We're on Our Way: My Year As A High Tech Entrepreneur: My Year As A High Tech Entrepreneur

By Michael Stockdell

In the 1980s the boom for start-up high tech companies began to explode, owing mainly to the development of the microprocessor. Though not yet with the ferocity of the Internet bubble around the turn of the millennium, the earlier period was wild enough that companies with little more than a wish and a prayer were able to attract millions in ven

• Language: English
• Publisher: Michael Stockdell
• Release date: Apr 18, 2020
• ISBN: 9781734684513

Michael Stockdell

After graduating from the University of Virginia with a Bachelor's degree in English, Mr. Stockdell began a career in systems work, which at the time included manual as well as computerized processes. After three years in Federal government employ, during which time he wrote what was considered to be the best systems report ever created in the Department of Agriculture, he took a job as a Systems Analyst for an internal computer department. For the next twenty-five years he was involved in a variety of jobs in high tech, both in business and engineering/ scientific computing. He was responsible for an integrated Order Entry System, which included Billing, Shipping, Inventory Control, Sales Analysis, and Accounts Receivable, which led to the company becoming first in its industry in order-delivery cycle. He was a Management Consultant for a (then) Big Eight accounting firm and founder of Digital Equipment Corporation's Management Consulting program. In the late 1980s he became Executive Vice President of Paralex, a startup Parallel supercomputer company. This is the origin of this book.

Book preview

Chapter ONE

An American Tragedy

My story is a tragedy. Though not the part about me. And not the kind of tragedy told by the contemporary media. No school shooting this. No fiery wreck on I-85.

I aim to tell a much older kind of tragedy, the sort that lit up the stages of ancient Greece, where the end might not be death at all and where the hero is far from innocent. It is by his own actions that an Oedipus or Agamemnon is brought down; this is why he is called not a victim but a protagonist.

Like those ancient heroes it was no small-minded action that brought Trevor Hicks to ground. This is no noir-story of a little man going after little stakes, but of a superior man, a king in his own way, a man who dared to challenge the arbitrary rules of civilization which go for gods today. A man who grasped at the moon and when he knew it was within his reach went after the stars just because they were more. A man so certain of the purity of his vision that he would rather see it destroyed – and everyone and everything around him - then to see it compromised by ordinary men. An Ahab for the modern world, a scourge of God, an ubermensch.

And, yes, a bit of madman. For who in his right mind would feel so compelled as Trevor seemed to be? Yet he was a madman who was sane enough to make me believe that it all was true. And it was not just me he convinced, not just the needy little men who were the founders of Paralex, but Governors of states, captains of industry - skeptical men, practical men, shrewd men - and he convinced them easier and more completely than he did me. How was this possible?

Because they were mortal and had needs too, not for things - they already had all the trinkets they could ever want - but for the myth of the great and powerful man. For if great men do not exist, then who are the governors and captains of industry themselves? And if we are all ordinary, if there is no divine spirit that endows some of us with a special destiny, then all the trappings of civilization are without meaning and futile.

Trevor was almost what we all wanted to believe we could be - psychologically dominant, extraordinarily attractive, endlessly fascinating, whose very excess drew you to him, whose Protean restlessness meant that he could be and seem anything. Which in the end was his downfall. So engaged was he with the next great conception of technology, with business, with life, that you could never really tell from day to day what he wanted – or, indeed, who he was. Not just because he was always dabbling with his identity – although he was - but because your judgement of who he was came down to who you were, so much did he play on the needs and desires of each individual.

Was Trevor a Romantic genius whose powerful will swept the most reasonable of men before him? Or was he a psychopath whose uncontrollable passions were bent on self-destruction and the dissolution of every relationship he had? It all depends on how he left you at the end.

* * *

Now, some might think that I am the real hero of this tragedy. This is a memoir, after all, and it is through my eyes and my intellect that all the events of that time are filtered. It is also true that I as much as Trevor was brought down by the events of that era. And it happened at a time when it seemed I was just about to make it, at a time when at last achievements, talent and recognition seemed about to come together into a close approximation of success. Yet I, the ultimate analyst, the quiet rationalist, the man who never took risks, I threw it all over because…

Because I was caught up in the seductive illusion that my own weaknesses were of no real consequence, that the misfortunes of my career and my life were mere mischance, that all my past was a prelude to my own higher destiny. If I could tame lightning, I thought, lasso a tornado, quench a volcano, control Trevor, I, too, might learn how to step outside the rules that bound other men to mediocrity. And for that presumption my career was ruined.

