What was happening was the personal computer. When Clark created Silicon Graphics, the computer world was a pyramid. At the top were people like him and his engineers, who played with the fastest machines. At the bottom was the personal computer. The PC had been created in Silicon Valley as a toy for hobbyists, a joke technology derided by the sort of hotshot who worked for Jim Clark. Now it had found a market for its services. Moore’s law, which stated that the price of computing power would fall by half every eighteen months, implied that the pyramid must collapse. The PC would soon be able to perform all the functions of a Silicon Graphics work station. Microsoft controlled the market for personal computers through its operating system, and so Microsoft would displace Silicon Graphics. Microsoft and Silicon Graphics sold shares in themselves to the public the same year, 1986. Silicon Graphics might have been stunningly successful, but Microsoft was taking over the world. “You could see a time when the PC would be able to do the sort of graphics that SGI machines did,” says Clark. “And SGI would be toast. Eventually, Microsoft would take over its business.”
The falling price of computing power was leading the computer into new markets. Moore’s law came with a social corollary: high-tech could not remain high-tech for long. You might be the smartest engineer in the Valley, and you might have built the most sophisticated computer, but it was only a matter of time before some schlepp with a PC wrote a program that let him do everything you could do, at a fraction of the cost. Technical vanity did not pay. If you wanted to make a great deal of money and acquire a great deal of power, you cultivated a more egalitarian outlook.
In the late 1980s and early 1990s people with a reputation for inventing the future spent a lot of time talking about where the trend might lead next. Jim Clark was one of these people. In December 1990 he joined a discussion at a conference called PC Outlook. One of the questions the panel considered was “Will personal computing and personal communications be combined, or will it just remain as science fiction?” In 1990 the idea that people would use their PCs to communicate with one another was outlandish. Yet it was still worth discussing, for if the computer ever did become a communications device it might transform not just the Valley but the economy. It would plug the masses into the thinking machine, and the thinking machine into the masses. At the conference Clark predicted that this would happen once the computer became fun to use. For the preceding two years he had argued that the new new thing was computer games, like Nintendo. He was wrong, but in an interesting way. He was groping toward a mass market. “Jim always was looking to democratize the technology,” says Dick Kramlich. “He was always thinking about how to make some high-end technology accessible to a larger audience. That’s what he’d done with the Geometry Engine. His chip cut the cost of real-time computer graphics from millions of dollars to tens of thousands.”
Clark saw only one solution: Silicon Graphics had to build a cheap computer to compete with the personal computer. Cheap machines meant mass markets, and mass markets meant great sums of money. In early 1987 he started arguing, in characteristically undiplomatic fashion, that SGI was doomed. “Jim was probably two or three years ahead of the rest of us in seeing what was coming,” Ed McCracken says. “He could see problems down the road and the problems become emotionally important to him.” “I was saying, ‘Goddamn we’re out of our minds,’” says Clark. “I was so worried about the PC. I was adamant that we had to build a low-end product, and that it had to be something that sold for under five grand.” It didn’t happen—largely because McCracken did not share Clark’s view. Fred Kittler, then an analyst with J. P. Morgan, recalls visiting Clark in his office at SGI in early 1990. “I was out here with a couple of analysts, and he was pacing back and forth like a man in a prison cell complaining about how his board wouldn’t let him create his cheap computer. It was clear by then he had no real power.”
Clark thought that Silicon Graphics had to “cannibalize” itself. For a technology company to succeed, he argued, it needed always to be looking to destroy itself. If it didn’t, someone else would. “It’s the hardest thing in business to do,” he would say. “Even creating a lower-cost product runs against the grain, because the low-cost products undercut the high-cost, more profitable products.” Everyone in a successful company, from the CEO on down, has a stake in whatever the company is currently selling. It does not naturally occur to anyone to find a way to undermine that product. Clark thought he knew how to become the agent of his own creative destruction, and he was prepared to do the deed. He wanted Silicon Graphics to operate in the same self-corrosive spirit.
More to the point, he wanted Ed McCracken to operate in this way. But Ed McCracken was not the man to roll the bones on the future—for which he could hardly be faulted, since as CEO he would get a lot more of the blame for any gamble that went wrong. Still, he manufactured a kind of contradiction in the heart of his company. All the good things that happened at Silicon Graphics happened because Clark had guessed that computer graphics had a commercial future. The company had been built entirely on Jim Clark’s foresight. But once it became a big company it had no room for Clark or his hunches. A big company—even a big company as highly charged as SGI—needs to believe its own internal propaganda: that its products are the best, that its technology will win, and so on. It has trouble entertaining the thought that it is doomed.
When it became clear to Clark that he could not force his own company to reinvent itself and lead this charge, he went into another dark hole of despair. He’d just married for the third time, and his new wife, the journalist Nancy Rutter, had started to complain about his behavior. He had to find something else to occupy his turbulent mind. He hired people to renovate his house. He bought a motorcycle and rode it too fast. He discovered model helicopters. At first, he was embarrassed at the mere possibility that someone would see him—it was such a typical technogeek thing to do. But they had these fantastically elaborate kits that let you assemble a machine that you could fly by remote control. So he would wait until the middle of weekdays, when the neighbors were all at work. Then he’d take out his helicopter kit and fly.
