The Soul of a New Machine

by Tracy Kidder

  Some of the engineers closest to West suspected that if he weren't given a crisis to deal with once in a while, he would create one. To them he seemed so confident and happy in an emergency. But as for this big crisis in the little world of Westborough engineering, although West had made it his own, no one could say he had invented it.

  Why had Data General failed to produce a rival to VAX? When trade journalists asked this question, Data General put on a bold face of course, and suggested in essence that all was proceeding according to plan. In fact, years before the appearance of VAX, Data General engineers had foreseen the advent of such machines. For about five years they had been trying to produce one of their own. But there had been problems. They had made some false starts, and the engineers involved had been arguing over who was going to produce this new machine and what it would be.

  Some computer engineers harbor strong feelings toward their new designs, like Cossacks toward their horses. Carl Alsing, a veteran engineer and one of West's cadre, told the fable of an engineer, who, upon being informed that his plans for a new machine had been scrapped by the managers of his company, got a gun and murdered a colleague whose design had been accepted. Alsing said he thought that such a murder really happened but that a woman was probably involved — yet it came, he said, to much the same thing.

  The history of what West and some of his associates called their "wars" began in the mid-seventies. Data General had gotten over its manufacturing problems; the fuss over the Keronix affair had long since died down and the company had been growing apace. They had followed up on the instantaneous success of their first CPU, the NOVA, by producing a whole line of compatible NOVAs, and while those computers kept selling and selling, they had created another and generally more powerful line of machines, called the Eclipse. Like the NOVA, it was a hit The Eclipse line was just beginning to grow, however, when the leader of the Eclipse Group began to leave his team. He went away to invent the next kind of Data General computer. As it developed, this became a grand project Among other things, it would solve one of the important technical questions then on the horizon. This problem lay in the question of how best to enlarge the minicomputer's "logical-address space." This was the problem that superminis like VAX would address, by achieving what one wag called "thirty-two-bit-hood."

  Computers, it is often said, manipulate symbols. They don't deal with numbers directly, but with symbols that can represent not only numbers but also words and pictures. Inside the circuits of the digital computer these symbols exist in electrical form, and there are just two basic symbols — a high voltage and a low volt age. Clearly, this is a marvelous kind of symbolism for a machine; the circuits don't have to distinguish between nine different shades of gray but only between black and white, or, in electrical terms, between high and low voltages.

  Computer engineers call a single high or low voltage a bit, and it symbolizes one piece of information. One bit can't symbolize much; it has only two possible states, so it can, for instance, be used to stand for only two integers. Put many bits in a row, however, and the number of things that can be represented increases exponentially. By way of analogy, think of telephone numbers. Using only four digits, the phone company could make up enough unique numbers to give one to everybody in a small town. But what if the company wants to give everyone in a large region a unique phone number? By using seven instead of four digits, Ma Bell can generate a vast array of unique numbers, enough so that everyone in the New York metropolitan area or in the state of Montana can have one of his own.

  Inside certain crucial parts of a typical modern computer, the bits — the electrical symbols — are handled in packets. Like phone numbers the packets are of a standard size. IBM's machines have traditionally handled information in packages 32 bits long. Data General's NOVA and most minicomputers after it, including the Eclipses, deal with packages only 16 bits long. The distinction is inconsequential in theory, since any computer is hypothetically capable of doing what any other computer may do. But the ease and speed with which different computers can be made to perform the same piece of work vary widely, and in general a machine that handles symbols in chunks of 32 bits runs faster, and for some purposes — usually large ones — it is easier to program than a machine that handles only 16 bits at a time.

  In this case, the main issue was the computer's storage system. Here, in packages of symbolic bits, are kept both information for the computer to manipulate and also many of the instructions that tell the computer what to do with that data. The situation resembles that of a region's telephone system; phone are of no use unless they are distinct one from the other, and an item in a computer's storage system is of no use unless it can easily be found. And the general solution resembles the phone company's; each compartment in the computer's storage has its own "phone number," its own unique symbol, known as an address. A 16-bit machine can directly generate symbolic addresses only 16 bits long, which means that it can hand out to storage compartments only about 65,000 unique addresses. A true 32-bit machine, however, can directly address some 4.3 billion storage compartments.

  Some of Data General's old customers and many potential new ones needed, or soon would, the large "logical-address space" of a 32-bit machine. Although other customers had no such needs as yet, a general feeling held that 32-bithood would become a de facto standard in the industry. You had to produce a 32-bit machine.

  It was 1976. By degrees West had taken command of the Eclipse Group. He and his small team were, as he put it, pounding out new 16-bit Eclipses. Meanwhile, the group's former leader and another team of engineers were working on the monumental new machine, which would solve the logical-address problem, among others. The monumental machine had taken on the code name FHP, an abbreviation for "the Fountainhead Project" The team designing FHP repaired to a suite in the Fountainhead Apartments — a local edifice that lends a touch of Miami Beach to the town of Westborough — in order to pursue their work. West's Eclipse Group carried on at headquarters, extending the successful line of Eclipses. And all might have proceeded this way, in relative harmony, but for politics.

