The Soul of a New Machine

by Tracy Kidder

  It was a game in which new hands were always being dealt, a little like poker perhaps. West and his staff had created the deadline of April and, in the act, had agreed at least to pretend to take it seriously. Many months later, Carl Carman would say that no one upstairs believed they would finish Eagle that soon. Some evenings downstairs, West seemed to say the same thing. "We're gonna finish this sucker by April, Alsing," he'd say.

  "Yeah, Tom. Sure we are," Alsing would reply.

  They'd smile at each other.

  Sometimes, however, when, for instance, Alsing came in and told West that the Microteam would probably miss some intermediate deadline, West would say, "Come on, Alsing, this schedule's real." Now, on an evening early in 1979, Carman brought West a piece of momentous news: North Carolina was going to miss their own deadline by a huge margin.

  What did this mean? That the game of internal competition had essentially ended and the game of outlandish scheduling had begun in earnest? West had always maintained that Eagle was crucial to the company. Well, events were proving him right. But that was no cause for celebration. Meeting that preposterous deadline of April, when Eagle should be free of all bugs, was no longer merely desirable or just a matter of pride: it had become a corporate necessity.

  At least that was the message West thought Carman gave him — the message West decided to receive, anyway. West sat in his office brooding about it for a long time.

  Depending on who you were, you saw West's narrow office either as the lair of an ogre or as a haven with a door behind which you could talk privately. And if it meant the latter, it was, depending on the day of the week and the time of the day, either a place to talk nervously about impending disasters or else to use as a kind of lounge. Steve Wallach, Carl Alsing and Ed Rasala looked forward to their collective weekly meetings with West. He held them in his office on Fridays at 3:00 p.m. They'd do some business, West would tell them the latest company gossip, and, like Alsing at his own meetings with subordinates, West might submit to what Alsing called "zingers from his troops." "We could be in a lot of trouble here," West might say, referring to some current problem. And Wallach or Rasala or Alsing would reply, "You mean you could be in a lot of trouble, right, Tom?" It was Friday, they were going home soon, and relaxing, they could half forget that they would be coming back to work tomorrow.

  In the mornings, if West took Alsing into his office, it was usually to ask such questions as: "Is this real, Alsing?" "Is this gonna work, Alsing?" Alsing called these episodes "time out to worry." In the evenings, however, West could ask Alsing those same questions with a wry smile, as if to say, "Weird trip we're on, eh, Alsing?" and the question of whether this project was "real" would usually lead to a little mutual rumination on the meaning of the word itself.

  Alsing sometimes encouraged members of the Microteam to visit West's office after six; they would enjoy it, he promised. None seems to have taken the advice, but Alsing was right. At that transitional evening hour, before hurrying away toward his farmhouse, West would leave his door ajar, like an invitation, and leaning back, his hands fallen still, he would entertain most any visitor.

  But on the evening after West heard the news about North Carolina, nightfall brought no relief. The game of what's-the-earliest-date-by-which-you-can't-prove-you-won't-be-finished scheduling had been switched on West. Or he had switched it on himself; there was no essential difference. The debugging wasn't going well, and West had a wild air, as if his office were a cage.

  He started out talking with his hands in his lap, keeping them busy by twiddling his thumbs. His hands kept getting away, though. He pushed back his hair. He drove his index fingers up under the bridge of his glasses. He made fists and his fingers exploded outward. "I was selling insurance before. Now it's the big crap-shoot, all on one number. Now I gotta get way out on the edge, full of anxiety about the company falling into a revenue hole if it's not there by April — ten thousand jobs on the line, and I gotta pretend everything's fine — and there's a lot of pressure in that. I can't afford to appear too scared around here. I don't talk about it. In the first place, no one outside of here is interested, and I can't talk to people around here and say, 'Here's how I'm manipulating you.'

  "Carman says the company's in a lot of trouble if it's not there by April. Suppose I quit? I could just say, 'Fuck it,' and go Fm

  not gonna do the next machine. I'm gonna give somebody else a chance to fail. I'm gonna get totally out of computers."

