by Tracy Kidder
The cause of this fretting lay mainly with the "instruction set." These are the basic operations that the computer's builders equip it to perform. Typically, instructions bear such names as ADD, which means the computer should perform addition, and Skip On Equal, which tells the machine to compare two values and if they are equal to skip the next step in a program. Today, there are a couple hundred instructions or so in most minicomputers' sets. A large part of the art of designing a computer's architecture lies in selecting just the right set of instructions and in making each instruction as versatile as possible. This art had advanced a great deal since the invention of the Eclipse. But in order to be fully compatible with Eclipses, Eagle would have to contain the same old instruction set in its entirety. DEC had not tried to do that VAX was not fully compatible with DEC's old 16-bit machines. DEC had given up full compatibility for the sake of giving VAX what Wallach called a wonderful — a "super" — instruction set. He was somewhat biased. EGO's instruction set had resembled VAX's. Both were examples of the state of the art. The Eclipse's instruction set was an example of what that state used to be.
For many months, Wallach would continue to mourn EGO outwardly and to tell anyone who was interested that if some expert asked him someday why he hadn't invented a better instruction set, he would spill the beans. He'd tell them he had been forbidden to use a mode bit. "I'll say it was because I was told that these were management's objectives and I was told that I couldn't work on the machine if I wouldn't fit in with management's objectives." But in his own mind, he was changing his tune. He was getting to like the looks of this architecture. He was starting to think of it not as a wart on a wart, but as a clean design with a wart on it. The wart was the Eclipse instruction set, virtually every part of which Eagle would have to contain, for the sake of compatibility. But there were some other empty corners of this canvas, aside from memory management and protection — chiefly, the new 32-bit Eagle instructions. Wallach came up with some that he liked a great deal. In fact, he even found a way to slip in a well- disguised equivalent of a mode bit — which would have allowed him to define a wholly new set of Eagle instructions, ones not at all derivative of Eclipse. But he didn't disguise the mode bit well enough. West found him out. "We're not gonna do that," he said to Wallach. Wallach went back to his office and punched the wall near his door.
But West did let his adjutant put into Eagle some new instructions that weren't Eclipse-like. Not all the ideas for new instructions came from Wallach, but sometimes it worked this way. They developed a routine. Wallach would bring West an idea for a new instruction. West would say that it looked like a win, but that it wasn't Eclipse-like. Wallach knew what that meant. If the wrong people saw this new instruction in the spec, it might cause a stir. People upstairs and in North Carolina might get the idea that Eagle was, after all, intended as a challenger to FHP — which it certainly was. So Wallach would take the idea for the new instruction to friends working in System Software, and the system programmers would approve the idea. They and Wallach would sit down and fully define the new, non-Eclipse-like instruction, and then Wallach would ask them to write a memo to the Eclipse Group, requesting that this instruction be put into Eagle. "They wrote the memo," said Wallach, "so that the idea would be perceived as coming from them, just in case we ever got called on it." Wallach went on: "A lot of things we did were unique to that environment. It's clear they weren't always the way things should be done."
"But you enjoyed doing things that way?" I suggested.
"We all enjoyed it," he said. "Anytime you do anything on the sly, it's always more interesting than if you do it up front."
After his discovery of the basic scheme for memory management and protection, Wallach had to work out the details, in many long and sometimes loud arguments with Ken Holberger, sublieutenant of the Hardy Boys, who were going to have to implement the architecture. For technical argument, Wallach found Holberger to be a worthy adversary, and on the whole he enjoyed those debates.
Next, all of the instructions had to be defined, as did the precise mechanism by which Eagle would move smoothly, and without the intervention of the user, from programs written for 16 bit Eclipses to ones made for a 32-bit machine. This was a tricky, time-consuming piece of work. Wallach also had to collect all these details and schemes in a document. He took great pains with this volume. He called it "my book," and refining it, rewrote it seven times over the following months. This book was some two hundred pages long, and at the beginnings of each chapter he placed a famous or semifamous quotation.
