That the Systems Development Department (it was no longer simply a “division”) was finally able to bring any product to market, much less the triumph of integrated system design known as the Star, must have struck some of its own employees as nothing short of miraculous.
The Star program’s duration and complexity, the personal tensions within SDD, Xerox’s ceaseless vacillation, and numerous other agonies had driven many engineers off the project long before it reached the promised land. Thacker abandoned ship to return to PARC. Simonyi left to sell Altos with Jerry Elkind. Bob Metcalfe quit in 1979 in search of the entrepreneurial main chance.
Even the machine’s code name demonstrated SDD’s need to clamor for Xerox management’s attention. It had been coined by Bob Spinrad in the hope of lending the project some luster in the eyes of Dave Culbertson, a group executive to whom SDD then reported.
“Culbertson was a sailing enthusiast,” David Liddle recalled, “so Spinrad decided to name it after a one-design sailing class.” After considering “Lightning” and “Sunfish” they settled on “Star,” which, as Liddle observed, “was a decent sailboat and a tolerable name for an office appliance.”
Engineers both inside and outside SDD expressed frequent doubts about the department’s course. At PARC, many computer engineers viewed the Star as Xerox’s attempt to yoke their inventions to its fading office monopoly, to the former’s disadvantage. Around mid-1980 Butler Lampson predicted to his SDD friends that they would never ship a product. “They had a system with a million lines of code in it built by a team of people hired off the street,” he said. “The whole thing took four years, and in my experience any project that had those properties had another property, which is it wouldn’t work. I predicted it wouldn’t work and they wouldn’t be able to ship it.”
He was wrong. On April 27, 1981, at the National Computer Conference trade show in Chicago, SDD formally unveiled the Star as the Xerox 8010 Information System.
With its unique seventeen-inch bitmapped screen and graphical interface, the product was an instant sensation. Its full-dress demos every hour on the hour “had people overflowing into the aisles,” recalled Charles Irby, a former Engelbart engineer who had been one of SDD’s first recruits from outside PARC.
Irby was particularly amused to notice among the repeat visitors Larry Tesler, then at Apple, and his Lisa design team. “They’d watch every demo, then go off into a corner and talk about what they had seen,” he recalled.
The Star’s success attested to the pertinacity of David Liddle, who had managed to keep his mind and his organization focused through years of indifferent and even hostile treatment from the Stamford headquarters. Year by year SDD got kicked around the corporate organizational chart—now under the Information Technology Group, now under Xerox Business Systems—until, as Bob Belleville recalled, “We just stopped paying any attention to where we were.”
In 1979 the division finally fetched up like a beached whale at the doors of the Office Products Division. This was the Dallas operation originally managed by the detested Bob Potter. But after Potter had moved to International Harvester with his patron, Archie McCardell, the division had come under the charge of an entrepreneurial firebrand.
Don Massaro had joined Xerox when it purchased Shugart Associates, a disk drive company he had co-founded. Brash, risk-oriented, abrasive, and persuasive, he seemed a throwback to the glory days of Shelby Carter. For his divisional symbol he chose the Road Runner from the Warner Brothers cartoons, the better to taunt the Xerox “coyotes” he maintained were constantly out for his tail. “I had not spent twenty years of my life climbing the Xerox ladder rung by rung, playing according to the rules,” he told an interviewer. “I was prepared to fail.” When Dave Liddle flew down to Dallas to show him the Star software, he was jazzed. “I said, ‘Fuck it! This is incredible technology and we’re going to bring it to the marketplace!”
Talk like that was just what the weary engineers of SDD needed to shake off their torpor. Massaro was the first Xerox executive they had met who displayed any business acumen at all. He made snap decisions, moved fast, and had more confidence in his own judgment than the rest of the executive roster put together. Rallying behind his drumbeat—“I think we have another 914 on our hands,” he crowed to Business Week—they redoubled their efforts to get the Star out the door.
Massaro also contributed some desperately needed rationality to Xerox’s treatment of PARC technology, much of which had been kept under wraps as though by reflex without any consideration given to how best to exploit it. For example, the company had long insisted that Ethernet be kept secret in case it chose someday to market the network as a proprietary product.
