by Steven Levy
His big project in that course was an attempt to “solve” the game Peg Solitaire (or HI-Q), where you have a board in the shape of a plus sign with thirty-three holes in it. Every hole but one is filled by a peg: you jump pegs over each other, removing the ones you jump over. The idea is to finish with one peg in the center. When Gosper and two classmates proposed to Minsky that they solve the problem on the PDP-1, Minsky doubted they could do it, but welcomed the try. Gosper and his friends not only solved it—“We demolished it,” he’d later say. They hacked a program that would enable the PDP-1 to solve the game in an hour and a half.
Gosper admired the way the computer solved HI-Q because its approach was “counterintuitive.” He had a profound respect for programs which used techniques that on the surface seemed improbable, but in fact took advantage of the situation’s deep mathematical truth. The counterintuitive solution sprang from understanding the magical connections between things in the vast mandala of numerical relationships on which hacking ultimately was based. Discovering those relationships—making new mathematics on the computer—was to be Gosper’s quest; and as he began hanging out more around the PDP-1 and TMRC, he made himself indispensable as the chief “math hacker”—not so much interested in systems programs, but able to come up with astoundingly clear (nonintuitive!) algorithms which might help a systems hacker knock a few instructions off a subroutine, or crack a mental logjam on getting a program running.
• • • • • • • •
Gosper and Greenblatt represented two kinds of hacking around TMRC and the PDP-1: Greenblatt focused on pragmatic systems building, and Gosper on mathematical exploration. Each respected the other’s forte, and both would participate in projects, often collaborative ones, that exploited their best abilities. More than that, both were major contributors to the still nascent culture that was beginning to flower in its fullest form on the ninth floor of Tech Square. For various reasons, it would be in this technological hothouse that the culture would grow most lushly, taking the Hacker Ethic to its extreme.
The action would shift among several scenes. The Kluge Room, with the PDP-1 now operating with the time-sharing system, which Jack Dennis had worked for a year to write, was still an option for some late-night hacking, and especially Spacewarring. But more and more, the true hackers would prefer the Project MAC computer. It stood among other machines on the harshly lit, sterilely furnished ninth floor of Tech Square, where one could escape from the hum of the air conditioners running the various computers only by ducking into one of several tiny offices. Finally, there was TMRC, with its never-empty Coke machine and Saunders’ change box and the Tool Room next door, where people would sit at all hours of the night and argue what to an outsider would be bafflingly arcane points.
These arguments were the lifeblood of the hacker community. Sometimes people would literally scream at each other, insisting on a certain kind of coding scheme for an assembler, or a specific type of interface, or a particular feature in a computer language. These differences would have hackers banging on the blackboard or throwing chalk across the room. It wasn’t so much a battle of egos as it was an attempt to figure out what “The Right Thing” was. The term had special meaning to the hackers. The Right Thing implied that to any problem, whether a programming dilemma, a hardware interface mismatch, or a question of software architecture, a solution existed that was just . . . it. The perfect algorithm. You’d have hacked right into the sweet spot, and anyone with half a brain would see that the straight line between two points had been drawn, and there was no sense trying to top it. “The Right Thing,” Gosper would later explain, “very specifically meant the unique, correct, elegant solution . . . the thing that satisfied all the constraints at the same time, which everyone seemed to believe existed for most problems.”
Gosper and Greenblatt both had strong opinions, but usually Greenblatt would tire of corrosive human interfacing, and wander away to actually implement something. Elegant or not. In his thinking, things had to be done. And if no one else would be hacking them, he would. He would sit down with paper and pencil, or maybe at the console of the PDP-1, and scream out his code. Greenblatt’s programs were robust, meaning that their foundation was firm, with built-in error checks to prevent the whole thing from bombing as a result of a single mistake. By the time Greenblatt was through with a program, it was thoroughly debugged. Gosper thought that Greenblatt loved finding and fixing bugs more than anybody he’d ever met, and suspected he sometimes wrote buggy code just so he could fix it.