Wallach said he never read technical tracts outside of work and that he shared West's suspicions about people who did. At home, he said, he mainly read Playboy. "I read the short stories. I really do! Yeah, I look at the pictures, but I like the stories." He laughed. He had not read widely in the classics. So the epigraphs for his book didn't come to him without effort. He took the quotes from Victor Hugo, Nietzsche, Shakespeare, T. S. Eliot, Santayana and FDA. Some were playful and some downright witty, if you understood the context. At the top of the chapter about the instruction set, for instance, he placed this quote from Macbeth:
We still have judgement here; that we but teach
Bloody Instructions [Wallach's cap.], which, being taught, return To plague the inventor.
For the chapter about his elegant scheme of memory management, he chose these verses from Tolkien's Lord of the Rings:
It cannot be seen, cannot be felt Cannot be heard, cannot be smelt, It lies behind stars and under hills And empty holes it fills.
Wallach spent about twenty hours in the Framingham Public Library, with his nose in Bartlett's Familiar Quotations and dipping into some of the actual works, just in order to add these flourishes to his spec. They added something. They revealed the class of feelings that Wallach brought to his job. If he was a Hessian, he was a passionate one, and with the quotations he signed his name to his piece of the new computer.
Wallach actually spent far more time looking up epigraphs than it took him to discover the right way to manage and protect the computer's memory. But that small golden moment colored everything else for him. As he saw it, the rest of the plan simply unfolded from that one idea. It was a good omen.
How do such moments occur? "Hey," Wallach said, "no one knows how that works." He remembered that during the time when he was working on the Navy computer for Raytheon — the one that got built and then scrapped — he was at a wedding and the solution to a different sort of problem popped into his mind. He wrote it down quickly on the cover of a matchbook. "I will be constantly chugging away in my mind," he explained, "making an exhaustive search of my data bank."
MIDNIGHT PROGRAMMER
It was the hour of insomniacs. In the basement of Westborough, the corridors and cubicles stood empty and in shadows. Carl Alsing's cluttered little area made a small rectangle of light. Strewn before me across the surface of his desk, like the relics of a party, lay dozens of roughly drawn maps. They consisted of circles, inside of which were scrawled names such as Dirty Passage, Hall of Mists, Hall of the Mountain King, Complex Junction, Splendid Chamber, Bedquilt, and Witts End. Webs of lines connected the circles, and each line was labeled, some with points of the compass, some with the words up and down. Here and there on the maps were notations — "water here," "oil here," and "damn that pirate In the midst of all this paper sat Alsing's computer terminal. On the screen of the tube in white letters, like the little voice that whispers in a wild gambler's ear, this message stood:
ARE YOU SURE YOU WANT TO QUIT NOW?
Alsing had offered to demonstrate "midnight programming," and now, early in the winter of 1979, he had made good on the promise. He was sitting beside his desk with his hands folded, wearing his soft, sneaky smile. For hours, he had been watching me play the game Adventure. "You really got into it," he said. "That's good."
When first invented, the program for Adventure had traveled widely, like a chain letter, from coast to coast among computer e
ngineers and buffs. It had arrived in Westborough just in time for the aftermath of the EGO wars. It was everywhere by now; grade schoolers were playing it.
In the game, the computer appears to create for you an underground world, called Colossal Cave. It moves you through it in response to your commands. The computer seems to act as game board, rule keeper and, when you foul up, as both assistant and adversary. You move by typing out directions on the keyboard at your terminal. If you spell out these directions in full a few times, a message will appear on your screen, saying:
IF YOU PREFER SIMPLY TYPE N RATHER THAN NORTH
"How did the computer know to do that?"
"I don't know," said Alsing, coyly. "Sometimes it's perceptive, other times just dumb."
