by Steven Levy
Computer manufacturers, particularly IBM, were not enthusiastic. It was clear that MIT would have to go about it pretty much on its own. (The research firm of Bolt Beranek and Newman was also working on time sharing.) Eventually two projects began at MIT: one was Jack Dennis’ largely solo effort to write a time-sharing system for the PDP-1. The other was undertaken by a professor named F.J. Corbató, who would seek some help from the reluctant goliath, IBM, to write a system for the 7090.
The Department of Defense, especially through its Advanced Research Projects Agency (ARPA), had been supporting computers since the war, mindful of their eventual applications toward military use. So by the early sixties, MIT had obtained a long-range grant for its time-sharing project, which would be named Project MAC (the initials stood for two things: Multiple Access Computing, and Machine Aided Cognition). Uncle Sam would cough up three million dollars a year. Dennis would be in charge. Marvin Minsky would also be a large presence, particularly in using the one-third share of the money that would go not for time-sharing development, but for the still ephemeral field of artificial intelligence. Minsky was delighted, since the million dollars was ten times his previous budget for AI, and he realized that a good part of the remaining two thirds would see its way into AI activities as well. It was a chance to set up an ideal facility, where people could plan for the realization of the hacker dream with sophisticated machines, shielded from the bureaucratic lunacy of the outside world. Meanwhile, the hacker dream would be lived day-by-day by devoted students of the machine.
The planners knew that they’d need special people to staff this lab. Marvin Minsky and Jack Dennis knew that the enthusiasm of brilliant hackers was essential to bring about their Big Ideas. As Minsky later said of his lab: “In this environment there were several things going on. There were the most abstract theories of artificial intelligence that people were working on and some of [the hackers] were concerned with those, most weren’t. But there was the question of how do you make the programs that do these things and how do you get them to work.”
Minsky was quite happy to resolve that question by leaving it to the hackers, the people to whom “computers were the most interesting thing in the world.” The kind of people who, for a lark, would hack up something even wilder than Spacewar and then, instead of playing it all night (as sometimes was happening in the Kluge Room), would hack some more. Instead of space simulations, the hackers who did the scut work at Project MAC would be tackling larger systems—robotic arms, vision projects, mathematical conundrums, and labyrinthine time-sharing systems that boggled the imagination. Fortunately, the classes that entered MIT in the early sixties were to provide some of the most devoted and brilliant hackers who ever sat at a console. And none of them so fully fit the title “hacker” as Richard Greenblatt.
Chapter 4. Greenblatt and Gosper
Ricky Greenblatt was a hacker waiting to happen. Years later, when he was known throughout the nation’s computer centers as the archetypal hacker, when the tales of his single-minded concentration were almost as prolific as the millions of lines of assembly-language code he’d hacked, someone would ask him how it all started. He’d twist back in his chair, looking not as rumpled as he did back as an undergraduate, when he was cherub-faced and dark-haired and painfully awkward of speech; the question, he figured, came down to whether hackers were born or made, and out came one of the notorious non sequiturs which came to be known as Blatt-isms: “If hackers are born, then they’re going to get made, and if they’re made into it, they were born.”
But Greenblatt would admit that he was a born hacker.
Not that his first encounter with the PDP-1 had changed his life. He was interested, all right. It had been freshman rush week at MIT, and Ricky Greenblatt had some time on his hands before tackling his courses, ready for academic glory. He visited the places that interested him most: the campus radio station WTBS (MIT’s was perhaps the only college radio station in the country with a surfeit of student audio engineers and a shortage of disc jockeys), the Tech Model Railroad Club, and the Kluge Room in Building 26, which held the PDP-1.
Some hackers were playing Spacewar.
It was the general rule to play the game with all the room lights turned off, so the people crowded around the console would have their faces eerily illuminated by this display of spaceships and heavy stars. Rapt faces lit by the glow of the computer. Ricky Greenblatt was impressed. He watched the cosmic clashes for a while, then went next door to look over the TX-0, with its racks of tubes and transistors, its fancy power supplies, its lights and switches. His high school math club back in Columbia, Missouri, had visited the state university’s batch-processed computer, and he’d seen a giant card-sorting machine at a local insurance company. But nothing like this. Still, despite being impressed with the radio station, the Model Railroad Club, and especially the computers, he set about making dean’s list.
This scholastic virtue could not last. Greenblatt, even more than your normal MIT student, was a willing conscript of the Hands-On Imperative. His life had been changed irrevocably the day in 1954 that his father, visiting the son he hadn’t lived with since an early divorce, took him to the Memorial Student Union at the University of Missouri, not far from Ricky’s house in Columbia. Ricky Greenblatt took to the place immediately. It wasn’t merely because of the comfortable lounge, the television set, the soft-drink bar . . . It was because of the students, who were more of an intellectual match for nine-year-old Ricky Greenblatt than were his classmates. He would go there to play chess, and he usually had no problem beating the college students. He was a very good chess player.
