Given the right equations and variables, four clicks tapped with the aid of a big toe will record everything required for roulette prediction by a Laplacian intelligence. Click the passage of a point on the rotor—the 00, for instance—in front of a fixed point on the rim of the wheel. Record the rotor’s second turn in front of the reference point. Mark the passage of the ball in front of the same reference. Note the ball’s second and subsequent revolutions past the marker. The more clicks for the ball, the greater the computer’s accuracy in predicting its deceleration, but two alone, along with the two already entered for the rotor, will suffice.
“While deriving the roulette equations, I took everything I had already done on the campus computer,” said Doyne, “which was written in the higher-level language of BASIC, and translated it into machine language. Then I had to think of ways to program the feedback stuff. What parameters did we need, and how were we going to set them? If you input data via microswitches under your toes, how does the computer unravel these clicks and decide what each of them means?
“It’s no easy task writing a program to predict roulette, even when you know how you want it written. Getting the computer to perform the calculations took only a quarter of the time I spent on the program. The rest went into teaching it what the inputs meant and how to get information back to the outside world. Just making sure the computer didn’t get confused about what it was supposed to be doing required a lot of care.”
A good part of Doyne’s program dealt with identifying the parameters that differ from one roulette wheel to the next. “There are basically five numbers you want to know. We devised a trial-and-error method for setting these five parameters. The computer is programmed in advance with ideal values. You click in data with your toes and compare them to what the computer expects to find. Then you fiddle with these predictions until you get each of them coinciding with reality.”
For this fiddling process, Doyne divided the computer program into eight specialized domains, or modes, five of which were devoted to calibrating parameters for the ball, rotor, and tilt of the wheel. Two other modes were used as mathematical scratchpads for keeping histograms on how well the computer was performing. A final mode—holding all the adjustable parameters—was reserved for playing the game itself.
It is easy to get lost in a program like this, especially when you have to communicate with the computer through your toes. So Doyne devised something called a mode map. This diagrammed, in a series of interlocking loops, the relationship among the eight modes. A specific pattern of toe clicks steered the computer from one mode to another in a procedure that came to be known as “driving around the mode map.”
Two big toes were actually required for driving around the map. The left toe looped the computer from mode to mode, while clicks made with the right toe incremented or decremented parameters. A complete tour of the map, with stops along the way for adjusting variables, took about fifteen minutes and ended with the computer being clicked into the playing mode. Once its parameters had been fine-tuned to the roulette wheel at hand, the computer’s predictive power was uncanny. Even Laplace might have been surprised by the intelligence available to a big toe.
4
Radios from Other Planets
The computer can’t tell you the emotional story. It can give you the exact mathematical design, but what’s missing is the eyebrows.
Frank Zappa
If the Eudaemonic KIM was a mother computer destined for re-production, its first offspring was going to be, relatively speaking, a giant. Four by five inches—roughly a quarter the size of a printed page—the prototype roulette computer would itself be cloned into later versions no bigger than a library index card. There was a simple practical reason for starting big: transistors are easier to count when spread out over surfaces larger than the head of a pin.
Juano was nominally in charge of building the prototype computer, although Doyne and Norman helped design it, and Jack Biles assisted in putting it together. They decided as a general rule to christen their new computers with the middle name of the primary builder. So the Project’s first homemade computer, constructed by John “Juano” Raymond Boyd, came to be known as Raymond.
Raymond caused a lot of trouble right from the start. Built with a microprocessor and chips bought off the shelf from parts houses in the Silicon Valley, it showed them they had a lot to learn about making computers from scratch. “We became a familiar sight to suppliers over in the Valley,” said Norman, “as we accumulated a nice supply of burnt-out chips.”
Even under the tutelage of Dan Browne, Juano had lost his shirt as a poker player. On returning to Santa Cruz, he was nominally looking for a job, but with black hair draped over his spectacles and halfway down his back, Juano—actually the most harmless of people—looked like a refugee from the Haight-Ashbury drug wars. After a job conducting telephone surveys, and another sorting parts for an electronics firm in Santa Cruz, he tried to capitalize on his B.A. in physics by selling himself in the Silicon Valley as a technician. While commuting to interviews on the far side of the Santa Cruz Mountains, he was able to shop in the Valley for chips, resistors, capacitors, diodes, crystals, and the other ingredients needed for making computers at home.
The microprocessor in Raymond was identical to that in the Keyboard Input Module. The major difference between the two computers lay in their memories. While the KIM stored its programs on cassette tape, Raymond needed something more compact. To come to life as a portable roulette computer, it had to incorporate a silicon chip for storing information known as a PROM, which is short for programmable read-only memory. To program this new memory chip, the Project required an electronic circuit known as a PROM burner, which can duplicate the memory on one chip—its orientation of 1’s and 0’s—and burn it onto another.
Doyne built a PROM burner onto the KIM. Juano drove over to the Valley for another handful of chips. They wired the components onto a circuit board and succeeded finally in burning the roulette program from the KIM into Raymond’s memory. “At that point,” declared Doyne, “Raymond was a computer in its own right and ready to go.