I could never be a tragic hero in the way Trevor was destined to be to be, though. For all my gifts and personal qualities, I am a man with ordinary ambitions. All I hoped for was survival – and survive I have.

This is not to pretend that the collapse of my business career was not traumatic. But my job was never my identity and what drove me on in that dark time was not recognition or even the prospect of great wealth - but the opportunity to build something that would show ever after that I had lived a life on earth. And that requires not a vision but an execution, a slow, plodding brick by brick construction.

And so, I suppose, my failure was not that I was irrationally drawn to enthusiasm, but that I was too much a creature of Reason, and too little of the pure emotional compulsion that drives men like Trevor on.

Then how is it that I aim to be the protagonist of this tale and not just a spear-carrier? How is it that I presume to see its events through my peculiar vision and not merely dramatize them as they happened?

Because it is a story and not a drama; because it is a history and only secondarily a tragedy. A character like Trevor is inevitably an archetype, the subconscious untrammeled, and the subconscious unleashed is a disturbing phenomenon. Greek tragedy arose for this very reason, as means of a communal purgation achieved through the audience’s affiliation with the protagonist.

We moderns are not like the ancient Greeks, though. We must calm the troubled waters of our psyche not just with purgation but with Reason. This is why this is my story, my memoir and not a play – or a novel.

You will see Trevor as a powerful figure, a dominant personality, but you will see him from the outside in and as an abstraction. Me, you will see from the inside out, as I reflect on events now sixteen years later. And since the largest part of who I am is my rational life, you will see my experience not just as I sensed it, but as I reasoned out its consequences on that larger world that impinged upon me.

* * *

Except for Guy Hines, who was semi-retired, all the founders of Paralex worked in the Digital Southern Area Office in Atlanta. Trevor and I had cubicles on the third floor of the main office building from which we could see the traffic on I-285, the highway around Atlanta called the Perimeter. David Gibson and Nolan Drevitch were consultants in the Applications Center for Technology, located in an industrial park a couple of miles across the interstate. And then there was Fred (whose last name I cannot recall), a tall, bland-faced young man who often accompanied Trevor as a sidekick and was later in and out of Paralex.

In 1987 Digital was still regarded as one of the outstanding growth companies in America and the Atlanta Area office covered seven of the eleven states of the old Confederacy and we were, as a consequence, all busy as hell. And although I had been an employee for ten months and Trevor’s cubicle was only a few rows away from mine, I knew him mainly by reputation and I didn’t know the others at all.

I doubt if I would have had much contact with any of them even if the Area office weren’t such a large and chaotic place. Digital was not the kind of company where you build close personal relationships, and this was especially true in the Atlanta office with its sprawling territory that required us to be on the road almost all the time. Besides, my work was on the business applications side while Gibson, Drevitch and Trevor were experts on the technical end and the split between the two disciplines was so great that the three of them might as well have been on another planet.

I suppose it would have been impossible not to be aware of Trevor and his outsized personality, however, and there were a few occasions in spite of our travel schedules when we were in the office at the same time and I ran into him striding down the hallways with that blown-by-the-wind air of perpetual urgency he affected. And sometimes late in the evening, when the office was almost deserted, I could overhear him holding court in his cubicle not too far from the bullpen in which I then resided. (You knew it was Trevor owing to his sort of English accent; I was later to discover he was born in what is now Zimbabwe.) He would invariably be declaiming loudly on some technical point or other, always pacing in and out of his cubicle, always with a cigarette in his hand and usually with an audience of two or three of his admirers who would be supporting themselves on the partition walls.

Trevor was not particularly prepossessing physically, although he bore a slight resemblance to Douglas MacArthur, which arose, I suspect, as much from the extreme self-confidence he exuded as from the jut jaw and patch of straight black hair which he slicked over his high forehead. This hint of five-star general authority was undercut, however, by a narrow-shouldered and not particularly fit-looking puffiness about the body that, along with a seamed and jowly face, made him look older than the forty-eight he claimed to be. And while he was always elegantly dressed in dark suits and regimental striped ties, his unathletic appearance made him seem rumpled even when his clothes were freshly pressed.

