The Soul of a New Machine

by Tracy Kidder

  Much of the engineering of computers takes place in silence, while engineers pace in hallways or sit alone and gaze at blank pages. Alsing favored the porch and staring out at trees. When writing code, he said, he often felt that he was playing an intense game of chess with a worthy opponent. He went on:

  "Writing microcode is like nothing else in my life. For days there's nothing coming out. The empty yellow pad sits there in front of me, reminding me of my inadequacy. Finally, it starts to come. I feel good. That feeds it, and finally I get into a mental state where I'm a microcode-writing machine. It's like being in Adventure. Adventure's a completely bogus world, but when you're there, you're there.

  "You have to understand the problem thoroughly and you have to have thought of all the myriad ways in which you can put your microverbs together. You have a hundred L-shaped blocks to build a building. You take all the pieces, put them together, pull them apart, put them together. After a while, you're like a kid on a jungle gym. There are all these constructs in your mind and you can swing from one to the other with ease.

  "I've done this in short intervals for a short period each year. There's low intensity before it and a letdown at the end. There's a big section where you come down off it, and sometimes you do it awkwardly and feel a little strange, wobbly and tired, and you want to say to your friends, 'Hey, I'm back.'"

  If simplicity is elegance, then Alsing's code for the Eclipse was a kludge. It was full of tricks and subtleties — but for a purpose: he had at his disposal a severely limited storage space. And his concoction worked well, in the end. It takes no imagination to see, however, that for someone trying to manage the invention of a computer, Alsing's style could provoke nightmares. Afterward, when the first Eclipse was sent to market, Alsing felt proud. "I did a damn good job," he told himself. He received what he still felt was a touch of glory. He was asked to attend the conference at which the company's salesmen were briefed about the new machine. Grinning, Alsing remembered one of the executives saying to the salesmen, "Okay, here's your brochures. You can jack off to them on the plane home." Alsing was asked to say a few words. The salesmen applauded. "That was pretty rich stuff to me, coming out of the cellar and receiving all that fame and attention." But his habit of putting off the creation of his code until all was almost lost probably did hurt him. During his months of procrastination, he wondered why he couldn't get down to work and came to believe that he was simply lazy, and he guessed that he let it show in little ways — an averting of the eyes, a slump in the shoulders. And partly because of that, perhaps, and certainly because of the anxiety he created, he did not receive a substantial gift of stock for his work on the Eclipse, although some other engineers got tidy rewards.

  One time West happened to mention Alsing's name to an executive upstairs, and the executive said, "You know, it's funny, but when you think about it, Alsing's written just about every line of microcode that's come out of Data General." Alsing did make himself easy to overlook. By his own account, he deliberately obscured himself in West's shadow. He went to work for West, he said, after West took over the Eclipse Group, largely because he felt that it would be safer to stand on West's side than not. West, almost overnight, had become formidable. For Alsing, West represented a buffer and a shield. He had some other reasons, though, for signing up to help West on the Eagle project.

  Kludge made Alsing imagine a wheel built out of bricks, with wooden wedges in between them; such a thing might work, but no sane engineer would be proud to have designed it. Alsing tended to agree with those who maintained that Eagle must, by definition, be something of a kludge. He did believe it when West said that Eagle could put a lot of money on Data General's bottom line, but past experience made Alsing doubt that he would ever see any of the loot. He even thought sometimes that if he went on helping to pound out 16-bit Eclipses and never worked on another big new machine, he would be a little bored but probably content. Nevertheless, Alsing stood with West from the start of Eagle, and he never asked for coaxing.

  "West's not a technical genius. He's perfect for making it all work. He's gotta move forward. He doesn't put off the tough problem, the way I do. He's fearless, he's a great politician, he's arbitrary, sometimes he's ruthless." Alsing laced his hands to gether so that his knuckles made an arch, on which he rested his chin. Why was West so obsessed about this machine? How would it all end? Sometimes, Alsing said, he felt he had joined up mainly to find out. He went on: "I screamed and hollered over NAND gates and microinstructions with the first Eclipse, but I'm too old to feel that way about computers now. This would be crashingly dull if I was doing it for someone else. West is interesting. He's the main reason why I do what I do."

