The Soul of a New Machine

by Tracy Kidder

  There are two general kinds of computer programs. The users write one kind themselves or hire consultants to write them. These "user programs" may look as though they are telling the computer what to do step by step, and a relatively simple program — for calculating a company's payroll, for instance — may look forbiddingly long and complex. But this is nothing like the length and complexity that the program would have if it really did work directly on the machine. Instead, nowadays, a series of other programs that are stored inside the computer breaks down commands such as "divide" into several more basic commands that the machine is equipped to obey. These intermediary programs, which serve as translators of user programs, are known collectively as "system software." It is usually the manufacturer's responsibility to create system software, and the customer's to buy it when purchasing a new machine.

  By the mid-1960s, a trend that would become increasingly pronounced was already apparent: while the expense of building a computer's hardware was steadily declining, the cost of creating both user and system software was rising. In an extremely bold stroke, IBM took advantage-of the trend. They announced, in the mid-sixties, all at one time, an entire family of new computers — the famous 360 line. In the commerce of computers, no single event has had wider significance, except for the invention of the transistor. Part of the 360's importance lay in the fact that all the machines in the family were software compatible.

  It cost IBM a true fortune and no end of trouble and anxiety to create system software for the 360 line. But all the machines in the family used that same software. So IBM had to create the stuff only once, and thus was able to amortize the cost of its development over all of the many thousands of 360's that it sold. Moreover, any user program that worked on one machine in the family worked on all of them. Users become attached to their programs and system software. Software is expensive. Getting it to function properly often takes time. Software that works is precious. Users don't idly discard it Obviously, this reluctance can present problems for computer manufacturers: how do you get customers to buy bigger, better machines? Total software compatibility made it easy for customers to do what IBM wanted them to do, which was to buy several different kinds of 360 computers. A customer could buy a small one now and later on buy a bigger one, or vice versa, without haying to re-create any software. Software compatibility strengthened IBM's already tight grip on its customers: they weren't likely to forsake IBM and take their business elsewhere when that meant assuming new expenses and problems with software.

  Soon every manufacturer of computers was employing some variation of IBM's 360 strategy of software compatibility. Data General had made all of its NOVAs compatible one with the other, and likewise all the Eclipses. Moreover, the designers had made the Eclipse "upwardly compatible" with NOVAs. This meant that while new programs written for Eclipse computers would not run on NOVAs, old programs written for NOVAg would work on Eclipses. This sort of compatibility was a useful tool for marketing, for it allowed customers to switch from NOVAs to Eclipses with relative ease — they could do so without discarding all their old software.

  Software compatibility is a marvelous thing. That was the essential lesson West took away from his long talks with his friend in Marketing. You didn't want to make a machine that wasn't compatible, not if you could avoid it. Old customers would feel that since they'd need to buy and create all new software anyway, they might as well look at what other companies had to offer; they'd be likely to undertake the dreaded "market survey." And an incompatible machine would not make it easy for new customers to buy both 16-bit Eclipses and the new machine. This was the gist of what West learned. He became increasingly interested. DEC'S VAX was only "culturally compatible" with the line of machines that preceded it. Data General should build a 32 bit machine that was fully compatible with Eclipses. "From a marketing point of view what a win that would be!"

  Matters proceeded so informally and with such speed that there would never be any way of telling afterward where all of the first technical solutions to the problem of making a 32-bit Eclipse came from. But it was West who gathered them together. Soon he began an in-house PR campaign that would not end for a long time.

  Carl Alsing participated in the beginnings of West's proselytizing. Among other acts, Alsing gave the new, unbuilt machine the code name Eagle. Mostly, though, he observed. Alsing was by temperament a watcher, a moviegoer. He had been with West longer than anyone else in the Eclipse Group. He felt that he knew West and then again he felt that he didn't. West launching Eagle was to Alsing something worth watching. For example, the meeting West set up with the vice presidents of engineering and software. West took Alsing along. The way it looked to Alsing, West brought two proposals to-the VPs. One of these was an obvious loser and the other was Eagle. "West's letting them pick Eagle," thought Alsing, and he smiled.

  "West's never unprepared in any kind of meeting. He doesn't talk fast or raise his voice. He conveys — it's not enthusiasm exactly, it's the intensity of someone who's weathering a storm and showing us the way out. He's saying, 'Look, we gotta move this way.' Then once he gets the VPs to say it sounds good, Tom goes to some of the software people and some of his own people. The bosses are signed up for this,' he tells them. 'Can I get you signed up to do your part?' He goes around and hits people one at a time, gets 'em enthused. They say, 'Ahhh, it sounds like you're just gonna put a bag on the side of the Eclipse,' and Tom'll give 'em his little grin and say, 'It's more than that, we're really gonna build this fucker and it's gonna be fast as greased lightning.' He tells them, 'We're gonna do it by April.' That's less than a year away, but never mind. Tom's message is: 'Are you guys gonna do it or sit on your ass and complain?' It's a challenge he throws at them. So he basically made us stop moaning about the demise of Westborough."