Other than these few mainly negative observations – I have never warned up to blowhards – all that I had learned about Trevor before he suddenly appeared in my cubicle came via the company rumor mill, where he was usually portrayed as an impossibly brilliant technical guru, and from one chance comment by Steve Latimer. Quite a guy, Steve remarked one evening in his cubicle when we were hatching some deal and Trevor could be overheard expatiating half a bay away. Who the hell understands him? But he sure brings in the deals. Five, ten million at a pop. And complicated… He let out a long sigh. I wonder sometimes if we can deliver what he promises. Though our technical folks say we can. There was no confidence in Steve’s eyes that the technical people were right.

Now, Steve was one of those energetic little computer salesmen you just love to be around. He was, in fact, the principle reason I had come to work for Digital. We had done a couple of deals together when I was still a consultant at Peat Marwick and he was impressed enough with my rapport with clients that he tried for nearly a year to persuade me to come to work at Digital. Then when it finally came together and I took a job in the Area office he did his best to make it work for me, although from the very beginning my intuition was screaming misfit.

* * *

My dissatisfaction arose initially from the very different surroundings in which I found myself. Digital, like many engineering-oriented companies of the era, had a bias against private offices. They were not democratic or utilitarian enough, I suppose. And while the company had not gone so far as, say, Intel, and eliminated private offices entirely, it had restricted them to managers only, somewhat apologetically justifying this exception by citing the manager’s need to confidentially counsel his people. All Digital non-managers no matter how senior were relegated to individual cubicles of a uniform size and height.

That this arrangement was a far cry from the cherry-credenzaed and plush-carpeted office I had been used to at Peat Marwick would have been a little discouraging under the best of circumstances. But the Southern Area was short of office space when I reported for work and I was assigned a desk in an unpartitioned bullpen area next to the very noisy tel-sales group. This, along with the small pay cut I had taken in order to take the job at Digital, was enough to set me on an attitudinal down slope from which I never entirely recovered.

Chapter TWO

I Meet Dr. Hicks

Naturally, I trusted Steve Latimer a lot and his less than ringing recommendation of Trevor had made me very chary of any relationship. So when Trevor cornered me in my cubicle in the late spring, I tried to avoid him by pretending to be busy reading e-mails. He was a very determined fellow, though, and waited patiently while I carefully reread the Management Summary of a proposal draft, which owing to its length and complexity, must have taken at least fifteen minutes. Not until I exited to the message menu did I acknowledge his presence, and Trevor responded by laying a large, warm paw on my shoulder, whispering conspiratorially, Got time to talk?

I couldn’t be so rude as to not offer him at least my perfunctory attention and so I replied, Sure.

Not here, he cocked his head away from the cubicle. In one of the conference rooms.

I was very busy and there was no way I was going to go off on some half-cocked bull session on such short notice. Besides, there was something ominous about his looming presence, almost like a fog bank sweeping in from the Bay. Is it urgent? I asked, turning back to my computer terminal.

Not really. Trevor gave my shoulder a comradely squeeze. Just wanted to get your insight on the direction of computing.

Where the hell was this going? I wondered. I had acquired a bit of a reputation as a visionary on the software side of things, but hardware engineering wasn’t my bag. Besides, the topic he suggested didn’t seem to have anything to do with the job at hand – or any other job in the Area office for that matter.

Well, maybe, later, I replied. I’m busy right now.

Okay, he said. But I will be back. And, oddly as I was later to discover, he left.

A couple of weeks went by and both he and I were in and out of the office and I had almost forgotten about his visit when I ran into him at the snack shop across the street from where I worked. Late spring had turned into an early summer and it was too hot to eat on the patio and I was looking around for an empty table inside, when Trevor just showed up.

It didn’t occur to me as odd that he would find me there – it was only later that I was to discover that Trevor’s tastes were considerably more patrician than ham-and-Swiss and bagged potato chips – but once again I experienced the unpleasant sensation of being about to be overwhelmed by a dense miasma. I could think of no excuse to deny Trevor my attention, however, and let myself be drawn into a mainly one-way conversation with him.

I don’t exactly recall what Trevor said during that luncheon, but two topics I know he didn’t talk much about were: (1)himself, since it wasn’t until months later that I was to learn any details about who he was or (2)the company. In fact, the entire conversation seemed to center on me.