  Looking around the basement, some of the team's brand-new engineers would sometimes wonder what would happen to them when they turned thirty. Being young, they could make light of the question, and say, as one did, "When a computer engineer gets old, he gets turned out to pasture or else made into dog food." Data General was a young company, and so its engineers tended to be young. There really was such a thing in the world as a practicing middle-aged computer engineer. It did appear, however — management handbooks say so — that many engineers experience a change of life when they reach the age of thirty or so.

  Among engineers generally, the most common form of ambition — the one made most socially acceptable — has been the desire to become a manager. If you don't become one by a certain age, then in the eyes of many of your peers you become a failure. Among computer engineers, I think, the wish to manage must be a virtual instinct. The industry's short product cycles lend to many projects an atmosphere of crisis, so that computer engineering, which is arduous enough in itself, often becomes intense. The hours are long. Emotions get taxed. Moreover, the technology of computers changes constantly; every year it's a struggle to keep up with the youngsters fresh out of school. What another of West's old hands called "a long-term tiredness" can easily creep over computer engineers in their thirties.

  From the start of Eagle, Alsing disengaged himself from much of the technical work on the machine. He was running the Microteam, but from a little distance. Eagle would contain more code than any Data General machine before it — as much code as Alsing had written in his entire career. Alsing could not write all of it, even if there were time. He simply could not generate the excitement he used to feel about gates and bits. Moreover, he believed that since he could not write all of the code, then he couldn't write any of it. These new kids, he saw, approached the job in a way he never had. They worked steadily, day after day, night after night. That was fortunate, for the sake of the team. Alsing admired their discipline. He believed that it exceeded his by far. So he left the writing of the code to half a dozen new recruits, and most of the supervision of their work to submanagers.

  Sometimes Alsing worried about his detachment. "Although I sometimes say I don't care too much this time around, if I were to lose this — if I were to be fired or transferred to another project more mundane — I would be, uh, very unhappy. Maybe I'm starting to take this place for granted," he said once.

  For a time, when he was still in college, Alsing had wanted to become a psychologist. He adopted that sort of role now. Although he did keep track of his team's technical progress, he acted most visibly as the social director of the Microteam, and often of the entire Eclipse Group. Fairly early in the project, Chuck Holland had complained, "Alsing's hard to be a manager for, because he goes around you a lot and ten's your people to do something else." But Holland also conceded: "The good thing about him is that you can go and talk to him. He's more of a regular guy than most managers."

  Alsing created the Microteam. He chose its members and he gave them their first training, with some help from Rosemarie Seale. Nowadays it takes a computer to build a new computer, especially when it comes to writing microcode for one. Alsing figured that before the Microkids did anything else, they must learn how to manipulate Trixie. He didn't want simply to give them a stack of
manuals and say, "Figure it out." So he made up a game. As the Microkids arrived, in ones and twos, during the summer of 1978, he told each of them to figure how to write a certain kind of program in Trixie's assembly language. This program must fetch and print out the contents of a certain file, stored inside the computer. "So they learned the way around the system and they were very pleased," said Alsing. "But when they came to the file finally, they found that access to it was denied them."

  The file in question lay open only to people endowed with what were called "superuser privileges." Alsing had expected the recruits to learn how to find this file and, in the process, to master the system. He was equally interested in seeing what they would do when they found they couldn't get the file.

  One after the other, they came to him and said, "I almost have it."

  "Okay," said Alsing, "but you don't have it."

  In the end, most Microkids went to Rosemarie. Alsing had conferred with her beforehand. She was to help the Microkids find the file, if they asked. They learned something, Alsing felt. "If a person knows how to get the right secretary, he can get everything. It was a resourceful solution — one of the solutions I hoped they'd find."

