The whole idea had been that the system would learn to make its animations more lifelike by imitating the behavior of real people injected as surrogate selves into the simulation. It had no way of knowing why the surrogates that it watched behaved in the ways they did—any more than they frequently did themselves. At the end of the experiment nobody would know, let alone have been able to specify beforehand, the precise structure of software structures and linkages that had self-organized to make such mimicking possible. The neural structures responsible for the complexity of human behavior in the real world had evolved by principles that were appropriate to carbon chemistry. Trying to duplicate them in code would have been as misguided as building airplanes that flapped feathers. Oz was designed to build, in ways appropriate to software, whatever structures it needed to achieve similar results. Nobody needed to know exactly what the final structures were, or how they worked. The aim was to achieve directed coevolution: the end-product, not the mechanism for attaining it, was the important thing.
That had been the theory, anyway. Whether it would work was what Oz had been set up to test. And from the bizarre goings-on going on in the world around him, Corrigan's first conclusion had to be that as far as its prime goal was concerned, the project had wandered somewhat off the rails. For, far from modeling themselves on the surrogates, the system animations seemed to be going off into self-reinforcing behavior patterns of their own, while—if his own and Lilly's cases were anything to go by—the surrogates had become misfits. That in itself didn't trouble him unduly. This was research, after all; perfection could hardly be expected from a first-time run—and especially in an undertaking as unprecedented and as ambitious as this.
Hence, it was no surprise that the animations fell short of true human emulation in some aspects. What was astounding was that they came so close. The empty stares and "flatness" were minor flaws compared to the extraordinary degree of realism—even if it did tend somewhat toward the eccentric—with which the personas that he encountered daily were able to act out their affairs and effect the continuity of leading consistent background existences offstage. So what if the system had overstepped the boundary of neurosis when it tried to make Jonathan Wilbur an embodiment of human criteria for personal success and failure; or if Maurice at the Camelot couldn't master a value system that didn't reduce to a simple profit-and-loss calculus? They had fooled Corrigan. It was sobering to realize just how effective the combined weight of suggestion and authority had been in persuading him that the defects he had perceived in the early stages were in himself and not in the world around him. Now so much seemed so obvious.
The universal ineptness at fathoming humor and metaphor that he had observed for years—processes that involved the associative genius of human intellect at its subtlest—should have given the game away. And if not that, then surely the curious and unnatural persistence of people like Sherri and Sarah Bewley when they pressed him for explanations of where they had missed the point. Of course. All the time it was the system—wanting to know how it could do better. Mind reading was not an option.
In the real world, when people acted strangely or unsociably, others tried to gain some insight to why by getting them to talk. In the same way, when the system sought deeper understandings of what motivated the surrogates, it put animations around them to ask its questions for it. And the closer the relationship, the more personal—and hence relevant to the purpose—their questions could naturally be. Maurice, his boss at work; Sarah, his rehabilitation counselor (and the earlier attempt in the form of Simon, which had failed)—both were examples of the computers trying to get close, wanting to discover what made him behave as he did. No wonder he sometimes found himself reflecting that Horace, Sarah, and Maurice sounded the same. They were the same. His house manager was the system in disguise, too.
And even his wife! For hadn't it been Sarah who first came up with the suggestion that marrying Muriel might be a good move—for "therapeutic" reasons? Weird but frighteningly effective, he had to concede. Acting through one manufactured personality, the system had insinuated itself into his personal world in the form of another. Corrigan could only marvel at the ingenuity of it. Already the project was surpassing anything he had imagined, even in his wildest moments of selling it to others.
And now there was the risk that just at this crucial stage when Oz was surpassing all expectations, Lilly, unless he could get to her, might jeopardize the whole thing. But he was not going to get to her this way, he admitted to himself finally. Until he figured out another way, or until she got to him again, the thing was to carry on acting normally. That meant going in to work today, just as if nothing had happened. He crossed back over the river to Downtown and decided that it wasn't worth going out to Oakland. On the other hand, it was too early to go straight to the Camelot just yet.
There was a bar not far from the Vista—the hotel where Evelyn had stayed when she came down from Boston for her job interview—that he used to frequent a lot during his time with CLC, but which he hadn't been in since his "breakdown." He knew every scratch on the countertop in that place, the prints and curios on the walls, all the scuff marks and stains in the pattern on the wallpaper. There was no way that the realscaping crews could have covered every place in town.
Out of curiosity, he made his way there. The street had acquired its share of changes over the years, but apart from a new door and a coat of paint, the bar still looked pretty much the same—outside. But that was the easy part. He went inside. . . .
And sure enough, it had all been remodeled. New counter, new walls, new everything. Corrigan sighed and ordered a Bushmills, straight up. It could have been his imagination, but he was sure he detected a hint of a knowing smirk on the face of the pudgy, balding bartender.
"Okay, you got me," Corrigan conceded, raising his glass.
"Ah . . . pardon, sir?"
"It doesn't matter."