In the fifteen minutes it took me to order, wolf down my meal – I am always a fast eater and when I am uncomfortable I barely chew at all – and try to digest my food, Trevor pestered me with a Q and A that consisted mostly of assertions about myself which I could either affirm or deny, taking no time to eat himself. He claimed he was trying to cut down. Just had a little stroke, you know. He was noticeably thinner than when I had first seen him back in the fall and, as if to confirm his recent spate of ill health, he smoked only a single cigarette during the entire time we were in the snack bar.

After I had finished my sandwich and allowed a decent interval more, I swept away the potato chip crumbs and, gathering my papers, excused myself. I suppose it was natural that Trevor would want to accompany me back to the office, since he worked in the same general area as I did, but if I had earlier found just his presence troubling, the fact that he kept up with me stride for stride, walking so close that he sometimes bumped into me, was positively unsettling. Nor was my uneasiness improved by his soothing, smooth-as-silk tone of voice, which was in stark contrast to his normally brash performances on stage in his cubicle.

I had been considered a fairly attractive fellow in my youth and I had experienced a more than ordinary number of advances by gay men and knew there was nothing sexual about Trevor’s approach. He was far too intense for that. It was clear nevertheless that there was some kind of a seduction in the works.

Whatever he was after, Trevor was not long in trying to get to second base. When the elevator arrived at our floor and I headed in the direction of my cubicle, Trevor grabbed me by the arm and steered me into a small conference room, which, as I later learned, he had reserved for this exact meeting.

* * *

For all its commitment to cubicles, Digital had recognized that they were completely ineffective for group collaboration and had constructed a number of smaller conference rooms that were available on short notice. Invariably situated on the interior of the building, often filled with detritus from earlier meetings, lit by those intense and unearthly fluorescent bulbs, the small conference rooms had all the warmth and charm of an interrogation chamber at Lubyanka prison.

It was into one of these that Trevor led me.

This was downright weird! I thought. Without any kind of preparation, without an invitation even, I was suddenly confronted with a one-on-one meeting with this…person. What was I to do? Well, I had nothing specific scheduled for that afternoon, and if Trevor didn’t go on for too long…

Trevor went right to the point. What do you know about high-performance computing? he said, drawing on the blackboard a diagram of a bunch of boxes connected by lines. It was clear to me that what he was trying to describe with the drawing was some kind of network.

I considered myself pretty knowledgeable as to the conceptual side of the computer industry and so I fatuously replied, Well, I know that IBM is having a tough time competing.

Trevor didn’t miss a beat. No, I mean real high-performance computing.

Digital, I said, makes systems that are almost as powerful as mainframes now and if you go to distributed processing, why there’s no reason that you can’t run almost any application on a VAX.

Hmm, he said. Well, that’s not exactly what I’m talking about. What do you know about Cray?

Oh, that, I said, finally catching on. Supercomputers? Well, I don’t have much use for that. I’m a business systems guy.

You’ve seen Crays, haven’t you? he asked, turning away from his diagram and, reaching inside his suit coat, produced a glossy photograph folded in quarters. He spread the glossy onto the conference table beside which I was by then seated.

The photograph showed a group of towers arranged in a circle with what looked like cushions around them. Vector processing, Trevor went on. Cooled by liquid nitrogen. All hand-tooled. Costs tens of millions of dollars to buy, millions to install and millions more for the consulting to make them run.

Who can afford…?

He took out a pack of cigarettes, looked wistfully at them and then put them back. National Labs, the Weather Bureau. Some consulting firms. George Lucas.

Lucas?

You know `Star Wars.’ Computer animation requires huge processing capability. Lucas bit the bullet and now he has a practical monopoly on computer animation. And that’s only a mini-Cray.

Well, what do you expect? I was anxious to show off my technical knowledge. The industry has about hit the wall with silicon. Look at IBM. They’re going to multiprocessing because they can’t build a faster CPU.

IBM’s a shell, Trevor said, a business front. Don’t expect anything revolutionary from them.

That’s why I came to DEC, I announced proudly. IBM is cooked. Its operating system is twenty years out of date and its computers aren’t that big of a deal price-performance-wise. And as to that bureaucracy…What real technician wants to work for IBM? A real snoozer. Having made my point, I rared back in the chair and folded my hands behind my head.