One of his chess victims was a UM engineering student on the GI bill. His name was Lester, and Lester’s gift to this nine-year-old prodigy was a hands-on introduction to the world of electronics. A world where there were no ambiguities. Logic prevailed. You had a degree of control over things. You could build things according to your own plan. To a nine-year-old whose intelligence might have made him uncomfortable with his chronological peers, a child affected by a marital split which was typical of a world of human relations beyond his control, electronics was the perfect escape.
Lester and Ricky worked on ham radio projects. They tore apart old television sets. Before finishing college, Lester introduced Ricky to a Mr. Houghton, who ran a local radio shop, and that became a second home to the youngster through high school. With a high school friend, Greenblatt built a gamut of hairy projects. Amplifiers, modulators, all sorts of evil looking vacuum tube contraptions. An oscilloscope. Ham radios. A television camera. A television camera! It seemed like a good idea, so they built it. And of course when it came time to choose a college, Richard Greenblatt picked MIT. He entered in the fall of 1962.
The course work was rigid during his first term, but Greenblatt was handling it without much problem. He had developed a relationship with a few campus computers. He had gotten lucky, landing the elective course called EE 641—lntroduction to Computer Programming—and he would often go down to the punch-card machines at EAM to make programs for the Hulking Giant 7090. Also, his roommate, Mike Beeler, had been taking a course in something called Nomography. The students taking the class had hands-on access to an IBM 1620—set in yet another enclave of those misguided priests whose minds had been clouded with the ignorant fog that came from the IBM sales force. Greenblatt would often accompany Beeler to the 1620, where you would punch up your card deck, and stand in line. When your turn came, you’d dump your cards in the reader and get an instant printout from a plotter-printer. “It was sort of a fun, evening thing to do,” Beeler would later recall. “We’d do it the way others might watch a sports game, or go out and have a beer.” It was limited but gratifying. It made Greenblatt want more.
Around Christmas time, he finally felt comfortable enough to hang out at the Model Railroad Club. There, around such people as Peter Samson, it was natural to fall into hacker mode. (Computers had various states called “modes,” and hackers often u
sed that phrase to describe conditions in real life.) Samson had been working on a big timetable program for the TMRC operating sessions on the giant layout; because of the number crunching required, Samson had done it in FORTRAN on the 7090. Greenblatt decided to write the first FORTRAN for the PDP-1. Just why he decided to do this is something he could never explain, and chances are no one asked. It was common, if you wanted to do a task on a machine and the machine didn’t have the software to do it, to write the proper software so you could do it. This was an impulse that Greenblatt would later elevate to an art form.
He did it, too. Wrote a program that would enable you to write in FORTRAN, taking what you wrote and compiling the code into machine language, as well as transforming the computer’s machine language responses back into FORTRAN. Greenblatt did his FORTRAN compiler largely in his room, since he had trouble getting enough access to the PDP-1 to work online. Besides that, he got involved in working on a new system of relays underneath the layout at TMRC. It seems that the plaster in the room (which was always pretty grungy anyway, because custodial people were officially barred entry) kept falling, and some of it would get on the contacts of the system that Jack Dennis had masterminded in the mid-fifties. Also, there was something new called a wire-spring relay which looked better than the old kind. So Greenblatt spent a good deal of time that spring doing that. Along with PDP-1 hacking.
It is funny how things happen. You begin working conscientiously as a student, you make the dean’s list, and then you discover something that puts classes into their proper perspective: they are totally irrelevant to the matter at hand. The matter at hand was hacking, and it seemed obvious—at least, so obvious that no one around TMRC or the PDP-1 seemed to think it even a useful topic of discourse—that hacking was a pursuit so satisfying that you could make a life of it. While a computer is very complex, it is not nearly as complex as the various comings and goings and interrelationships of the human zoo; but, unlike formal or informal study of the social sciences, hacking gave you not only an understanding of the system, but an addictive control as well, along with the illusion that total control was just a few features away. Naturally, you go about building those aspects of the system that seem most necessary to work within the system in the proper way. Just as naturally, working in this improved system lets you know of more things that need to be done. Then someone like Marvin Minsky might happen along and say, “Here is a robot arm. I am leaving this robot arm by the machine.” Immediately, nothing in the world is as essential as making the proper interface between the machine and the robot arm, and putting the robot arm under your control, and figuring a way to create a system where the robot arm knows what the hell it is doing. Then you can see your offspring come to life. How can something as contrived as an engineering class compare to that? Chances are that your engineering professor has never done anything half as interesting as the problems you are solving every day on the PDP-1. Who’s right?
By Greenblatt’s sophomore year, the computer scene around the PDP-1 was changing considerably. Though a few more of the original TX-0 hackers had departed, there was new talent arriving, and the new, ambitious setup, funded by the benevolent Department of Defense, nicely accommodated their hacking. A second PDP-1 had arrived; its home was the new, nine-story rectangular building on Main Street—a building of mind-numbing dullness, with no protuberances, and sill-less windows that looked painted onto its off-white surface. The building was called Tech Square, and among the MIT and corporate clients moving in was Project MAC. The ninth floor of this building, where the computers were, would be home to a generation of hackers, and none would spend as much time there as Greenblatt.
