Dealers of Lightning

by Michael Hiltzik

  Goldman's initial reaction was outrage. He viewed the new president as a novice—"my junior on the board of directors," he fumed—floundering in the murk of a corporate reorganization without understanding the importance of technology or that top researchers were mobile assets who could vote with their feet. He stormed directly into McColough s office and threatened to resign on the spot. "I pounded on his desk," he recalled, "and said, 'You can't do this shitty thing!'"

  Summoning all his powers of appeasement, McColough managed to get Goldman calmed down, but he did not rescind Kearns's order. Regaining his composure and examining the situation pragmatically, Goldman realized he was overmatched. The forces arrayed against him extended well beyond David Kearns, and the battle was more than merely personal. A political drama was unfolding at Xerox, with tech­nology and research the pawns.

  He now had a new goal: to keep his beloved research labs—his legacy—out of his enemies' clutches. "The independence of the research organization is what enables you to attract the kind of people you attract," he explained later. "And certainly putting it under an O'Neill type of guy would kill it from Day One." He understood that to stop this from hap­pening he would have to fall on his sword. When he called George Pake in Palo Alto that Friday, it was to ask him to pick up the mantle of research as it fell from his own hands—by stepping in as research chief.

  Pake felt deeply for his boss. "If I'd have been Jack I'd have been totally shocked that with no warning at all they would just yank the three research centers away from me," he said later. (Goldman's jurisdiction covered PARC, Webster, and a third lab outside of Toronto.) At Gold­man's urging he flew to Connecticut that weekend. "We had breakfast together on Monday and he said, 'George, this is a play by the engineers in Rochester to gain control of the digital technology at PARC.' He was indicating that he had lost that round. And he told me the only way to keep the first-rate science we had in research was for me to agree to take the job of overseeing the three labs."

  Pake detected a few flaws in the scheme. The job had not been offered to him and he had no knowledge that it would be. Second, he could not see himself functioning any more cozily under O'Neill and Sparacino than Goldman had. Finally, he was concerned about his health. Back in 1974 he had accepted a one-year staff appointment at headquarters and relocated to Stamford, leaving Hall behind as acting PARC director. The stress of corporate politics had driven his fragile blood pressure sky high and brought him to the verge of a stroke, forcing him to return home before the full year was up. He was not sure he wanted Goldman's job under any circumstances, and certainly not if it meant working in Stam­ford again.

  Goldman bulled through all Pake's objections by invoking the threat to Xerox research. He ushered him in to meet with Kearns, to whom Pake dutifully delivered Goldman's dire message. "I told David, 'Jack feels and I feel that the research scientists will just abandon this sinking ship if you put O'Neill and Sparacino in charge of the labs. We worked so hard to build this research enterprise that that would be a terrible tragedy.'" He offered to take over as head of research—on condition he could do the job from Palo Alto. "I'll just agree that whenever you want me here I'll get on a plane and come," he said. ("Many hundred airplane trips later I kind of regretted that," he remarked later.)

  Kearns said he would think it over and let Pake know. Pake left Kearns's office in the same frame of mind in which Bob Taylor had left his own in 1970, convinced the deal was dead. Instead, his offer evoked widespread approval in Stamford, where Pake was viewed fondly as a high-caliber scientist and a consummate gentleman. He defended his positions but never turned them into personal crusades like Jack Goldman. Further­more, Pake had always been content with the opportunity to create good science and technology at PARC. "He never had a focused ambition to turn the world or Xerox on its ear like Jack did," George White observed. "He didn't challenge these other experts' in their own fields, like mar­keting and finance. In short, he wasn't uppity."

  Two weeks later Kearns called to welcome him back onto the corpo­rate staff.

  Even though he was staying in Palo Alto, Pake's new responsibilities ruled out any possibility he could remain PARC's director. Of the can­didates to replace him, one stood out. He was Bob Spinrad, the genial New York-born electrical engineer who had risen from a post at Max Palevsky’s Scientific Data Systems to a corporate staff job under Gold­man. He was now head of the Systems Development Division, which was building the Star.

  Spinrad seemed to have all the qualities Pake valued most in a research manager. His scientific and research credentials had been earned at Columbia, MIT, and Brookhaven National Laboratory, a government nuclear research center. He was an old hand at navigating the shoals of digital computing, having served as SDS's software chief and managed the large-scale engineering team at SDD.

  Best of all, Spinrad was popular on both coasts. He had served with dozens of PARC and SDS people on corporate task forces (including Odyssey, which put Xerox's computer business out of its misery), and frequently dealt face to face with Jim O'Neill. "Goldman used to send me to talk to him when he couldn't because they were fighting," he recalled.

  What no one could have predicted was that Spinrad's biggest prob­lem would not be Jim O'Neill, but George Pake.

  About a year after his accession as director July 1, 1978, PARC's inter­nal battle over research resources took a turn for the worse. The catalyst, ironically, was the corporation's consent to the most significant expansion of the research center since its founding. This was the establishment of a program in the new technology of silicon-based integrated circuits. Tak­ing the science of VLSI a few steps beyond the work Lynn Conway and Doug Fairbairn were doing with Carver Mead, the new lab would actu­ally manufacture devices on an experimental fabrication line. This was not a trivial commitment. It meant millions of dollars in capital expendi­tures and the recruitment of an entirely new professional staff. But it was a particularly gratifying victory for Pake, for whom it meant that PARC would be doing cutting-edge research in his own academic specialty, solid-state physics.