  This first game led to others. Not long after the recruits arrived, the "Tube Wars" began. As a rule, it was the kids against Alsing. In one commonly used gambit, a Microkid would sit down at a terminal and order Trixie to open up Alsing's files. The Microkid would then move the files to a new location. Returning from coffee or lunch, Alsing would find his files gone. He'd hear tittering from the cubicles nearby. And he would know he'd been "tube- warred."

  "What did you do to me?" he'd cry.

  "Find out, stupid," a voice would answer.

  The Microkids weren't the only ones playing games with the computers in the basement. A young woman worked for Rosemarie. She was unmarried and, by general consensus, goodlooking. Every day for a couple of weeks during the Eagle project, she was "assaulted" at her desk. She would be doing her electronic paperwork when suddenly everything would go haywire, all her labor would be spoiled, and on the screen of her cathode-ray tube would appear cold, lascivious suggestions. "Whoever was doing it," said West, had "the mentality of an assassin."

  West put Alsing on the case. Alsing had some members of the team lay traps inside the computer system — traps designed to leave a trail back to the masher's terminal. But the masher spotted all of these; one time he made his escape by bringing to an abrupt halt the entire system on which most of the engineering departments relied. He had to be stopped, and eventually Alsing found a strong suspect, a young man outside the group. Alsing had a casual chat with him about all the marvelous tricks that could be played with the in-house computing system; afterward, the obscene messages ceased. Wholesomeness, in this regard, returned to the basement. Indeed, said one young engineer, the place seemed antiseptic.

  The masher's game had been especially nasty and unfair, Alsing pointed out, because the victim could not fight back. But Tube Wars pitted worthy adversaries against each other. The jousting did no harm, and, on the contrary, released tension. One day Alsing came back from lunch and went to work at his terminal. Everything looked right, all his files seemed to be in place — until he tried to do something with them. Then, to his surprise, he found that all of them were vacant. "It was like opening a filing cabinet and finding all the folders empty. They were dummy files. It took me an hour to find the real ones. So now I can never be sure, when I log on the system, that what I see is real."

  Alsing struck back. He created an encrypted file and tantalized the team, "There's erotic writing in there and if you can find it, you can read it." They tried, and ultimately all gave up, including Bob Beauchamp. Alsing taunted Beauchamp, though. So Beauchamp tried again. This time he wrote a program that broke Alsing's encryption system. "He beat me," Alsing said. "But I think he was too much of a gentleman to read what was inside."

  Alsing double-encrypted the secrets in his files after that, and for many months he assumed they were safe. Beauchamp abandoned his first approach, feeling that it was a little crude. Now he made a slight revision in Trixie's operating system. In essence, he instructed the machine that whenever Alsing encrypted a message, the operating system should send to Beauchamp's files an unencrypted copy. This was the ultimate victory in Tube Wars, not least of all because Alsing never learned what Beauchamp had done to him until Beauchamp himself spilled the beans.

  Tube Wars died out slowly. At their height, whenever I visited Alsing, I'd take a look at the screen of his cathode-ray tube and "almost always see something peculiar written on it, some message or picture sent to him by a young engineer at play. I'd come into Alsing's cubicle and there on the screen would be a picture of a fist with the middle finger extended; or there'd be a little story on his screen:

  SEX LIFE OF AN ELECTRICIAN (PART 3) FULLY EXCITED MILLIE AMP MUMBLED OHM! OHM! OHM!...

  Alsing arranged several sorts of social gatherings, among them a weekly meeting of the Microteam held around a table in a barren little conference room that contained one tiny window. I went to a couple of these convocations. Alsing would call the assembly to order, read a few announcements, and then submit to teasing.

  "I read about someone who did a study of his company and discovered that he was the least important employee. So he quit," Alsing said.

