by Rudy Rucker
One of the King Kong Original Hackers of the Valley was a guy called Bill Gosper. Back in the 1970s, I’d read about him in a column by Martin Gardner, describing how Gosper had discovered a special pattern called the Glider Gun for an early cellular automaton known as the Game of Life. Gosper and his friends were among the very first computer types at MIT, majoring perforce in mathematics (as there as yet were no computer-science courses). It was they who’d first adopted the word “hacker” as a proud name for their ilk.
One night in 1987 when I was high, I mustered the courage to phone up Gosper, and he cheerfully agreed to have me visit him at his office. Although he was five or ten years older than me, he sounded like an excitable kid.
Gosper was working for a company called Symbolics that made computers the size of refrigerators. His office was filled with beige plastic artifacts: an ellipsoidal electric pencil sharpener, a stack of computer monitors, odd-shaped modern chairs.
So far as I could make out, Gosper’s job at Symbolics was to work on a program that could in some sense behave like a mathematician—manipulating such things as algebraic equations, geometrical constructions, matrices and integrals. But he spent most of his time devising weird graphics involving cellular automata, fractals, prime numbers and the like.
At one point he held the world record for computing the most digits of pi. Being Gosper, he didn’t compute pi as a series of decimal digits like 3.1415926…Instead he computed it as a so-called continued fraction, that is, as deeply nested expression that has fractions within fractions within fractions.
3 + 1/(7 + 1/(15 + 1/(1 + 1/(292 +…))))
Although Gosper didn’t have a PC-clone machine, the desk of his absent secretary did, so I opened up her computer case and jacked in my axe, showing Gosper the new wave of East coast cellular automata. He was only mildly impressed. The old Game of Life was world enough for him. He’d recently devised a hack that let him explore spaces with trillions of cells at a time.
I came to love hanging around with the guy, and occasionally he’d join our family for holidays. He always reminded me of a giant prehistoric bird, with his beaky profile, bouncy gait, and discordant voice. About ten percent of Gosper’s words weren’t standard English. Instead they were odd phrasings peculiar to him. Like he’d use “mumble” as an ordinary word, as shorthand for technical expressions that were too dull to actually say out loud.
“Let’s set mumble to mumble,” he once remarked to me, leaning over his screen. “Ooh! It’s not converging. What the hell’s going on? 572??!!! It’s supposed to be 570! God help us. It’s batshit. Oh, this should be a Taylor series, right? I have to stun it, I have to neutralize it. Now we can crank up the value. And this is the right answer.” Gosper paused and gave me a sly smile. “Now let’s see if I can earn my nerd merit badge.”
“How?” I asked, greatly enjoying myself even though I had no idea what he was talking about.
“By typing in this number,” answered Gosper, the keyboard clicking. “The nineteen digits of two to the sixty-fourth power.”
Eventually he’d tire of discussing computer science with me. “That’s enough about square things,” he’d say. “How about round things?”
By this he meant that we should switch to his other obsession: throwing his Aerobie flying rings. He had several hundred of them, each one labeled with a number. Gosper knew the flight characteristics of each Aerobie as if they were cattle in a herd. He liked hacking the physics of toys.
In addition to keeping Aerobies in the trunk of his car, Gosper usually carried a Superball that he’d toss at things, putting some weird spin on the ball that made it bounce right back to his hand.
From time to time I still meet Gosper for lunch. Invariably he insists on the buffet at an Indian restaurant. When I saw him in 2008, he showed up carrying a pizza box, and with a three-foot tall man at his side. The box held a steel packing puzzle of Gosper’s design, known as the Twubblesome Twelve. And the little person was a random youth he’d met at a donut shop. Gosper was teaching him math.
My cellular automata machine card took me all kinds of places. I was at the first-ever workshop on artificial life, in Los Alamos, New Mexico, in the fall of 1987. The town of Los Alamos is weird and quiet, like a Twilight Zone movie set. No less a man than William Burroughs went to a boarding school there.