  The IC lab, however, was far from universally popular on Coyote Hill, where it was viewed as a carpetbagging rival for money and head count. CSL trotted out the same arguments used against the VLSI program— that it was unnecessarily duplicative of work done by other companies that were in the business of making integrated circuits.

  "Xerox didn't have any strategic need for integrated circuits research,"

  Butler Lampson argued. "You could buy it perfectly well: That was the crux of the argument against it. There would be a very good chance that spending all this money would not only lead to no substantial payoff but would actually hurt you, because you would be attempting to do things internally that were better to do externally, and you'd end up with worse components. Meanwhile it seemed obvious to me that if we took that money and spent it on hiring more computing researchers we'd get a lot more mileage out of it."

  The IC lab added a difficult new factor to Spinrad's struggle with the eternal question of how best to balance the resources of PARC. Almost from the start he found Taylor in his face, entreating his new boss as only he knew how. Taylor recalled: "I was making sure Spinrad was briefed and encouraged him to get briefed by others to decide how to allocate PARC's resources. I'd say to him, 'Do you diink PARC's resources are allocated to the best benefit of the corporation?' He'd say, 'No,' and I'd say, 'I think you're right.'"

  Spinrad did agree that computer science gave Xerox the best bang for the buck at PARC. But he disagreed that CSL should receive the lion's share of the budget at the expense of the Systems Science Lab, for he was quite taken by some of programs Sutherland had under­taken as head of SSL. "Some of those projects were beginning to probe the edges of important things about user interfaces and social sys­tems," Spinrad recalled. "Studies of the applications of the systems in offices. The ethics and etiquette of e-mail. Some didn't work out and some did, but I felt it was important. Taylor's lab was
narrowly hard sci­ences and unambiguous results and measurable performance and communications reliability, and it didn't get into what clearly was the important area: How the hell are you going to use this stuff?"

  Nevertheless, he did share Taylor's general opinion that the physics labs had been overfunded. Perhaps failing to recognize that virtually since the day of PARC's opening the physics lab had played the role in Pake's mind of a political counterweight to Bob Taylor, Spinrad in March 1980 took a step that forever marked him, unfairly or not, as Taylor's cat's-paw. This was his preparation for Pake of a five-year plan in which he proposed reallocating PARC's budget in favor of the com­puter labs (including SSL) and reducing the money spent on the Gen­eral Science Lab.

  "I figured if I had a zero-sum game"—that is, if PARC's budget were to remain static overall—"I was going to have to cut back slowly in some areas," he recalled. "It would not be sudden, but some people's oxen were going to get gored more than others. I was going to change the status quo."

  Spinrad's plan violated PARC and Xerox orthodoxy in at least one important respect. Xerox's corporate culture always treated budget cuts as burdens to be shared equally by every cell of the organism. If a 10 per­cent cut was indicated, every division and branch office took a 10 percent cut whether it was a marginal contributor to the company or an indis­pensable cog in the machine. "I was probably the first one not to be egalitarian about cuts," Spinrad recalled. "I had prejudices, and I thought one of the few roles management has is to make choices and judgments."

  But inside PARC many people found it hard to distinguish Spinrad's prejudices from Bob Taylor's. "Spinrad succumbed to Taylor's unre­lenting pressure," was Harold Hall's judgment. Pake agreed. "My per­ception was that Taylor, being in complete ascendancy in the political jockeying between the two labs, enlisted Spinrad." That the realloca­tion plan took direct aim at the physicists in Pake's pet laboratory—and Taylor's bete noire—only reinforced that impression. Pake had spent ten years defending the General Science Lab from Taylor's carping. He was not about to sit by and let it be gutted now.

  Pake viewed the situation even more urgently because he harbored growing doubts about Spinrad's overall performance. For several months he had been fielding complaints from within the research center about Taylor's ambitions—complaints that would never have reached him if Spinrad had kept Taylor on a properly short leash. Moreover, he believed Spinrad had deliberately dragged his feet in recruiting a director for the new integrated circuits lab, which consequently had not yet gotten off the ground.

  On March 21 Pake summoned Hall to his office and, clearly anguished, outlined his concerns along with what he called a "really zany solution."

  "Maybe I can split the center in two parts," he told Hall. "That might solve the problem."

  Specifically, he would divide PARC into two independent research centers, manifestly configured to keep Taylor isolated. One, the "Sci­ence Center," would comprise SSL, GSL, and the new IC lab and be headed by Hall. Spinrad would retain jurisdiction over the "System Center," which was limited to CSL and the Optical Science Lab.

  Hall assented to return to line management. Within hours after Pake first broached it, the change was official. "I suppose if I had been a sci­entist when this was happening I wouldn't have known what was going on," Pake acknowledged later, "because all of a sudden everyone gets this memo through the internal mail saying PARC is now two PARCs."

