Grant's Pet Shop Chapter 5

Professor Carl Eiger was sweating even more than usual this afternoon because General Pratt Houston was en route from Washington to Livermore. As the head of Lawrence Livermore Laboratory's Smart Weapons Group, Eiger would receive the General's full attention as soon as routine diplomatic courtesies were attended to—maybe even sooner, knowing the General. Eiger was very much aware that the real reason for the General's visit had nothing whatsoever to do with official business. Eiger hurried his preparations accordingly.

Still superficially resembling the young computer hacker that he had been before the General had recruited him to work at the lab, Eiger sported a neatly trimmed beard and shoulder length hair tied in a thick pony tail. He typified the Silicon Valley high tech executive look—establishment suit coat and tie above the waist, blue jeans and loafers below. His exceeding poor eyesight, a consequence of years spent staring into computer monitors, demanded that he wear heavy horn rim glasses which rested uncomfortably on a thin angular nose. Eiger constantly sponged at his forehead and wiped his glasses to soak up a never-ending deluge of perspiration which he emanated in spite of the lab's air conditioning—a new trick that his body had played upon him since his conscription by the General.

Eiger's graduate studies had been conducted at the Massachusetts Institute of Technology Media Lab, a powerhouse for innovation in computer technology. His genius resided in what was then the arcane field of artificial intelligence—specifically in creating programs that processed data in parallel fashion. Most computers, even the Cray supercomputers, performed their calculations one at time in series. Parallel processing computers, however, are able to perform many functions simultaneously—just like the human brain. This radically new kind of computer architecture provided the foundation for the newly developing field of artificial intelligence.

The Media Lab's Connection Machine was an experimental parallel processing computer endowed with great speed and awesome calculating power. Containing 64,000 processors in a multi-dimensional array, the "CM" offered the first opportunity to quickly solve enormously complex problems, like assembling and analyzing the vast amount of data needed to accurately make weather predictions.

Eiger's doctoral research was an attempt to teach the machine a simple, but vitally important attribute of the human brain—the ability to learn through trial and error experimentation. His work plan consisted of programming a host of parameters into the CM then providing a set of rules which specified exactly how the CM could manipulate the parameters. He also designed a performance testing sequence by which the CM could evaluate whether the outcome had merit or should be discarded. The topic to be learned was the fastest way to New York City from Cambridge under a wide range of traffic conditions; rush hour, weekend, winter weather, etc. The CM tried thousands of alternate routes under a given set of circumstances until it found the optimal path. Once the CM had "learned" the procedure, the task was changed to Chicago then Los Angeles.

The research proceeded much faster than anticipated—so fast that Eiger feared that his work might be considered trivial and inconsequential. Therefore, Eiger following his well developed instinct for survival in graduate school, he stalled for time. For several weeks he fooled around with the program, tweaking it to increase its performance. At the end of this period, he began to speculate that there might be an immediate practical application for his program. And nothing could be more practical than to have the CM use his new program to design a faster microprocessor for itself.

In simple terms, Eiger designed a set of software algorithms that taught the CM how to analyze the layout of circuitry on an existing semiconductor chip, then to experiment with variations in the layout (using a mathematical model) until it improved the original design—at least in theory. Programs like Eiger's would eventually serve as the architect of a future new generation of supercomputers.

Eiger assembled the massive documentation for the CM's Intel master microprocessor and scanned it page by page into the computer. Then he loaded his designer software and activated the program. By dawn Eiger was astonished to discover that the procedure was a success. In addition to thousands of attempted modifications that failed and were rejected, one version appeared in theory to hold the promise of astronomical improvement—at least in a simulated mockup design.

Naturally, he yearned to test the results to see if they would actually improve the CM's performance—or even function for that matter. But by now he had diverged so far from his thesis topic in artificial intelligence that he was reluctant to reveal his new program to his advisor. Eiger had already been warned by his doctoral committee not to get side-tracked. During his five years at the Media Lab, he had developed a reputation for not completing a project. Now he was gun shy. He couldn't risk any kind of unusual dialog with his committee. He merely had to type up his results and submit them for review. But this time things were different. He had finished his project and now he wanted to go onto something else.

