Hoff was thirty-three years old when he invented the microprocessor. Not long out of Stanford—one among hundreds of bright students graduated from the physics department into the high tech factories along the Camino Real—he was eager, as he said, “to get out into the commercial world and see if my ideas maybe didn’t have some commercial value.” Noyce had turned him loose on a problem raised by some Japanese manufacturers of desk-top calculators. They wanted a calculator with math and logic circuits etched onto no more than eleven chips. It was while lost in the maze of designing this circuitry that Hoff came up with a new way of thinking about the geometries of silicon itself.
By adding extra dimensions to an already infinitesimal universe, he saw how to condense the eleven chips into one microelectronic circuit that would constitute the central processing unit, or CPU, of a new kind of computer. This “computer on a chip,” as Intel called it, required the support of additional chips providing it with memory and a program, as well as input-output circuits and a clock to synchronize its operation. But Hoff’s microprocessor—built on a single chip of silicon measuring one-eighth by one-sixth of an inch, and holding no fewer than 2250 microminiaturized transistors—sprang to life as the fully developed mind of a working computer.
At first not even the military knew what to do with the microprocessor. They had yet to comprehend the truth in Noyce’s prediction that “the future is obviously in decentralizing computer power.”
Intel, which previously had specialized in making semiconductor memories for computers, quickly became the world’s largest producer of computers themselves. With no idea as to who might buy such an exotic device, they christened their first microprocessor the 4004, wired it onto a plastic board the size of a paperback book, attached a clock, control devices, and four more memory chips, and launched the MCS-4 out onto the market as the world’s first “microprogrammable computer on a chip.”
Its beauty lay in its flexibility. Teach a microprocessor how to add, instruct it in the solving of differential equations, program it with an algorithm capable of integrating Newton’s laws of motion, and it can land a spaceship on Mars, or play roulette. “An individual integrated circuit on a chip perhaps a quarter of an inch square,” wrote Noyce, “can now embrace more electronic elements than the most complex piece of electronic equipment that could be built in 1950. Today’s microcomputer, at a cost of perhaps $300, has more computing capacity than the first large electronic computer, ENIAC. It is twenty times faster, has a larger memory, is thousands of times more reliable, consumes the power of a light bulb rather than that of a locomotive, occupies 1/30,000 the volume, and costs 1/10,000 as much. It is available by mail order or at your local hobby shop.”
On September 8, 1976, in a package addressed to Mr. “Dwang” Farmer, Eudaemonic Enterprises received in the mail its first computer. Shipped from MOS Technology, Inc., in Norristown, Pennsylvania, the package contained a KIM, or Keyboard Input Module, computer development kit. For the price of two hundred fifty dollars, the kit included an Intel 6502 microprocessor—the same one later used to make Apple computers—a second chip to serve as a memory, two more chips specialized for inputting and outputting data, a crystal clock, an interface that allowed the computer program to be stored in a tape recorder, a primitive Korean keypad, a panel with enough light-emitting diodes to display one line of numbers, and a plastic board on which to solder all these parts together. With the addition of four extra memory chips, Eudaemonic Enterprises had purchased, for a total of four hundred dollars, a computer intelligent enough to navigate the Newtonian cosmos.
Taking over the work done previously by the university’s PDP 11/45, the KIM would serve as the Project’s mother computer. The reason for switching from mainframe to micro lay in the latter’s adaptability. For another few hundred dollars, the Project could build onto the KIM a device for burning—or programming—secondary chips. Capable of reproducing itself ad infinitum, the KIM, once successfully programmed with a roulette algorithm, could be cloned into the smaller computers that would actually be carried into the casinos. The KIM was going to be the Ms. Big of the operation, masterminding the work of her minion computers in Las Vegas while cooling off under the redwood trees in Santa Cruz.
When soldered onto its plastic card, the KIM functioned as an eight-bit microcomputer with five kilobytes of random-access memory. In learning the language of bits and bytes, remember that digital computers operate in the rudimentary world of base two mathematics. They manipulate binary digits, or bits of information. They think in spindly strings of 1’s and 0’s, which are themselves the symbolic representation of electrons passing through transistors. Thousands of transistorized locations are etched into a minute piece of silicon. Each of these locations in turn can be oriented as either “on” or “off,” 1 or 0. Permanently fix the magnetic charge that orients these transistors, and you have a chip that functions as a ROM, or read-only memory. Allow for the transistors to be reoriented, and you have a more interactive chip known as a RAM, or random-access memory.
Governed by a crystal clock that oscillates at a million cycles or more per second, electrons pulse through silicon circuits to produce the binary digits that are a computer’s smallest and, in some sense, only unit of data. These pulses are measured in nanoseconds—one thousand millionth of a second—but to speed the process even further, computers clump together bits and shuffle them in packages of four, eight, sixteen, sixty-four, or, most recently, two hundred fifty-six bits at a time. The incremental sizing of these packages is due to the crystalline geometries of silicon. A group of eight adjacent binary digits clumped together and shuffled as one unit constitutes a byte. What makes a byte important is the fact that an alphabetic character can be represented by one of them. A kilobyte is equal to 210, or 1024 bytes, although in common parlance this number gets rounded off, in this case, to a thousand bytes, or 1K.