  West stopped once in this long, unusually bitter monologue to say: "No, we wouldn't have a disaster. We'd back and fill." He seemed to be saying that he had to make himself believe that dire consequence's would ensue if his team missed its deadline. In a moment, his reason seemed clear. "It just gets so scary I can't even talk about it. I may have to take Rasala into the lab and work on that thing myself twelve hours a day. When you're doing that, you carry it with you everywhere you go. At some level purely me faces me. I better be up here, just crackerjack every morning, fixing that thing. Every morning I gotta wake up and think, 'Oh, Christ, can I do it?'" West took hold of the arms of his desk chair, as if he were about to get out of it and head for the lab right then. "Rocking back here in my chair and talking about doing it is one thing, but it makes me worry. It gives me a nauseous feeling, because I'm not doing it."

  West didn't go into the lab that night or on the working days. One Sunday morning in January, however, when the team was supposed to be resting, a Hardy Boy happened to come by the lab and found West there, sitting in front of one of the prototypes. Then, one Sunday, West wasn't there, and after that, they rarely saw him in the lab, and for a long time he did not even hint that he might again put his own hands inside the machines.

  Whether West accomplished anything or not on his Sunday trysts with Eagle, no one ever knew. But some time after he called an end to them, he said: "We're way beyond what any one person can do. It's too complex."

  A senior member of the Microteam had wanted to create special microdiagnostic programs for Eagle, and West had turned him down, thinking that the regular, higher-level diagnostics would suffice, as they had for debugging Eclipses. Now he relented. Later, the microdiagnostics would become very important. Opinions varied on how much they assisted then, but it seems that they did help the Hardy Boys surmount the first awful barrier how to fix the machine enough so that they could really fix it.

  For a while West continued to take Alsing aside and ask him if they really were making progress in the lab. Gradually, even those queries ceased. Alsing, who often had the feeling he was watching a movie, believed that a scene ended there. His leading man had finally mastered restraint, and in the running commentary Alsing was keeping through me, he recorded: "This week Tom finally gripped the arms of his chair and decided to trust Rasala."

  When West described the other engineers on his staff — Wallach and Alsing — he boasted about their technical attainments. Of Ed Rasala, he said: "His biggest strength is, the guy does not give up. He's in there when most other people go belly up." Rasala was West's trustiest lieutenant.

  From the top of his head, the dome of it bald, to his feet, usually shod in construction boots, Rasala looked big. He wore a thin mustache and beard. He delivered firm handshakes and hearty hellos. "How ya do-in?" he'd say. "He talks very fast," observed Alsing. "It's almost a speech impediment. Sometimes I feel like giving him a shot of Valium and then telling him, 'Okay, Ed, say it again.'" It was odd, but sometimes, even when just saying hello, Rasala's voice carried a sarcastic sound, as if he was mocking himself for what he was saying.

  On an evening at the end of January, down in the basement, Rasala walked up to one of the doors labeled Restricted Area, unlocked it, and led the way in, down a corridor, around a bend, through another door and into a room with yellowish cinderblock walls on one side and on the other a movable partition. The chamber wasn't much bigger than most suburban living rooms, and it was more cluttered than many. It had a floor of linoleum tile and no ceiling
, but instead a tangle of metal braces, heating pipes and, hanging down into the room, many thick black electrical cables. On a gray metal table lay a number of large bound books and looseleaf binders with such labels as Eagle Micro Seq II and ATU PALS. In a gray metal bookcase next to one wall stood published volumes with titles such as Bipolar TTL Data

  Rasala stood near the center of this chamber of exotic initials, arms folded on his chest, and, nodding across the room, said, "Here are two state-of-the-art computers, sitting there."

  And there they were at last, the object of anxiety: a pair of Eagles, standing a few yards apart from each other along the cinder-block wall. They were just two blue metal frames, their tops about level with Rasala's shoulders. Machines in this state are said to have their skins off. Inside each frame, exposed to view — and somehow that didn't seem right — was a shelf full of what are called wire-wrapped boards — thin plates, each with one side covered by a profusion of tiny wires. Small cables, flat like tapeworms, ran among the boards, and farther inside, below the shelf, hung many bundles of multicolored wire. Oh my, there were a lot of wires. At the base of the frames, in metal housings, were the fans that keep the machines from overheating. The fans' steady roar filled the room.

