Fractal algorithms provided a method for generating a lot of material from minimal information—the way Nature compresses its assembly directions in DNA. The principal drawback was that to produce a convincingly realistic output, some kind of randomizing capability needed to be introduced, which meant no two results of applying the same input formula would, in general, be alike. Hence, two separate runs to generate a tree, say (or leaf, or rock, or mountain, or snowflake), from the same set of starting parameters would both look like a tree, but they wouldn't be the same tree. This would probably be all right for representing things like forests, skies, or general scenery where precise details didn't matter too much; but other situations demanded a different approach.
So in these latter cases there seemed to be no alternative to having to store all the detail that might be required. Even with the kinds of faster processing methods and special-purpose hardware that Barry Neinst had been exploring, this imposed severe limitations on how large a world they could hope to simulate accurately. For the world to be sufficiently varied and interesting, not all of it could be represented everywhere, all the time. Nor, of course, did it need to be. So the system was organized to concentrate on the parts making up the immediate experiences of the individuals experiencing the simulation. Thus, indeed, in these worlds the Moon was not there when nobody was looking at it; and the question of trees falling in deserted forests didn't arise, since with nobody around, forests ceased to exist.
For smoother continuity, and to reduce the occurrences of "black-hole glitches"—as when somebody opened a door to find themselves staring into a featureless void for a perplexing moment—various ways were tried for getting the system to anticipate and pre-access the branches that were most likely to be needed next. But none of them proved wholly satisfactory. Following chess-playing-machine parlance, this became known as the "look-ahead" problem. Solving it, along with "realscaping" more of the Pittsburgh area, was the main focus of the EVIE group's work through the second half of the year.
During that period, Ivy Dupale resigned from the company and got a job on the West Coast.
* * *
With Christmas approaching, Corrigan and Evelyn drove out one evening to Eric Shipley's house in Franklin, north of the city. Tom Hatcher and several others from the project were also due. It was a homey, unpretentious place, nestled in a fold of wooded hillside that provided seclusion, yet with the township center conveniently less than a mile away. It was the main house of what had once been a farm. The outbuildings were now converted or demolished, and most of the land sold off, except for a couple of acres forming a shady, somewhat overgrown garden bordered by a creek at the rear. Shipley lived there with his graying, genially disposed wife, Thelma. They had two sons and a daughter, all of them grown and gone in different directions. The children's rooms were always kept the way they had been for their frequent visits home.
Hatcher was already there when Corrigan and Evelyn arrived. With him were two of his programmers, Charlie Wade and Sue Lepez, and also Bryan Reed, one of the electronics technicians working on EVIE. There were sodas, coffee, and beer. Later on, a couple of pizzas arrived. By the middle of the evening the talk had ranged over shop topics and settled on the do's and don'ts for surviving in a popular VR game called "Sniper." Shipley drew Corrigan away to go and see an old nautical chronometer of polished teak and gleaming brass that Shipley had in his study. It was not long, however, before they were back to the subject of developments within CLC. Corrigan sensed that this was the real reason why Shipley had taken him aside.
Most people—including Corrigan himself—had seen the consolidation of EVIE under his direction as an effective promotion, and an indicator that he was solidly on his way upward to better things. Shipley, however, wasn't so sure.
"You said yourself, once, that in the long term EVIE doesn't lead anywhere," he reminded Corrigan. "You called it a short-term stunt, a hybrid mishmash. So what does that say about anyone doing you a favor by putting you in charge of it?"
But Corrigan was still riding the wave. "Ah, come on, Eric," he said lightly. "You wouldn't want me think that you're having an attack of sour grapes, now, would you? Don't you remember, too, that EVIE was to be the main-thrust program for two years? Whoever runs it now will automatically pick up whatever comes next."
"That could be changing, Joe. There's been a lot going on involving Tyron and Pinder at the division level that we're only getting parts of. My guess is that corporate thinking has been turning away from EVIE ever since that business with Feller and Faber. In other words, they're saddling you with a lame duck."
"What else would they have in its place, then?" Corrigan challenged, his voice a touch sharper. They were still a year or more away from shifting up to the thalamus—and there was no guarantee that it would work even then.
Shipley shrugged and showed his hands. "You tell me."
"Why assume that they've got anything else at all?"
"Then look at it this way. If EVIE really is a sinking ship, whose name is being quietly dissociated from it and who'll be the skipper who goes down?" Shipley paused to let Corrigan think about that. "Then ask yourself what Tyron and his people were really brought in for. It certainly wasn't just to take over Ivy's section. That's a holding operation."
This time Corrigan said nothing but stared hard at him for several seconds. Shipley waited, holding his eye questioningly. Before they could resume, however, the long, loose-limbed figure of Tom Hatcher sauntered in from the living room, holding a can of beer in one hand and licking pizza grease off the fingers of the other. Evelyn was behind him, looking fresh and casually appealing with her long, fair hair, white top, and red, clinging slacks.
"Not interrupting anything, are we?" Hatcher drawled. " 'Cause if we are it's too bad. This is a party."