“But how would the Xerox Corporation make any money by proprietarily pulling coaxial cable?” as Liddle asked rhetorically. He, Massaro, and Metcalfe proposed an alternative. If other electronics companies could be persuaded to adopt Ethernet as an industry standard, Xerox could profit from what was sure to be an exploding market for the peripherals that were already part of its product line, like laser printers. This would also break IBM’s stranglehold on the networkable equipment market, which it maintained by promoting its inferior “token ring” network—a system that, once installed, compelled users to buy only IBM-made peripherals.
This argument finally prevailed in Stamford, which in 1979 granted Massaro and Liddle approval to make Ethernet public by enlisting Intel and Digital Equipment Corporation in the effort to turn it from an experimental system into one of commercially viable robustness. The new industrial-strength specifications were published in 1980 as the joint Xerox-Intel-DEC Ethernet standard. Xerox’s liberal licensing rules, which allowed any company to manufacture Ethernet cards, cables, transceivers, and peripherals after paying of a one-time $1,000 license fee pledging to support the standard as written, turned Ethernet into the most widely used local networking technology in the world.
Don Massaro’s enthusiasm for the work of SDD was requited by the ultimate product. The Star workstation he shepherded to launch was an amazing accomplishment. Enclosed in a squat beige-colored box which, like its ancestral Alto, slid on casters under a desk, the machine came packed with features no one had ever seen before and few envisioned in a commercial office machine. These included a bitmapped screen (in “muted blue,” as Xerox promotional literature described it at the time), a mouse (“an electronic pointing device”), windowed displays, and “What You See Is What You Get” document preparation. The bundled functions included text processing, a drawing program, the first integrated “help” program, and electronic mail.
By far the system’s most striking feature was its graphical user interface, the stylized display that communicated with the user via the bitmapped screen. This arrangement of icons and folders built around what the Star designers called the “desktop metaphor” is so familiar today that it seems to have been a part of computing forever. But its pioneering implementation on the Star included some capabilities that had yet to resurface on the market nearly two decades later. Text, formulas, and graphics could all be edited in the same document. (Compare today’s “integrated” software, in which a drawing imported into a text document can no longer be altered, but must be changed in the original graphics program and reintroduced into the text document.) Out of the box the Star was multilingual, offering typefaces and keyboard configurations that could be implemented in the blink of an eye for writing in Russian, French, Spanish, and Swedish through the use of “virtual keyboards”—graphic representations of keyboards that appeared on screen to show the user where to find the unique characters in whatever language he or she was using. In 1982 an internal library of 6,000 Japanese kanji characters was added; eventually Star users were able to draft documents in almost every modern language, from Arabic and Bengali to Amharic and Cambodian.
As the term implied, the user’s view of the screen resembled the surface of a desk. Thumbnail-sized icons representing documents were lined up on one side of the screen and those representing peripheral devices—printers, file servers, e-mail boxes—on the other. The display image could be infinitely personalized to be tidy or cluttered, obsessively organized or hopelessly confused, alphabetized or random, as dictated by the user’s personality and taste. The icons themselves had been painstakingly drafted and redrafted so they would be instantaneously recognized by the user as document pages (with a distinctive dog-eared upper right corner), file folders, in and out baskets, a clock, and a wastebasket. Thanks to the system’s object-oriented software, the Star’s user could launch any application simply by clicking on the pertinent icon; the machine automatically “knew” that a text document required it to launch a text editor or a drawing to launch a graphics program. No system has ever equaled the consistency of the Star’s set of generic commands, in which “move,” “copy,” and “delete” performed similar operations across the entire spectrum of software applications.
The Star was the epitome of PARC’s user-friendly machine. No secretary had to learn about programming or code to use the machine, any more than she had to understand the servomechanism driving the dancing golf ball to type on an IBM Selectric typewriter. Changing a font, or a margin, or the space between typed lines in most cases required a keystroke or two or a couple of intuitive mouse clicks. The user understood what was happening entirely from watching the icons or documents move or change on the screen. This was no accident: “When everything in a computer system is visible on the screen,” wrote David Smith, a designer of the Star interface, “the display becomes reality. Objects and actions can be understood purely in terms of their effects on the display.”
What was even more remarkable was that much of this was accomplished over the objections of Xerox marketing experts, whose kibitzing about even trivial matters slowed the development process by months. Irby recalled a particularly trying confrontation over the mouse with a marketing man from the Dallas division named Ron Johnson.
