Each night Doyne and Browne moved the wheel over to the experimental physics laboratory, which Doyne had access to as head teaching assistant responsible for setting up daily experiments. Here they worked from eight at night until two or three in the morning. Using Norman’s electronic clock, as well as infrared detectors and high-speed film, they measured the physical forces at work in roulette. The ball’s trajectory, deceleration, scatter, bounce, and relative position vis-à-vis the spinning rotor were all quantified into the numerical stuff of physical prediction.
“I have fond memories of those late nights in the physics laboratory,” said Browne. “Afterwards we went skinny-dipping in the university pool, and then made it down to Ferrell’s doughnut shop just as the cinnamon rolls were coming out of the oven. We had only one close call after we started carrying our crate around campus. A janitor by the name of Fred Faria, who called Doyne ‘Mr. Professor,’ was dying to know what we did every night in the physics laboratory. He finally caught us with the wheel uncovered. He was impressed, but not overly taken aback. He thought we were running a gambling ring to pay our way through school. No big deal.”
As his academic progress reports slipped from “excellent” to “satisfactory,” Doyne finally came clean with his adviser. “I told him what was going on, that I wasn’t interested in doing astrophysics but wanted to take a year off to do roulette. I explained that that’s why I had been screwing around the past four months, because I had been spending all my time working on roulette.”
George Blumenthal, Doyne’s adviser, was sympathetic. He reviewed the equations and thought the scheme looked promising. He then told Doyne about his own career as a blackjack counter, which had ended abruptly on his being wiped out by a cheating dealer.
Late that spring there gathered in Doyne’s apartment a nucleus of people prepared to work throughout the summer building a predictive machine to beat roulette. Norman Packard and Jack Biles came down from Portland to join Dan Browne, who was already camped in Santa Cruz on Doyne’s floor. Another physics student, John Boyd, arrived from Moscow, Idaho. The lone humanist among them was Steve Lawton, a friend of Doyne and Letty’s from their undergraduate days at Stanford.
Tall, balding, gregarious, and more athletic than his spacy demeanor might otherwise have indicated, Lawton was a specialist in utopian literature. That spring he organized a reading group on the subject, and when not acting as gofer on shopping trips to the Silicon Valley, he led discussions on how best to organize utopian communities. This mix of theory and practice characterized life in Doyne’s apartment during the early stages of what was code-named Project Rosetta Stone, or the Project, for short.
“I used to think a lot about community,” said Doyne. “How to bring people together and make things happen. We wanted to build up a network of people we could trust and a set of tools. We started the Project thinking it would be a way to organize this sort of community. It would finance a home base for everyone. I may not touch home very often, but there has to be someplace to go back to. Otherwise it’s crazy in this society, where people drift off to Timbuktu and Philadelphia just because that’s where they get jobs. There has to be a better way for people to control where they live and how they stay in touch with each other.”
“We imagined the roulette project as a cash cow that would support our other interests,” said Norman. “Ever since our days in Silver City we had had the idea of starting a company to finance our electronics projects and travel to faraway lands and the study of physics. It’s been a long-standing pipe dream to put together a community that would allow us to live off the fruits of our ideas.”
Tom Ingerson, then on sabbatical at the Cerro Tololo Observatory in La Serena, Chile, was busy tracking astral bodies known as Seyferts across the night skies of the Southern Hemisphere. His spirit was often invoked in these early discussions; his practical advice took longer to arrive by mail. In dense epistles that filled both sides of five or more typewritten pages, he melded philosophy with physics and made numerous practical suggestions that were later adopted by the Projectors.
Their immediate tasks ranged from basic research to the design and construction of a computer. They needed investors, lines of credit, a checking account, stationery. Who would direct the Project, and how, eventually, would they reward all the people contributing either time or money? They began by taking out papers of incorporation. In spite of its official status, this company was to be run democratically, with all decisions arrived at by consensus and all future profits divided equitably between investors and workers.