After you have moved, a message appears on the screen telling you where you are and what you are confronting. You must respond, in two words or less, both to opportunities — treasure or tools lying on the floor of some chamber — and to threats and challenges — the hatchet-hurling dwarf, the snake, the troll who guards the bridge, the dragon. If, for instance, you want to get past the rusty door in one of the chambers, you have to think of what will conquer rust, then you have to remember where h was you saw that pool of oil, then you have to type in step-by-step instructions to get back to that oil, and then, because the computer will let you carry only so many things, you may have to drop one of your tools or treasures — DROP GOLD COINS, you might write — and then type in, TAKE OIL. Of course, you must al ready be holding a container for the oil. Then you have to retrace your steps-back to the rusty door and type, OIL DOOR. This method of travel and maneuver takes some getting used to, but after a while it's as easy as driving a car. The game is a harrowing of Hell.
Earlier that night, traveling more or less at random and without maps, I had stumbled into a place that the message on the screen described this way:
YOU ARE IN A MAZE OF TWISTY LITTLE PASSAGES, ALL DIFFERENT
So I typed, NW, in a hurry, hoping to get out of there by going back the way I'd come. But the screen responded:
YOU ARE IN A TWISTY MAZE OF LITTLE PASSAGES, ALL DIFFERENT
If you are susceptible to Adventure, you get worried at this point. I had the feeling I was lost in a forest, and I acted as no smart woodsman would, heading off in this direction, then heading off in that, and getting nowhere.
Then I heard Alsing chuckle. "Ahhh, I love it."
I thrashed around in that maze awhile longer. Finally, Alsing said, "Look carefully at the messages on the screen."
"They're all the same."
"No, they're not."
Each chamber of this maze within the big labyrinth had a slightly different and unique address, formed by a particular arrangement of the words twisty, little, passages, and maze.
"And what do you do when you get lost?" asked Alsing.
"You make maps, of course."
Alsing sat back and nodded, smiling — the complacent schoolmaster, one of the roles he played at the beginning of Eagle.
Later on, though, I wandered into a maze that really scared me. YOU ARE IN A MAZE OF TWISTY LITTLE PASSAGES, ALL ALIKE
You have to find your way around this maze if you hope to begin to master Adventure, because this one contains the vending machine with the batteries for your indispensable flashlight and, moreover, harbors the lair of the kleptomaniacal pirate who is forever sneaking up behind you and snatching away your treasures. But how do you find your way around a maze in which all the chambers are the same? You must do what Hansel and Gretel tried to do, more or less —drop something on the floor of every chamber to leave a trail marking where you've been.
That isn't the worst maze, however. You can get caught in Witts End and think that you'll never get out. Some of the engineers at Westborough who had come close to mastering the entire game believed that the only way out of Witts End was to tell the computer you want to commit suicide — AXE ME. That works; you get reincarnated shortly afterward. But you lose points; suicide isn't the best solution.
I myself did not get as far as Witts End,' however, but quit while in the maze that was all alike. And though the computer seemed reluctant to let me go — DO YOU REALLY WANT TO QUIT NOW? — I stood firm. Alsing got up and led the way toward the cafeteria. He got us lost in the corridors of the basement. No doubt that was his plan, though he denied it. Intentional or not, our getting lost allowed him to cry out, "The twisty little passages of Data General!"
It was the time of night when the odd feeling of not being quite in focus comes and goes, and all things are mysterious. I resisted this feeling. It seemed worth remembering that Adventure is just a program, a series of step-by-step commands stored in electrical code inside the computer.
How can the machine perform its tricks? The general answer lies in the fact that computers can follow conditional instructions. They can take two values and compare them — that comes down to simple arithmetic — and, if so commanded, can perform one action if the values are equal and another if they are not. In this ability to follow conditional instructions — an ability built into the machine — lies much of the computer's power. You can set before it, in sequence, bifurcating webs of conditional instructions, until the machine appears to make sophisticated decisions on its own.
When we had returned from our adventure finding coffee, I asked Alsing how he felt about the question — twenty years old now and really unresolved — of whether or not it's theoretically possible to imbue a computer with intelligence — to create in a machine, as they say, artificial intelligence. Alsing stepped around the question. "Artificial intelligence takes you away from your own trip. What you want to do is look at the wheels of the machine and if you like them, have fun."
