Cyber Rogues Read online

  “If you accept the idea of evolution,” Laura reminded him pointedly.

  “I don’t want to go into all that again,” he muttered, then resumed his former tone. “Computers didn’t evolve from survival-dominated origins. They were designed to do very complex, very specialized things, very efficiently. They can mimic Man’s intellectual feats superbly well. Not only that, they’re a lot better at some of them than we are . . . for instance they’re faster, more accurate, and don’t get tired or fed up. But they don’t possess any of the commonsense awareness of what they’re doing or what’s going on around them that animal ancestors had to evolve in order to stay healthy. That’s what I meant when I said they’re evolving backward. They’re good at what we ended up with, but they don’t have what we had to start with.”

  “So that’s what you’re doing?” Laura conceded grudgingly. “Trying to teach them how to tie what’s going on all around into a picture that means something?”

  “You could put it that way,” Dyer said with a nod. He returned his attention to his meal and began eating at last.

  “So what’s the point of it?” Laura asked after a while. “Okay. You’ve spent millions of dollars and ended up with a computer that’s smart enough to know how to fry an egg. What are you supposed to do with it?”

  “All kinds of things,” Dyer replied, sounding deliberately nonchalant. He shrugged while he finished chewing. “Give it a fusion power plant to run. Manage a space mission . . . take charge of New York City air-traffic control. Whatever . . .” He knew he was being provocative and took quiet pleasure from observing the desired effect.

  “What!” Laura almost choked. “Put that imbecile we just saw in charge of a power plant? It can’t even take charge of a kitchen. Tell me you’re not serious.”

  “I am serious. The computers that run all those things right now are a lot dumber than the one you just saw . . . if you insist on judging them by human standards, anyway. On the other hand, if you base your opinion on the ability to crank through fifty million calculations in a second then they’re quite smart.” He paused, unable to contain a smile, and added, “Your problem, you see, Laura, is that you’re too much of a chauvinist.”

  “I’m a what . . . ?” The conditioned reflex in her started to respond but she saw what he was doing and checked it deftly. Dyer complimented her inwardly. “They’re labor-saving gadgets, sure,” she continued. “They’re good for doing all the repetitive mechanical stuff—I’ll buy that. But you’ll always have to have people in charge. You’re not telling me you think you can come up with a machine that’s capable of exercising human judgments too . . . not after what I’ve just seen. That I won’t buy.”

  “But programming the computers is labor too,” Dyer pointed out. “And when you want them to do more complicated jobs, it gets to be hard labor. So why not have the computers generate their own programs?”

  “Because they don’t understand the problems that the programs have to solve.”

  “Exactly.” Dyer nodded in satisfaction. “They don’t understand the problems because they’re not equipped to be able to understand them. They don’t have the basic capability to learn and connect things together that any newborn baby has . . . or they didn’t have until HESPER machines came along. But supposing you could educate FISE to the level where he knew enough about real-world concepts to be able to make commonsense decisions for himself reliably. Then you could put him through a specialist course on—I dunno, say something like steel-making—so he knows all the things you have to aim for in order to run a steel plant efficiently. Then you let him practice for a while, maybe by connecting him to another computer that’s pretending to be a steel plant. Because he’s smart he can learn from his experiences and because he’s a computer he can learn fast. Pretty soon you’ve got a hotshot manager who can run rings around any team in the business. Then you ship him out into the real world, give him a real plant to run and let him get on with it. The beauty of it is he’ll do all the right things, but you haven’t had to go through the hard labor of programming in every specific detail of every situation that might ever arise and every specific detail of what to do if it does. All you gave him was the basic capability to learn. The rest he figured out for himself.”

  Laura continued to eat in silence for a while, keeping her eyes directed down at the plate before her. Her fashionable clothes, meticulously styled hair and faultless grooming made her look out of place among the casual shirts, denims and well-worn traditional jackets of a university restaurant. There was no doubt, Dyer thought, that in purely physical terms she was stunning. He found himself trying to picture what she would look like stripped of the close-fitting velvyon dress that changed its hue from midnight blue to silver as she moved.

  Laura looked up at him. “If FISE is a learning computer, what’s a HESPER computer?” she asked. “I thought HESPER was supposed to be some kind of learning computer too.”

  “It is,” he replied simply, “Or more precisely, it’s a programming technique. It stands for HEuristic Self-Programming Extendable Routine—a set of interrelated programs that form a structure that can learn as it goes.”

  “I’m not sure I see the difference.”

  “It’s a question of degree,” he said. “HESPER systems are specialized to handle one particular kind of application. You could set up a HESPER system that will optimize itself over a period of time, say . . . play a game of chess. The more games it plays, the better it gets until you can’t keep up with it. But that’s all it’s good for. But something like FISE would possess a broad base of general concepts. It could learn to handle anything. So all you’d have to do is develop it once and get it right instead of having to set up thousands of different HESPER systems all the time. It would supersede HESPER programming in the same kind of way that HESPER is taking over from the classical distributed parallel programming that’s been around since . . . aw, the 1980s, 1990s.”