The KIM—an eight-bit microprocessor with 5K of RAM—shuffled eight electronic pulses at a time through a memory holding up to five thousand bytes. These numbers alone are not impressive. A game of Space Invaders operates on only a slightly smaller scale. But while the video game is frozen into perpetual intergalactic strife with ROM, or read-only memory, that in the KIM was wide open to random access. Within the limits of computer logic, it could be programmed to do anything imaginable.
After their summer at Professor Nauenberg’s, the Projectors moved that fall into a house of their own. Doyne, Norman, and Letty had searched the county for someplace large enough to hold the first Eudaemonic household. They finally found a rambling, wood-frame structure at 707 Riverside Street, a few hundred yards from the beach and just back from the levees that keep the San Lorenzo River from flooding the town built along its banks. The house and its barn had once presided over this stretch of riverbank as their sole occupants. But the acreage had long since been sold off for beach bungalows and condominiums, the barn was sagging, and the house itself was in need of cosmetic, if not structural, attention.
The Riverside neighborhood, in its democratic receptivity, held a smattering of every element found in this sun-drenched town of fifty thousand. Tourists unloaded children and baja chairs into cottages rented by the week. Retired couples turned their gardens into mini-citrus groves or Shangri-las overrun with bougainvillea and fuchsia. High-tech employees from Intel, after an hour’s commute over the mountains, wheeled their Porsches into the front yards of otherwise unadorned condominiums. Other citizens, surviving somehow in an economy dependent on fish, Brussels sprouts, the university, a Wrigley’s chewing gum factory, food stamps, silicon chips, and tourism, used their front lawns for planting snow peas, fitting skylights into Dodge vans, rigging Windsurfers, grilling vegetables over hibachis, or reading Good Times, the local newspaper whose masthead slogan is “Lighter than Air.”
The flower-lined mall and cafés of Santa Cruz lay just across a bridge spanning the San Lorenzo, or one could stroll instead to the harbor, an expanse of blue water situated wh
ere Monterey Bay takes a final nip in the coastline before rejoining the Pacific at Lighthouse Point. Surfers off the Point shot the curl in Steamer Lane, one of the best surf breaks on the coast, while back in the quieter waters of the Bay one found a yacht harbor, a wharf with fishmongers selling the catch of the day, and a boardwalk complete with arcades and a roller coaster. The only incongruity in this pleasant neighborhood—which soon went unnoticed by its residents—was the screaming of riders on the roller coaster as they took the big plunge.
Besides its location, 707 Riverside had much to recommend it. A stone foundation, having already survived numerous earthquakes, supported a bank of stairs, a pillared porch, and a clerestory gable whose eaves and upturned roof made the house look vaguely like a Chinese pagoda. Despite its loftiness, the structure contained only one habitable floor, although one so extensive that it contained along its perimeter six bedrooms, as well as a living room, dining room, and kitchen built on a grand scale. The basement held two more rooms, with windows facing out onto a large back yard and the barn.
Then in her third year of law school at Stanford, Letty paid slightly over fifty thousand dollars for the house. “Norman and I were considering buying it ourselves,” said Doyne, “but no one at the bank would give us the time of day. They thought Letty was pretty suspicious, too, until she produced her stock certificates. It was clear sailing from there.”
Norman moved that fall from Portland to Santa Cruz and began his first year as a physics graduate student at the university. Letty came down from Palo Alto as often as possible. Juano, on being wiped out as a poker player in the card rooms of Montana, drifted back into town. The house filled up with other residents that included, over the years, scientists, teachers, lawyers, a pianist, a nurse, a volleyball coach, two Dutch film stars, and an Italian leftist from Milan. A way station for travelers and the headquarters of Eudaemonic Enterprises, 707 Riverside acquired the air of a commune, a physics laboratory, and a casino all rolled into one.
The Eudaemonic family fenced in the yard and planted a garden. They built tables and beds and bought other furniture at the Sky View Drive-In flea market. In a small white chamber off the front hall, Doyne set up the new computer in what came to be known as the Project Room. He lined the walls from floor to ceiling with shelves that he filled with shoe boxes containing electronic parts, technical manuals, spare chips, wiring diagrams, and other paraphernalia needed for assembling and programming the KIM.
The filigrees of silicon in a computer, its keypad, electronic circuitry, and clock are known as hardware. The second-level abstraction of a computer program—the set of instructions that actually endows the hardware with memory and logic—is called software. The KIM and other early microprocessor kits were shipped from the factory without software. And none existed.
As a computer the KIM was a tabula rasa, knowing nothing of language and numbers or their symbolic manipulation. This meant that it had to be addressed down at the brute level of electrons. In its state of near idiocy, the computer had to be spoken to in machine code, which is a combination of electronic bits no more articulate than one or two fingers wagged in the face of a gurgling baby. But from enough such repetitions even a dumb machine can wire its synapses into the pattern recognition of names and numbers. Playing with the computer ten hours a day, seven days a week, Doyne at the end of a month had taught the KIM how to multiply.