To the two computers that the Eclipse Group used, the engineers had given the names Woodstock and Trixie, after characters in comic strips. They often spoke about these computers as if (hey had personalities. When especially frustrated, one Microkid would walk into the lab where Trixie resided and yell at the machine. Alsing said: "A lot of people are really tired of anthropomorphizing computers, but it sure is an easy way to talk about them. You can anthropomorphize your car and the analogy works, and then at some level it doesn't. We anthropomorphize big business, the military and so on, as some strange creatures with alien personalities. I think that's sane, I think that's normal. You tend to have to anthropomorphize the computer. It presents a face, a person to me — a person in a thousand different ways."
He drew his chair up to his terminal and typed a few letters — a short code that put him in touch with Trixie, which was the machine reserved for the use of his microcoding team. "We've anthropomorphized Trixie to a ridiculous extent," he said.
He typed, WHO.
On the dark-blue screen of the cathode-ray tube, with alacrity, an answer appeared: CARL. WHERE, typed Alsing.
IN THE ROAD, WHERE ELSE! Trixie replied.
HOW.
ERROR, read the message on the screen.
"Oh, yeah, I forgot," said Alsing, and he typed, PLEASE HOW.
THAT'S FOR US TO KNOW AND YOU TO FIND OUT.
Alsing seemed satisfied with that, and he typed, WHEN.
RIGHT FUCKING NOW, wrote the machine.
WHY, wrote Alsing.
BECAUSE WE LIKE TO CARL.
One of Alsing's Microkids had programmed Trixie to deliver these impertinent responses. In a real sense, Alsing was conversing with a member of his team. I think — one of them said as much — that if the Microkids' computer had ever started talking back to them all on its own, they would have ripped its wires out.
Anthropomorphizing Trixie was just a game, and by no means the most sophisticated that Alsing played with his team. They were the imps of the Eclipse Group and their computer showed it.
Assembled from the comments of a few of the young men on the team, the typical engineer, an imaginary creature perhaps, wears a white undershirt and a plastic pouch (known to some as a "nerd pack") in his breast pocket, in order
to keep his pens from soiling his clothes. An electronic calculator — used to be a slide rule — hangs like a ring of janitor's keys from his belt Jim Guyer of the Hardy Boys, who wore a beard and drove a large motorcycle, added, "A lot of people think that an engineer is shut off in a little corner and doesn't give a damn about anything else except his own little thing," Guyer said. "They exist. And they're the most obvious engineers, because in their isolation they're obvious."
When first encountered, Alsing might seem to fit this description. He could be taken for one accustomed to dark corners. In this respect, as in others, he is deceiving. Alsing is tall, over six feet, but he doesn't seem to realize it and he doesn't look that big. He is neither fat nor thin. He keeps his hair cut fairly short. Often his dress looks sloppy — not deliberately or extremely so, but slightly careless. He speaks softly, as a rule, and the pitch of his voice, though not at all squeaky, is high. About his hands, the way he folds them in his lap or puts them together under his chin, there is something delicate. One acquaintance of his said, "He looks uncoordinated." In fact, Alsing does not play many physical games, though on a vacation in the Caribbean he took up scuba diving. Ham radio is an old hobby of his. I thought that I could see in him the lonely childhood behind him — he would have been the last boy picked in every schoolyard game, the one who threw a ball like a girl. But Alsing is gregarious. He set out some years back to master social gracefulness, and it shows. He will sit down in a strange living room, fold his hands neatly in his lap, and listen. You can forget that he is there, until gradually, so delicately you hardly notice, he enters the conversation. After several such occasions, people who had just met Alsing said to me, "He's really smart and interesting, isn't he?"
By the time the Eagle project began, Alsing had attained the age of thirty-five, which made him very elderly within the group — but he was a clever and playful old codger. His eyes would dart around. His eyebrows would dance; they'd do his winking for him. When he'd close his mouth and let a little smile creep across his cheeks — in each one was carved a line in the shape of a half-moon — then it was time to be on guard, for a game, a trick or a joke.