Trevor whipped off his suit jacket and whisked it cape-like over one of the chairs. Then he leaned very far forward so that I could see his eyes across the table. They were not piercing exactly; the intensity was much greater than that. It was like being in a deep forest where there was a far off blaze. A campfire? A cabin? A forest fire set off by lightning? While you wondered about that, the forest closed in on you.

And so you decided to come to work for the Norwegian Engineer? The President of Digital and the founding genius behind the company was named Ken Olsen. Whether he was of Norwegian heritage or Swedish or perhaps even Danish, I didn’t know and I don’t suppose Trevor did either. But Norwegian sounded more exotic and thus a lot funnier.

He’s a real visionary and this is a great company, I insisted, insincerely.

Some such, he said. But this is not what I want to do for the rest of my life. Do you?

Then what? I asked.

Do you know what the computer industry wants? He turned to the blackboard and punched his diagram with the chalk. "Power, real power, unimaginable power. Not twenty megaflops or a hundred (a megaflop is a million floating point operations per second, a measure of power for mathematical computing). Not a gigaflop (a billion floating point operations per second). But a teraflop. Think of the applications that you might be able to do with that kind of power. Aircraft design, crash test simulation, signal processing, much more elegant animation, really accurate weather forecasting, non-explosive weapons testing. Mapping the human genome in five years rather than thirty. A cure for cancer. SDI. It’s all feasible but we can’t do any of it because we lack the compute cycles.

And how do you think we’ll get there? Not through serial mainframes, that’s for sure. And not through vector processing. We’ll do it with parallel computing." The pitch of his voice went up at the end as if he were announcing the next President of the United States.

But how? I asked. And why isn’t IBM doing it? Or anybody else, for that matter?

Tech weenies, he replied, contemptuously. Parallel computing is a curiosity right now. A parlor game for smart computer engineers. A bunch of backroom types at Intel. Ncube. Sequent. A bunch of others. And as to IBM, why would they build a cheap big computer when they can get ungodly rich off small costly ones.

Then he turned away from the blackboard, propped his leg against the wall and began unloading on me. He began by telling me what I already knew - that the growth in power in single processors had slowed and that the current alternative was multiprocessing which involved a very complicated operating system that spread the load over several systems.

But there was a problem with multiprocessing. To the external world several computers might look like one, but to the operating system it was still four or six discrete systems with one being the master and the rest slaves. The more processors that were added the more the one with the operating system was occupied with optimizing the balance. At a fairly small number of processors, adding an additional one actually slowed the whole system down.

This didn’t happen with parallel computers, Trevor explained.

Chapter THREE

Parallel Processing

Trevor proceeded to the point of his diagram.

He circled the whole thing with his chalk. What makes a parallel supercomputer different is that its processors are networked. This allows the building of a computer with a really large number of CPUs, sometimes numbering in the thousands, with little decline in performance.

Then he punched a box way off in the corner that was connected to only one system. This gain in performance is achieved at a cost, however. All parallel computers require a front-end system. But since that system is there mainly to load the programs, it can be quite small, so small, in fact, that it is often a Personal Computer.

Then there’s the additional effort to program the beast. Trevor circled just one of the boxes. Since each CPU in any network design must operate more or less independently, a parallel program must be divided into a number of subprograms, each of which runs within its own dedicated processor and each of which is designed to run on its own.

* * *

Then Trevor laid the chalk on the blackboard and, thrusting his hands in his pockets, turned to face me. His posture was casual, almost indifferent, and his argument was presented methodically and without passion, and yet there was something about his delivery, the English accent perhaps, his obvious intelligence, that, in spite of the highly abstruse nature of his material, sucked me in.

Parallel computers can’t be used for just any application, though, he went on. "They are designed to run only one program at a time. This practically excludes most business applications and the kind of scientific processing that’s largely data gathering. Fortunately, no one buys a supercomputer unless they have at least one application that will hog the entire machine.

A parallel application must also be the sort that can be subdivided. This limits you a bit, though you’d be surprised how little. There are a fairly large number of processes that can be designed around the architecture, enough to build a business around, anyway.

And if you can… he grew thoughtful, "well, if you can, look at all you would gain. The simplification of the operating system, for one. Since each parallel processor contains its own kernel operating system, identical to the kernel of all the other processors, it can be far smaller than the operating system of a serial system.