“The first time he’d ever used a mouse he’d had a bad experience—apparently he’d used a dirty one that didn’t track right,” Irby said. “So for two years he was against our using it, while we spent all our time on user studies and tests to show him it was the right thing. We spent at least $1 million of Xerox resources proving that it was better than a cursor button or touch screen, which is what he wanted. Finally we presented all these findings to him at a meeting—and he still wouldn’t go for it!
“That was one of the very few times when I totally exploded. I got out of my chair and towered over him and yelled about what an idiot he was being. I screamed, ‘We’re going to use the mouse, goddamn it!’ and walked out. We never got a complaint from him again.”
Had the Star performed up to the level of its dazzling first impression, Xerox might have been able to to establish and hold that beachhead in office computing craved by dozens of executives ranging from Jim O’Neill to David Liddle.
But the glow faded fast.
The first shortcoming users noticed was its speed. The elaborate system ran, as one of its designers acknowledged, “like molasses.” While the Dandelion processor was a marked improvement over Thacker’s Dolphin, it was still overwhelmed by the pure tonnage of a million lines of heavy-duty Mesa code running under the surface. “The Star software was built to consume all available computing resources in the universe,” cracked Smokey Wallace, an SDD engineer.
Another hurdle was its cost. The Star workstation reached the market at a retail price of $16,595. This might have made sense for equipment aimed at a high-performance engineering market. But it was far more than most commercial businesses would spend to furnish a secretary or clerical worker with capital equipment. Furthermore, nobody could buy just one Star workstation any more than one can eat just one potato chip. A meaningful installation required two to ten workstations, plus a highspeed laser printer and Ethernet to link it all together. That raised the per-user cost to at least $30,000 and the price of the whole integrated system to a quarter of a million dollars or more. Some experts forecast that the Star would not sell until Xerox reeducated it customers to use it properly and made it cheaper. “It’s a good product,” one said, “for the second half of the 1980s.”
Within a few months of its launch the Star began to look like an egregious marketing blunder. It was the old story of engineers building a system that only engineers could love—except that instead of building one too complicated for average users, SDD had built one too big.
It seemed as though SDD as an organization had been driven by designers lacking any counterweight of sales and marketing professionals. As Lampson observed later, “It was kind of amazing that this company whose biggest single strength was marketing set up an organization composed entirely of engineers to get them into a whole new line of business.”
In truth, SDD did have marketing advice. The problem was that, possibly for the first time in Xerox history, the marketing experts were so overawed by the system they were examining that they were themselves swept up in the engineers’ enthusiasm.
The upshot was a series of surveys known internally as the “Wave” studies, on which the company spent hundreds of thousands of dollars to analyze its customer base. Undertaken during Spinrad’s stewardship of SDD, well before the division priced and launched the Star, the Wave studies compiled data from telephone and face-to-face interviews with decision-makers at nearly 100 companies, as well as on-site surveys at another fifteen businesses that lasted several weeks each, into a shelf full of thick loose-leaf binders.
Wave concluded that the Star was so good and latent demand so strong that customers would clamor for the technology regardless of price. To the monopoly-minded executives of Xerox, this was familiar and gratifying territory. A machine that could be sold or leased for any price—it was the 914 all over again!
The SDD engineers therefore considered themselves free to create absolutely the best office system they could imagine. Blinded by the almost religious fervor that seizes software and hardware developers set loose in a boundless design space, they shoveled every sophisticated function they could contrive into the Star without giving a moment’s thought to the one real-world player whose opinion was critical: the buyer.
“The techies were given their head to make the best system they could,” Spinrad lamented years later. “There were no constraints of any substance put on us as to the cost of the product, as far as I remember. But what the Wave studies missed was that there were other things coming along they didn’t recognize.”
This was an understatement. In 1975, or even 1979, one might have argued that the Star’s technology would place it in a class by itself, that it would blow away every other office machine on the market on its way to becoming an instant de facto standard. In 1981 the same argument was dangerously presumptuous. For at a secret skunk works in Boca Raton, a couple of miles from where Xerox had held its spectacular company picnic, an IBM team had slapped together a machine that would annihilate the market for big, integrated office systems.