There was a pay phone in an alcove by the cigarette machine. Corrigan changed a twenty into quarters and sauntered over. He set his drink on top of the phone and tapped in the number for Information International. The codes were all different from the ones used in the real world—the change had been explained as necessitated by changes in procedure by the phone companies. Corrigan couldn't remember offhand why the Oz designers had done that.
"How can I help you?" a voice inquired.
"I'd like a number in Ireland, please."
There was a confused pause. Corrigan smiled at the thought of the drastic axing and reassembly of a whole section of the system's pointer tree that those few simple words would have caused.
"Why do you want a number there?" the voice demanded, sounding belligerent. Oh, yes, Corrigan thought to himself, it all seemed so obvious now.
"What the hell does it matter why I want it?" he retorted. "Would you please just do your job."
Another pause, then a different voice, a woman's, with a believable brogue: "Directory, which town, please?"
"Dun Laoghaire."
"Yes. And who would you be wanting there?"
"There's a grocer's shop on the corner of Clarinda Park and Upper George's Street, called Ansell's, that stays open late. What's their number?" It would be approaching ten P.M. in Ireland—five hours ahead of Pittsburgh time.
A long pause. "Ah, I don't seem to have them listed anywhere. Are you sure it's still there? It might have changed."
"How about the New Delhi? It's an Indian restaurant along the street."
"No. I don't have that either."
Corrigan grinned. The system was throwing every obstacle at him that it could come up with. "Then tell me the number of the Kingston Hotel at the bottom of Adelaide Street."
"A hotel is it, you said?" The system was trapped. Corrigan could sense it, there in the voice.
"Yes, the Kingston, on Adelaide Street. If that's gone too, give me the number of the police station around the corner."
He got the number, and after parting with a fistful of co
ins was through. "Is this the Kingston?" he inquired.
"Yes, it is," a young woman's voice replied.
"And are you at the reception desk there?"
"I am. Who's this?"
"Just somebody who would appreciate it if you could help settle a small bet we're having here. I wonder, would you mind stepping across the hall for a moment and looking out the front door to your right, and then describe to me what you can see?"
"Oh, no, I couldn't be doing that."
"Why not?"
"It's, er . . . not company policy."
Corrigan had to stifle a laugh. His eyes were watering. "Then tell me what the large picture is above the main bar in the lounge."
"That was taken away, I'm afraid."
"I'd like to have seen them do it. It's painted on the ceiling."
A sudden shrill tone announced a disconnection. "There seems to be a technical fault," Corrigan was informed when he checked with the operator.
Still smiling, he went back to the bar with his drink. In a niche among the shelves of bottles, standing between a darts trophy and jar of ticket stubs, there was a figurine that he hadn't noticed before. It was of an Irish leprechaun, complete with hat and pipe. "So, you're still haunting me, eh, Mick?" he grunted as he sat down on the stool. It was uncannily like the one he had in his hallway at home.
Chapter Seventeen
The hills behind the Bay to the east looked invitingly sunbaked after the chill and wet of winter in Pennsylvania. Below, as the plane descended on its final approach into San Francisco International Airport, fingers of houses and marinas creeping outward along the water's edge formed complex, convoluted patterns like frost on a windowpane.
Corrigan, looking casual in an open-neck shirt, light windbreaker, dove-gray jeans, and sneakers, slipped a hand over Evelyn's and leaned closer. He had been more relaxed than she had ever known him, telling stories and cracking bad jokes all through the flight. "You know, Eric was right," he said. "We've been cooped up inside CLC for too long, worrying about its politics. It's not worth it. This is the kind of thing we should be making more time for. There might be something to be said for those old books of his after all. People need to get their values straight."
She smiled and treated him to a look of mock superciliousness. "Why go back two thousand years to find that out? I've been telling you the same thing for ages."
"Have you? I never noticed."
"My point exactly."
"Then you're right too. Let the world be advised that Joseph M. Corrigan is switching to a lower-wattage lifestyle. The high-power stuff, I'll leave to the Pinders and the Tyrons. And the blood pressure that goes with it."
"Half your time would be empty," Evelyn pointed out. "No. On second thought, most of it."
"Great."
"What would you do with it?"
He kissed her on the cheek and pretended to think about it. "Oh, I'd find something."
* * *
They spent the next couple of days sight-seeing around the city. They went to the aquarium, planetarium, botanical gardens, and museums in Golden Gate Park, rode cable cars, and ate the best at Fisherman's Wharf, Japantown, and Broadway. They rented a car and drove north across the Golden Gate to the wine country, around the Bay to visit some of the researchers at Berkeley, and back across the Bay Bridge in the evening to see the SF Symphony, playing the winter season.
When they got back to the hotel, Corrigan called Hans Groener, his onetime colleague from MIT days, to confirm their visit to Stanford for the next day.
"Yes, that will be fine, Joe," Hans said over the phone. "Also, I have a surprise for you."
"Oh? What's that?"
"I talked to an old friend of yours who's out here now, and who would like to say hello again. So I invited her to join us for dinner tomorrow night and make it a foursome—Ivy Dupale."
"Hey, terrific!" Corrigan called across the room to Evelyn. "How about this. Hans knows Ivy. She's joining us for dinner tomorrow night."