"Processor independence can also mean I/O independence. A disk drive controller can be hooked to one processor today and another tomorrow. Or alternatively each drive can be direct-attached to its own CPU. A 64-processor parallel system can have sixty-four disk drives, each with nearly instantaneous access to the system. This could permit really high bandwidth data transfers.

With a parallel system each CPU can have its own graphics system – or several parallel processors can be logically arranged to permit really high performance graphics. And a parallel system can be built with a very small footprint. A 128-processor Intel system can be housed in four standard computer cabinets without any of the special coolant requirements of a Cray supercomputer. And the thousand-processor Ncube system is housed in a cabinet about the size of a large toy box. Best of all a parallel system can be built cheaply. While a twenty-million dollar Cray is able to deliver half a billion megaflops, a 128-processor parallel system can deliver the same power for a couple of hundred thousand…

* * *

Playing the dispassionate technician to the hilt - and all the more impressive for it - Trevor went on in this vein for quite a while. I was to see Trevor deliver this same pitch in this same way at least a dozen times over the next few months. And, oddly enough, it was this, the technical detail that wowed them, computer jocks and management types alike. His audience would lean forward and go dreamy-eyed and nod their heads all together like whitecaps on an angry sea.

To the technical people, I suppose, the effect of Trevor’s exposition was the same as it had been on me. It was as if we had suddenly learned that there was another universe that was much more interesting than the one we’d believed was all there was.

If Trevor’s words came off as a blaze of language to the rest of his audience, the content of which they only dimly understood, it was so logical it sure sounded possible – and if it were, now wouldn’t that be wonderful.

Then when I - and they – were lulled into the impression that this was merely a technical presentation, albeit by a master, Trevor would stop, take his hands out of his pockets and white-knuckle the nearest piece of furniture, often a lectern, in this case a Steelcase conference room chair.

Not that parallel computing is about to take over the world, he would say, his voice rising to a crescendo. The industry is far too immature for that. And with that immaturity lies opportunity.

He let that notion gestate for a moment, long enough for me (in this case) to wrap my mind around what he was trying to do. It was a business he was talking about, a business that somehow involved me.

That thought piqued my interest, but also made me uneasy. I was never much of a risk-taker, one reason I was only a middling employee at Digital, and I was strongly inclined to be loyal to my employer.

Trevor went on with his spiel.

How do I know this?

"Because there are least three problems nobody’s solving.

"The first is a consequence of a screw-up in hardware design. Most everybody builds their parallel system so that the I/O is funneled through the front-end system. This forces all peripherals to communicate to the powerful main system through a, would you believe it, a PC! for God’s sake. And, if that weren’t bottleneck enough, they attach the front-end through a relatively low speed Ethernet cable. So instead of the really high bandwidth I/O I talked about, you get really low bandwidth.

"It’s like having to empty Lake Erie by sucking it through a coffee stirrer.

"Now, why in the world would computer designers build a system with such a bottleneck?

"Well, the designers aren’t really crazy, just a little shortsighted.

"All the currently available parallel computers have been developed either by small companies or by intrepreneurial ventures within larger enterprises such as Intel. These organizations are usually undercapitalized and need to get to market as quickly as they can. A PC front-end means one less technology an engineer has to invent. And since most supercomputer applications are low I/O, batch systems, they’ve got a ready market, too.

"What nobody seems to grasp is that there is a much larger market out there just waiting to be exploited.

"The reason, you see, that CDCs and Crays are limited in their use is that they are huge, finicky machines that can only be installed in environmentally controlled data centers. That excludes such rough environments as factory floors, research laboratories or remote defense facilities…

"…While parallel systems are so simple and durable that they can be installed anywhere.

"If, that is, if you redesigned them to permit really high bandwidth I/O, you could approach such mission critical applications as real-time process control, crash test simulation, and command and control systems. A huge market.

"So why on earth did their inventors limit themselves this way?

"Part of the reason is the practical barriers to entry into business. If you want to start a computer company you must meet at least four requirements:

(1) you build what you know,

(2) you build it for a market you understand,

(3) you build it around existing applications, and

(4) you build it so that its easy to use and understand. Anything less and you risk having no market for a product. More and you overcomplicate your business.

"Current parallel systems meet all these criteria. A market has been created and a number of companies are successfully competing in it. Mightn’t that be good enough to kick-start the industry?

