A few days before the party I had received a phone call from Doyne. “Are you interested in playing roulette?” he asked. “We have a trip coming up soon, and I think we could use you.” Other than inviting me to play roulette with a computer in my shoe, Doyne had a favor to ask of me. He was going to be spending the year in Los Angeles at USC, and the rest of the Chaos Cabal was deep into their own research on strange attractors. This would leave Mark working alone on the Project down in the basement of the Riverside house. Would I mind dropping by the Shop now and then to say hello? Surrounded by RAMs, ROMs, and the twitter of a 150-page computer program stored in a tape cassette, Mark might appreciate the human contact.
Doyne and I were fellow students at UC Santa Cruz. This is primarily an undergraduate institution, with so few graduate students—about three hundred altogether—that it is not unusual for writers and physicists to talk to each other. I had known about the existence of Project Rosetta Stone for several years, but I was nonetheless surprised to learn how far advanced and well financed was the most recent of the Project’s incarnations. Doyne phoned me at a moment when I was suffering from an advanced case of dissertationitis. This disease attacks graduate students in the terminal stages of writing their theses. The sickness, which can prove lethal, is compounded out of a mixture of ennui, insomnia, and other psychophysical symptoms ranging from hair loss to satyr-ism. No wonder I leapt at the chance to play roulette in Las Vegas with a computer in my shoe. This was a far cry from the burden of consciousness. Doyne assured me I would have no trouble getting good enough to work the system in my sleep. Of course there would be some danger involved. But it would be far easier, I thought at the time, to confront the Mafia than my dissertation committee. A pit boss might take me into the back room and ask a lot of tough questions. But when it got to that point, the trick in talking to the boys in Las Vegas would be to divulge as little information as possible. How refreshing! After being so forthcoming, responsive, verbal, and informed, I was delighted by the prospect of clamming up, going underground, stonewalling it. When it came to playing a part in this film noir, I assured Doyne I was a shady character he could count on.
The first opportunity I got to drop by the Riverside house, I found Mark out in the back yard pulling scraps of copper-covered fiber glass out of a deep-purple acid bath. “I’m making printed circuit boards,” he told me. “I’ve never done it before so I thought I’d experiment first with small pieces.”
If you want to make a computer from scratch, as Mark was in the process of doing, you go to the store and buy some logic and memory chips, a quartz clock, and a few transistors and other components that you then wire together into a circuit. Among the various ways to make these circuits, the simplest is wire-wrap. The chips get mounted into sockets with pinlike legs, plugged into a fiber glass board, and then wrapped with strands of conductive wire. The Project’s early computers—Raymond, Harry, Patrick, Renata, and Cynthia—had been made by the wire-wrap method, but this technology was too bulky for producing something as dense as a computer sandwich. So Mark planned to switch to the more sophisticated technology of printed circuit boards.
A finished PC board consists of thin lines of copper foil laid over a fiber glass base. Into this map of lines, which represents an electronic circuit, computer chips are plugged directly without intermediary sockets or connecting wires. Hewlett-Packard and IBM, once their design departments have come up with a workable circuit, order their PC boards by the millions. Their production facilities are spiffier, but their methods for manufacturing PC boards are essentially the same as those employed by Mark in his back-yard acid bath. First comes design. You gather together all the components needed for building an electronic city—its memory banks, data libraries, buses, and central processing units—and imagine linking them to each other in close proximity. Then comes layout. You tumble the chips onto a mock electronic terrain made out of Mylar, push them around, and figure out the best system of highways and access roads for connecting them into a working metropolis. Next comes artwork. You make a black and white street map of what the city will look like when finished, complete with dots representing the copper pads into which the chips will eventually be plugged. Then comes photo reduction and the process of masking the artwork onto the PC board. Finally, there is the etching of the board in acid, so that with most of its coppery surface eaten away, all that remains on the face of the board are the swirling lines of a metal circuit overlaid on a fiber glass ground. “When you reach this stage,” said Mark, “all you do is plug things in.”
Immediately after Halloween, Mark had run into what he called the PC problem. Very few computer manufacturers make their own boards. They subcontract the work to smaller companies that specialize in the design, layout, artwork, or photo-etching of printed circuits. Eudaemonic Enterprises had also planned to subcontract work on its PC boards, but on leafing through the Yellow Pages and phoning a dozen companies in the Valley, Mark was surprised to find that they charged more than $2000 for artwork alone, and another $500 for a prototype board. This was more than the Project had counted on spending; so Mark undertook to design and manufacture the boards himself. With the exception of the internal components, the Eudaemonic computer sandwiches would now be completely homemade. Mark was embarking on high tech as a cottage industry.
Using pencil and paper, he started designing a circuit. “It’s like finding your way through a maze,” he said. “You’re working with a three-dimensional tangle of lines that you have to fit onto a two-dimensional surface. In the Valley there are entire layout departments that specialize in this stuff. They even have computers programmed to solve the maze problem for you. A computer can generate more versions of a circuit than a person can stomach. By the time I’d finished, I was absolutely sick of it, but a computer has no capacity to feel nausea.”