Corrigan hesitated, then grinned. "No, it's okay, Tom. Just shop as always." It was a good time to ease things up a little anyway. Evelyn squeezed past Hatcher to hand Corrigan a sausage on a cocktail stick, then snuggled close while he slipped an arm around her. Hatcher went over to look at the chronometer that Corrigan had been examining earlier.
"Say, that's some piece. They don't make 'em like that these days."
"Not a shred of plastic in it, and the knobs don't come off in your hand," Corrigan agreed.
"Can't say I'd want to carry it around on my wrist, though."
"You didn't have to. You had a ship to carry these around."
"Where do the batteries go?" Hatcher moved to admire a highly polished period revolver, mounted as a display on a board fixed to the wall nearby. "Looks like a .44 Dragoon Colt," he commented. "Probably from the Civil War."
"Right on," Shipley said, nodding.
"I didn't know you were into that kinda thing."
"I'm not. Thelma picked it up at a yard sale for five dollars."
"Does it work?"
"No—just an ornament."
"Too bad." Hatcher's interest in guns was well known.
Shipley nodded in the direction that Hatcher and Evelyn had come from. "What's going on back there?" he inquired.
"Shop," Evelyn said. "Is it ever different with this bunch?"
"Charlie's talking about his accelerator for the new look-ahead tree," Hatcher explained. "I had enough of it all day. I came here to get away."
They still hadn't found a reliable way of paralleling the human intuition for knowing what people were apt to do next. In a test the previous day, the system had properly anticipated all the things that one of the experimenters could reasonably have been expected to do with a magazine when he picked it up and rolled it—except use it to swat a simulated fly. Hence there was a hangup upon impact, in which time the velocity of the magazine fell to zero, and the fly was able to walk with impunity onto the object that was supposed to have flattened it.
Machines were good at organizing the world into neat hierarchies of computed probability. The real world, however—essentially because of the way that the people i
n it behaved—didn't work that way.
"Charlie's still an idealist," Hatcher said. "He just won't accept that the world isn't logical."
"Well, it doesn't work by formal, Aristotelian logic," Shipley agreed. "You see, that's purely deductive: you start with what's true, and from that the way the world has to be follows. That's what machines are good at. But in real life you start with experience of the way the world is, and then infer the reasons why and hope they come close to being true. Inductive: that's what people do—and even they aren't sure how. That's why textbook science and real science aren't the same thing."
"Is philosophy a hobby of yours, Eric?" Evelyn asked. She had been running her eyes over the shelves of books around the study.
"Oh, I dabble in a bit of everything," Shipley replied affably.
"So the universe is inductive," Corrigan concluded.
"Isn't it obvious?" Shipley said.
"I thought that philosophers have been having a problem with induction for centuries," Evelyn commented.
Shipley shrugged. "It's of their own making—as are most of humanity's problems. They started by assuming that the universe couldn't work inductively—because they couldn't reduce it to formal rules—when it obviously does."
"So we have to teach the simulator how to be inductive," Corrigan said. "How does real-world logic work, then, Eric?"
"Being ninety-percent right, ninety percent of the time," Shipley replied. "It's what science, business, war, and evolution are all about."
"What about sex?" Hatcher asked, looking away from the Colt and taking a swig from his can.
"Oh." Shipley smiled. "That's made up of all of the above."
A thoughtful expression came over Hatcher's face. "Maybe the way isn't to try and teach the system how to be inductive at all," he said. "I mean, if we're not really sure how we do it ourselves, we're hardly in a position to spell out the rules, are we?"
"What other way is there?" Corrigan asked.
"Maybe the thing to do is turn it the other way around."
There was silence for a couple of seconds while the others puzzled over this. "How do you mean?" Shipley asked finally.
"Let it learn in the same way as we do: by observing the behavior of real people in the environments that it creates. With EVIE, we've got all the technology you need." Hatcher paused, then went on, more excited visibly as he warmed to the idea, "Instead of the inhabitants of a world evolving in response to the environment, the environment learns to get better by watching the reactions of the inhabitants. See what I mean—it would be turning nature upside down."
"Hmm." Shipley drew back, frowning. "I'll have to give that some thought. . . . It's interesting, that, Tom. Very interesting."
Corrigan had taken down one of the books that Evelyn had noticed and was turning the pages idly. "Epictetus? I've heard of him."
"Greek slave, taken to Rome," Shipley said, moving over.
"Got freed and became a philosopher," Corrigan completed.
"He's the reason why I've never been interested in politics or prestige," Shipley told them.
"Really?" Evelyn said.
Shipley grinned. "Oh, I was kidding. But he does say some interesting things."
"Such as?" Corrigan asked curiously.
"That you shouldn't seek happiness through things that other people have control over," Shipley answered. "Otherwise you end up being enslaved to them."
That didn't seem to leave very much, as far as Corrigan could see. "What else is it that you should want, then?" he asked.
"Live for your own values and beliefs: things that nobody can take away," Shipley answered. "Then nobody can own you."
The veiled reference to their private conversation earlier would have been enough on its own to goad Corrigan into dissenting, even without his innate Irish argumentativeness. "It sounds like a pretty empty cop-out, if you ask me, Eric," he opined. "The kind of thinking of somebody who would never try going for anything for the fear of losing it. Where's the challenge and satisfaction in living a life like that?"
