Dealers of Lightning

by Michael Hiltzik

  "All the computers were happy and everyone could walk around in their suits—this was Xerox, after all," Ellenby recalled. (After the show was over Boggs and Sproull tried to see if they could use the truck to pro­duce snow in the ballroom. They turned it up until one of the compres­sors blew out, ending their experiment. "That was another expense we had to slide through the system as something else," Ellenby recalled. "Fairly expensive, too.")

  Disaster averted, the executives and their wives were encouraged to spend as much time as they wished familiarizing themselves with the Altos. They could type and draw on the bitmapped screens, meet a remarkable pointing device called a "mouse," send their handiwork off to the networked printers, and store it in electronic file cabinets known as "file servers." They could prepare documents on computer and send them by fax anywhere in the world—without generating a single sheet of printed paper.

  The demo room with its thirty Altos stayed open until late in the evening. "Lots of people came by and they all said how fantastic it was," Ellenby remembered. Most of the PARC scientists on the scene seized the opportunity to witness for the first time how nonengineers would react to their fabulous technology. The results were mixed. On the one hand, it was clear they were "very impressed that the stuff all worked, because they'd heard PARC was sort of a flaky place," Ellenby recalled. "There was sort of a feeling around that PARC was a luxury, so it was a relief for them to see some real value and even some things they might want." Yet clearly they were also perplexed and mistrustful about the effect these mysteriously overeducated contraptions might have on their careers.

  Xerox's top executives were for the most part salesmen of copy machines. From these leased behemoths the revenue stream was as tan­gible as the "click" of the meters counting off copies, for which the cus­tomer paid Xerox so many cents per page (and from which Xerox paid its salespersons their commissions). Noticing their eyes narrow, Ellenby could almost hear them drinking: "If there is no paper to be copied, where's the 'click'?" In other words: "How will I get paid?"

  For Geschke, the most discomfiting revelation was the contrast between the executives' reactions and those of their wives. "The typical posture and demeanor of the Xerox executives, and all of them were men, was this"—arms folded sternly across the chest. "But their wives would immediately walk up to the machines and say, 'Could I try that mouse thing?' That's because many of them had been secretaries—users of the equipment. These guys, maybe they punched a button on a copier one time in their lives, but they had someone else do their typing and their filing. So we were trying to sell to people who really had no concept of the work this equipment was actually accomplishing.

  "It didn't register in my mind at that event, but that was the loudest and clearest signal we ever got of how much of a problem we were going to have getting Xerox to understand what we had."

  There was at least one other harbinger of the coming letdown. Toward the end of the evening McColough, Kearns, and a few of the executive staff materialized in the demo room. Their appearance had been prearranged. "They were there to have an opportunity to say, 'Well, now we're going to do something, guys,'" Ellenby recalled. "But they didn't take that opportunity. They just said, 'Thank you.'

  "I was expecting a bit more than that," he said. "We'd developed a camaraderie that was quite unusual. My people felt pumped up and hyped, like a sporting team. Instead what we got was, 'Thanks, boys, the war is over, and you can take your horses back.'"

  Thus did the doubts surface almost before the euphoria of a flawless demonstration had a chance to run its course. Despite McColough's ringing re-endorsement of "the architecture of information," his and Kearns's equivocal farewell told Ellenby and his team that they were naive to think Xerox would exploit this technology anytime soon.

  And in this beleaguered and distracted corporation, Ellenby knew, time was the enemy.

  CHAPTER 19

  Future Plus Due

  Not all the signals PARC received from headquarters after Futures Day were as deflating as the initial one. In fact, the event did have two promising offshoots.

  The first was an unusual visit to Coyote Hill Road the following Janu­ary by Peter McColough and his nine top subordinates, including Archie McCardell, David Kearns, and Jack Goldman. Their purpose was to attend a two-day crash course in the art and science of software—the brainchild of Bob Taylor, who managed to get it sold to the executives while they still felt some gratitude, however begrudging, for PARC's bravura performance in Boca Raton.

  Taylor believed that the unique success of the 914 copier had incul­cated Xerox management with the doctrine that good things derived only from hardware. He was determined to show them that this idea was obsolete. As PARC envisioned the office of the future, a single piece of equipment could be made to serve multiple uses simply by changes in its software. As long as the brass found this idea an alien one—and it was revolutionary to most people in 1977—they would remain blind to the superiority of programmable Altos over narrow-minded electromechan­ical word processors like the 850.

  Taylor asked Alan Kay to help him break up what might otherwise be two days of lectures in the nasal drone of Computer Science Lab engi­neers by designing a hands-on software demo for the visiting brass. "You can have them for one and a half hours each day," he said. "See what you can do."

  Kay delegated the task in turn to his educational expert, Adele Gold­berg. The deadline was about the same as Ellenby had faced for Futures Day, nine weeks. But the challenge was possibly even more daunting: Bring ten stuffy male executives face to face with the unfamiliar rigors of computer programming, and make it fun. The strangeness of the experi­ence for these men could not be underestimated. Even Jack Goldman, the paterfamilias who battled so tirelessly to get PARC's technology into the Xerox product stream, would be doing serious work on a computer for the first time in his life.