“In our spare time, Norman and I had been designing the hardware for the computer. Once you come up with a program, you have to imagine a circuit for getting it to work and a way to input and output data. We devised a plan for this and drew up schematics. But still in January and February, and even later into the spring, we were struggling to get Raymond running.”
Computer troubleshooting is known as debugging, which Anthony Chandor, in The Penguin Dictionary of Microprocessors, defines as “the process of testing a program and removing faults. Ideally, a single phase in the development of a program in which the program is run with test data to test all branches and conditions that may exist in the program. Unhappily, debugging can often continue throughout the working life of a program.”
Chandor only hints at the misery of it. He neglects to mention the fright of debugging a program for which there is no test data, or the appalling fact that bugs can exist in the hardware itself. Computers can be lousy with bugs, so thick with them that one suspects—in the ultimate nightmare—that bugs are shuttling back and forth at will from software to hardware. This is comparable in human pathology to the mind-body problem, where disorders in one realm delight in sneaking into the other. Some philosophers use this fact to argue against the Cartesian splitting of mind and body, and so too do the more existential hackers, when confronted with the mysteries of debugging, sometimes suspect the identity of software and hardware.
As Doyne bluntly put it, “Debugging Raymond was hell. I remember spending at least a month on it, and doing practically nothing but that. It was really very nerve-racking.”
Norman’s thoughts were no more fond. “We had a miserable time troubleshooting that sucker. We got really depressed at one point, because it looked as if the computer was just not working. So we built a separate machine that clamped onto Raymond and made it run through its progra
m one step at a time. That’s when we discovered we had left out a three-cent resistor. The problem was as simple as that. We popped it in and the computer worked fine.”
With Raymond up and running, the Project jumped ahead to build its first casino-model microcomputer. “Raymond was our prototype,” said Doyne. “It was meant as something to debug on. We thought it was too big to take into the casinos, although later it actually did get taken in.”
It requires more than a handful of chips to make a computer. They are inert without electrons flowing through them. To generate and control this flow, the chips need to be wired into an electronic circuit made of transistors, resistors, and other components. Fragile and little bigger in size than the head of a pin, silicon chips come housed in plastic or ceramic cases known as DIPs, which is short for dual in-line packages. These resemble black centipedes with golden legs. Electrons flow up and down the legs, but only after the DIPs have been mounted onto circuit boards that hold them in place and organize the flow of current among them.
There are various ways to load chips onto a board. For Raymond, the Project had used a technique known as wire-wrap. The DIPs, after being mounted into sockets with pins of their own, had been plugged into a circuit board, and then a spider’s web of interconnecting wires had been wrapped on the underside of the board from pin to pin.
“We wanted to make the second computer much smaller,” said Doyne, “and we didn’t think wire-wrap would work. It looked good for prototyping, but not for production. That’s when we heard about a new technique, some kind of wonder wire whose insulation was supposed to melt off when you hit it with a solder gun. It did, but it also shorted out and melted all over the board. The wires were so ridiculously thin that they’d break if you sneezed on them. We ended up with a huge mess of Hansel and Gretel twine wound around the computer. We were jumping ahead too far when we built that one.”
Raymond’s first progeny—junked in the chip horde without having shown any signs of life—went unnamed. The Projectors returned to wire-wrap and started building another computer. By fitting the chips more snugly together, and clipping the pins on the bottom of the DIP sockets, they shaved a half inch or more off each dimension. After several weeks of soldering and debugging, they detected the first signs of life in the new computer, named Harry, after its paterfamilias, Norman Harry Packard. Raymond and Harry would take their places in this family history as the first Eudaemonic computers to cross the border and confront the roulette tables of Nevada.
Beyond building and programming computers, a third part of the Project’s “global plan” remained, at best, fuzzy in outline. Thinking their presence in the casinos would be less suspicious with tasks divided between a data taker, standing near the wheel, and a bettor, positioned farther down the layout, the Project had opted for a two-person system using computers run in tandem. The data taker, while setting parameters and clocking the ball and rotor, would play penny-ante stakes, and might even choose to place losing bets. The second player, wired with a computer that received and decoded signals sent from the data taker’s computer, would play the high-stakes game. Standing far from the wheel while racking up an obscene pile of money, the bettor could foil suspicion with any number of innocent disguises.
Chips of the gambling variety have a magnetism of their own. They draw energy like a short circuit. Stack enough of them in front of a player and crowds gather to stare. This is when the management gets worried and turns on the heat. At either elbow come the inquiring eyes of croupiers, pit bosses, shift foremen, shills, floormen, and dicks. They ply you with alcohol, distract you with questions, chew gum, sneeze, rub their crotches. “What,” they ask themselves, “is causing this untoward good fortune? Is this guy on the level, or is he doing us dirt?”