  Alsing went on: "West brought us out of our depression into the honesty of pure work. He put new life into a lot of people's jobs, I think."

  Not everyone associated with the Eclipse Group liked the looks of this proposed new machine. They thought it would be just a refinement of the Eclipse, which was itself a refinement of the NOVA. "A wart on a wart on a wart," one engineer said. "A bag on the side of the Eclipse." Some even said that it would be a "kludge," and this was the unkindest cut Kludge is perhaps the most disdainful term in the computer engineer's vocabulary: it conjures up visions of a machine with wires hanging out of it, of things fastened together with adhesive tape.

  So some engineers dropped out of the project right away. To the remainder and to new recruits, West preached whatever gospel seemed most likely to stir up enthusiasm. It would be an opportunity for them to "get a machine out the door with their names on it." When Alsing came up with the code name, Eagle, West felt pleased, because, he said, it was impossible to say "Eagle" without it sounding as if you were saying "EGO." Not that West cared himself; he did not feel at all vengeful toward North Carolina. But some others did. There would also be something here for technology bigots. Eagle would not be, as the saying goes, "a clean sheet of paper"; but there would be some clean corners to work in. It might look at first glance like a VW Beetle, but think what they could put under the hood. This wasn't going to be just a slight variation on the Eclipse CPU, but a wholly new and fast machine that would happen to be compatible with Eclipses. It would put money on the bottom line, of course. Lots of money. They were going to do it in record time, because the company needed this machine desperately. And when they succeeded, they would be heroes. If, after EGO had been canceled, you had left the Eclipse Group moping in their comer of the basement, and had returned a year later, you surely would have failed to recognize the place. At some moments, their corner of the building now had the air of a commuter train about it, and at others, the silent intensity of a university library on the eve of exams — many new young faces peering into cathode-ray tubes, leafing through fat documents. In the conversation around there you heard words and phrases such as these: A canard was anything false, usually a wrongheade
d notion entertained by some other group or company; things could be done in ways that created no muss, no fuss, that were quick and dirty, that were clean. Fundamentals were the source of all right thinking, and weighty sentences often began with the adverb fundamentally, while realistically prefaced many flights of fancy. There was talk of wars, shootouts, hired guns and people who shot from the hip. The win was the object of all this sport and the big win was something that could be achieved by maximizing the smaller one. From the vocabulary alone, you could have guessed that West had been there, and that these engineers were up to something.

  Rosemarie Seale, the group's main secretary, felt excitement in the air. She, for one, took West's exhortations to heart. She resolved to do whatever she could in order to keep bureaucratic and trivial affairs from distracting these youngsters on their crucial mission. Her spirit never flagged, but from time to time she did wonder why, if this project was so important to the company, so few people in other departments seemed to recognize the fact. Why, for instance, was it allowed that the mailroom be moved in the midst of the project, creating the risk that vital mail would be held up? To prevent that sort of small catastrophe, for several weeks she would go to the mailroom daily and sort through the mail herself. Likewise, why would the carpenters be allowed to come in during a particularly delicate phase of the Eagle project and totally remodel the Eclipse Group's office space?

  The answer, one possible answer, was that West had two ways of describing Eagle. One way made it sound important and glorious; the other, like something routine. West explained: "Yon gotta distinguish between the internal promotion to the actual workers and the promoting we did externally to other parts of the company. Outside the group I tried to low-key the thing. I tried to dull the impression that this was a competing product with North Carolina. I tried to sell it externally as not much of a threat I was selling insurance; this would be there if something went wrong in North Carolina. It was just gonna be a fast, Eclipse-like machine. This was the only way it was gonna live. We had to get the resources quietly, without creating a big brouhaha, and it's difficult to get a lot of external cooperation under those circumstances."

  West went on, theorizing now: "The company would have been just as happy if we hadn't done it. De Castro thought he had it covered in North Carolina. But once you got somebody saying, 'Hey, we want to do this,' then you're in a position where you gotta say no, and that's a different proposition. A bunch of good engineers were getting ready to quit from being told no too many times already; that's another problem. I went to de Castro about Eagle at some point. I said, 'We'll do it in a year,' and he may have said, 'Okay.' But it was clear we had to do it in a year to have any chance."

  Years before, according to local legend, West's mentor, the former leader of the Eclipse Group, had said that he could build a NOVA on a single printed-circuit board. Told that he was suffering from delusions of grandeur, he did the job at home on his kitchen table and wound up producing the best-selling of all the NOVAs. Before that, de Castro and two of the other founders had cut loose from their old company, in order to build a new computer. West was doing nothing so extreme. He seemed to have the full support of Data General's vice president of engineering, Carl Carman. He was given money to hire new recruits. He had de Castro's laconic permission to give Eagle a try. But even after the project had begun, West would say (for example), "De Castro won't take a piece of paper from me on this machine." He would remark, "There's a lot of people pretending that this project doesn't exist." Some others in the group expressed the same contradictory feeling: that they were building a machine absolutely essential to the company but were doing it all on their own. "I think we're doing it in spite of Data General," one of the older hands in the group said. The circumstances fostered such an attitude. So did West. Deliberately, somewhat surreptitiously, he separated his team from the rest of the company.