The idea behind the a-life conference was: Forget artificial intelligence, let’s do artificial life, let’s make a pool of simple programs that interbreed and compete and get more interesting as time goes on! It was another of my science fiction dreams coming to life, for in Software I’d written about a race of self-replicating, evolving robots who shared and recombined their programs like DNA.
The conference had a relaxed vibe, with a lounge outside the lecture hall where lots of computers—including one the hosts had lent me—were running artificial life and cellular automata demos. I spent most of my time in the lounge, tweaking with the hackers.
A big turning point came when, later in the fall of 1987, I attended an annual Silicon Valley event called the Hackers Conference. As I keep insisting, hacker was still a good word, and these guys were Silicon Valley programmers and hardware tweakers. Some of them were even fans of my books. The fact that I’d written a science fiction novel called Software had put me on the hackers’ radar.
I brought my computer with its CA axe, and I stayed up all night with the hackers, drinking beer, smoking pot, and admiring the weird things on their computer screens. In the wee hours of the night, they examined my cellular automaton machine card and told me how it worked.
A guy named John Walker remarked that it should in principle be possible to get rid of the card and accelerate the CA programs with pure software.
By way of spreading the word about CAs, I gave a special Christmas talk and demo at the IBM research lab in San Jose in 1987. I plugged my cellular automata card into one of their IBM PCs and connected it to this monster projector that they had. Nobody outside high-end labs had computer projectors back then, so it was incredibly exciting for me to see my images get so big. I would have liked to take off my clothes and let the sparkling little squares of the CA graphics slide across my bare skin.
The guys loved my realtime animated images, and when I was done I got more applause than I’d heard since being on a panel with star writer Larry Niven at a world science fiction convention one time in 1980. But now, after my IBM talk, some execs took me into a conference room and started asking me, “What are CAs good for?”
I kind of hoped they’d offer me a consulting job, but I didn’t really have a convincing answer to their question. I mean, yes, I could make promises about CAs eventually being able to simulate air turbulence or the heat flow in a jet engine. But, to me, it was evident that the main thing about CAs was that they were beautiful art that had permanently changed my way of seeing the world. The execs weren’t into that.
No matter. As it turned out, John Walker was going to pull me deep into a business project for creating software to look at CAs.
Computer Hacker
My new hacker friend John Walker was a founder and the CEO of a booming Sausalito corporation called Autodesk, and in 1988 he asked me if I’d be interested to come work for him. Autodesk had done very well with their drafting software, and they had a big surplus in the bank. Walker wanted to explore some radically new kinds of software products.
I wasn’t sure whether I should accept Walker’s offer. But, as it happened, in February, 1988, a computer science student at San Jose State had gone over the edge and had mass-murdered seven people at the lab where he worked. Richard Farley. I’d in fact sat next to Farley in the assembly language class that I’d audited during my first semester of teaching. He was a real asshole, always arguing with the teacher over picayune points that made no sense, and sleazing all over any women students who committed the error of sitting near him. Farley’s rampage set me to wondering whether San Jose State was paying me enough to be associating with the
se kinds of people. If I could go into industry and make some real money—why not?
So when Walker and one of his software engineers, Eric Lyons, showed up at my house in the spring of 1988 with a computer to show me a new Mandelbrot fractal zoom program they’d written, I decided to pursue Walker’s suggestion that I find work at Autodesk. The actual job offer came through in the fall of 1988, and I worked at Autodesk until 1992.
Fortunately, San Jose State liked the idea of having faculty go out and work in industry for awhile. It was, after all, by far the best way for a computer science professor to pick up new skills. Whatever you know about computers is continually becoming obsolete, and the only people who stay abreast are the commercial software engineers. So the math and computer science department agreed to let me go on half-time leave, which I later increased to full-time leave.
