Dealers of Lightning

by Michael Hiltzik

  This so-called "asynchronous" architecture had never before been a problem, which is why Thacker had re-implemented it for the new machine. In simple terms, the Dolphin and Alto varied the number of instructions they performed each clock cycle according to the proce­dure they were executing. More complicated procedures required more instructions (and more time) than easy ones. Therefore the sys­tem could only guess when the processor would be available to service the next needy peripheral, whether that was the Ethernet transceiver or the disk controller. Since the processor speed of the Alto was more than fast enough to handle all these functions together, nothing ever went awry. Given the stepped-up demands of the Dolphin machine, however, such an informal arrangement would not do.

  Lampson's "synchronous" design differed by feeding a fixed number of instructions through the processor and memory per clock cycle, which enabled him to synchronize all the peripheral devices to the processor and eliminate the guesswork about when an incoming bit could be processed. With everything ticking along according to the same clock, a much smaller volume of data had to be held in buffers—which relieved the design of the Dolphins hardware bloat.

  Belleville and the others saw instantly that the design Lampson had sketched out on seven sheets of lined yellow paper would run faster, more efficiently, and cooler than the Dolphin. After making a few minor changes, they renamed it "Dandelion" (most of the machines being turned out by Xerox's Palo Alto engineers in this period got names start­ing with "D" for "Digital" and became known collectively as the "D machines") and installed it in the Dolphin's place as the heart of the Star. A major hurdle had finally been cleared.

  But the Star was still more than two years from its launch—and that would make all the difference in its future. For while the Star was designed to be the office professional's personal nirvana, meticulously assembled from the finest technological components money could buy, it was destined to get blindsided in the marketplace by a new species of machine that was quite explicitly less than nirvana, assembled quick and dirty from the cheapest components available, and so low-tech it bore almost the same relationship to the Star that a roller-skate does to a Mercedes-Benz.

  A legend in its own time, boasting capabilities still unmatched nearly twenty years after its launch, the Star would end up as one of the most resplendently obsolescent machines ever sold.

  By 1975, the fifth year of PARC's existence, George Pake's prediction had come true: The center had outgrown its original quarters, and then some. There were two hundred employees spread among three locations ranging from Porter Drive up to Hillview, where two build­ings were now rented. In March Jack Goldman's original vision of a free-standing shrine to basic research was realized with the opening of PARC's ten-million-dollar permanent home at 3333 Coyote Hill Road, directly across the street from Building 34.

  The new building had a difficult birth. Its ground-breaking had been delayed for more than two years, until August 1973 (Goldman and sev­eral other dignitaries took home silver-plated shovels marking the occa­sion), by an old land-use controversy enflamed by a local conservation group's discovery that Stanford University, the site's owner, had illegally subdivided the land. The so-called Committee for Green Foothills sued, tying up all development on the site until Xerox, desperate to get its con­struction project under way, brokered a settlement.

  As part of the deal Xerox and Stanford agreed to stringent construc­tion restrictions. The PARC building was to be no more than 200,000 square feet (its first phase actually came to only 120,000) and situated out of view of motorists passing on most of the major local thorough­fares. Stanford's penance was a pledge to leave several neighboring (and potentially lucrative) acres permanently undeveloped.

  Designed by Gyo Obata, an architect who had made a name for him­self with the arresting design of the National Air and Space Museum in Washington, D.C., PARC was a long rectilinear concrete building adorned with hanging gardens. The building cascaded down the hillside in steps, arranged like an upside-down cake so that visitors entered on the third floor, where the administrative staff was housed, and descended to the second (occupied by the computer and system science labs) and first (the general and optical science labs). There at the bottom was a staff entrance, gratifyingly outfitted with a sheltered bike rack. Responding to suggestions Rick Jones had dutifully collected from the building's future occupants, Obata generously fronted PARC's northern elevation with horizontal ribbons of glass to bathe its interior in ample sunlight, and fur­nished it with numerous open-air atriums ideal for the tranquil contem­plation of nature, technology, and the nexus thereof.

  The completion of the Coyote Hill building represented a milestone in PARC's public profile. The local newspapers featured fulsome interviews with Pake, glimpses of the office of the future ("scientists at the Xerox Research Center in Palo Alto aren't even working with copiers .. . "), and optimistic talk about the blending of the new sciences with the old.

  The contrast could not be greater with the grim saga playing itself out a few hundred miles to the south. This was the death watch over SDS.

  "McColough's Folly" had been an invalid virtually since the moment it changed hands back in 1969. Periodically it would come under the microscope of one task force or another (Jeny Elkind served on one such group in 1972), but no one could ever find a way to stanch the flow of red ink. Instead, Xerox management tried to hide the losses deep within the balance sheet by reorganizing the company into three functional organizations—sales and marketing, engineering and manufacturing, and strat­egy and planning—and distributing pieces of the computer business among them. As part of this program, Xerox in 1972 dissolved SDS as a separate entity and fired Dan McGurk, the former Palevsky lieutenant who had been running it ever since the sale.