“We needed a front for dealing with the outside world so that we could do things, like buy electronics,” said Norman. “We had to come up with a name and get a checking account under that name and keep track of all the money involved, because we were trying to be conscientious about money. We had in front of us the vivid image of profits. We wanted to keep the record straight about who would be getting what, so there wouldn’t be any conflict later on when the Project was earning hundreds of thousands of dollars.”
Doyne was leafing through the dictionary one day when he found the word eudaemonia. Aristotle posits the existence of many daemons, or presiding spirits, but his favorite among them was the eudaemon, or spirit of rationality. Eudaemonia describes that special happiness resulting from an active, rational life. Doyne’s dictionary defined it as “a state of felicity or bliss obtained by a life lived in accordance with reason.” When put up for a vote against Utopian Ventures and Amphibian Productions, Eudaemonic Enterprises (the first word is pronounced something like a Wagnerian recitative, yoo-die-mahn-ick) was chosen unanimously as the most suitable name for the new company.
Having at this stage of its existence no capital, retained earnings, or other financial assets, the best Eudaemonic Enterprises could offer its worker-owners in terms of payment was a share of future profits. This share would be sliced out of something called the Eudaemonic Pie. Filled with the totality of all future earnings, the Pie would be served out in portions calibrated to the number of hours or amount of money invested in the Project. All tasks, from building microcircuits to spinning roulette balls, would be rewarded equally. Outside advice on designing equipment, technical reviews of the physics involved, suggestions on how to implement the Project—these and other ideas would also be assigned hourly values and given a piece of the Pie. No one doubted that this conceptual dish would someday be rich enough to feed everyone who helped to make it.
Eudaemonic Enterprises imagined itself ultimately developing a number of pies. Project Rosetta Stone would provide the capital for launching other ventures, which ranged from the design of dirigibles and weightless cubes to the building of energy-efficient houses and a utopian lending library. Money from roulette would also go toward buying land in the Coast Range of Washington or California. Here, E.E. would construct its own utopia: a technological commune of friends and tools gathered for the purpose of putting science to human use. Between visits to Timbuktu and Philadelphia, all Eudaemons would have someplace to call home.
“The assembling of all the Projectors was an ecstatic occasion,” said Norman. “We were electrified and ready to blast off on this promising endeavor that wasn’t just another kinky way to spend our summer, but potentially a gateway to new levels of life.”
While speculating on the fruits of Eudaemonia, the Projectors more immediately set to work studying the game of roulette. “We began with a thorough feasibility study,” said Doyne. “We had to figure out the trajectory of the ball. Then we needed to pin down the problems of bounce and scatter. At the same time we had to research the hardware necessary to implement the system. What kind of computer would we build to input data and output predictions?” There was the ultimate problem of the equations themselves. What algorithm would integrate all the forces in the game with enough accuracy to predict its outcome?
When not working nights in the physics laboratory, the Projectors slept on Doyne’s floor or out under the redwoods. T
om Ingerson had given Doyne the Blue Bus as a graduation present, and Norman now moved into it with Lorna Lyons, a friend who had come to visit from Portland. Statuesque and imposing, Lorna has a face that looks from certain angles like a Leonardo portrait. Her voice is more American in scope. Imagine a Chicago grain report on sexism or mind-altering drugs. While shuttling around campus between the university’s long- and short-term parking lots, Lorna and Norman inaugurated what would become a long-lasting affair.
Later in the summer, when Professor Nauenberg in the physics department rented Doyne his house on Laurent Street, the Projectors moved to more spacious quarters overlooking the town and Monterey Bay. They divided tasks among themselves as follows: Dan Browne, Steve Lawton, and John Boyd would undertake physical studies of the ball’s trajectory and bounce. Norman Packard and Jack Biles would research computers and hardware. Doyne Farmer would analyze data and derive the roulette equations. “We expected to finish by the end of the summer,” said Norman. “Certainly by the following Christmas we planned to have the whole thing working and ready to take into the casinos.” Beating roulette proved more difficult than Norman imagined. His estimate was overly optimistic by twelve months. It was not until the following Christmas that they reached Nevada with their first computer.