Soon enough Clark had become obsessed with his little choppers. He would put them together in the garage, then take them out and fly them around the neighborhood. He’d buzz other people’s homes. Nancy would come out to the garage and see this six-foot-three-inch, forty-seven-year-old man bent over the ground piecing together a toy designed for a small boy—at least in the beginning the choppers resembled children’s toys. Clark started with the small models; once he got the hang of it, he moved on to the bigger, more realistic machines. The bigger machines kicked up huge clouds of dust in the front yard, so naturally he built a landing pad in his driveway. He’d stand out there for hours in the middle of the day in the middle of the week, executing takeoffs and landings of these giant toy helicopters.
One day some workers he’d brought over to move a swimming pool, or build a hill, made the mistake of parking their cars near his landing pad. Clark went over to them and told them, in a slightly embarrassed tone, that they might want to move the cars, as he planned to fly his helicopters. The workers laughed, and went about their business. Clark seized the controls of his largest helicopter and sent it up, up, up over the neighborhood. And then something went terribly wrong. As he guided it down toward the landing pad, the controls ceased to function, and instead of gliding backward and down the helicopter came zooming straight for his head. Clark dropped the remote control and dived out of the way just in time. The helicopter whipped along the side of the workers’ cars with a horrible whack whack whack, as it dented the doors and peeled the paint.
Obviously, he couldn’t spend all of his time flying toy helicopters around his neighborhood. He wrote the workers a check for the two grand to cover the damages, and took up computer programming.
5
Inventing Jim Clark
When I asked Kittu Kolluri what he thought of his first encounter with Jim Clark, he tho
ught for a moment and then said, “So…” This was not unusual. When a computer programmers answers a question, he often begins with the word “so.”
“Why did you come to Silicon Valley?”
“So…I’m from this small town in Iowa…”
“So” cuts across the borders within the computing class just as “like” cuts across the borders within the class of adolescent girls. It’s the most distinctive verbal tic manufactured by the engineering mind. Silicon Valley engineers for whom English is a second or even third language acquire it as readily as native speakers. Nobody knows why. Some say that “so” imposes the semblance of logic on an essentially illogical event, human conversation. After all, “so” implies that the answer follows directly from the question. Others claim that “so” just buys you time to think.
“So,” said Kittu Kolluri, who learned to speak English where he learned to program computers, in Hyderabad, India. “Jim Clark walks into our office at Silicon Graphics. It is 1990. I have just joined the company. And I’m thinking: This could not be! The chairman of SGI has just walked into our office! He walked right by my cube! He’s standing right there next to Pavan! [Pavan is Pavan Nigam, the Indian man who hired Kittu to work at Silicon Graphics]. Jim Clark didn’t say anything; he was looking at our group. Finally, he turns to Pavan and says, ‘Pavan, we got to get some more Indians around here.’ I was so shocked. I am thinking: You cannot say such a thing in California. It must be against the law!”
Clark then asked to see what the Indian engineers were working on. Pavan and Kittu showed him. Silicon Graphics was still having problems making its machines easy for people to use. Clark could relate. Looking for some way to occupy himself, he had taken to programming at home on a Silicon Graphics work station. He was experiencing some of the same frustrations as SGI’s customers and was hungry for anything that made it easier for him to write programs that worked. And that, it turned out, was why he had paid his call on Pavan and Kittu. Clark was a physicist by training. When he returned to programming computers again, he wrote his code in the language of physics, mathematics. The Silicon Graphics machine required the programmer to translate his math into computer language. For example, if you wished to square a number (x2) in a program, you needed to write it as x**2. “Jim was speaking German, and the machine only knew Spanish,” said Pavan Nigam, “so there was this extra layer of thought required to turn Jim’s German into Spanish. This drove Jim crazy; he thought the machine should bloody well know German, since German was the language of first principles. Since he was the chairman, we would say, ‘Good idea, Jim! Good idea, Jim!’—and then we would talk to product development people, and they would say, ‘Why the hell do you want to do that? That is about number 300 on a list of 300 things we need to do right now.’”
Clark turned the young Indians into his private software tutors. The young Indians were both flattered and impressed. Two minutes with Clark and the young wizard always fell under his spell, as many engineers before them had done. They felt that Jim Clark understood their value as no other important person did. “Jim Clark is not one of these flakes who starts companies,” said Kittu. “He’s not some manager who doesn’t know what he’s talking about. When he first saw the tools, he knew exactly what they meant. He knew we were onto something.” But it was a few months before Kittu discovered why Clark had taken up computer programming, which seemed a rather odd activity for such an important man. Then one day Kittu’s phone rang. He picked it up, and a female voice on the other end of the line told him that Jim Clark was calling. Kittu felt a surge of excitement. He pictured Clark sitting at his fine desk in his fine office in the finest of the dozen red brick buildings on the Silicon Graphics campus. There was a long pause and then came Clark’s voice, distant and crackling. He sounded as if he was in Outer Mongolia.