Greenblatt was getting paid (sub-minimum wages) for hacking as a student employee, as were several hackers who worked on the system or were starting to develop some of the large programs that would do artificial intelligence. They started to notice that this awkwardly polite sophomore was a potential PDP-1 superstar.
He was turning out an incredible amount of code, hacking as much as he could, or sitting with a stack of printouts, marking them up. He’d shuttle between the PDP-1 and TMRC, with his head fantastically wired with the structures of the program he was working on, or the system of relays he’d hacked under the TMRC layout. To hold that concentration for a long period of time, he lived, as did several of his peers, the thirty-hour day. It was conducive to intense hacking, since you had an extended block of waking hours to get going on a program, and, once you were really rolling, little annoyances like sleep need not bother you. The idea was to burn away for thirty hours, reach total exhaustion, then go home and collapse for twelve hours. An alternative would be to collapse right there in the lab. A minor drawback of this sort of schedule was that it put you at odds with the routines which everyone else in the world used to do things like keep appointments, eat, and go to classes. Hackers could accommodate this—one would commonly ask questions like, “What phase is Greenblatt in?” and someone who had seen him recently would say, “I think he’s in a night phase now, and should be in around nine or so.” Professors did not adjust to those phases so easily, and Greenblatt “zorched” his classes.
He was placed on academic probation, and his mother came to Massachusetts to confer with the dean. There was some explaining to do. “His mom was concerned,” his roommate Beeler would later say. “Her idea was that he was here to get a degree. But the things he was doing on the computer were completely state-of-the-art—no one was doing them yet. He saw additional things to be done. It was very difficult to get excited about classes.” To Greenblatt, it wasn’t really important that he was in danger of flunking out of college. Hacking was paramount: it was what he did best and what made him happiest.
His worst moment came when he was so “out of phase” that he slept past a final exam. It only hastened his exit from the student body of MIT. Flunking out probably wouldn’t have made any difference at all in his life had it not been for a rule that you couldn’t be a student employee when you were an exiled student. So Greenblatt went looking for work, fully intending to get a daytime programming job that would allow him to spend his nights at the place he wanted to spend his time—the ninth floor at Tech Square. Hacking. And that is exactly what he did.
• • • • • • • •
There was an equally impressive hacker who had mastered the PDP-1 in a different manner. More verbal than Greenblatt, he was better able to articulate his vision of how the computer had changed his life, and how it might change all our lives. This student was named Bill Gosper. He had begun MIT a year before Greenblatt, but had been somewhat slower at becoming a habitué of the PDP-1. Gosper was thin, with bird-like features covered by thick spectacles and an unruly head of kinky brown hair. But even a brief meeting with Gosper was enough to convince you that here was someone whose brilliance put things like physical appearance into their properly trivial perspective. He was a math genius. It was actually the idea of hacking the world of mathematics, rather than hacking systems, that attracted Gosper to the computer, and he was to serve as a long-time foil to Greenblatt and the other systems-oriented people in the society of brilliant foot soldiers now forming around brand-new Project MAC.
Gosper was from Pennsauken, New Jersey, across the river from Philadelphia, and his pre-MIT experience with computers, like Greenblatt’s, was limited to watching Hulking Giants operate from behind a pane of glass. He could vividly recall seeing the Univac at Philadelphia’s Franklin Institute churn out pictures of Benjamin Franklin on its line printer. Gosper had no idea what was going on, but it looked like great fun.
He tasted that fun himself for the first time in his second MIT semester. He’d taken a course from Uncle John McCarthy—open only to freshmen who’d gotten disgustingly high grade point averages the previous term. The course began with FORTRAN, went on to IBM machine language, and wound up on the PDP-1. The problems were nontrivial, things like tracing rays through optical systems with the 709, or working routines with a new flo
ating-point interpreter for the PDP-1.
The challenge of programming appealed to Gosper. Especially on the PDP-1, which, after the torture of IBM batch processing could work on you like an intoxicating elixir. Or having sex for the first time. Years later, Gosper still spoke with excitement of “the rush of having this live keyboard under you and having this machine respond in milliseconds to what you were doing . . .”
Still, Gosper was timid about continuing on the PDP-1 after the course was over. He was involved with the math department, where people kept telling him that he would be wise to stay away from computers—they would turn him into a clerk. The unofficial slogan of the math department, Gosper found, was “There’s no such thing as Computer Science—it’s witchcraft!” Well then, Gosper would be a witch! He signed up for Minsky’s course in artificial intelligence. The work was again on the PDP-1, and this time Gosper got drawn into hacking itself. Somewhere in that term, he wrote a program to plot functions on the screen, his first real project, and one of the subroutines contained a program bum so elegant that he dared show it to Alan Kotok. Kotok by then had attained, thought Gosper, “godlike status,” not only from his exploits on the PDP-1 and TMRC, but from the well-known fact that his work at DEC included a prime role in the design of a new computer, a much-enhanced version of the PDP-1. Gosper was rapturous when Kotok not only looked over his hack, but thought it clever enough to show to someone else. Kotok actually thought I’d done something neat! Gosper hunkered down for more hacking.