Gosper had a more public style of hacking. He liked to work with an audience, and often novice hackers would pull up a chair behind him at the console to watch him write his clever hacks, which were often loaded with terse little mathematical points of interest. He was at his best at display hacks, where an unusual algorithm would evoke a steadily unpredictable series of CRT pyrotechnics. Gosper would act as tour guide as he progressed, sometimes emphasizing that even typing mistakes could present an interesting numerical phenomenon. He maintained a continual fascination with the way a computer could spit back something unexpected, and he would treat the utterances of the machine with infinite respect. Sometimes the most seemingly random event could lure him off into a fascinating tangent on the implications of this quadratic surd or that transcendental function. Certain subroutine wizardry in a Gosper program would occasionally evolve into a scholarly memo, like the one that begins:
On the theory that continued fractions are underused, probably because of their unfamiliarity, I offer the following propaganda session on the relative merits of continued fractions versus other numerical representations.
The arguments in the Tool Room were no mere college bull sessions. Kotok would often be there, and it was at those sessions that significant decisions were made concerning the computer he was designing for DEC, the PDP-6. Even in its design stage, this PDP-6 was considered the absolute Right Thing around TMRC. Kotok would sometimes drive Gosper back to South Jersey for holiday breaks, talking as he drove about how this new computer would have sixteen independent registers. (A register, or accumulator, is a place within a computer where actual computation occurs. Sixteen of them would give a machine a heretofore unheard-of versatility.) Gosper would gasp. That’ll be, he thought, the greatest computer in the history of the world!
When DEC actually built the PDP-6 and gave the first prototype to Project MAC, everyone could see that while the computer had all the necessary sops for commercial users, it was at heart a hacker’s machine. Both Kotok and his boss, Gordon Bell, recalling their TX-0 days, used the PDP-6 to demolish the limitations that had bothered them on that machine. Also, Kotok had listened closely to the suggestions of TMRC people, notably Peter Samson, who took credit for the sixteen registers. The instruction set had everything you needed, and the overall architecture was symmetrically sound. The sixteen registers could be accessed three different ways each, and you could do it in combinations, to get a lot done by using a single instruction. The PDP-6 also used a “stack,” which allowed you to mix and match your subroutines, programs, and activities with ease. To hackers, the introduction of the PDP-6 and its achingly beautiful instruction set meant they had a powerful new vocabulary with which to express sentiments that previously could be conveyed only in the most awkward terms.
Minsky set the hackers to work writing new systems software for the PDP-6, a beautiful sea-blue machine with three large cabinets, a more streamlined control panel than the One, rows of shiny cantilevered switches, and a winking matrix of lights. Soon they were into the psychology of this new machine as deeply as they had been on the PDP-1. But you could go further on the Six. One day in the Tool Room at TMRC the hackers were playing around with different ways to do decimal print routines, little programs to get the computer to print out in Arabic numbers. Someone got the idea of trying some of the flashy new instructions on the PDP-6, the ones that utilized the stack. Hardly anyone had integrated these new instructions into his code; but as the
program got put on the blackboard using one instruction called Push-J, to everyone’s amazement the entire decimal print routine, which normally would be a page worth of code, came out only six instructions long. After that, everyone around TMRC agreed that Push-J had certainly been The Right Thing to put into the PDP-6.
The Tool Room discussions and arguments would often be carried over to dinner, and the cuisine of choice was almost always Chinese food. It was cheap, plentiful, and—best of all—available late at night. (A poor second choice was the nearby greasy spoon on Cambridge’s Main Street, a maroon-paneled former railroad car named the F&T Diner, but called by hackers “The Red Death.”) On most Saturday evenings, or spontaneously on weeknights after 10 P.M., a group of hackers would head out, sometimes in Greenblatt’s blue 1954 Chevy convertible, to Boston’s Chinatown.