“Kittu,” Clark said, “I have this bug I can’t find.” He described the problem he was having. Kittu could barely hear him, he sounded so distant.
“Jim,” he asked, “where the hell are you?”
Long pause. More crackling on the line. Then Clark’s deep voice, “I’m not sure.”
“What do you mean you’re not sure?” asked Kittu.
Nervous laughter. “Off the coast of New Zealand somewhere.”
“Jim, what are you doing off the coast of New Zealand?”
“I’m on my boat.” Clark had bought a second-hand sailboat. The boat came equipped with a personal computer, which Clark had chucked over the side, on principle. In its place he’d installed a Silicon Graphics work station.
“Jim, what are you doing programming a goddamn computer off the coast of New Zealand?”
More nervous laughter. “I’m programming the boat.”
And this was the first inkling that Kittu, and so also Pavan, had that all was not right with Jim Clark. Soon Kittu figured out that Clark called him from everywhere except his fine office in his fine building at Silicon Graphics. The man had stopped showing up for work. He spent all of his time writing a navigation program for the sailboat he had just bought. He was doing this, he once told them, because he had ideas about the future of high technology that no one would listen to. Teaching his boat to sail itself diverted his mind from the fact that others, not he, were about to inherit the earth. On his boat he could think about how to inherit the earth from its current heirs.
For about nine months Clark called Kittu and Pavan regularly to pester them about his software. Then he went silent.
About that time—late 1990, early 1991—the cardboard boxes in his guest room stuffed by his devoted secretary with his old papers suggested, his life took another unexpected turn. Clark’s motorcycle skidded out from under him as he rounded a bend near his home in Atherton. It crushed his leg against the street and laid him up in bed for six weeks. It gave him another reason to be angry with Ed McCracken. If McCracken would do what Clark told him to do, he, Clark, wouldn’t be forced to amuse himself by riding around on motorbikes on wet city streets in the middle of the week.
In bed waiting for his leg to heal and with nothing better to do but curse McCracken, he wrote a paper. It summarized his thinking of the past few years, as he groped for a solution to what he viewed as Silicon Graphics’ inevitable doom. He was right about the future of the company. Through its monopoly of the operating system Microsoft already controlled the personal computer. Microsoft would one day overrun the high end of computing where Silicon Graphics made its money. Microsoft had made it clear that the only way to preserve your station in Valley life was to create a monopoly. If you created a monopoly, you were at least partially exempt from the ordinary rapid cycle of creation and destruction. In computing, a monopoly took the form of a toll booth. Bill Gates had his toll booth, the PC operating system. Jim Clark wanted his own toll booth.
In 1991 people who had computers on their desks used them mainly for financial analysis and word processing, but that was sure to change as the price of computer memory fell. For the moment, though, the reach of the personal computer was nothing like that of the television set. About 95 percent of American households owned at least one television set; only one in ten owned a computer. To Jim Clark this looked like a huge opportunity: turn the television set into a computer. He finished his paper in bed and called it “The Telecomputer.”
Technically, the telecomputer was feasible. The United States had a tantalizing new infrastructure into which such a device might plug. In the preceding decade the cable television industry had laid pipes leading into 75 percent of all American homes, and those pipes could carry information into and out of a telecomputer. The question was: Could you make a telecomputer cheaply, and persuade ordinary people to buy it? If so, you could control an exciting new chunk of the American economy. The computer had an important trait that the ordinary television did not: it could interact with the viewer. You could tell it what you wanted, and it could go out and fetch it for you. You could shop through it. You could send instant messages through it, and receive
messages back. You could order up local news from anywhere in the world. You could order any movie you wanted, when you wanted it, and pause it when you went to the toilet. The machine could be the conduit for all information into and out of the home.
The general idea hardly originated with Clark. Back in the late 1970s the corporate ancestor of Time Warner, called Warner-Amex, had created a pilot for an interactive television in Columbus, Ohio, called Qube. Qube was supposed to let cable TV viewers send messages to the networks, and request the programs they wanted to see, via remote control. But the machine didn’t work particularly well, and it cost a fortune to build. The company spent $30 million trying to build it, and then quit. Then there was an idea, called Videotext, cooked up by researchers at the British post office, also in the 1970s. It hooked a black box between the telephone line and the television set and enabled people to send messages to each other while they watched the BBC. Nobody wanted that either. A few years later AT&T and Knight Ridder came together to deliver the newspaper each morning to the masses on their television set. The masses yawned and went back to bed.
In every case, at least a part of the problem was that the wires entering and exiting the average home were unable to transmit data with sufficient speed. The wires could handle printed text, which required much less “bandwidth,” but they could not send moving pictures. Moving pictures contained a lot of digital data, and so required vast computing power. Never mind. The engineering problem was so interesting that it seemed rude to spoil it by asking who, exactly, wanted to read text on his television.
The New New Thing: A Silicon Valley Story Page 7