Finely machined and oiled, the disk of a roulette wheel suffers little decay in its velocity. A roulette ball, on the other hand, confronts numerous opportunities for entropic degradation. As it drops from orbit and arcs toward a rendezvous with the spinning rotor, it passes through a veritable minefield of galactic booby traps. It encounters friction from the track on which it turns. It faces wind resistance, drag, and the ineluctable pull of gravity. Analyzing the ball’s trajectory would still be relatively straightforward were it not for further complications. Several metallic diamonds, pitched either horizontally or vertically, decorate the stator of a roulette wheel. On hitting a diamond, as the Projectors had already discovered, a roulette ball will be knocked off course. It may be lofted higher in its trajectory, or dropped straight toward the wheel, where it faces even more interference from the metal frets that separate the cups.
“There are several questions involved in whether or not the game is predictable,” Norman explained. “First you have to ask whether the wheel itself is predictable, which means that if you input two clicks into the computer, can it tell what the position of the wheel is going to be several revolutions hence? It’s clear that if you input the clicks accurately enough the computer can do it, but then you have to find out whether the clicks plus the error we introduce by being a little spastic, because we’re human beings, are still accurate enough for prediction.
“Then you have to ask the same question for the ball. Does the ball, as it’s running around the track, slow down uniformly? If the track is rough or irregular, this may not be the case. The last and major question: If we can predict the rotor motion and ball motion so that we know exactly when and where the ball is going to fall off and exactly where it’s going to hit the rotor—suppose we give ourselves that predictive power—then what happens? What happens is that the ball bounces around a lot. If it bounces around too much, then all of your predictive power up to that point is useless.”
Set up in the basement and covered with electronic gadgetry, the B. C. Wills roulette wheel underwent a series of tests that left it looking like a coronary patient in intensive care. The experimental wing of the Project began by stationing photo resistors around the wheel. In conjunction with little light bulbs beamed onto the track, the resistors recorded the ball’s rolling trajectory. “Running our experiments at night,” said Doyne, “it looked like Halloween, but we still weren’t getting sharp enough signals to trigger the timer on Norman’s clock.”
They then discovered optrons: tiny infrared-sensitive devices that work at close range like radar. These incorporated light-emitting diodes that sparkled ruby red as the ball passed in front of them. “We set these up at eight stations around the wheel,” said Doyne. “Because of the jungle of wires feeding from the optrons into an amplifier and then into Norman’s clock, each station was about as big as a cigarette pack. The optrons triggered whenever the ball passed in front of them. By noting the elapsed time of the ball from one station to the next, we could easily compute its velocity.”
The Projectors measured the ball’s scatter and bounce by means of a device they called the guillotine. This consisted of a scaffold built over the wheel on which were hung a camera and stroboscopic flash. The camera, the flash, and Norman’s clock were triggered by a second amplifier that picked up the sound of the ball as it hit metal. These photographs recorded the first strike and subsequent landing of the ball on the rotor. By plotting on two axes this relationship between strike and land, one could graph the ball’s average displacement, either backwards or forwards, along the wheel.
This graph of a frequency distribution is known in statistics as a histogram. Variances in behavior are plotted on the x axis, and their frequency of occurrence on the y axis. Other than a strike-land histogram, another variety of obvious interest to the Projectors—especially when they had finished building their computer—would record the ball’s displacement either backwards or forwards from the predicted number. Over the life of the Project they would record hundreds of these predicted-actual histograms.