"It's being free, Joe. Fearing nobody. Look at the antics of some of the people we see every day and ask how many of them can say they have that."
Corrigan shook his head. "You live your way, I'll live mine. I couldn't accept a philosophy like that."
Shipley seemed unperturbed. "Maybe you should go back and get in touch with your roots again, sometime, Joe," he said. "To Ireland. There's a tradition there, too, that understands the kind of things I'm talking about."
"Oh, you don't buy that load of rot, too, do you, Eric?" Corrigan groaned. "Thieves, rogues, and scoundrels, the lot of 'em. They'd sell their grandmothers for the price of a pint—and then leave you stuck with the tab if you look the other way."
"I'd still like to go there," Evelyn said. It was something they had talked about a number of times.
Corrigan looked at her. "Well, maybe it is about time that you and I took a break somewhere." He raised his eyebrows. Her face split into a smile, and she nodded eagerly. "How about Florida, or maybe Mexico?" he suggested.
"Somewhere a bit sunnier than Pennsylvania in December, anyway," she said. "It's no better than Boston."
Corrigan thought for a few seconds longer. "Then let's make it California," he said. "There's a string of places dabbling in neural stuff on the West Coast that I've always been meaning to check out. And there's an old friend of mine from MIT called Hans Groener who's doing things at Stanford on sleep and dreams that sound interesting, but I've never had a chance to see it."
"Sure, California'll do. Why not?" Evelyn said. "I've never seen Yosemite."
"Do it," Shipley told them. "Everything's slowing down here for the holidays. And you've probably got some leave that you need to take before the year's out, Joe."
Why not? Corrigan thought. "I'll call Hans tomorrow and see what we can do," he promised.
Chapter Sixteen
Despite his fatigue and having been up all night, Corrigan did not sleep well. He awoke halfway through the afternoon, still feeling woolly headed and groggy. All he could remember from his disjointed recollections of the early-morning hours was that Lilly's place was north of the river, somewhere near the Allegheny Center. He cleaned up and put on some fresh clothes, then fixed himself a snack. Computer-injected hunger signals felt just the same, even if his real body was in repose, getting its nutrients from dermally transfused solutions. After that, he left without turning Horace on again, and caught a bus to the North Side.
But nothing that he saw jogged his memory as he wandered up and down the streets of the district contained in the crook of the I-279 Expressway, north of Three Rivers Stadium. Any of a score of apartment-block entrances that he passed could have been hers. Any of the streets that he walked along could have left the hazy image that was all he could piece together of unremarkable frontages glimpsed in predawn shadows.
It made sense to him now why recent years should have seen so much redevelopment around Pittsburgh. For every part of the old city that was "demolished," new, simulated scenery could be substituted that would not have to conform to anybody's real-world experiences. Nobody could walk around inside the Camelot, for example, and be puzzled by not finding things the way they used to be. The "realscaping" task was thus considerably eased.
He wanted to tell Lilly that she had been right, but everything was okay—the experiment was going as it should. Yes, their memories of the actual commencement of Oz had been suppressed, and alternative stories given to mask the transition from the real world to the illusory. But it didn't follow that something sinister was going on. Some such provision would have been necessary to ensure that the responses of the surrogates—the real-world participants coupled into the simulation—would be natural and valid.
And boy, had that part of the scheme worked as planned! Until Lilly waved the facts in front of his nose, he himself—one of the principal creators of the simulation—had failed to realize that he was inside it. She had thought to q
uestion where he had not because she had known less. He had been involved in the planning of Oz. Hence, if any deception were intended, he would have known about it. Since he didn't, there couldn't be any; and once the impossibility was established in his mind, there was no place for the possibility to coexist. The irony was that it had been able to work in his case only because of his knowledge that it couldn't work.
The main cause of Lilly's distress and anger was not so much the deception—she was a military volunteer, and things like that happened and could be compensated for—but the twelve years that she saw as stolen from her life. And who could blame her for that? But what he knew, and she almost certainly would not, was that those twelve years were also an illusion. The system coupled directly into post-sensory brain centers, which enabled data to be coded in a prereduced, highly compressed form that eliminated delays associated with preprocessing in the perceptual system. This meant that time inside the simulation ran about two hundred times faster than real time in the world outside. Hence, the actual time that they had spent hooked into the virtual world would be closer to three weeks than the twelve years that they remembered subjectively. Although even that was longer than the durations projected for the test runs that Corrigan had expected to be taking part in, it wasn't outrageous. They were all scientists and volunteers, after all. They would have had little problem agreeing to something like that.
He hoped that if he could find her and reassure her of at least that much, she would see things in a different light and be less likely to start doing anything rash that might disrupt the experiment. There was no reason for the test conditions to be affected by the mere fact of their knowing what they knew now, as long as they continued to act as if nothing had changed. The system could only monitor external behavior: what a surrogate did and said. Since nobody possessed the knowledge to tell it how, the system was not able to decode inner thought processes from deep inside the brain and read minds. If it could, there would have been no need for Project Oz in the first place.
Realtime Interrupt Page 13