  But this tactic could not disguise the losses in computing, any more than a farmer can obliterate the stench of manure by spreading it over more acreage. By early 1975 Xerox's poor overall performance had destroyed the last of Wall Street's confidence in its management. The stock, which had hit an all-time high of SI79 a share in 1972, dropped to $50. Corporate legend has it that one day in the men's room at Stamford headquarters Archie McCardell, the company's finance-oriented presi­dent, told McColough he thought it was time to give up on the computer business altogether. In response, McColough convened his final com­puter task force. Its purpose was to find the best way to cut adrift what was left of SDS, which may be why it was given the Homeric code name "Odyssey."

  Among Odyssey's members was Rigdon Currie, who tried to fight a rearguard action against the extinction of his old division. "1 put in a pitch to cut the operation back to our original customer base in scien­tific and real-time computing, instead of competing with IBM in the business market," he recalled. "I got stepped on."

  Odyssey's conclusion, instead, was that the computer business could be neither saved nor sold at any price. Currie was handed the disagreeable assignment of cutting the best possible deal for its remaining odds and ends with Honeywell, which had made a practice of scavenging the shards of other companies' failed ventures in computing. To cover the expense of getting out of the business, Xerox took an enormous $84.4 million write-off. The action produced the company's first annual loss since the introduction of the Model 914 copier, fifteen years earlier.

  Nothing was left of the long misadventure of SDS other than the pla­toon of researchers now installed in their imposing new hillside palace in Palo Alto. If Xerox was to be persuaded that digital technologies were the key to its future, it would be up to them alone to make the case.

  PART III

  Messengers

  CHAPTER 18

  Futures Day

  During the course of 1977 Peter McColough began to fear that his corporation was coming apart at the seams.

  The previous few years had been a skein of disap­pointments and embarrassments for Xerox. In 1975 had come the unrav­eling of the SDS purchase—"McColough's Folly"—and the resulting $84.4 million
write-off. The foreign invasion had turned into a full-scale rout, with a hundred thousand inexpensive Japanese copiers sold in a market segment for which Xerox had no competing product. Then IBM and Kodak took direct aim at the heart of Xerox's traditional customer base by introducing fast new copiers of their own. There were antitrust lawsuits and patent infringement battles. To top it all off, the U.S. econ­omy was in terrible shape, squeezed by the pincers of high inflation and savage recession—an unprecedented combination known as "stagflation."

  The Xerox chairman knew something had to be done to stem the erosion of morale or his best executives and salespersons would flee. His solution was to throw one huge, historic party.

  McColough envisioned the Xerox World Conference of November 1977 as a sort of company-wide revival, a last attempt to restore the pas­sion that had fueled its rise nearly twenty years earlier, when Joe Wilson bet the farm on an untested technology and won the world. The guest list would comprise the top 250 executives of the worldwide organization and their spouses. The setting would be Florida's exquisite Boca Raton Country Club, where the food, accommodations, and entertainment through four days of celebration, exhortation, and rebirth would be first- class all the way. In McColough's imagination, the affair would evoke the atmosphere of the previous World Conference, held in 1971 when Xerox's wealth and hubris were at their very peak.

  It is not clear which individual first proposed that PARC play a major role at the World Conference. In any case, once McColough heard the idea of using Boca Raton to introduce the technologies PARC had invented to the sales force, he embraced it whole heartedly. PARC would not only get a chance to show off; McColough decreed that one entire day of the four in Boca Raton would be devoted to PARC alone.

  One morning that summer, the SSL's director, Bert Sutherland invited John Ellenby into his office. Ellenby had worked in the Com­puter Science Lab for three years, during which time he had gained a well-deserved reputation as a man who Got Things Done—schematics realized, prototypes built, projects completed—not at all an easy proposition amid the clash of egos and the religious wars raging unceasingly over engineering and design at PARC.

  "If you had a free hand, John, what would you do to show off our work?" Sutherland began. "You see, there's this big conference coming up for managers from all over the world, and we've been invited."

  Ellenby thought for a moment, then asked, "Just what do you mean by a free hand?"

  John Ellenby was a slender, pale Briton with a fringe of brown hair crowning his high forehead. His unusually varied education and work experience had included the study of economic geography at the Univer­sity of London, a teaching post at the London School of Economics, and, following a course in systems engineering from IBM, a dual appointment as a lecturer in computer science at the University of Edinburgh and con­sultant in computer architecture and graphics to Ferranti Ltd., the pio­neering British computer maker. His first encounter with PARC had occurred back in 1971 thanks to Dan Bobrow, who was passing through Edinburgh on a fellowship. Bobrow described PARC to Ellenby and, when he returned stateside, described Ellenby to PARC. Not long after that Ellenby happened to be passing through the Bay Area; an invitation was issued for him to address a Dealer, and in September 1974 he was hired by Jerry Elkind, who no doubt saw in Ellenbys ability to shuttle so easily between academia and industry a reflection of his own skill in bal­ancing good science and solid pragmatic judgment.