  Data General had built a new research facility in a place called Research Triangle Park, in North Carolina, a state that had made itself comely to industry, partly by keeping taxes low. While lauding the government of North Carolina, Data General's spokesmen rather bitterly denounced Massachusetts, where taxes of all sorts run high. Edson de Castro himself joined in the criticism. One company spokesman went so far as to lament the fact that Data General had chosen to grow up in the Bay State. In reporting such comments, none of the Boston papers bothered to point out that the many universities and institutes of technology, existing tax free in Massachusetts, had produced much of the technology and many of the technologists that had made a Data General possible in the first place. But Data General's front office wasn't complaining without reason. In Massachusetts, as elsewhere, 1976 happened to be an election year, and the company was throwing its weight against several propositions on the ballot that threatened to increase both Data General's operating expenses and the personal income taxes of its better-paid employees.

  As it turned out, most of the company's favored causes won. No doubt Data General's campaign had an influence. The opening of the facility in Research Triangle Park, North Carolina, and the political campaign, however, had some unfortunate side effects on some of the company's engineers.

  For one thing, word had come down that the FHP project was being transferred to North Carolina. Some of the engineers who had been working on that grand new machine refused to pack up their families and go south. At least some of them felt robbed. "You gotta understand," said West later. "FHP was the one thing in the world they wanted to do most, the biggest-thing-the-world's-ever-seen kind of thing. Somebody told those guys that they would have seventy-two uninterrupted hours with the girl of their dreams. The thing they most wanted to do was dangled before them and then pulled away. And some people were pissed."

  Then th
ere was the newspaper story. One morning, after FHP's departure had been announced, a number of the engineers at Westborough, who felt of course that they were good, productive, "can-do" engineers, picked up the Boston Globe and saw in it an article about themselves. It read in part:

  Speaking to the Boston Security Analysts Society, de Castro said his company, the second largest minicomputer producer in the world, was finding it "a great deal easier to staff [its research-and-development center] in North Carolina. People are more willing to move into that area than Boston."

  ... In addition to this 20 percent cost-of-living difference, which is a combination of taxes, insurance, housing, food and other costs, de Castro said the Research Triangle Park area has "a different feel" than the research facility the company maintains in Westborough. "The ambition level is different There is a can-do attitude and that environment is contagious."

  West remembered the aftermath, laughing low and shaking his head: "De Castro called us all together and in his inimitable fashion totally confused everyone. He said that the press distorts things and I hope you know that even if I believed that quote I wouldn't say it. Then he gave us the march to victory speech and left." West added, "Morale hit an all-time low at Westborough."

  Various accounts of how matters stood among the engineers there suggest that many believed Westborough's days as an interesting place to work were about over. Sure, said West, people there could continue to build NOVAs and Eclipses. "But what fun is that?" Some of those who were going south with FHP boasted that they were going to the place where the action would be. And indeed it looked as though the important new machines, the sorts of projects that a significant number of Westborough's engineers wanted to work on, would thereafter be pursued in North Carolina.

  Some of the engineers who had chosen New England over FHP fell under West's command, more or less. And the leader of the FHP project suggested that those staying behind make a small machine that would solve the 32-bit, logical-address problem and would at the same time exhibit a trait called "software compatibility."

  Some of those who stayed behind felt determined to build something elegant. They designed a computer equipped with something called a mode bit. They planned to build, in essence, two different machines in one box. One would be a regular old 16-bit Eclipse, but flip the switch, so to speak, and the machine would turn into its alter ego, into a hot rod — a fast, goodlooking 32-bit computer. West felt that the designers were out to "kill North Carolina," and there wasn't much question but that he was right, at least in some cases. Those who worked on the design called this new machine EGO. The individual initials respectively stood one step back in the alphabet from the initials FHP, just as in the movie 2001 the name of the computer that goes berserk — HAL — plays against the initials IBM. The name, EGO, also meant what it said.

  The people working on EGO, nominally working for West but in fact amenable then to no one's control, truly labored. They worked nights. They worked weekends. They argued hotly with each other. "It was the most incredible, soaring experience of my life," said one of them later. And they worked with astonishing speed. Within two months they had a fairly complete specification. Then they took it to de Castro.

  To a disinterested observer it might have seemed obvious that Data General wasn't going to field both EGO and North Carolina's machine. The costs of supporting the two radically new CPUs would be prohibitive. Data General could sensibly afford only one major new machine, and Data General is almost always sensible about money. According to West, and others, de Castro told him and his troops to work out their differences with North Carolina. They didn't. To some of the engineers who worked on EGO, what ensued was a "war," and the first open battle, which was fought at a Howard Johnson's motor inn down south, was "the big shoot-out at HoJo's." Carl Alsing, who was not a participant but an interested observer, said: "I imagine the great EGO wars as a pen-and-ink drawing. Snarling engineers are shown hurling complexities at each other."

  Which processor was better? Which deserved to receive the support and resources of the company? Those were the stakes.