  "So long, Carl!" cried one of the Microkids.

  There was a lot of talk about Alsing's idea that they should hand out Honorable Member of the Microcode Group Awards.

  "I think we oughta give one to West," said one of the team. "So that when we get pissed off at him, we can take it back."

  "No," said another, "because now he'll be a member of the Microteam. He can solve his own problems." They did do some business, using the beautiful and, to me, inscrutable language of the microelectronic era: , default redix, floating-point mantissas, swapbites, sys log, sim dot, scratch pad.

  "The scratch pad doesn't come alive on CPD until one-sixteen," said one.

  "That means all the stack tests go away!" cried another.

  "That's right."

  There was laughter all around the table.

  When it died away, a Microkid said, "Look, we can speed up Eagle's stack stuff by putting in a scratch pad."

  "Can we," asked another, "also plant a little bomb in the thing?"

  Alsing sat with his hands folded, smiling subtly. He looked like a big contented cat. No one actually called an end to the meeting. It simply petered out in laughter.

  And yet there was bad feeling among some of them, much of it directed toward Alsing; and even then, petty intrigue was in progress. Some of the team would eventually describe the weekly micromeetings as "Alsing's weekly No-Op" — No-Op being the name of an assembly-language instruction that accomplishes nothing. For a long time, however, almost all of the recruits enjoyed these meetings, the Tube Wars and other entertainments that Alsing arranged. He made a point of sharing lunchtime with some of them several days a week. And they appreciated Alsing's friendliness; they could always talk to him.

  Alsing believed that the team's managers, in handling the new recruits, really were practicing what was called "the mushroom theory of management." It was an old expression, used in many other comers of corporate America. The Eclipse Group's managers denned it as follows: "Put 'em in the dark, feed 'em shit, and watch 'em grow." It was a joke with substance, Alsing felt; and he believed that their mushroom management needed an occasional antidote. Alsing in effect had signed up to provide the kids with some relief from their toil. West warned him several times, 'If you get too close to the people who work for you, Alsing, you're gonna get burned. But West didn't interfere, and he soon stopped issuing warnings.

  One evening, while alone with West in West's office, Alsing said: "Tom, the kids think you're an ogre. You don't even say hello to them."

  West smiled and replied, "You're doing fine, Alsing."

  FLYING UPSIDE DOWN

  West often said that th
ey were playing a game, called getting a machine out the door of Data General with their names on it. What were the rules?

  "There's a thing you learn at Data General, if you work here for any period of time," said West's lieutenant of hardware, Ed Rasala. "That nothing ever happens unless you push it." To at least some people upstairs, this condition took the name "competition for resources." As a strategy of management, it has a long lineage. "Throw down a challenge," writes Dale Carnegie in that venerable bible of stratagems dressed up as homilies, How to Win Friends and Influence People.

  In a sense, the competition between Eagle and North Carolina was institutionalized; each project lay in the domain of a different vice president. But that may have been accidental. West's boss, who was the vice president of engineering, Carl Carman, remarked that he had worked at IBM and that compared to competition among divisions there, rivalry among engineering teams at Data General resembled "Sunday school." Moreover, Carman said, in a company with a "mature product line" like Data General's, situations naturally occur in which not enough large new computers are needed for every team of computer builders to put one of its own out the door "And yeah," Carman continued, "the competition is fostered." He said that de Castro liked to see a little competition stirred up among teams. Let them compete with their ideas for new products, and bad ideas, as well as the negative points of good ones, are likely to get identified inside the company and not out in the marketplace. That was the general strategy, Carman said. What it now meant downstairs, to the Eclipse Group, was that they not only had to invent their new computer but also had to struggle for the resources to build it. Resources meant, among other things, the active cooperation of such so- called support groups as Software. You had to persuade such groups that your idea had merit and would get out the door, or else you wouldn't get much help — and then your machine almost certainly wouldn't get out the door.