If I present a proposal for continued studies, most likely they'll just say "Here's your degree, son. You have graduated. You can pursue this at your next faculty position."

But no one else has a CM and my designer program won't run on anything else. I just have to find out if it works before I leave...maybe there's a way.

He found one. The Innovation Branch at the Media Lab was heavily underwritten by RICOH of Japan, one of world's foremost manufacturers of semiconductors and one of the few companies which bothered to invest in the expensive development of one-of-a-kind prototypes. Every week dozens of chips were microscopically wired by hand and sent by courier to the Media Lab for testing. Diskettes with design modifications were sent back and new chips were returned for another round of testing. There was a lot of coming and going.

But these are simple chips, like those used in PC's. Nothing as complex as a CM microprocessor. But if I broke my new design into components they might not notice anything out of the ordinary. I could pop the subunits into a special card and achieve the same result as if with one master chip. I'm sure that it would work. Now how do I get them to make the subunits without finding out what this is all about?

Eiger broke the security system that protected the Media Lab's inventory management database. It was an easy matter to set up a fictitious account and to enter his design specs into the traffic flow to RICOH. A month later he picked up the packet from the delivery room without so much as a question asked.

During the last two weeks Eiger had shifted his time slot on the CM from midnight until 6 am in order to minimize the risk of discovery. During the daytime he poured over the volumes of circuit diagrams that depicted the CM's inner workings until he had the component names and functions indelibly inscribed in his memory. On several occasions he had pulled open the CM lid and run through a dress rehearsal. It was like looking into a pot of multi-colored spaghetti—each strand of wire twisted into a tangle of capacitors and resistors. One slip and he would have a serious problem. He was basically a programmer not a computer engineer, but pulling out the existing processor card and substituting another seemed easily within his ability.

Still Eiger waited. He evaluated for days on end the potential liabilities of installing his new processor. What if it shorts out the entire processor array? He decided a hundred times that the dangers were too severe, the reward not worth the risk, but the card with the new chips sat in his bureau drawer enticing him to give it a try. And another hundred times he decided to do it after all. Finally he knew that walking away from the lab, degree in hand and not knowing if his project would have worked was worse than not having a degree at all. Tonight he would install the new processor.

He waited until three, then casually checked the outer lab. A rather pretty girl was slumped forward in her chair in front of a Macintosh apparently asleep with her head on her folded arms. Joe, the janitor, was mopping the floor around the coke and candy machines. This was the most deserted he had ever seen the lab—even for the wee hours of the morning. He returned to the CM room and took a deep breath. In less than two minutes the new card had been installed and the CM was operating smoothly. He ran the accessory program that monitored its calculation rate. The CM's capacity to perform, measured in operations per second, had leaped from its normal one billion OPS to over a hundred billion! His new chip was an unqualified success.

By four in the morning he was back in his apartment lovingly fondling the new card. But the moment of exaltation was short-lived. Now he wondered what would happen if he ran his designer program with the new card installed. Would it work again, could the CM be driven even faster? Eiger went to the bathroom and looked in the mirror. He pulled off his glasses and had to lean closer to see his face. His eyes were bloodshot streams of molten lava surrounded by deep circles of volcanic ash.

I gotta get out of here. I need time to think this over. I'm turning into a Jyckll and Hyde spook show. I need some fresh air. I haven't been for a hike in the mountains in a year. I gotta go home and let this rest a while.

A week of strenuous hiking and cycling in the thin air of the hills around Boulder restored his body, but not a moment went by in which the temptation to try his program once more failed to consume his every thought. Finally the compulsion to know simply exceeded his fear of detection. Upon returning to MIT, he immediately installed the board with the new chip and ran the designer program. In less than an hour he had another improved design. He surreptitiously slid the diskette with the new design into the out going mailbox labelled RICOH.

Again the new chips exceeded even those incredibly fast ones just designed. In too deep now to quit, Eiger sent a third generation design to RICOH for prototyping and continued to repeat the design prototyping cycle until the CM could map a new generation chip in only a matter of minutes. The CM now ran faster than any device ever built by man. It had transformed into a machine rushing to fulfill its own destiny—the CM was evolving. And by now, because of the billions and billions of trial and error iterations, even the CM's vast memory lacked the vaguest recollection of the path that it had taken to create the latest chip complex. Certainly no human could ever re-create the nearly infinite number of experiments that the CM had conducted nor reconstruct the monumental flow of decisions that the machine had made. There was no trail to follow, no gigantic room sized flow chart tracking the latest processor's circuitry. It was if the final, super chip had simply materialized out of a vacuum.