  Fights like these often take place inside computer companies. So it is said. Here the winner was foreordained. When it came to rating the talents of the engineers, everyone had to agree that some of the company's finest stood behind EGO, but North Carolina's leader was generally acknowledged to be the company's stellar designer after de Castro. Moreover, the company had made a substantial investment in North Carolina as a research-and-development facility, and Data General is a company that likes to see its investments pay off without undue delay. "EGO was a group of five people, FHP had fifty. You're not gonna kill a group of fifty. The company's not gonna send FHP to North Carolina and then do EGO," said West. But that was later on. At this time he argued strongly for EGO. At a meeting before de Castro in September 1977, the two sides traded promises. Westborough would do EGO in a year. Well, then, North Carolina would do a version of FHP in a year. West remembered: "De Castro looked around the room and said, 'It's a dilemma.' It's a famous quote for him, never heard before or since. I said, 'Okay, we'll stop building EGO,' and de Castro walked out of the room."

  West told himself that his team hadn't lost; they'd merely retreated from a battle they couldn't win. Thinking of the promises North Carolina had made, West explained later on: 'They'd signed up to do the impossible. We weren't signed up to do anything. Right then, that looked like a pretty good position to be in." But EGO's designers took it hard.

  Rosemarie Seale, the Eclipse Group's secretary, watched the proceedings from afar. "Engineering is a man's world. I don't know how much territorial interests have to do with it, but they're all fighting for that piece of the pie. There are some who can't admit it, but all of them are," she said. She was sad for her troops. She had made special arrangements to get the document describing EGO nicely typed up. "They wanted it to look really good for de Castro." She had wished them luck when they had gone upstairs with it. Now she watched them trudge back down the hallway. "I knew the minute they hit the department Such depression. It was terrible, absolutely terrible. Ed de Castro didn't want them to do it." Certain of the engineers now entered what West called "the first off-the-wall period." A few quit. Others went on vacation immediately. Still others spent the next couple of weeks playing a game called Adventure, in which you travel by computer into an underground world, wandering through strange, awful labyrinths, searching for treasure that's guarded and sometimes snatched away by dragons, dwarfs, trolls and a rapacious pirate who mutters: "Har. Har."

  The period after EGO got killed was remembered as West's speechifying period. "Speeches of rising and falling expectations," one engineer called West's orations. West told his group that from now on they would not be engaged in anything like research and development but in work that was I percent R and 99 percent D; they'd pound out Eclipses and put money oil the bottom line and that was that and if they wanted to build "sexy" machines that "the technology bigots" would like, then they'd have to look elsewhere. Then West turned around and said that although they did not have the charter to build a new 32-bit machine, they could still have some fun and a challenge; they'd create a 16-bit minicomputer faster by a factor of two, or maybe even four, than any the world had ever seen. The name of this project was Victor, the Mature Eclipse. "Victor was a canard," said West. But it gave his people something to do.

  Dreams of EGO died hard. According to West, some software engineers at Westborough had felt despondent about EGO's demise; they faced the prospect of having no major new CPUs to write system software for. They liked EGO. Thus bolstered, EGO's creators got West to try to attempt an EGO revival. West got the impression from the vice president of engineering that EGO might win de Castro's approval this time. But it didn't Made Castro basically said, 'Do something to extend the addressing capability of Eclipses, but don't use a mode bit,'" West remembered. West made more speeches of rising and falling expectations. The second off-the-wall period hit their corner
of the basement. Privately, West felt very angry.

  "No mode bit." It seemed to him that de Castro was asking the Eclipse Group to work with only one hand. West calmed down directly, though. "De Castro's always very vague," he reasoned, "but when he says something, you don't figure he's just shooting off his lip. You find out what's wrong with mode bits." West asked around. He came to the conclusion that, once employed, mode bits tend to proliferate in a company's product line, leading toward unnecessary development costs. "You get hung up in your own underwear." But de Castro had said only what he didn't want. What did he want them to do?

  When FHP had gone south there had been some talk of "a 32 bit Eclipse." Some time before that, a person in Marketing had asked some members of the Eclipse Group if they couldn't just tweak a few things inside an Eclipse and give it the extended logical-address space of a 32-bit machine. Back then, they had brushed off the suggestion; it wasn't that simple. Now West virtually camped out in this executive's office. He wanted to know what it was that customers most desired. VAX had by now gone to market, with all the usual hoopla, and was selling briskly. Other minicomputer companies had entered 32-bit machines in the supermini sweepstakes. And a lot of Data General's old customers wanted one, but it seemed clear that they would also want a machine that displayed the quality called software compatibility.

  An old-fashioned automobile that can be started only with a crank requires a person to make a series of adjustments more or less directly to the engine; in a modern car, of course, you need only turn a key and a system of electrical and mechanical devices does the rest. In the modern computer, software has developed in such a way as to fill this role of go-between. On one end you have the so-called end user who wants to be able to order up a piece of long division, say, simply by supplying two numbers to the machine and ordering it to divide them. At the other end stands the actual computer, which for all its complexity is something of a brute. It can perform only several hundred basic operations, and long division may not be one of them. The machine may have to be instructed to perform a sequence of several of its basic operations in order to accomplish a piece of long division. Software — a series of what are known as programs — translates the end user's wish into specific, functional commands for the machine.