Chinese food was a system, too, and the hacker curiosity was applied to that system as assiduously as to a new LISP compiler. Samson had been an aficionado from his first experience on a TMRC outing to Joy Fong’s on Central Square, and by the early sixties he had actually learned enough Chinese characters to read menus and order obscure dishes. Gosper took to the cuisine with even greater vigor; he would prowl Chinatown looking for restaurants open after midnight, and one night he found a tiny little cellar place run by a small family. It was fairly dull food, but he noticed some Chinese people eating fantastic-looking dishes. So he figured he’d take Samson back there.
They went back loaded with Chinese dictionaries, and demanded a Chinese menu. The chef, a Mr. Wong, reluctantly complied, and Gosper, Samson, and the others pored over the menu as if it were an instruction set for a new machine. Samson supplied the translations, which were positively revelatory. What was called “Beef with Tomato” on the English menu had a literal meaning of Barbarian Eggplant Cowpork. “Wonton” had a Chinese equivalent of Cloud Gulp. There were unbelievable things to discover in this system! So after deciding the most interesting things to order (“Hibiscus Wing? Better order that, find out what that’s about”), they called over Mr. Wong, and he jabbered frantically in Chinese disapproval of their selections. It turned out he was reluctant to serve them the food Chinese-style, thinking that Americans couldn’t take it. Mr. Wong had mistaken them for typically timid Americans—but these were explorers! They had been inside the machine, and lived to tell the tale (they would tell it in assembly language). Mr. Wong gave in. Out came the best Chinese meal that any of the hackers had eaten to date.
So expert were the TMRC people at hacking Chinese food that they could eventually go the restauranteurs one better. On a hacker excursion one April Fools’ Day, Gosper had a craving for a little-known dish called Bitter Melon. It was a wart-dotted form of green pepper, with an intense quinine taste that evoked nausea in all but those who’d painfully acquired the taste. For reasons best known to himself, Gosper decided to have it with sweet-and-sour sauce, and he wrote down the order in Chinese. The owner’s daughter came out giggling. “I’m afraid you made a mistake—my father says that this says ‘Sweet-and-Sour Bitter Melon.’” Gosper took this as a challenge. Besides, he was offended that the daughter couldn’t even read Chinese—that went against the logic of an efficient Chinese Restaurant System, a logic Gosper had come to respect. So, even though he knew his order was a preposterous request, he acted indignant, telling the daughter, “Of course it says Sweet-and-Sour Bitter Melon—we Americans always order Sweet-and-Sour Bitter Melon the first of April.” Finally, the owner himself came out. “You can’t eat!” he shouted. “No taste! No taste!” The hackers stuck to the request, and the owner slunk back to the kitchen.
Sweet-and-Sour Bitter Melon turned out to be every bit as hideous as the owner promised. The sauce at that place was wickedly potent, so much so that if you inhaled while you put some in your mouth you’d choke. Combined with the ordinarily vile bitter melon, it created a chemical that seemed to squeak on your teeth, and no amount of tea or Coca-Cola could dilute that taste. To almost any other group of people, the experience would have been a nightmare. But to the hackers it was all part of the system. It made no human sense, but had its logic. It was The Right Thing; therefore every year on April Fools’ Day they returned to the restaurant and insisted that their appetizer be Sweet-and-Sour Bitter Melon.
It was during those meals that the hackers were most social. Chinese restaurants offered hackers a fascinating culinary system and a physically predictable environment. To make it even more comfortable, Gosper, one of several hackers who despised smoke in the air and disdained those who smoked, brought along a tiny, battery-powered fan. The fan was something kluged up by a teenage hacker who hung around the AI lab—it looked like a mean little bomb, and had been built using a cooling fan from a junked computer. Gosper would put it on the table to gently blow smoke back into offenders’ faces. On one occasion at the Lucky Garden in Cambridge, a brutish jock at a nearby table became outraged when the little fan redirected the smoke from his date’s cigarette back to their table. He looked at these grungy MIT types with their little fan and demanded the hackers turn the thing off. “OK, if she stops smoking,” they said, and at that point the jock charged the table, knocking dishes around, spilling tea all over, and even sticking his chopsticks into the blades of the fan. The hackers, who considered physical combat one of the more idiotic human interfaces, watched in astonishment. The incident ended as soon as the jock noticed a policeman sitting across the restaurant.