  The terms of Ellenby's employment amounted to a pay cut. Owing to the good salaries he pulled down from two full-time jobs, he lived in rel­ative luxury in a large old stone house in Edinburgh with his sculptress wife, Gillian, and two young sons. "But that was immaterial," he said later. "The chance was to work at PARC, which was absolutely the top-notch computer sciences lab in the world. So I joined, and we all moved."

  He had worldliness enough to discern at a single glance the dividing line separating PARC's virtues from its flaws. The former included the work that had produced the Alto and the Ethernet, which impressed him as marvels of elegant design and superb feats of engineering. The latter included Taylors dogmatism, which Ellenby believed discouraged thoughtful dissent in the Computer Science Lab. He was dismayed to find genuinely farsighted projects such as Shoup's Superpaint and Bobrow's work in artificial intelligence, both of which challenged CSL orthodoxy, hanging on by their fingernails.

  "Computer architecture in those days was a major battleground for reli­gious wars," he recalled, "and Xerox had them big-time." His first assign­ment would have him interceding between two of its contending armies.

  The task was somehow to get the Alto manufacturing process jump-started. The machine had been designed and prototyped but as yet there were only five in existence. The construction program, it seemed, had mysteriously stalled somewhere between Palo Alto and El Segundo.

  As it happened, Ellenby had distinguished himself as an industrial con­sultant in Great Britain by transforming dysfunctional programs into operational ones. Turning his experienced eye to the Alto, he recognized instantly that the machines were indeed hostages of a religious war—this one between the Computer Science Lab, which designed them, and

  SDS, whose downtrodden factory staff was tasked with building them to PARC specifications in El Segundo, five hundred miles away.

  "El Segundo was a product organization with a lot of pride," Ellenby recalled. "It had a lot of good guys from the days when they had built a lot of quite impressive machines. And now they were getting fucked over by this copier company that knew how to put powdered coal onto drums that went whistling around and transferred it to paper but didn't know shit about electronics. So there was a religious problem right there. Then there was this funny group of weird Northern Californians they had to deal with while they were trying to solve their other problems. Mean­while there was not a lot of respect at PARC for SDS. And nobody was really assigned at PARC to make it all happen. The Alto was kind of a baby looking for its mother."

  Ellenby stepped in to referee. His first achievement was getting El Segundo to complete twenty Altos stuck in the pipeline. Then he took a radical step. Organizing a small cadre of product engineers into an inte­grated engineering and manufacturing unit he called the Special Programs Group, he arranged for Chuck Thacker's time machine to be reengineered into an object that could be efficiently mass-produced. The Special Pro­grams Group replaced all the Thackeresque shortcuts, which looked like virtues when the goal was hastily to turn out a serviceable machine with spare parts, but were now merely the sources of annoying glitches.

  "No fault of Chuck's, but the machine was just flaky," Ellenby said. "I had come from a pretty rigorous background because the machine for which I had been consulting designer at Ferranti, the Argus 700, was designed for very high-reliability process and communications control. I thought that somehow or other a machine that stops for no reason was not a good machine."

  Ellenby's group added a memory error-correction system similar to the one Thacker had designed for MAXC (but had left off the Alto). This substantially cut the manufacturing cost of the machine by allowing the SPG to use more error-prone, but cheaper, memory chips without com­promising the machine's reliability. The original Alto was almost unmain­tainable ("In order to get to something you had to take a lot of other stuff out," Ellenby recalled); he ordered the innards redesigned so every com­ponent would be easily accessible just by opening the cover, as in todays desktop PCs. The so-called Alto II was both durable and easy to manu­facture on a small production line. "We just popped em out," Ellenby said proudly. This was the machine that proliferated throughout PARC as a springboard for some of the most striking technological innovations the world has ever seen.

  Soon after the new machines started rolling off the fabrication line in early 1976, however, Ellenby came face to face with the realities of tech­nology politics at Xerox. Heady from the triumph of the Alto II, he forged ahead with a plan to design and manufacture an Alto III. This would be the Hol
y Grail: a mass-marketable, programmable computer that would exploit the snowballing manifestations of Moore's Law (such as faster and cheaper memory chips) by offering user-friendly word processing, pro­fessional database programs, and more. The goal was for the Special Pro­grams Group to design the machine for manufacture by Xerox's Office Systems Division, a Dallas-based unit that turned out electric typewriters and other non-copier office machines under the leadership of a former Webster lab chief named Robert Potter.

  That July, Xerox's Display Word Processing Task Force endorsed the plan. For a few short, glorious weeks, official Xerox policy was to ser­vice the growing market for electronic word processing with the Alto III, a programmable personal computer that would bear the same rela­tionship to the competition's glorified typewriters as a Harley does to a tricycle. Ellenby's group was on target to engineer an inexpensive computer-cum-word processor and printing system for shipment to cus­tomers by mid-1978. Had it done so, Xerox would have beaten the IBM PC to market by three years—with an infinitely more sophisti­cated machine.