“First I had to teach it how to count to ten and back,” he said. “The 6502 microprocessor manipulates data eight bits at a time, which means that it recognizes only 28, or 256 numbers between 0 and 255. This makes it complicated when you want to multiply 256 by 257. You also have to realize that for the early microcomputers ‘multiply’ was not one of the instructions. They understood only ‘add’ and ‘subtract.’ So you had to be able to multiply and divide any number that you were ever going to come up with by having the computer break it down into additions and subtractions of numbers between 0 and 255. And that’s not a trivial thing to do. It took me a month.
“Then I had to teach the machine how to do sequences of operations with variable data, which is necessary for solving equations. More time went into teaching the computer how to handle logarithms, which are mathematical functions not expressed exactly by any finite combination of ‘add,’ ‘subtract,’ ‘multiply,’ and ‘divide,’ although they can be approximated by series of these steps. I needed to learn how to program a computer, and there was no software anyway; so it made sense for me to write my own routines. I was also concerned about speed. When I calculated the order of magnitude to produce something like a logarithm, I got worried, because it looked like a significant chunk of time. I knew I had to get the final answer in less than a second, and I was shooting for a tenth of a second. As it turned out, I brought it in under a tenth of a second with no problem.”
Doyne drew up a “global plan” for organizing his work in the Project Room. He devoted part of every day to speculating on the theory of roulette, but the bulk of his time went to learning about computers. After soldering together the KIM and teaching it how to do arithmetic, he next had to instruct it in the logic of thinking.
“Microprocessors had been around for a couple of years, but the chip we were using hadn’t been available for more than five or six months. The manual was freshly printed, with numerous errors, and no one had written any higher-level languages or assemblers or software support. I had to learn to program by directly entering binary numbers into the computer. A computer program is nothing more than a sequence of these numbers. The first of them is an instruction, which corresponds to one of the two hundred fifty-six things that the computer knows how to do. Now it understands what to expect from the next number, which will be either another instruction or a piece of data for it to act on. Further along in the sequence, the computer can make decisions, perform arithmetic calculations, or stroke its input-output devices. It can also handle ‘interrupts,’ which instruct it to jump around the program and look for other sets of instructions. The computer doesn’t necessarily move through its program linearly. It can make loops and branches and jump among the numbers in some fairly complicated ways. And that’s why computers can do nontrivial things.”
As mapped out in his global plan, Doyne’s work in programming the KIM represented only a small, if important, part of the project’s attack on roulette. He also needed to solve the equations required for actually beating the game. “It was an ongoing process of figuring out how to make the predictions, along with careful derivation of the formulas and their testing through error analysis. It took me a fair amount of time just to come up with the basic idea of the program.”
The game of roulette, with a ball revolving around a spinning disk, represents a model universe governed by the laws of Newtonian mechanics. Planetary ball circles solar disk until gravity sucks it out of orbit and pulls it down to stasis. The equations of motion governing this galactic drama are accessible to any freshman physics student who understands the meaning of F = ma. But various stumbling blocks have stood in the way of calculating this heavenly rendezvous and made it a classic problem in physics from the time of Pascal, who was Newton’s contemporary, to the present.
Although the roulette cosmos works within the laws of gravity and planetary motion, its initial conditions alter every time the game is launched into play. This is comparable to the God in Newton’s watchmaker universe reaching down fifty times an hour to tamper with the mechanism. To attain predictability in a world as fickle as this, one needs to clock the velocities and chart the relative positions of ball and rotor at the start of each game, and then calculate their eventual rendezvous within the ten to twenty seconds between the ball’s cosmic launch and fall from orbit.
Humans not being fast enough, the only device competent for this act of celestial navigation is a computer. But when thought about in its totality, programming this computer becomes a daunting assignment. One has to derive and, if possible, solve the equations of motion governing roule
tte. The functions describing the game’s individual parts have to be integrated into one comprehensive function, or algorithm, capable of making a split-second prediction. Other constraints make designing this computer even more of a challenge.
The machine requires, even in attenuated form, what are known as peripheral interface devices. This is the generic name for keyboards, terminals, joysticks, thumpers, microswitches, voice simulators, LED displays, and other means by which humans get information in and out of the central processing unit, or brain, of the computer. Through some such device, or combination of devices, the computer, at the start of each session, needs to be informed about the game’s initial conditions. And then, on calculating the ball’s trajectory and final point of contact, the computer needs a means of outputting its information. This computer with its peripheral attachments has to be battery operated, long lasting, concealable, silent, reliable, undetectable, and smart.
To predict the future and unfurl its mysteries out to the end of time, claimed the French mathematician and astronomer Pierre Simon, marquis de Laplace, all one needs to know are the position and velocity of matter in the universe at one instant in time. The same principle pertains to the mysteries of roulette. To become its Laplacian intelligence—its god of predictability—one needs to do four things. Predict how far the ball will travel before it falls off the track. Determine when the ball will drop from orbit. Predict how far the numbered cups on the rotor will have traveled by the time the ball falls into one of them. Add together the distance the ball travels and the distance the rotor travels to correlate their relative motion and the timing of their final conjunction.
The Eudaemonic Pie Page 8