IBM launched its Chess machine, renamed simply the Personal Computer, in August 1981, a scant four months after the Star. Judged against the technology PARC had brought forth, it was a homely and feeble creature. Rather than bitmapped graphics and variable typefaces, its screen displayed only ASCII characters, glowing a hideous monochromatic green against a black background. Instead of a mouse, the PC had four arrow keys on the keyboard that laboriously moved the cursor, character by character and line by line. No icons, no desktop metaphor, no multitasking windows, no e-mail, no Ethernet. Forswearing the Star’s intuitive point-and-click operability, IBM forced its customers to master an abstruse lexicon of typed commands and cryptic responses developed by Microsoft, its software partner. Where the Star was a masterpiece of integrated reliability, the PC had a perverse tendency to crash at random (a character flaw it bequeathed to many subsequent generations of Microsoft Windows-driven machines).
But where the Star sold for $16,595-plus, the IBM PC sold for less than $5,000, all-inclusive. Where the Star’s operating system was closed, accessible for enhancement only to those to whom Xerox granted a coded key, the PC’s circuitry and microcode were wide open to anyone willing to hack a program for it—just like the Alto’s.
And it sold in the millions.
The introduction of the IBM PC changed the business computer market the way Hiroshima changed the world’s conception of battlefield weaponry. The PC demonstrated that the business user would gladly forego graphical bells and whistles and seamless system integration and would tolerate a large dose of flakiness in order to save on price. IBM proved correct everyone who had warned that the Star was too big, too complicated, too expensive—too good.
How serious a blow the IBM PC represented to large-scale systems like the Star would become increasingly evident over the next year, as the Star’s sales lagged behind Xerox’s projections and the PC’s surpassed IBM’s.
The Star was to enjoy just one more moment of triumph. In December 1981 the Japanese version was introduced at the Tokyo Data Show by Fuji Xerox, the company’s Japanese partner, with Liddle in attendance along with Bill English and his engineering partner Joe Becker, who had created the Japanese display system. The price was 4 million yen per workstation, or about $16,000. The acclaim was even greater than its domestic cousin had received in Chicago eight months earlier.
“The presentation aimed at the most dramatic effect; it seemed to disembowel the other exhibitors,” wrote the correspondent of the Japanese computer journal ASCII. “Comparing Star with Japanese computer makers’ office computers in the 4-million-yen class, their capabilities are as different as clouds from mud.”
Clouds from mud? Upon hearing the phrase from a Japanese interpreter reading the review aloud, Dave Liddle burst out with a delighted laugh. “Boy,” he said, “you can’t do any better than that!”
The exodus from SDD picked up steam after the Star introduction. One day Bob Belleville got a call from Steve Jobs, who badgered him into quitting by yelling, “Everything you’ve ever done in your life is terrible, so why don’t you come work for me?” Massaro and Liddle left in 1982 to found Metaphor Computer. Metaphor soon brought out a workstation that resembled the Star on the surface, but ran on a lightning-fast Motorola 68000 microprocessor, the same one that would soon turn up as the heart and brain of the Apple Macintosh. A dimly remembered name now, Metaphor experienced great success for more than five years, until it was bought out and its nameplate retired—by IBM.
As for the Star, it lived on for many years as the crowning glory of a niche market, a specialty product and a legend, somewhat like the fancy futuristic cars—time machines, so to speak—which Detroit auto makers manufacture for the big car shows but seldom actually get on the road. Instead of sales in the hundreds of thousands, as Xerox once dared anticipate, only about 30,000 Stars were ever ordered.
By 1989 the architecture that Liddle once foresaw lasting ten years was already a relic. That year Computer magazine published an article by several of the machine’s original designers entitled, “The Xerox Star: A Retrospective.” Among its features was a roll call of lessons the team had learned from bitter experience: “Pay attention to industry trends…Pay attention to what customers want…Know your competition.”
Blinded by their own technology, the Star’s designers had been almost entirely unaware of the coming revolution of cheap PCs—the equivalent of scaled-down Altos, as opposed to the scaled-up Star. They did not see it coming until the moment IBM announced its blockbuster. At that point it was too late.
“It was a disaster beyond words,” Belleville said later, “because the world was already different.”