  Goldberg figured the attention span of top corporate officers with much on their minds would be about the same as that of a classroom of fidgety adolescents. To hold their attention, she developed a demon­stration program called "SimKit" and loaded it with plenty of animation and music (one tune sounded automatically when class time was over). But underlying the bells and whistles would be a sophisticated system able to simulate situations the executives would recognize from the real world. Its basic format was a generic workplace scene populated by workers and customers. This could be altered to suggest the copy cen­ter of a large office, or a mechanized production line, a payroll depart­ment—whatever setting the executive-pupil might find comfortable and familiar.

  "We really were teaching them that while there are independent pieces in a program you can build something out of, there's also a context in which different kinds of objects are interacting," Goldberg recalled. "They'd have stations with workers they could select and specialize to do one function or another. The animation would let them actually see the customers come in and queue up and get service. We gave them the abil­ity to draw in their own workers and say what they looked like: This one's frowning, that one's smiling. And we had it all structured so they would never touch the keyboard. It was all mouse-pointing and mouse-clicking, because we knew these guys wouldn't type. In those days, that wasn't macho."

  As the event approached a few glitches cropped up, including one that no one on the youthful LRG team could have anticipated. During a dry run staged with ten middle-aged secretaries from the third-floor admin­istration offices, it developed that none of their subjects could read the Alto typefaces. "The small fonts our thirty-something eyes were used to didn't work for those in their fifties," Kay recalled. Goldberg made a virtue out of necessity by rewriting the curriculum so each student's first task would be to select a custom typeface and font size to use for the two- day session. This enhancement had the dual merits of instantly acclimat­ing the executives to the mouse and impressing them (it was hoped) with the unprecedented flexibility of the Smalltalk interface.

  A more harrowi
ng crisis arose the week before D-Day.

  This was a glitch in "Ooze," Smalltalk's intricate memory management system, which had been designed by Ted Kaehler. Ooze worked by rapidly shuttling data and objects between the Alto's scanty main mem­ory and its spacious hard disk, based on an algorithm that made the disk seem merely an extension of the memory. Nowadays this stratagem of fooling the computer into thinking it has more memory than really exists is a standard feature of desktop computers known as "virtual memory"; in 1978 implementing the system on such a small computer was a great pro­gramming feat. Ooze had its limits, however. Goldberg had loaded the system with so much new material that Ooze's capacity was strained to the breaking point. With five short days left before the seminar, Kaehler delivered the bad news.

  "Adele, the Ooze design is wrong," he said. "We didn't plan for all these subclasses."

  "Yeah?" she replied warily. "What does that mean?"

  "It's not clear it'll hold up during the seminar. And if it crashes, we're in trouble."

  "Well, what's the alternative? Should I simplify the curriculum?"

  Kaehler had nothing so modest in mind. "It seemed obvious to me," he said later, "that I should just fix the problem in Ooze."

  Goldberg grew fidgety at the very notion. This was no mere bug, she figured, but a major design change. "Ted," she said, "something tells me this is not a good idea."

  He waved his hand. The fix involved rewriting code in about seven places in the program to double its capacity. Easy enough in principle, he recalled later—assuming he could be sure of finding every place that needed fixing. In any event, he figured the team would have five days for testing. To Goldberg he said, "Don't worry about it."

  Easier said than done. Nervous about his tinkering with such an impor­tant component of the system so close to its make-or-break audition, Goldberg and the other team members put Kaehler on a short leash. "Adele, Alan, Dan Ingalls, and Larry Tesler each sat down with me sepa­rately to hear what the fix was and impress on me that it had to work," he recalled. "That was fairly intimidating." Goldberg told him that if it worked the first time, she would use it. Otherwise, they would restore the original Ooze and she would pare down her lesson plan.

  Kaehler finished his reworking in one evening and demonstrated the fix to his colleagues. To Goldberg's relief and amazement, it worked perfectly. "It was like you're building the Taj Mahal," she said later, "and just as you're about to put the final cap on it you decide that the foundation brick isn't right and you need to replace it. Just jack it all up, replace it, and put it back down again. That's what he did. Pretty remarkable stuff, and also a remarkable guy who could do that."

  The executive hands-on, as Kay recalled, proved "a howling success." To ensure victory, each corporate officer was assigned an LRG tutor to sit by his side during Goldberg's simulation exercise. Nine out of ten fin­ished their programs before the bell rang. One, who was in charge of a large Xerox manufacturing arm, even uncovered a flaw in his own real- life production line by plugging actual figures he carried around in his head into his computerized model.

  But whether the executives took away impressions of lasting value is another question. Diana Merry, who served as a tutor, doubted they had a very good idea of what they were looking at—that is, the remark­able engineering achievement underlying the simple and intuitive ani­mations playing over the Alto screen.