Mark worked at a makeshift light table. Using tracing paper, he drew dozens of trial circuits for the components he had to stack and connect to each other. “Given the number of lines you’re dealing with, you can’t solve all the problems at the same time. So you do many versions of the circuit, and with each version you solve some problems and introduce new ones. You may get three lines properly connected, only to discover that another one is trapped. You can’t cross lines over each other without their shorting out. So you work on the circuit one unit at a time, and when you get part of it looking all right you copy the lines onto tracing paper, before going back to fix another part that looks disastrous.”
On solving the maze problem, Mark started the artwork. He used thin black tape to mask a prototype circuit onto a piece of Mylar that resembled a sturdier version of cellophane. He was working big, making a mask that would later be photographically reduced by half. Even on this larger scale, the tape was no bigger than .032 inches wide. After laying down the lines and making certain that none of them touched, Mark attached stick-ons representing the little doughnuts or pads of copper into which the chips would ultimately be pinned.
When he finished weaving a tidy maze of lines and pads, Mark took his artwork to a photo shop for reduction. They produced a negative of the layout exactly the same size as the prospective PC board, except that the black lines and pads in the original now appeared as transparent filigrees laid over a darkened background. Transferring this negative onto a board is exactly like making a contact print in photography. Light is shined through the negative onto a surface that is then developed, washed, and fixed. But instead of using photographic paper, electronic circuits are “printed” onto copper-covered wafers of fiber glass one-sixteenth of an inch thick.
Mark had spent a month designing the circuit; now he had to manufacture the boards themselves. Working in a darkened room with light-sensitive lacquer, a 150-watt light bulb, a solvent bath, a sun lamp, and an etch solution of ferric chloride, he nursed his materials through the half dozen steps required to transform a negative mask into the positive tracings of a computer circuit. On pulling the first PC board out of it
s acid bath, Mark looked at the lines swirled on it and thought to himself that all the design problems involved in building a computer in a shoe had been solved.
“That day I was very high. As soon as I saw the first board come out of the bucket, I knew it was going to work.” Drilling, loading, and soldering a circuit this small requires painstaking effort. The attached components would be hard to troubleshoot. Mark had yet to figure out exactly how he would fasten the boards together into sandwiches. And there was the final problem of building the computers into false-bottomed shoes. These “minor technical details” would take him another year of full-time work to solve, but as far as Mark was concerned, the Project was already a complete success.
13
The City of Computation
Uncertain fortune is thoroughly mastered by the equity of the calculation.
Blaise Pascal
The acacia trees are in bloom and the air sweet with the smell of freesias and other spring flowers when Doyne phones to ask if I want to go shopping with him in the Valley. He is not referring to food or clothing, but another item sold there at dozens of outlets large, small, discount, and deluxe. He is talking about shopping for silicon chips. The Project’s homemade PC boards, after being drilled with holes and trimmed to size, are ready for loading with the hash of RAMs, ROMs, transistors, diodes, and other components out of which computers are made.
Leaving Santa Cruz early one afternoon, Doyne and I drive through the redwood groves along Highway 17 and over the Santa Cruz Mountains to San Jose. This megalopolis anchors the lower end of the Santa Clara Valley, now better known as the Silicon Valley. We turn north on Highway 101 and penetrate deeper into the pink smog of Santa Clara proper. Passing through the contiguous cities of Sunnyvale, Mountain View, and Palo Alto, we make stops along the way by exiting off the highway onto the wide boulevards of old farm towns that are now homogenized into bedroom communities and support services for Hewlett-Packard, Intel, Memorex, Teledyne, Synertek, Siliconix, and dozens of other companies with names assembled out of electronic acronyms. The computer factories themselves are housed in massive sheds built windowless for the sake of air conditioning. Other than tile roofs and glass entryways, they offer little more to the eye than acres of prestressed concrete. Surrounding the sheds are landscaped parking lots and fences with gates that control the flow of cars. Departing vehicles stop at the gates, while guards stationed in air-conditioned booths come out to search the briefcases of passengers, and sometimes their pockets.
“Theft is a big problem in the Valley,” Doyne tells me. “A briefcase full of chips can net you thirty thousand dollars.”
Rather than stopping at corporate headquarters, Doyne and I pull the Blue Bus into the unguarded parking lot at Halted, a surplus components store in Santa Clara that looks inside like the garage of a hacker gone wild. Cannibalized TV sets, radios, and antennas line the walls, while the rest of the building is filled with gray metal shelves overflowing to the ceiling with boxes of electronic parts. The contents of each box is marked in pen, with specifications given in ohms and farads, although many boxes are also labeled “Miscellaneous” or “?.”
Doyne hands me a carton of “Assorted Resistors” and tells me to pick out the smallest of them, with values measured down in the range of mili-ohms. The cylindrical bodies of the resistors are color-coded in a rainbow of stripes, so that I hold in my palm what look like dozens of miniature African trade beads. Doyne gives me other boxes to sift through for capacitors measured in pico-farads, or 10-12 of a farad. Resembling tiny lollipops with paper handles, the capacitors also come color-coded in greens, blues, and purples. After buying several spools of hair-thin antenna wire, we walk out the door with $75 worth of merchandise slipped into a paper bag the size of a Mars Bar.