"Well, the fact that no one company has broken out and dominated the market, the fact that parallel processing is still largely a playtoy for academics says probably not.

Then what’s wrong?

Here Trevor stretched, waved his arms as if to clear the kinks out of them, and began pacing about the room like a caged lion aroused by memories of hunting on the veldt.

"The personalities of the entrepreneurs, almost all of whom are West Coast design engineers. Technical purists much more interested in the elegance of the product then the business, West Coast engineers have little interest in people management, believe administration to be a necessary evil, and despise marketing, the venue of deceptive advertising and the slick salesman. Nor are they particularly interested in how their invention is used. We’ll build the system, they seem to say, let someone else design the applications. And, oh, yes, we want to get rich, too, but on our terms.

"Nor does talent and intellect make any difference. Quite the reverse. The more cutting edge the technology, the more likely it is that its founders will think this way.

"It so happens that the parallel main system, its components, operating system and interconnections are about as cutting edge as it gets. So far the most part the inventors of parallel systems don’t care about markets; they don’t care about the applications; they don’t really care about the technology. All they care about is their damn invention.

"And that brings me to the crux of the matter, namely, will we have a vibrant parallel computer industry, or won’t we?

That requires a little history lesson. For you can best understand the state of parallel computing today if you compare it with the PC industry of eight or ten years ago.

* * *

Trevor interrupted his pacing to pat the pocket where he kept his cigarettes. Then he smashed his right fist repetitively into his left hand, as if reflecting on what he had said.

This gave me a moment to absorb all I’d heard. Remarkable, I thought. The man is obviously so technical and a student of history, too…So unlike the technical people I had known, who’d been indifferent to the past, believing it irrelevant to the future, the only place they wanted to live.

Then Trevor proceeded with his history lesson. "Arising out of the Bay Area counterculture, the techno-hippies of the seventies had an instinctive aversion to private ownership of ideas. Instead of businesses they organized specialized conferences where papers were presented and models demonstrated. Anyone could use any part of anyone else’s idea – or none of it. In this freewheeling environment the technology was allowed to grow without interference from patents and copyrights.

"This unique social organization worked well in the early days of PCs when nobody knew what was feasible and what wasn’t. But a few members of the PC scene, perhaps a little greedier, perhaps a little more practical, realized that the technology would go nowhere unless it could be made functional for a non-computer clientele.

"In the case of PCs, this clientele was the general public - and since the general public was terrified of anything technical, functionality meant the masking of the engineered innards of the PC with a standard – and much more intuitive - user interface.

"The user interface these entrepreneurs eventually came up with included software, ergonomically designed keyboards, and sensual display terminals. Virtually everything, in other words, except mass storage and the processor itself. And since mass storage had to be made both compact and portable and the processors had to be powerful enough to accommodate the software, these eventually became a part of the package, too.

"Building a computer that contained this interface required the coordination of large, highly specialized design teams, assembly line manufacturing and mass marketing. Organizing a business, in other words. This led, of course, to the weeding out of the more extreme of the technical purists.

"Early parallel computer entrepreneurs came out of the same milieu. But they were not so lucky as to have an interface problem to deal with. Their method for accessing the main system was a PC and the PC was already acceptably user friendly. And since their clientele were engineers and scientists - and engineers and scientists were supposedly much less concerned with user friendliness than the average citizen, they made no attempt to pretty things up. Why some engineers actually like a little technical dirtiness! they reasoned. Keeps them closer to the machine where things are real.

"So the parallel computer designers are still technical purists."

* * *

This is, Trevor insisted, where so many parallel computer entrepreneurs get caught up in their pajamas. Because the purist view is naïve – and it is a trap.

"It is naïve because engineers, especially those in large corporate settings, are also users and have no more energy for tinkering with technologies than the rest of us poor non-technical slobs. This is why the workstations made by Sun and Apollo have standard operating systems and icon-driven user interfaces.

And it’s a trap because it allows computer engineers to concentrate on their first love, the technical internals of the parallel system while ignoring the applications world.

* * *

Tech weenies! I interrupted. Aha, I was thinking, this explains it all. That narrow technical focus, that difficulty in comprehending the finer aspects of human affairs. Intelligence and myopia everywhere. Digital is just like the West Coast design engineers. Only Digital was founded in the late fifties and it’s a big company now. But what did this mean about my future there?