  "We're building what I thought we could get away with," West said.

  BUILDING A TEAM

  The basement of Westborough, subterranean at the front of Building 14A/B and at ground level in back, was one of the places in Data General's widening empire where machines were conceived, designed, labored over in prototypes, and sometimes brought to life. Tom West led the way down into this region one night, through confusing corridors. I would try to learn my way around by noting small landmarks — a copying machine at a corner, a bulletin board trimmed with company news about dental programs and new Data General disk drives. Off some of the hallways were mysterious doors, locked up and bearing signs that said in loud letters Restricted Area.

  Then the hallways opened and all around under fluorescent light lay fields of cubicles without doors. Their walls — made of steel, some of them covered with cream-colored cloth — did not reach the ceiling, but stood about five and a half feet high. You could look over them. They created no privacy. Most of the cubicles were empty now, but each would contain one person during the day. Most had a desk with a computer terminal on it, and a little bookcase. Some held a drafting table and many had a houseplant or two. Many green plants poked their heads, like periscopes, above the tops of the cubicles' walls. "The great statement," said West, gesturing at the foliage, wearing a little grin that puckered one cheek.

  The arrangements looked temporary, and in fact they were. As one of the company's PR men explained, cubicles laid out as in a maze allow a greater density of workers per square foot than real offices with doors. Easily movable walls allow management to tinker with that area-to-people ratio without incurring enormous expense — for instance, to create fully enclosed offices where cubicles have been when it turns out that some jobs are most efficiently performed behind doors. It was said that the company's vice president for manufacturing could turn Westborough into a factory overnight, and maybe the joke had some substance. The last headquarters — compared to which this one was plush — had in fact been turned into a factory.

  Westborough seemed designed for quick changes. Smiling his wry smile, West offered some additional theories: "We can change it all around. It keeps up the basic level of insecurity It's basically a cattle yard What goes on here is not part of the real

  world"

  "How so?"

  "Mmmmnimmmmmh. The language is different."

  Some of it was, and a phrase book, such as the Penguin Dictionary of Computers, could be useful. ECO — each letter pronounced— meant "engineering change order." Hence this remark: "A friend of mine told his girlfriend they had to ECO their relationship." Give me a core dump meant "Tell me your thoughts," for in the past, when computers used "core memories," engineers sometimes "dumped" the contents of malfunctioning machines' storage compartments to see what was wrong. A stack is a special small compartment of memory, a sort of in-box inside a computer; it holds information in the order in which the information is deposited and when it gets overfull, it is said to "overflow." Hence the occasional complaint, "I've got a stack overflow." "His mind is only one stack deep," says an engineer, describing the failings of a colleague, but the syntax is wrong and he rephrases, saying: "See. He can push, but when it comes time to pop, he goes off in all directions" — which means that the poor fellow can receive and understand information but he can't retrieve it in an orderly fashion.

  The basement, it seemed, was never empty. Even in the small hours of the morning someone would be sitting in a little pool of light in a cubicle, working. By day, the place held a throng. I saw them all collected once, out back in the parking lot, during a fire drill. I counted only a couple of black faces, but I saw many women, many of them in skirts, and I presumed most were secretaries, because I knew that here as in the industry generally female engineers were scarce. Men were numerous. Most looked as though they were in their twenties. Only a few wore jackets and ties; the rest were dressed casually and, on the whole, neatly. Once, down in the basement, I saw an engineer who wore not just long but shaggy hair and who was wearing Army-surplus clothes. He was slouching off down a corridor car
rying a canteen cup. His appearance was sufficiently unusual that one of West's team took pains to point him out to me.

  West led the way into the Eclipse Group's quarters. They were indistinguishable from any other group's, except at night perhaps, during the Eagle project, when, as a rule, more lamps burned on in their cubicles than elsewhere in the basement. You could tell that an engineer enjoyed some rank if he had an office with a door. West had one of these. It was tiny and windowless. A thick, jacketed pipe and a steel girder descended through it, down the face of a cinder-block wall. There were some gray metal chairs, a gray metal bookcase, a couple of small gray metal tables and a gray metal desk, the top of which was absolutely clean save for a single stack of papers with their edges perfectly squared. A Magic Marker board, at the moment displaying some incomprehensible diagram, hung on one wall. For adornments there were an old clock in a beautiful oak case, and on the wall behind West's back, a picture of a square-rigged sailing ship. On the wall beside him hung several photographs of computers.

  * * *

  Physicians hang diplomas in their waiting rooms. Some fishermen mount their biggest catch. Downstairs in Westborough, it was pictures of computers.