Driving north up the Valley, we pass what might be a storage park with dozens of cubicles for rent, until I see from the sign out front that this is a chip factory. We drive in front of another building constructed out of adobe arches strung next to each other in what looks like a mile-long McDonald’s. After Halted, our next stop is Anchor Electronics, which turns out to be nothing more than a storeroom with a glass partition through which a woman dispenses chips. Doyne has phoned ahead, so the woman is waiting for us with a selection of CMOS RAM and EPROM chips packaged in two antistatic plastic tubes, each a foot and a half long. CMOS is the generic name for a family of chips designed to work at low power and under a wide range of temperatures and conditions. Good for cruise missiles and B-1 bombers, these chips also perform quite nicely in shoe-mounted roulette computers. Doyne needs more capacitors, so he stands at the window for twenty minutes running down a checklist of values. “We want the smallest you have,” he repeats to the woman over and over again. She asks no questions, but soon stops supplying plastic bags for the capacitors. “At two cents apiece,” she explains, “the bags are too expensive to keep handing out.” Doyne pays her with a check from Eudaemonic Enterprises, and we walk out the door with two tubes of chips and a sack of capacitors for $187.
Trying to beat closing time, we push the Bus up the Valley to Zack Electronics in Palo Alto. Zack’s looks like a hardware store with a long counter running the length of one wall. But the nuts and bolts being dispensed are of the high-tech, high-priced variety. Palo Alto is a classy neighborhood; so antenna wire that costs $2 at Halted goes here for $11.99. Doyne keeps two people busy behind the counter sifting through boxes of components. He buys a solder gun with an extra-fine point, a roll of antenna wire, four miniature 15-volt batteries, and a handful of resistors for $150. The last customers out the door, we filter through rush hour traffic and head back over the mountains to Santa Cruz.
Several weeks later, when I next see these chips and components, they are wired together into a prototype computer known as a breadboard. The word is also a verb, and the meticulous work of assembling computers is known as breadboarding. Norman walks into my house one afternoon carrying what looks like a box of Kodak photographic paper. “You might be interested in looking at this,” he says, opening the box. “It’s a breadboarded computer. We have here just about everything that goes into one of these gizmos.” I look inside the box to see two pieces of white Styrofoam into which a handful of black chips and multicolored components have been pinned. They remind me of entomological displays in the Museum of Natural History, where one sees exotic insect species captured from microhabitats radically different from our own. Missing from Norman’s collection are the name tags, but in their place are filigrees of wire running from specimen to specimen. “When the chips are finally loaded onto a PC board,” says Norman, “they won’t need any wires. The board itself will connect them into a circuit, and everything will be packed much more tightly together.”
The components pinned to the two pieces of Styrofoam have been divided according to which half of the computer sandwich they will occupy. Segregated into memory chips and logic chips, the former species is the larger and more impressive of the two. The largest chip of all—obviously the queen of the collection—is a black-bodied rectangle with no fewer than forty legs. “That’s a MOS Technology 6502 microprocessor,” says Norman. “You might say it’s the brains of the operation.”
The microprocessor is the one component so important to the operation of a microcomputer that the words microprocessor and microcomputer have become almost synonymous. “There aren’t a great many microprocessors to choose from,” Mark later informed me. “There might be twenty different kinds, but most are part of a family of components. Of the five major families of microprocessors, each family speaks its own language. The largest family grew out of the 8080 made by Intel, which in turn developed into the Z 80 made by Zilog in Cupertino. This is one of the fifteen companies in the Valley owned by Exxon. The Eudaemonic microprocessor is part of what’s known as the 6500 series. Originally made by MOS Technology, it got second-sourced by Mostek, a spinoff from Texas Instruments, which in turn got gobbled up by United Technologies.”
W
ere its plastic or ceramic case cracked open, the black slug of a microprocessor, which is also known as the CPU, or central processing unit, of the computer, would reveal under a microscope a gray lattice of silicon. Wave after wave of this lattice is serried into the ranks of what are known as registers or store locations, some of which are devoted to arithmetical and control functions, while others are capable of memory.
“The main task of the microprocessor,” says Norman, “is to shuffle data around, which it does through these forty gold pins. They’re sticking in Styrofoam right now, but later they’ll be soldered into a printed circuit. Each pin serves a different function, although the data running up and down it is limited to nothing more complicated than either a 1 or a 0. A pin can carry one bit of information, which means that it can be either ‘on’ or ‘off.’ But you can also put the bits together and shuffle 1’s and 0’s in chunks. A chunk of four 1’s and 0’s is called four bits. A chunk of eight is called eight bits, and so on. Each microprocessor has a characteristic word size, which is how many bits it can shuffle around in one fell swoop. Ours is an eight-bit microprocessor, which means that it manipulates eight-bit words. Other computers use words of different lengths, and word length is one of the main differences between microcomputers and larger machines. The IBM 360, for instance, uses sixty-four-bit words. A word that long requires huge amounts of processing, which is one of the reasons why those machines were never miniaturized.”
The Eudaemonic Pie Page 24