  The two prototypes looked identical, except for their names. Taped across the top of each frame was a piece of paper. One read Coke, the other Gollum. When the machines were first put together, the team had planned to call them Coke and Pepsi, but Ken Holberger insisted that one be named Gollum, after the sinister, spidery creature in Tolkien's Lord of the Rings; and for some time that winter and spring, Gollum was in fact the more interesting machine, the one on which they met the most intricate problems.

  "They're not too impressive at this point in time," said Rasala. He meant the machines' looks, not their abilities, though he would have been right in either case. Equipment much fancier-looking than the prototypes themselves surrounded Coke and Gollum. On top of each sat a microNova, Eagle's therapist, with its blue plastic skins on. A system console — a chair, typewriter and printer — attended each prototype, and a magnetic tape drive stood next to Coke. On TV and in the movies, a tape drive often appears in order to signify the presence of a working computer. That must be because the reels of tape spin jerkily and thus show that something is happening, but in fact, mag tapes represent some of the least important and slowest parts of a computing system. The real action takes place inside the boards. To get a look at it, you need special tools, and they had them there, of course — boxy little machines called logic analyzers.

  The analyzers sat on low wheeled carts, like pieces of surgical gear. They were covered with switches, and each had a small oscilloscope screen; at this moment, on one screen, a strange shape etched in white lines stood frozen and on another danced shifting geometrical patterns. In essence, Rasala explained, these machines were cameras. They took pictures of what happened inside the computer. The analyzers could not look into the chips, but you could hook their probes to the pins of a chip or to wires connecting boards and take several different kinds of snapshots, as it were, of the signals coming and going. The analyzers also contained little memories. When the clock inside Coke or Gollum ticked, every 220 nanoseconds — every 220 billionths of a second, that is — the analyzer would take a picture. It could take and save 256 snapshots during a given foray and play them back on demand.

  "It's funny," Rasala said, "I feel very comfortable talking in nanoseconds. I sit at one of these analyzers and nanoseconds are wide. I mean, you can see them go by. 'Jesus,' I say, 'that signal takes twelve nanoseconds to get from there to there.' Those are real big things to me when I'm building a computer. Yet when I think about it, how much longer it takes to snap your fingers, I've lost track of what a nanosecond really means." He looked at me. "Time in a computer is an interesting concept."

  Then he excused himself. Time in all varieties weighed upon him, and out here it was time to go to work.

  We had come in at the changing of the shifts. The second shift, which often became a graveyard shift, was just coming on duty, a little early, as usual — everyone came early and stayed late these days. Ken Holberger, who is short and handsome, was sitting in front of Gollum. Rasala joined him; Rasala looked like a giant beside him. Holberger was Rasala's second in command, and you could see from the easy grin Rasala turned on him that they were friends.

  "It doesn't loop on fetch, Ed," said Holberger, in greeting.

  Shirt sleeves, jeans and corduroys; mustaches, beards and neat, shoulder-length hair; jogging shoes and hiking boots were in fashion inside the lab. I stood near the center of the room. I heard Rasala cry out: "Whoa. What did you put in for an address?... I'll give it all F's for the hell of it... Whoops, that didn't help your float much." Questions would be asked and never answered out loud. An engineer sitting alone at Coke said: "The program blew away on me. I don't know what happened to it." But he seemed to be talking to himself.

  Gradually, the scene shifted. Some of the Hardy Boys put on coats and departed. A few others came in. Then Rasala sat down at the central table and took up a sheaf of papers. These were engineering change orders — ECOs. They contained descriptions of recent "fixes" that the team had made in the boards of the prototypes. The ECOs mattered because the engineers were working on different problems in each of the machines. After they fixed a problem in one, they wrote up an ECO so that they'd know how to make the same repair in the other. In his list of seventeen rules for the lab, Rasala had written this commandment: "All boards should be updated each morning so that we can insure that each board is in the same state." Apparently, the crew hated this job. Rasala said they were always putting it off. "For the sake of forging ahead with new problems," he said, "I'm thinking of enforcing the rule."