Eiger was acutely aware that he was treading on very thin ice in tampering with this particular computer—the Connection Machine was the personal invention of the Media Lab director who took extraordinary pride in its creation. Eiger religiously replaced the now archaic original processor after each of his sessions.

But something went wrong. When the fifth generation chip was removed and the original version replaced, the CM exhibited an acute form of aberrant logic—almost as if it had developed withdrawal symptoms. Its internal diagnostic systems had apparently adjusted themselves to the vastly improved processor. The machine didn't "want" to operate a million times faster on Eiger's chips, then be reduced to just an ordinary supercomputer when the souped up processor was removed.

One late session, perhaps Eiger had become careless, the Lab director caught Eiger new card and screwdriver in hand. The director suspected sabotage and called security while shouting down Eiger's futile attempt at explanation. Since the Connection Machine was sponsored by the U.S. government, specifically the Office of Technology, the matter became a federal offense. The FBI was brought in and Eiger was arrested and taken to Washington, D.C. The lab director was infuriated to learn several months later that Eiger had quietly received his Ph.D. from MIT—at the General's insistence—and that the standard requirement for a dissertation had been waived. The director even dared to call Houston to complain, but Houston curtly told him to mind to his own affairs and had hung up on him. In light of the Media Lab's dependence upon the General's continued goodwill for mega projects like the Connection Machine, the director merely grimaced when it was announced a month later that Eiger had been appointed to a major position at the Lawrence Livermore Lab.

Unlike the director, General Houston had listened to Eiger...very carefully as a matter of fact. Houston was more than willing that Eiger continue his research. He even suggested that Eiger could have his own Connection Machine as well as funds to live in luxury or he could spend the rest of his life in prison and be publicly branded as a traitor. Of course, Eiger also swore his personal allegiance to the General as part of the deal...and then be began to sweat.

Not only was Carl Eiger an expert in artificial intelligence, he was also gay. In the state of Massachusetts in the year 1980 that was a very intense situation for the young scientist. His response was to suppress his sexual tendencies—even to deny sex entirely. This posed long lasting problems for Eiger. Many fervent emotions were looping through his brain, many of which he had no conscious awareness. But he was smart, very smart. And it didn't take a computer genius to realize that he had been had by the General. Something had gone desperately wrong with his life. Always a pacifist at MIT, he was now at the beck and call of a general who possessed immeasurably twisted plans and the most sinister of intentions. What little pleasure Eiger derived from his work, like the Government Defense Network Project which he undertook within a week of his arrival at Livermore, was overburdened by a deep despair for his life—even to the point of not caring if the General succeeded or not in his wild plot. Eiger chain smoked and never exercised, Even still he lost weight until his clothes hung loosely on his frame like a sagging tent in a stiff breeze. Chronic insomnia combined with the General's persistent pressure had led to incipient depression. Eiger no longer cared much about anything except his one friend in the city. Eiger hated and feared the General and visibly cringed in his presence. He knew that Houston would object and most likely interfere if he found out about Matt.

Eiger's co-workers at LLL thought that he was quite odd. They had never seem him with a female companion at any of Livermore's compulsory social events, and his stark appearance drew attention—especially his constant nervous perspiration. Eiger looked physically used and wasted despite his relative youth. Not especially liking him, and envious of his apparent privileged status, his colleagues were eager to mention Eiger's quirks to the CIA agents during the incessant clearance checks. But, Eiger was brilliant and he had demonstrated his genius in his handling of the Government Defense Network "Big Scare" fiasco. But more importantly, General Houston personally reviewed all security reports before they were submitted to the Pentagon for processing. Eiger was both trapped and protected at the same time.

The massive olive drab Huey helicopter throbbed to a landing on the concrete pad fronting Lawrence Livermore Laboratory's Main Gate. The General was halted by immaculately polished, rifle bearing MPs at the entrance checkpoint. The General admired their weapons as they carefully inspected his identification papers.