Autodesk’s core business was a product called AutoCAD, an electronic drafting program used worldwide by architects and industrial designers. But Walker didn’t want me to work on that. Instead he was starting a small Advanced Technology Division, headed by Eric Lyons, and, for the moment, me and one or two other guys.
My first project was to produce some cellular automata software with Walker. He was an insanely talented programmer. He worked at the level of a grand master in chess, or at the level of a mathematician at the Institute for Advanced Study. Over Autodesk’s one-week-long Christmas break, Walker wrote an assembly language program that eliminated any need for a special card such as the so-called cellular automaton machine I’d been carrying around.
My role in this was to create some sample CA rules for our new software to run, and to write a manual explaining it all. I got deeply into the task. Walker and I and finished our project over the course of several months. When we were done, we’d produced a slick, boxed software package called CA Lab: Rudy Rucker’s Cellular Automata Laboratory, which sold for about $50 and went on the market in 1989. In those pre-Internet days, some people were actually willing to buy software of this kind on disks. I did demos of it at a number of computer trade fairs, always having to parry the same old question.
“What’s it good for?”
“You stare CAs for hundreds and hundreds of hours and they eat your brain, okay?”
We sold a decent number of copies, and Walker had the idea that we could develop a whole line of software packages for hackers to enjoy. These packages were meant to be like books, but interactive, illustrating new aspects of science. Walker wanted to call the line the Autodesk Science Series.
The second package in the series was James Gleick’s Chaos: The Software, designed to let users play with some of the programs mentioned in Gleick’s bestselling book Chaos.
What was the chaos craze all about? Chaos is another new idea whose true origins lie in computer science. We all know about simple, deterministic processes that do something utterly predictable—like a cannonball flying along a predetermined parabola through the air. We also know about completely messy natural processes, such as the crackling static we might hear on a radio.
Chaotic processes lie midway between the extremes of predictability and randomness. On the one hand, a chaotic process doesn’t settle into any kind of dull and simple pattern. On the other hand, a chaotic process isn’t actually random. It’s generated by some fairly simple and deterministic law of math or physics.
There’s a certain overlap here with Bennett’s notion of logical depth that I mentioned before. A chaotic pattern is logically deep in that it’s generated by a concise rule that uses a long computation in order to produce the patterns that you see.
The hard thing to grasp about a messy-looking chaotic process is that it is in fact deterministic. If, for instance, you set about computing the successive digits of pi, you’ll always end up with that same number sequence, 3.1415926…So all the digits of pi are in some sense predetermined. But yet—and this is the subtle point—the digits aren’t predictable, at least not predictable by any rapidly-acting rule of thumb. Yes, someone like Bill Gosper can compute the billionth digit of pi, but he needs to run a powerful computer through quite a few cycles in order to come up with the answer. There are some good formulae, but there’s no quick and dirty pencil-and-a-scrap-of-paper shortcut for finding the billionth digit of pi. Pi is gnarly, pi is chaotic, pi is logically deep.
In his Chaos book, Gleick talked about some mathematical systems that were known to generate chaotic patterns. Among these were the Lorenz attractor and the Mandelbrot set, and we put simulations of these into the Autodesk Chaos software, along with some other funky things.
Working on this second Autodesk program all through 1989 and 1990 was a lot harder for me than working on CA Lab had been. The big difference was that this time Walker didn’t step in and write the bulk of the code. Instead I worked with another Autodesk programmer, a knowledgeable and irascible guy called Josh Gordon. Truth be told, my own programming skills were still pretty rudimentary. I was in over my head. And Josh was never shy about telling me this. But somehow we struggled to a conclusion and in 1991 we shipped this second product, too.
Later, on my own, I’d write a third science series program called Artificial Life Lab, which would be published as a disk with a book in 1993, not by Autodesk, but by the low-end Waite Group Press in the North Bay.