  Rasala rocked back in his chair. "Let's figure out what we gotta do." The others turned toward him. He leafed through the thick sheaf of ECOs, saying: "ATU, ATU, ALU, ALU. And then of course we have the anonymous problem." He made a sour face. The other engineers had come near —warily, it seemed. Holberger said, "Well, commander, what do we do?"

  Rasala looked at them from over the tops of his wirerimmed glasses, and with a faint, Westian smile, he held aloft the sheaf of ECOs.

  Josh Rosen, a black-haired young man said: "Oh, no. Not me. I did that last night."

  As Rosen turned away, Rasala stared at him — a lingering gaze, his eyebrows slightly lowered. Then he lifted them and looked at Holberger, who said, "Okay, I'll do the ATU."

  Rosen was over at Coke. The others — Rasala and Holberger and a technician — sat down before a long bench on the opposite wall. Each laid a wire-wrapped board down in front of himself.

  On one face of a board lay neat rows of the little boxes that housed the chips. The other side, though, was pandemonium. Snarls of thin wires were attached to the hundreds of pins that came out of the chips and extended through holes to these undersides of a board. The wire-wrapped boards rendered a vision of the Jekyll-and-Hyde phenomenon. Each engineer laid his down with the face of Hyde up, and each focused a high-intensity lamp on it. Holberger would pick up the board he was working on and peer into the spaghettilike maze as if he were nearsighted. Rasala left his board on the table and bent low over it. They all followed the same careful routine: locate the pins from which a wire is supposed to be detached; make sure these are the right pins by testing with an ohmmeter; unwrap the wire gently with a small, hollow- tipped tool; throw that wire far away; locate the pins to which a new wire should be attached; wind the new wire on; test again with the ohmmeter, and go on to the next ECO.

  About ten minutes went by in silence. Then Holberger said, while peering at his board: "I've got a two-A here. Is that meaningful?"

  "Yeah," said Rasala, but he didn't even look up.

  Half an hour went by. Rasala pushed back his chair, lifted his glasses, and rubbed his eyes. "The thrill of debugging the machine," he-said to me. "We do this ourselves, rather than have a wire person do it. You're so pro
ne to make mistakes with this. We may be a little more cautious. Then again" — he pulled himself back to the bench — "we may not be."

  The fans in the two Eagles droned on behind the bent backs of the engineers-turned-technicians. In a while, Rasala turned to me again, holding up before my eyes a pair of tweezers. I looked closely,. The tweezers held a tiny strand of wire. "One of the problems," he said. It was awfully easy, he explained, to drop a sliver of wire this size in among the thick nest of wires on the back of a board. And if that little fragment should get lodged in the wrong place, it could cause the machine to commit failures of the sort known as "flakeys." Debuggers could spend hours, even days, finding the cause of such a problem.

  I said, "It's like surgery."

  "Well, not quite," said Rasala. "Most problems we create are repairable."

  I laughed.

  But he wasn't finished. "Most surgery isn't," he added, and went back to his labor.

  When it came to working on the machine, Rasala had no truck with incomplete thoughts.

  "I come from a conservative background, from the beer-drinking generation — I'm into sports, my wife's into being a housewife — so I don't have dreams of grandeur fundamentally about going out and building the next evolution of the computer."

  The first few times I talked to him, Rasala described himself in long rapid sentences like that one, and he spoke them with an edge in his voice that seemed to say, "You probably think I can be sized up like this."

  "I'm a poor Polish boy, I come from the slums of New York, I have a high-school equivalency diploma and would rather play softball," he said. In fact, as he later told me, his father came from Poland, and Rasala grew up in an apartment in Brooklyn. But it wasn't the slums, and what he really remembered was a home where you never went hungry and you knew from an early age that you were going to college. Both of his parents worked. His father, a waiter for Child's Restaurants who read a great deal on the side, taught himself from a book to fix TVs and made a parttime business of it. "Both my parents are very smart," said Rasala. His older brother was "a math whiz" and wound up valedictorian of his class at Columbia University. "Pretty good," said Rasala, "considering my mom has a high-school equivalency diploma and my dad never went past fifth grade."