In the early strike-land histograms, where only the behavior of the ball was being studied, how far and in which direction it bounced were irrelevant. The Project cared only about consistency. “What we wanted to see in the data,” said Norman, “was a sharp peak, a tendency on the part of the ball to displace itself with regularity in one direction or the other. Instead, we got something looking more like a hump. Predictability in roulette is inversely proportional to the width of that hump. If the hump spreads out over more than nineteen numbers, that is, over more than half the wheel, it gets very hard to predict where the ball will land.”
The Projectors also examined the nature of roulette balls themselves. These are made out of substances ranging from ivory to human bone, although more common materials include nylon, Teflon, acetate, and the composite material of billiard balls. These substances vary greatly in their characteristics. Teflon decelerates 100 percent faster than the billiard ball composite, while nylon has more bounce than the others. The standard ball in Las Vegas is made of acetate, which was lucky for the experimental team. A good reflector of infrared light, acetate balls are easy to track.
“Along with gathering data, I was thinking a lot about the theory of roulette,” said Doyne. “What kind of equations does the roulette ball satisfy? A rolling ball on a circular track is no problem, but in most physics classes you don’t work much with friction. You seldom get beyond an incredibly simplistic model that takes account of either static friction or sliding friction. I had assumed the main friction would be between the ball and track, until it occurred to me to think about wind friction, which is proportional to the square of the velocity. I estimated what this coefficient of friction would be and discovered to my surprise that wind resistance alone is mainly responsible for the ball slowing down.”
The Projectors next discovered the importance of tilt. An ideal roulette wheel spins on a perfectly level plain, although no such surface exists in the casinos of this world. Having set their wheel on stainless steel jacks, the Projectors had spent hours trying to level it into a semblance of perfection. “Then I did a little calculation,” said Doyne. “Given x amount of tilt on the track, what difference does it make in the fall-off velocity of the ball? It turned out that a tenth of an inch was enough tilt to make a significant difference in where the ball came off the track. It’s very hard to get the wheel level within a tenth of an inch. So it was obvious from that moment that tilt would also play an important role in the prediction.”
Given these variances in tilt, bounce, scatter, and drag, the Projectors discovered further wrinkles in the roulette cosmos when they rented a second wheel from a novelty store in San Fr
ancisco. Rigged up under the guillotine for high-speed filming, the wheel revealed two further singularities. The San Francisco wheel was a wreck of its former self. A wobble in the central spindle slowed it down more quickly than the B. C. Wills wheel. The track was badly pitted, so that balls spinning on it would again slow down and fall off more quickly. This meant that the final algorithm for predicting roulette, if flexible enough to account for differences such as these, would require what are known in physics as adjustable parameters.
“When we rented the second wheel,” said Doyne, “we discovered that the parameters for wheels on which we might play were going to be different. That meant I had to sit down and derive mathematical functions for these parameters which would be adjustable. And then I had to think of a way to interact with the computer program so that it could change these parameters every time we went up against a new wheel.”
Doyne isolated no fewer than five variables in need of adjustment. There were two for the ball itself: one for measuring the rate at which it slowed down, and the second for establishing the mean velocity at which it fell off the track. He called the first of these the ball deceleration parameter. The second was known as the time of fall parameter. These shifted according to the kind of ball in play, or the curvature and condition of the track itself. Another variable measured the rate at which the cups on the central rotor slowed. Two final parameters dealt with tilt. They approximated its magnitude and swung the computer’s prediction around to the high side of the wheel, which is where a roulette ball tends to come off the track.
While the roulette equations were being worked out in detail, another team, directed by Norman Packard and Jack Biles, researched their implementation. It was one thing to have a wheel sparkling with optrons, but something more subtle was required for playing roulette in Las Vegas. Attracted to the exotic in technologies, Jack suggested they use lasers or radar. A simpler solution, however, lay in microswitches operated by fingers or toes. Two clicks recording successive passes of the ball in front of a fixed reference point would establish its position, velocity, and rate of deceleration. “Assuming,” as Norman put it, “that we weren’t too spastic in our clicking.”
The Eudaemonic Pie Page 6