Rather than using my literary agent, Susan Protter, for this minor book project, I pigheadedly made the deal on my own. But then, annoyingly, the Waite Group refused to pay me my royalties on the Japanese edition of the book. And without the relentless force of La Protter by my side, I never did get that money. As I mentioned before, it’s not so much that I need an agent for getting deals—it’s more that I need the agent for making sure the contract is fair, and for making sure that the publishers honor it.
The three software packages that I worked on, CA Lab, Chaos, and Artificial Life Lab, are all long out of print by now, but you can download them for free from my website, www.rudyrucker.com. In searching my site, note that eventually we had to change the name of CA Lab to Cellab. The robotic greedheads who run some boring company called Computer Associates were threatening to sue us for infringing on their sacred trademarked initials CA. As if cellular automata hadn’t been around much longer than them.
As Autodesk was located in Sausalito, some seventy miles north of where I lived, I only went there physically about once a week. The rest of the time I’d stay in touch via email, which was something brand new to me.
I quickly learned some painful lessons about email. Never write and send a message when you’re angry or, even worse, drunk. And don’t send the message to everyone in the company. And especially don’t use curse words. Fortunately, hackers tend to be resilient and forgiving types. As Walker once put it, “Don’t worry too much about flaming me. I have thick scales.”
Going up to Autodesk in person was always a kick. Some days it would feel like grabbing hold of a live electric wire with a million volts coursing through it. They always had the latest software and hardware, and the engineers were weird and smart, with awesomely wild plans.
Sometimes after work I’d visit with some spacey fans who lived in Mill Valley. One day early in 1989 they gave me a large marijuana plant, live and growing in a big pot that I could nurture in my back yard. I’d gotten a maximum new Mac computer from Autodesk that same day as well, along with Stephen Wolfram’s brand-new computer algebra software, Mathematica. And thanks to my fat new salary, I’d bought myself a peppy red Acura car. Driving home across the Golden Gate Bridge that day with all my goodies, I was, like, “Yeah!” It was one of those rare moments when everything seems to come together.
I took to listening to CDs of environmental sounds on the long drives home. I missed rain so much that I’d play a CD of a thunderstorm. I had other nature CDs too: a rain forest, a blizzard with a banging shutter, and a walk along a brook in a meadow followed by a ride in a sailboat. My mind would drift and I’d think about computer programs—not as code, but as patterns of sh
apes and connections, sort of like the way I’d formerly thought about the mathematics of hyperspace and infinite sets.
Things were always changing at Autodesk. By the time I stopped working there, in 1992, the Advanced Technology division employed about twenty or thirty people. As well as my little Science Series programs, they’d set up a virtual reality lab.
We had sets of Jaron Lanier’s new VR goggles, which held a little TV screen for each eye, also his stretchy, optical-fiber-equipped data gloves that tracked the position of each finger joint, creating images of your hands in the virtual world that the goggles were showing you. I wrote a demo that immersed the user in a flock of artificially alive birds that were continually wheeling and regrouping around the user’s position in cyberspace. Rather than making them look like actual birds, I made the birds look like three and four dimensional polyhedra that I called topes—and thus my demo was called Flocking Topes.
The lab’s goal was to develop a VR operating system to be called Autodesk Cyberspace®. They’d picked this name on their own, and without asking me about it. My old cyberpunk writer friend William Gibson was a little annoyed by this development. After all, he’d coined the word cyberspace, and now Autodesk wanted to trademark it? A few months later, when I saw Gibson at a San Francisco virtual reality fair called the Cyberthon. He half-jokingly threatened to trademark the name of Autodesk’s original virtual reality programmer, a talkative hipster named Eric Gullichsen.
Although sales of our Chaos program were even better than CA Lab’s had been, the profits from these relatively low-priced packages were negligible compared to Autodesk’s income from their flagship product, AutoCAD. And Autodesk Cyberspace® was shaping up to be a dud. And then the company’s stock price dropped.