  Laura looked at him quizzically for a moment as if she expected him to draw some conclusion from his own words. Then she sighed and shook her head.

  “Can’t you see how irresponsible the whole thing is?” she asked.

  “Irresponsible?” There was no surprise in Dyer’s voice. Everything had been going too smoothly.

  “Criminally! They’ve been plugging HESPER machines into the TITAN network all over the world for over a year now, haven’t they? So those things are out there, going through their learning processes and being put in control of manufacturing plants, transportation systems and everything else, yet from what you’ve just told me they’re even dumber than FISE is. How can you say it isn’t irresponsible to give idiots like that a fusion plant?”

  “Because they’re not the same thing,” Dyer insisted. “HESPER machines are designed simply to be able to get steadily better at doing a particular job. They’ve been thoroughly tested, they’re well understood and there’s nothing mysterious about them. FISE is a first step toward something radically different. You can’t judge them both by the same criteria.”

  But Laura was only just warming up.

  “How can they be well understood when they’ve only been going into TITAN for a year?” she demanded. “You said yourself they need time to learn and that they don’t have any common sense anyway. What’s to stop them starting to do things that don’t make sense?”

  “They can only work inside the limits they’re designed for,” Dyer told her. “If a HESPER machine is set up to coordinate the communications traffic across part of the net, it can only learn how to do the job better. It can’t make things worse because it isn’t programmed to, and it can’t do anything else because it doesn’t have any generalized capabilities.”

  “But it extends its own programs as it goes along,” Laura retorted. “That’s what you just said the last time I was here. So machines are out there that are putting stuff into those programs that nobody knows about. So how can anybody know what they might do? You have to admit that nobody can claim
to understand them completely anymore. That means there’s a whole planetful of people being used as guinea pigs. Who ever asked them whether or not they wanted all these machines running everything anyhow? Nobody asked me.”

  “Aw come on,” Dyer replied gruffly. “You’re not gonna give me one of those back-to-the-good-old-days speeches, are you? How did they live fifty years ago? People living like zombies, doing the same thing day in, day out, five days a week, fifty weeks a year, right through from when they left school to when they got put out to pasture . . . and being conditioned into accepting it as normal. Think they want to go back to that? Not on your life . . . no more than they want to put their kids back down in coal mines.”

  “Okay, okay,” Laura held up a pacifying hand. “I don’t want to go back to that either. I didn’t say anything about the good old days. You’re always twisting things, Ray. What I’m concerned about is the future. We’ve put all these machines everywhere and connected them all together and, yes I agree, they’re doing a pretty good job. Nobody starves these days, nobody goes without much, people don’t fight about the things they used to, everybody does his own thing in life and hey, isn’t that nice.

  “But people have always been in control. This business you’re talking about sounds like handing control over to machines as well, and I’m just not convinced they can handle it. HESPER is just a first step. You’d be perfectly happy to hand the whole shooting match over to a bunch of morons because the thought just doesn’t cross your mind that they might screw everything up. Then the whole setup would collapse in a heap and we’d all be right back in the bad old days that you’re so delighted to be out of.”

  “It’s just progress being taken to its next logical—” Dyer began, but Laura was not through.

  “It isn’t progress at all. It’s abdication. But the people who have the say in what happens are all people who think the way you do. You shape the world to suit yourselves and the rest of us have to live in it. I don’t like it.”

  “I disagree,” he told her bluntly. “People like you do get a say in it. Everybody gets a say in it. Society evolves the way it does because it reflects the net result of billions of individuals all pushing and pulling in different directions. In other words it’s what best suits most of the people most of the time. Therefore things always get better because better is automatically defined by the process. It’s the way most people want to go. If they didn’t, then they’d go some other way instead and then something else would automatically become better.”

  “Suppose I don’t want to be stampeded along with the herd?” Laura challenged.

  “Then don’t be. Go live alone someplace and do your own thing your own way. Who’s stopping you?”

  “You’re being ridiculous.”

  “I’m not. There are plenty of places you could set up a shack and try it the way people used to. You’d soon be knocking to come back in out of the rain, though. Then you’d see why they gave it up to go our way and why the old days changed into what we’ve got now.”

  Laura leaned forward to rest her elbows on the table and fixed him with a steely glare.

  “You’re twisting everything I’m saying again,” she accused him. “I never said anything about wanting to tear down civilization. I’m a big girl now and grew up in Detroit and came to the big city ten years ago and I happen to love it. I don’t want to see it torn down. That’s my whole point. You want to put machines in charge of running it and I say it won’t work. Why not keep them in their place and leave things the way they are? That way we know it works.”

  Dyer sat back and shook his head in a way that said he wasn’t buying.

  “You could have said that at any point in history,” he replied. “When you start thinking like that, that’s when you stagnate. Gotta keep moving.”

  “Why? Why do I have to keep moving if I’m satisfied out where I happen to be? Why can’t I just stay and enjoy it?”