  In truth, everyone at PARC regarded the arrangement of two research centers sharing the same building—and in some cases the same floor - as unsustainable over the long term. But Pake could see no other solution to the dual problems of Taylor's imperialism and the stalled progress on the IC lab. Spinrad, philosophical as ever, accepted the rebuff com- plaisantly. But he was clearly chastened, and within a year accepted a reassignment back to the corporate staff.

  Meanwhile Hall assumed responsibility for recruiting a chief for the integrated circuits lab. After several months he was convinced he finally had his man: a physicist from the University of Kansas named William J. Spencer.

  The physically imposing Spencer's academic credentials were less than sterling—before taking his Ph.D. from Kansas he had gotten his bache­lor's degree in physical education—but at the age of fifty-six his profes­sional career was distinguished by management posts at Bell Labs and the Sandia and Lawrence Livermore national laboratories. Although Spencer had been Hall's third choice for the job (the two other candi­dates had turned him down), he felt better about his find as time went on. Heading the IC lab, he thought, would only be the start of things for Bill Spencer. As he reported to Pake, "When he's ready for the big job I'll let you know, but I can't help thinking that I've hired not only my successor, but yours."

  Meanwhile, the pressure of the outside world was being felt more and more inside PARC. "The only problem with PARC was a law of physics," Charles Simonyi observed. "A star that bright eventually has to blow up."

  Simonyi began to sense the impending supernova in 1980. One day that fall he found himself wandering the hallways of Xerox headquarters in Stamford, having his lowest expectations confirmed. The disparity between the opulence of his surroundings and the paltriness of the brain­power housed therein beggared the imagination. He felt wholly irrele­vant, like a wayward tourist rather than an employee being interviewed for a corporate staff position.

  Simonyi had been inveigled into making this trip by Jerry Elkind, who sensed that the Advanced Systems Division was losing its charm for his young subordinate. He was right. The excitement Simonyi had savored in getting Altos out to the world had worn thin. With BravoX nearing com­pletion he was unsure of his next step, especially given the absence of any sign that Xerox meant to follow up ASD s market probes with a full-scale merchandising program.

  He had only grown more restive when a friend showed him an Apple II running VisiCalc. The spreadsheet program was new to him but daz­zling in its power. One typed numbers or formulas into the cells of a grid and linked them, so the answer from one cell could be part of the formula of another. This allowed anyone to tabulate data in an infinite number of permutations. It was particularly valuable for businessmen and engi­neers, who could perform "what-if analyses simply by altering a figure here or there and letting the grid automatically calculate the myriad ram­ifications of the change. Sure enough, within months VisiCalc had trans­formed the Apple II into a commercial sensation.

  By contrast, at PARC, where funds had flowed so limitlessly that no one ever felt the urge to run "what-if' budget scenarios, the spreadsheet idea had not even occurred to the greatest software engineers in the world. What Simonyi found even more depressing was that VisiCalc was simple, intuitive, and fast—all the qualities he and his colleagues had strived for in their work over the past decade.

  "That alarmed me a lot, how good it was," he remembered. "They were using even fewer cycles than the Alto to run it on the Apple II, way fewer."

  Like Larry Tesler, Simonyi had discovered the power of low-end com­puting. Until the day he saw VisiCalc on the Apple II he too had dis­missed the hobbyist machines as a joke, as absurd in their triviality as the Star was in its bloat. Now he recognized in them a future that PARC had missed.

  As a guide to the new world Simonyi turned to his former SDD boss, Bob Metcalfe, who was heading his own startup, an Ethernet equip­ment maker named 3Com Corporation headquartered in Santa Clara, a few miles south of Palo Alto.

  Metcalfe rather relished the role of trailblazer for his old PARC and SDD colleagues ("I was the one who had gone out into the world and didn't the," he observed). He invited Simonyi to lunch, and over appe­tizers handed him a list of ten young entrepreneurs who he thought had a chance of propelling the computer industry toward its exciting future and who might make good use of Simonyi's talents. The first name on the list was someone Metcalfe described as "a crazy guy," which in Simonyi's eyes bathed him with a perverse appeal. His name was Bill Gates. Simonyi would never meet any of the other
s.

  A few weeks following that lunch, Simonyi happened to be oversee­ing the installation of an Alto at the Seattle headquarters of Boeing, one of the VIP customers granted a shipment of ASD machines. On his last afternoon in town he dropped in on Gates's little company. Microsoft's thirty or so employees occupied half of the eighth floor of the Old National Bank building in Bellevue, just across Lake Washing­ton from the city of Seattle.

  Carrying a portfolio of his work, Simonyi entered Suite 819 relaxed and confident, thanks to his mistaken impression that Metcalfe had already called to smooth the way. In fact, he was an unexpected visitor. Bill Gates being tied up at the moment with a delegation from a Japanese manu­facturing company, Simonyi was escorted instead into the office of Steve Ballmer, a friend of Gates's from Harvard. Unlike Gates, Ballmer had stayed at Harvard to graduate, after which he signed on to be Microsoft's maniacal chief salesman and hyper-motivational troop leader.