spencer’s memo and his own response with his senior researchers (denied), as well as with Xerox CEO David Kearns and other top executives (permitted). In a note to Kearns and two senior executives, Taylor wrote, “It is absolutely clear that I can no longer report to Spencer and, through him, to Pake.” Taylor did not know it, but Kearns was well aware of the content of Spencer’s original memo. Before confronting Taylor, Spencer had discussed its major points with Kearns, warning him that Taylor might quit in response.¹⁸

After receiving Taylor’s note, Kearns deputized his second-in-command to meet with Taylor and Spencer at Xerox’s East Coast headquarters. After four hours, Taylor understood that Xerox senior management was backing Spencer. The antagonists shared a silent ride back to the airport in a Xerox corporate car.¹⁹

When he returned to California, Taylor knew what he had to do. He briefed a few members of his inner circle and then called a lab meeting for the next morning.

Standing in front of his team, with Bill Spencer tall at the back of the room, Taylor read a short message that he had handwritten in tiny looping letters on a page torn from his Xerox notepad:

Thirteen years and 19 days ago, I formed a laboratory with the hope of creating an entirely new form of Information processing that would someday serve millions of people. You have created that new form, and I will always be indebted to you. . . .

By the mid 70s, you had created an excellent base upon which I expected corporate research to build. That expectation has not been fulfilled. PARC continued to invest less than 20 percent of its resources in computer systems research . . .

So I will leave and have no immediate plans. [I will be on] vacation until 30 Sept. I do not know what I will be doing next.

It has been an incredibly great experience to work for you.²⁰

Stunned silence followed, as the researchers watched Taylor lower his notes and walk away. Bill Spencer stepped into the space where Taylor had stood, ready for questions. What he got instead was fury. With the exception of the few lead researchers whom Taylor had briefed the night before, Taylor’s firing—anyone who knew him knew he would never voluntarily resign this way—came out of nowhere. The researchers began shouting at Spencer. The passion of the best Dealer meetings turned black. What had just happened? What had Spencer done to Taylor? Was the lab being dissolved?

Chuck Thacker stood up. He said he was quitting and left the room. Later that day, he submitted his formal resignation, telling Spencer that he doubted “your motives as well as your wisdom” and explaining that he did not wish to “witness or participate in” the dissolution of the lab that would inevitably follow Taylor’s departure.²¹

Spencer, meanwhile, was dealing with his own shock at how Taylor had handled the situation. Though he says “my intent was never to force Bob to leave PARC,” deep down, he did not think it would be the worst thing for Xerox if the computer science lab got a new manager and even a new core research staff. “I think they had made almost all the contributions they were going to make,” he later reflected about the computer science lab under Taylor. “I think their accomplishments probably came between 1971/72 and 1980, and then they were kind of fine-tuning the things they had done.”²²

But Spencer had not expected Taylor to play his hand the way he had. “I had not thought he would leave abruptly. I thought we’d sit down and talk, and if we couldn’t reach a compromise, we’d have to say, ‘Okay. Now what?’ ”²³ That misapprehension, more than anything, shows how little Spencer understood Taylor. Compromise was not Bob Taylor’s style.

A rumor began to circulate through the computer science lab’s email lists that Spencer had been ordered to shut down the lab. Within a few days, copies of Spencer’s memo and Taylor’s response had been posted electronically (by whom remains a mystery). Soon both posts were quietly taken down.

Several of the most senior computer scientists flew to Xerox headquarters to urge CEO David Kearns to keep Taylor at Xerox.²⁴ They wanted Kearns to understand that if Taylor left, others would leave, too. Moreover, even if the lab somehow remained intact—and it would not, they assured Kearns—Spencer would never be able to run it. It was vital to understand that Taylor “works for the laboratory, rather than vice versa,” they said. Moreover, by “deflecting external distractions,” Taylor freed the researchers to focus on research at the highest levels.

Spencer, by contrast, possessed a different—and unsuitable—management style, the researchers told Kearns. “He has preconceived ideas about technology transfer, does not understand systems research, does not take account of CSL’s demonstrated success, [and] commits the lab to research without consultation.”

Taylor, they promised, “is manageable,” and they, as a team, had found ways to “compensate” for his “imperfections.” The group’s suggestion for salvaging the situation was to induce Taylor to return, have him report to someone other than Spencer, and increase the funding of the computer science lab.²⁵

It is difficult to imagine that the researchers thought their appeal would have any real effect. To bring Taylor back to PARC with more resources and more power would have exacerbated the problems perceived by Xerox management.