That’s part of the reason for it, Trevor went on. The other part is the scientists’ bias towards advancing the stock of knowledge. If you share the wealth everybody benefits. Then he let out a cat-ate-the-canary grin. "Won’t catch me doing anything like that.

We’re paying a hell of price for that bias. Maybe, even our national lead in computing. I wouldn’t be surprised if the Japs aren’t busy coding away at parallel applications right now.

And then in a more ruminative vein, This brings me to the second reason why parallel supercomputing has made little dent commercially, namely, a lack of applications.

* * *

"The reason no one has built parallel applications is because software designers are, for the most part, unaware that the architecture even exists. And those few that know about it are reluctant to undertake the effort. To be forced to reorganize every program into a series of independent calculations when you can’t be sure the application will work let alone deliver the promised performance…

"There are a few industry pundits who disagree with me on this. They are convinced that a new generation of parallel software developers is about to spontaneously rise up just like they did in the computer industry a few years back.

"But I don’t believe that, not for a moment…

"…Because parallel systems designers are also ignoring the programmers who would build the applications. There are very few college courses being offered in parallel programming and no manufacturer-sponsored training. Nobody’s building any programming tools, either.

"This is a hell of an oversight…

"Because the PC industry was not actually built by the IBMs and Apples – or even the Intels and Microsofts - but by the dozens of small software firms. And the growth of these firms was made possible by the fact that programming a PC is not all that different from programming a mainframe and programmers had had thirty years to learn proper design techniques.

How is an applications software industry going to spring up, spontaneously or otherwise, without parallel programmers? And what business is going to take a chance on a parallel supercomputer for a bet-your-company application with no software available?

* * *

"…Especially when users find out that parallel computing is no new idea, that it has been considered and found wanting several times in the forty-year history of the computer industry.

"Although the giant boxes of the day made them unwieldy monsters, people built parallel systems as early as the fifties. As late as 1969 a sixteen-processor system would have filled a large warehouse.

"The technology first became feasible with the growth of microprocessor technology in the 1970s. But the industry was still dominated by business systems and IBM. Scientific computing was a niche market dominated by single-purpose minicomputers and high-performance scientific computing was relegated to a niche within that niche. The latter was the bailiwick of the Control Data Corporation, which built elaborate mainframes - and of IBM which had a presence everywhere.

"It was only when Seymour Cray broke off from CDC that the situation began to change - and not just for giant institutions with huge applications. ‘Crayette’ companies, such as Convex, have grown rich off of selling minicomputer versions of Cray supercomputers.

"But Crays are based on vector processing and vector processing is really only a stopgap and very expensive solution to a huge problem, namely, that existing silicon technology is about to hit the performance wall. This is why IBM and other vendors have converted to multiprocessing for their biggest systems.

It is also, Trevor insisted, why the industry ought to consider parallel computing.

(It turns out that Trevor and - a lot of other industry pundits - were wrong about this. The growth in computer power was not about to come to a sudden end in 1987 and the fifteen and twenty megaherz processors of that era have gradually grown into the gigaherz CPUs of today.)

Chapter FOUR

More of THE PITCH

T hen why, I asked, "haven’t one of the big industry players entered the fray.

Partly, Trevor replied, interrupting his pacing so that he could launch a withering glare my way, not out of anger, but to intimidate me, because why’d you want to sell inexpensive systems when you can make a bundle by selling fabulously costly ones? His voice had gone up a couple of decibels and he was gradually working himself into the lather he’d used to cajole his troops from his cubicle. "And partly because it would immediately obsolete a vendor’s entire installed customer base.

"But mainly because it means building an entirely new organization around a technology that nobody is really sure will work.

This is why Seymour Cray devotes the lion’s share of his research toward radical semi-conductor designs and new materials. And why Cray continues to plod along the well-worn vector processor path even as non-silicon materials prove unstable and radical designs prove infeasible.

"Admittedly, there are huge obstacles in the way of commercial acceptance of parallel computing. But in those obstacles lies…opportunity." Trevor had whipped himself into such a state that he intoned the last word like the knell of doom.

"If parallel computing is to be successful some entrepreneur is just going to have to build an entire industry. And to build an industry, someone will have to found a company that is willing to do it all - product development - and not just the computer system but the applications also – and not just the applications but the training and technical support. And administration, too.

"But marketing, most of all, since the company that can build the industry will be one that