Every LLL engineer, physicist, secretary and maintenance worker was required to wear a blue plastic badge with his photo and signature prominently displayed. Livermore security was intense. The inviolable process continued without interruption every hour of every day. Two armed security guards at each of the four gates rigorously checked all vehicle and personnel moving on and off the premises.

One MP chattered continuously into a handset linked to Security Central located deep within the complex—a dreary bunker enmeshed in steel and concrete. The guard's never-ending monolog served to relay to the main office every development and movement at the gate. If there was trouble, a small well-equipped army of trained combat soldiers was on call only seconds away. Even though the General was an internationally known figure, the guards checked his ID and cautiously inspected his leather brief case before admitting him into the compound. Houston was impressed. Once cleared, the General selected a route that would avoid the administrative offices. He and his escorts proceeded directly to Professor Eiger's lab.

Eiger stood at the door of the colorless hallway on pins and needles, chain- smoking furiously, sweat drips making dark spots on the dirty grey carpet. Per standard operating procedure, two armed guards waited with Eiger, his lab's single entrance was guarded twenty-four hours a day by very intense young military men with a rank never less than that of lieutenant. At last, looking just a little annoyed about having his ID checked once again, the General was cleared into Eiger's section. Impatiently he stormed into the laboratory and was immediately overwhelmed by the thousands of multicolored indicator lights that glowed like a miniature city at night spanning the width of the dark room. His heavy footsteps reverberated on the dull metal floor. The room seemed to held captive by the machine which had expanded its domain until it smothered the cubicle. Cables and flexible conduits sprouted from a honeycomb of ports forming a web which silently transmitted a vast stream of data and supplied electronic nourishment to the Connection Machine.

The laboratory was dark except for the radiance of the flickering lights and a green glow that emanated from a central monitor. The room was also frigid. A cloud of frost billowed from a duct overhead and seeped slowly downward forming a grey fog that sank slowly towards the floor. Eiger had often explained that the lab must be maintained just above the freezing point in order to keep the behemoth from overheating. Even the CM's internal circuitry was flooded with liquid nitrogen. Wispy vapors eerily snaked from dozens of vents at the top of the machine. As the frozen mist sank past the indicator lights, it momentarily glowed red, green, and blue.

A thick metal door braced by well oiled stainless steel locking cylinders slid smoothly into place sealing the lab. There was a whoosh of air and their ears popped. After several visits, the General knew that the CM would now search his body and brief case—as well as the premises for hidden microphones or electronic bugs. The CM would repeat this procedure at random intervals whenever an outsider was present.

The General gratefully accepted the silver coated down filled parka and zipped up the front. Eiger, in just a white lab coat, offered Houston a steaming cup of hot coffee and the General relaxed into a high-backed red leather chair at the console next to Eiger. It amused the General to see his breath even without smoking.

Blowing steam, the General exaggerated, "Damn Eiger, didn't you pay your heating bill? Its cold enough in here to freeze the ass off an Eskimo! Teach the old gal to talk yet?"

Although the CM was perfectly capable of speech communication, Eiger had insisted that it not be equipped with a voice synthesizer. He wanted all interactions to be personal and private. He was adamant that others not be allowed access to his baby. But strangely enough, he had never given his beloved computer a nickname.

Eiger shrugged and closed his notebook. He moved to a bank of red filing cabinets at the rear of the room and placed his index finger in the fingerprint lock which acknowledged his ID with a beep. He pressed number one on the digital pad and the top drawer opened. The microchip in the lock automatically logged the time that the drawer opened and the code name of the file that was removed. Some files, especially those stored on computer tape or diskette, could not leave the lab unless Eiger manually overrode the system. Otherwise as soon as they were taken through the door, a detector alerted security and the ever present guards with loaded M-16s and no sense of humor took aim right between the ears of the careless violator.

Although this was definitely a file that contained extremely sensitive information, Eiger had made certain that its classification was rated benign enough to allow the General to take possession of its contents. Not being much for words, Eiger handed the file to the General and said, "Yes, it can be done."

GRANT'S PET SHOP