While the group was making its case to Kearns, other members of the lab contacted the world’s leading computer scientists to ask them to write to the CEO and pressure him not to let Taylor leave. Within days, the letters were on Kearns’s desk.

“A mind-boggling event,” Donald E. Knuth, the distinguished computer scientist at Stanford, called it. “Bob has accomplished such miraculous things. I had always assumed he was considered one of Xerox’s most important people.” With Taylor gone, Knuth wrote, the world’s finest assembly of computer scientists would likely dissolve, which would “spell the end of Xerox’s leadership in computer systems.”²⁶

Similar letters arrived from Dana S. Scott at Carnegie Mellon (“The matter is absolutely crucial to the future of Computer Science, and I could not remain silent”²⁷), Richard Karp at the University of California, Berkeley (“I know that a great many people throughout the computer science community share my concerns”²⁸), and Brian K. Reid at Stanford (“From my perspective, the very survival of Xerox as a company depends heavily on Mr. Taylor, and you obviously do not realize this or you would have fought harder to keep him”²⁹). Taylor’s mentor, J.C.R. Licklider, wrote a long letter from his laboratory at MIT, worrying that “if the resignation is allowed to become effective, it will surely disrupt and could even destroy what many computer scientists (including me) believe is the best computer science laboratory in the world.” The lab, Licklider wrote, was “much more than the sum of its parts,” thanks to the “atmosphere, esprit, critical mass, and favorable interpersonal interactions” Taylor had fostered. “If Taylor does leave and you are nevertheless able, somehow, to keep together the group Taylor assembled, you are going to need another Taylor. And I am afraid that there may not be another Taylor.”³⁰

At best, progress in computer science would be delayed as the PARC researchers either dispersed or regrouped elsewhere. At worst, the momentum within the lab would be lost for good. Computing progress in the United States, at a time when the Japanese were feared as posing a threat to the nation’s technical primacy, would suffer.

The onslaught of letters stunned Kearns—and enraged him. “If I had known this would happen, I wouldn’t have agreed to the changes you proposed,” he told Spencer. Spencer reminded him that he had known it might happen; Spencer had told him so.³¹

But Kearns had said he would support Spencer, and he did. Bob Taylor was gone.

Less than a month later, Taylor founded the Systems Research Center for Digital Equipment Corporation in Palo Alto. “I said I would consult for them, and the next thing I know, some guy shows up asking where I want to build the lab,” Taylor says. His pay was $145,080, a 22 percent increase over his PARC salary. He received 2,500 stock options, and he reported to an accomplished computer scientist he liked and respected: Sam Fuller, Digital’s vice president of research and architecture.³² Today Taylor calls Fuller “the best boss I ever had.”

Back at the PARC computer science lab, an anonymous author circulated a gleeful “Parable of Rabbits” beautifully printed on a PARC printer. In the story, a group of rabbits (some experts in software, others in hardware) rejoice to hear that the “one amber rabbit, their leader, who knew all about how to help rabbits work together in harmony, so that they cooperated with one another and built wonderful things,” had gone to an “Other Company”—and now every rabbit could “hold the warm thought in his heart that he would go to the amber rabbit and say, ‘May I work for you in the Other Company?’ ”³³

Once year-end bonuses had been issued at Xerox, the rabbits hopped away. Within eight months of Taylor’s departure, twenty-eight people, more than half the technical staff of the computer science lab, had resigned.³⁴ Of those, fifteen joined Taylor at Digital Equipment Corporation. Once again, researchers under Taylor’s direction helped to lay the foundations for a new wave of technical innovations based on easy-to-use distributed computing systems. Among other projects, his team at DEC helped to develop reliable computer networks, multiprocessor machines, and an electronic book. One researcher, Mike Burrows, was also an essential contributor to AltaVista, the world’s first blazingly fast search engine. AltaVista came to market three years before Google was founded.³⁵

---

I. Taylor’s list further included a systems implementation language with a matching optimized machine architecture; programming environments; personal workstations; interactive text formatters; server stations for printing, filing, and mail; bitmap displays; electronic mail with transport and authentication facilities; and “a complete, distributed systems architecture accommodating all the above.”

II. Taylor also needed to alert Spencer of all staff and planning meetings, so Spencer could participate. Moreover, Taylor was required to meet with members of the recently dissolved imaging lab to assure them they could join his lab if they wished (whether Taylor wished it was irrelevant).

III. The Star was unveiled on April 27, 1981. It sold for almost $17,000 but had to be installed as a system with laser printer and Ethernet, bringing the price to roughly $30,000. Apple’s Lisa sold for $9,995 when it debuted on January 19, 1983.