  Some may have been struck more by PARC's unstuffy ambiance. At one point Jack Goldman barged into an office on the first floor and almost tripped over Adele Goldberg, nursing her infant daughter inside. In 1978 such informality was unheard-of in a corporate setting and entirely unex­pected. Goldman backed away, embarrassed, never to forget the moment. Goldberg’s reaction was sharper: "It served him right. The door was closed and he shouldn't have walked in without knocking."

  As for David Kearns, Xerox's future chairman and chief executive officer, his most vivid impression was of PARC's unkempt and self- indulgent culture, perpetuated by its isolated and tranquil campus. "The place," he said later, "just sort of drifted along on its own course." This impression would not work to PARC's advantage when Kearns s help and understanding would be needed in a crisis.

  The second important consequence of Futures Day was, oddly enough, the creation of a formal program to commercialize the Alto. Led by Jerry Elkind, who chose John Ellenby as one of his top lieutenants, the new venture was empowered to dribble Altos into the marketplace by offering them to select customers on stringent terms.

  As crafted by Jack Goldman, the charter of the Advanced Systems Division, or ASD, would be the mirror image of that of the SDD unit building the Star. Where David Liddle's team was assembling an arsenal of maximum firepower to deploy in the market only after everything was in place—quite similar to the bottomless money pit Ellenby had so tact­lessly described to Peter McColough—ASD could weave and dart like a PT boat, using whatever ammunition it found handy to keep a few key buyers intrigued. Ellenby liked to compare the arrangement to the Nor­mandy invasion, with Liddle as a General Eisenhower tolerating an experimental foray or two to reconnoiter the beaches. Liddle implicitly acknowledged the metaphor: "We were very deliberately strategic," he said. "ASD was tactical."

  The new division solved two of PARC's most pressing problems. First, it indulged the veterans of Futures Day by giving them a chance to prove the Alto's commercial viability. Perhaps more important, it gave Jerry Elkind a place to go.

  Elkind had only recently returned to PARC from a temporary assign­ment to the corporate engineering staff. It was not a joyful homecom­ing. His absence had not only distanced him further from the daily routine of the computer science lab, but also underscored the gulf between his management style and Taylor's.

  For a long time it had been evident he could not hope to compete with Taylor for the affection of most of the lab personnel. One prob­lem was that he comported himself as their equal in technical ability, something which Taylor would never presume and which was bound to annoy the egotistical staff even if it were true. Moreover, he continu­ally reminded the Computer Science Lab that for all its independence and arrogance, it belonged to a larger organization, PARC, which was in turn beholden to yet a greater entity, Xerox. Taylor, by comparison, behaved as though CSL was the sun around which PARC, Xerox, and indeed all known computer science revolved.

  "I took a much less aggressive stance with respect to the resource issues, and I would hope a less antagonizing stance, than Bob did," Elkind said later. "In times when you're competing for resources I always had to make the argument with Bob that there was such a thing as a fair share."

  Perhaps the greatest friction point was Elkind's frequently blunt and condescending temperament. One could argue that in this he scarcely differed from Butler Lampson or, indeed, Taylor himself. But as long as Elkind lacked the former's intellectual charisma and the latter's sym­pathetic paternalism, many at PARC would not consider him authori­tative, just officious.

  "Jerry had a shorthand in his management style I never could appre­ciate," said Ed McCreight, as mild and complaisant a person as anyone who ever walked CSL's hallways. "He'd come into my office and say, in effect, 'I don't think you're doing anything important.' It was his way of saying, 'Explain how this fits into the grander scheme,' but it made you get down on yourself."

  Some of McCreight's colleagues like Jim Mitchell did not feel demeaned, just infuriated. "Jerry would always say, 'Let me play devil's advocate,'" he said. "But that was all he goddam did! Every time you went to him he was always telling you what was wrong with your ideas, so most of us stopped going to him. I told him once, 'Jerry when that's all you do its kind of a downer. You need to be going "rah, rah" sometimes, too, not just being a devil’s advocate.' He was technical but he wasn't as smart as any of us, and we knew it. He should have, too."

  Elkind's manner would not have polarized the lab as much if he had not had his fans, too. Several CSL staff members were quite comfortable with Elkind's sty
le. This group included Dan Bobrow and Warren Teitelman, both recruits from Elkind's old firm, Bolt, Beranek & Newman, as well as Peter Deutsch and Bob Metcalfe, the last of whom had known Elkind at MIT and considered him "a fine man, a gentleman, and an intellectual."

  But they were a minority. Considered strictly as the laboratory's chief, Jerry Elkind was an increasingly isolated figure. While he was away for an extended assignment to a corporate task force in 1977, the notion arose that the lab might function just as well without him. One day after Taylor had been running CSL unhampered for several months, McCreight remembered, Bob Sproull walked into his office.

  "Ed, have you noticed anything different in the atmosphere recently?" he asked.

  "You know," McCreight replied, "I sort of enjoy having Bob run the place."

  "You're not the only one who feels that way."

  As Mitchell recalled, "Every one of us noticed the difference in feel­ing in the lab. We all felt more productive, and we sure as hell liked the atmosphere a lot better."