  Dyer reflected on the question for a few seconds.

  “Because everybody else will keep moving anyway,” he said at last. “When you find you’ve been left behind, you don’t feel so satisfied anymore. That’s when you remember how you got to where you are.”

  CHAPTER FIVE

  Dyer arrived at Sigmund Hoestler’s office a few minutes before 2 p.m. He was shown straight in and to his mild surprise found that Vincent Lewis, the Dean of the Faculty, was there too. Hoestler, a big man with sagging fleshy cheeks and a shock of uncontrollable wiry hair, motioned Dyer into an empty chair next to where Lewis was sitting, and leaned forward to come straight to the point.

  “I’m afraid we have some very serious problems that are going to affect you directly, Ray,” he said in his usual throaty voice. “It looks as if we may be forced to close down your unit.”

  Dyer was halfway through the process of sinking back into a characteristically relaxed posture. The bombshell made him sit up again as if the chair had suddenly acquired a few kilovolts. He knew that Hoestler was a man of few words, but even so, the bluntness of the statement had caught him totally unprepared. He had barely begun opening his mouth to frame a question when Hoestler spoke again.

  “I only found out about it myself this morning. Vince was in Washington over the weekend with the Secretary for CIM and some of his people. So don’t get the idea that it’s just petty local politics or anything like that. Vince, you could probably tell Ray about it better than I could.”

  Dyer turned expectantly toward Lewis, his features contorted into a frown of disbelief. Communications And Information Management was a comparatively new executive department of state, formed eighteen years previously in 2010. Originally it had been instituted in response to the need for a single authority to assume overall responsibility for operation of the integrated data communications and computing network that emerged when the military systems were declassified and merged into the already integrated commercial-industrial-scientific complex to form the EARTHCOM net. When HESPER nodes were later incorporated to transform EARTHCOM into the Totally Integrated Teleprocessing and Acquisition Network, TITAN, the Department of CIM automatically became the administrative authority for the NORAM Sector of the global system. As Hoestler had in effect said, the Department of CIM didn’t mess around with interdepartmental university politics.

  Lewis was impossibly tall and impossibly thin. He sat splayed in his chair at all angles like a marionette whose limbs had come out of joint everywhere, leaving him held together only by his clothes. When he was standing up he never failed to cut a distinguished figure, with his elegant crown of white hair, deeply lined face and inevitable immaculate, dark three-piece suit. Dyer had always found him something of an aloof and remote kind of person, but right now Lewis was showing every sign of distress and genuine concern.

  “Certain events have happened recently, Ray, that have caused CIM to reconsider the whole philosophy of adding HESPER capability into the net,” he said. “Some very senior people are pressing for TITAN to be reverted back to EARTHCOM until we get firm answers to some important questions. In a nutshell, they’re saying that the move to upgrade EARTHCOM was premature, that we didn’t know enough about HESPER at the time and we still don’t, and that HESPER ought to be pulled out until we do.”

  Dyer looked from one to the other and spread his upturned palms.

  “Events . . . ? What events?”

  “About a week ago, TITAN came within a hair’s breadth of killing five people,” Lewis told him somberly. Dyer stared at him incredulously. Before he could say anything, Lewis went on. “It appears that HESPER program structures are capable of integrating to a far greater degree than anybody thought. They’re starting to link things together in ways they were never supposed to and the results in behavior are impossible to predict.”

  Hoestler explained, in response to the still bemused look on Dyer’s face. “It used the Maskelyne mass-driver to bomb an ISA survey team on the Moon. Could have wiped them out.”

  “What?” Dyer turned
an incredulous face toward Lewis but the Dean nodded regretfully to confirm Hoestler’s words.

  “One of the HESPER-controlled subsystems in the Tycho node was given the job of shifting a piece of terrain that was forming an obstruction,” he explained. “It was supposed to use normal earth-moving equipment to do it, but nobody bothered to tell it that. Somehow it managed to connect together information from several subsystems that shouldn’t have been connected, and came up with what it thought was a better shortcut to solving the problem. According to the people who analyzed the system dump afterward, it seemed quite proud of itself.”

  Lewis went on to describe the incident on Luna in greater detail. As Dyer listened, his initial astonishment changed to growing concern. In 2020 he had moved out of neurological research in order to apply his knowledge of learning psychology to the field of self-adaptive programming and, after spending some time at M.I.T., had come to CUNY to set up the HESPER Unit, which had since gone on to spearhead development of the very techniques that were now being applied worldwide to transform EARTHCOM into TITAN. His knowledge of the technicalities of HESPER programming was shared by fewer than a handful of people. If it was anybody’s, it was his baby.

  “Unfortunately there happened to be an ISA team sitting practically on top of the target,” Lewis continued. “But naturally, that didn’t mean very much to the computers.”

  “Twenty sixty-pound packages of rock coming down at over a mile a second,” Hoestler commented. “Every one was roughly equivalent to a two-thousand-pound bomb.” He shrugged and made a face.