While beneficial in terms of security, a two-person roulette system calls for some tricky engineering. It requires the communication of signals, by radio or other means, between one computer and another. Once transmitted, these signals have to be decoded into some humanly understandable form—such as an LED read-out built into a pair of spectacles, or tones sounded in the ear canal, or shocks or thumps received on various parts of the body. The problems involved in designing a two-person system such as this—one demanding both computer-to-computer and computer-to-human communication—lie in the realm of electrical engineering. The only person among the Projectors with experience in this domain, coming from his childhood days as a TV repairman in Silver City, was Norman.
To begin with, Norman tackled the problem of intercomputer communication. He decided that a radio link, which was simple enough to build, was also too obvious. Large casinos are wired from end to end with detectors for bombs, cameras, guns, and other objects identifiable within the normal band of radio frequencies, and one could expect to find sensors in the Eye in the Sky above every roulette table.
Having vetoed a radio link, Norman explored other options, including ultrasonics, which are untraceable without special detectors, and an optical system designed around infrared lasers. Built into the heels of shoes, lasers could transmit signals through a sheet of infrared light invisible to the human eye. “Both ultrasonics and lasers worked perfectly well,” Norman reported, “so long as no one was standing in the way.”
It was Tom Ingerson who suggested the solution to Norman’s problem. While on sabbatical in Chile, Ingerson maintained a steady correspondence with Santa Cruz in letters full of everything from philosophy to wiring diagrams. “I suggest you go the James Bond route,” he wrote, “and put a radio receiver in someone’s tooth. It isn’t hard these days, and one can send signals by biting down, and receive actual numbers to make the bets on by bone conduction, which cannot be heard by anyone else.”
As an alternative to the James Bond route, Ingerson advised the Project to transmit the signals by means of Faraday, or magnetic, induction. This works on the same principle as a transformer, where current passed through a coil of wire alters the magnetic field and thereby creates a current in a second coil of wire. The flow of voltage through proximate wires is described by Faraday’s law; hence the name for this kind of signal.
Unlike radio waves, which propagate through space in a radiation field, Faraday induction creates slowly varying magnetic fields that lose very little energy to the outside world. “They get weak so fast,” Norman discovered, “that beyond ten or fifteen feet you can’t detect them at all, even with the best of instruments. This was obviously a big plus. Our signals would be virtually undetectable to the Eye in the Sky, or anyone else in the casino not standing directly next to us.”
Norman flipped through his textbooks on electricity and magnetism and sat down to build a set of transmitters and receivers. Little did he suspect that they would be the bane of his existence for years to come. He would spend thousands of hours troubleshooting the mysterious malfunctions that plagued their circuitry. What was intended as a bright idea for slipping through casino security would strain, in its complexity, even Norman’s exemplary patience.
He was working simultaneously on the second of his two assignments: communication between computers and humans. One usually talks to computers through keyboards and visual displays, but for obvious reasons the Projectors required less obtrusive devices. For entering data into the computer, they had already replaced the keyboard with toe-operated microswitches. Now they needed some way to get information back out again. The options available to Norman included visual, audio, electrical, and tactile outputs.
He rejected the visual and audio as too obvious. Watches with LED read-outs or hearing aids sounding tones in the ear canal were both detectable. To examine tactile outputs, he got in touch with a company in Palo Alto that made devices for the blind. Their most promising product was a machine that translated printed text into buzzes readable by fingertips. But Norman found the machine too fragile and inflexible about the kinds of voltage it required.
Having ruled out visual, audio, and tactile devices, Norman turned to look at electrical
outputs. “The perfect solution, we decided, were shockers, which would be totally flat and easily concealed.”
Electrical shocks sent to the body could be decoded as signals identifying a particular octant on the wheel. Roulette wheels are conveniently etched with black lines that slice their central disks into eight pie-shaped wedges. The thirty-eight pockets on the wheel are not evenly divided by eight; so each wedge contains between four and five numbers. A computer narrowing the outcome in roulette to a particular octant, and allowing time to cover these four or five numbers with bets, would give someone a killing advantage in the game—if Norman’s shocks had not already done mortal damage of their own.
Using the tops of tin cans coated with a “special medical conducting goop,” he wired a ground electrode into the small of his back and four shockers to various parts of his body: one on each leg and two on the stomach. With combinations of shockers vibrating together, he could generate more than enough signals to identify the eight octants and a ninth “no bet” signal.
“The idea was to send voltage through the system, causing a little sensation, presumably not an unpleasant sensation, although it was hard to get the voltage right.” This was Norman’s mild-mannered way of admitting that the Project Room at one point resembled death row at Sing Sing, with human guinea pigs writhing on the floor in the early stages of electrocution.
After giving up on tin cans, Norman ordered from Hewlett-Packard special medical sensors used for measuring EEGs and cardiograms. “These,” he declared, “didn’t work worth a damn, even after we shaved our skin and gave each electrode its own ground. The current was too hard to control as it flowed over our bodies. Once it entered, it could travel around and come out all sorts of places. So after many months, we finally gave up on shockers.”
The Eudaemonic Pie Page 9