That was an exception to what were usually convivial gatherings. The talk revolved around various hacking issues. Often, people would have their printouts with them and during lulls in conversation would bury their noses in the reams of assembly code. On occasion, the hackers would even discuss some events in the “real world,” but the Hacker Ethic would be identifiable in the terms of the discussion. It would come down to some flaw in a system. Or an interesting event would be considered in light of a hacker’s natural curiosity about the way things work.
A common subject was the hideous reign of IBM, the disgustingly naked emperor of the computer kingdom. Greenblatt might go on a “flame”—an extended and agitated riff—about the zillions of dollars being wasted on IBM computers. Greenblatt would go home on vacation and see that the science department at the University of Missouri, which allegedly didn’t have any money, was spending four million dollars a year on the care and feeding of an IBM Hulking Giant that wasn’t nearly as nifty as the PDP-6. And speaking of grossly overrated stuff, what about that IBM time-sharing system at MIT, with that IBM 7094 right there on the ninth floor? Talk about waste!
This could go on for a whole meal. It is telling, though, to note the things that the hackers did not talk about. They did not spend much time discussing the social and political implications of computers in society (except maybe to mention how utterly wrong and naive the popular conception of computers was). They did not talk sports. They generally kept their own emotional and personal lives—as far as they had any—to themselves. And for a group of healthy college-age males, there was remarkably little discussion of a topic, which commonly obsesses groups of that composition: females.
Though some hackers led somewhat active social lives, the key figures in TMRC-PDP hacking had locked themselves into what would be called “bachelor mode.” It was easy to fall into—for one thing, many of the hackers were loners to begin with, socially uncomfortable. It was the predictability and controllability of a computer system—as opposed to the hopelessly random problems in a human relationship—which made hacking particularly attractive. But an even weightier factor was the hackers’ impression that computing was much more important than getting involved in a romantic relationship. It was a question of priorities.
Hacking had replaced sex in their lives.
“The people were just so interested in computers and that kind of stuff that they just really didn’t have time [for women],” Kotok would later reflect. “And as they got older, everyone sort of had the view that one day some woman would come along
and sort of plunk you over the head and say, you!" That was more or less what happened to Kotok, though not until his late thirties. Meanwhile, hackers acted as if sex didn’t exist. They wouldn’t notice some gorgeous woman at the table next to them in the Chinese restaurant, because “the concept of gorgeous woman wasn’t in the vocabulary,” hacker David Silver later explained. When a woman did come into the life of a serious hacker, there might be some discussion—“What’s happened to so-and-so . . . the guy’s just completely falling apart . . .” But generally that kind of thing was not so much disdained as it was shrugged off. You couldn’t dwell on those who might have fallen by the wayside, because you were involved in the most important thing in the world—hacking. Not only an obsession and a lusty pleasure, hacking was a mission. You would hack, and you would live by the Hacker Ethic, and you knew that horribly inefficient and wasteful things like women burned too many cycles, occupied too much memory space. “Women, even today, are considered grossly unpredictable,” one PDP-6 hacker noted, almost two decades later. “How can a hacker tolerate such an imperfect being?”
Maybe it would have been different if there had been more women around TMRC and the ninth floor—the few that did hang around paired off with hackers. (”They found us,” one hacker would later note.) There were not too many of these women, since outsiders, male or female, were often put off by the group: the hackers talked strangely, they had bizarre hours, they ate weird food, and they spent all their time thinking about computers.