"How wonderful!"
"Fine, Hans. That'll be just great."
* * *
The following morning they left San Francisco again and headed southward this time, to Stanford University. Hans was involved in sleep and dream research, which Corrigan looked on as something different and quite probably interesting, but not really relevant to his own line of work. It would be good to see Hans again anyway, even if the visit did turn out to be mainly social. But after he and Evelyn arrived at the sprawling campus with its Spanish-inspired facades of rounded arches and shady colonnades, and had talked with Hans in his laboratory for only half an hour, he realized that he had been mistaken. Hans's work could turn out to be very relevant indeed.
The DINS technology used by Pinocchio and EVIE used a configuration of electrodes inside the collar to create a dynamic pattern of ultra-high-frequency electric fields that penetrated the lower brain regions and brain stem. The fields superposed and were precisely shaped to add or cancel in different spots that could be very finely localized, which was how desired neural centers were activated selectively. However, attenuation and dispersion increased with penetration depth, reducing selectivity and hence the effectiveness of the technique. This was the main factor drat had restricted direct coupling to the medulla.
For several years, papers had been appearing in the scientific literature, reporting on an alternative approach using intersecting photon beams tuned to several narrow-frequency windows at which body tissue was found to be surprisingly transparent. It was only in talking with Hans that Corrigan first came to realize how rapidly these investigations had consolidated and were advancing. The new, emerging field was known as "Deep Selective Activation."
"The window allows photons to penetrate coherently and maintain a tight focus, below disruptive ionization energies," Hans explained. "On top of that, frequency tuning to specific neural states provides an additional dimension for fine probing. There's no need for flooding the cells with huge numbers of photons." He was lean and narrow-framed, with straight blond hair and a pale, thin-lipped countenance. The movies would have cast him as an SS officer whose sadism was a compensation for a physique that fell short of the Wagnerian Nordic ideal. In reality, Hans played American folk guitar and bred parakeets. Most of his equipment was being rebuilt currently, and his staff hidden away in offices or at computer terminals: there wasn't a lot, really, to show that day.
"I knew you were dabbling in this, but I never realized it had come so far," Corrigan confessed.
"DSA has had a boost from a lot of government work that was declassified," Hans told him. "We've got quite a club springing up here on the West Coast. SRI are putting a team together. Todd's group up at Berkeley."
"We hoped to see him yesterday, but he's away this week."
"Hughes and Lockheed are in on it. Some department in the Air Force has been very active."
The significance of the remark didn't hit Corrigan just then. He was still telling himself inwardly that he would have to make a point of keeping more up to date with the literature in future. Get back to being a scientist again, and forget trying to turn into a corporate politician. Being back in academic surroundings was reawakening his appetite for intellectual excitement.
Evelyn was studying some charts of neural organization fixed to one of the walls. Corrigan's background was software rather than interfacing, and his personal expertise at the working level lay in the area of self-modifying associative nets. What Hans was describing came closer to the kind of work that Evelyn had had experience of at Harvard and was now doing with Shipley.
"Hans, what are these references to `resonance modes' here?" she asked curiously. "I use this mapping system practically every day, but I've never come across those before."
Hans stepped across and looked pleased, rubbing the palms of his hands together and showing teeth with lots of metal. "Ah, yes, very good. You spotted it." He nodded approvingly at Corrigan. "You've found a smart lady here, Joe."
"And what else would you expect?"
Hans looked back and forth, taking in both of them. "This is something fairly recent that we've discovered through DSA. It's quite exciting—something that I think you will find particularly interesting, Joe. We call it associative neural resonances."
Corrigan's eyebrows rose. "Which are? . . ."
"Shortcuts to generating complex pictures inside the brain. We've found that triggering just a few, precisely selected, neuronal groups can activate entire chains of connected imagery."
"Wilder Penfield's experiments, back in the forties," Evelyn tossed in.
"Yes," Hans agreed. "Except we can do it from the outside." He glanced back at Corrigan. "You know how extraordinarily lucid dreams can be, yes? The images can be so rich in detail that it's often impossible to tell whether one is asleep or not."
"Sometimes I've been awake for five minutes before I realized I wasn't awake yet," Corrigan said.
"Exactly." Hans nodded and went on. "Obviously that information isn't coming from anywhere outside. It was already present there, in the mind. Random firings can set off whole trains of them that are linked together, which we experience as dreams—or it may be firings that are predisposed by recent repeated activity due to worry, intense emotional contexts, and that kind of thing."
"Like the way a bell rings," Evelyn said. "The complexity of the sound has nothing to do with how you hit it, or what with. It was already there implicitly, in the bell's structure."
"And with language," Hans said. "Words are just a code system to trigger associations already established in the listener's neural system from the experience of living. The information is in the listener, not the speaker. It seems to be a general characteristic of the neural system. And that is what we are learning to control. Activating just the right set of primitives can cause amazingly detail-rich images to be generated in the visual system. By `playing' the input combinations like a keyboard, we can induce complete event-sequences to order, inside the subject's mind, without having to inject huge data streams to specify every detail. We simply reactivate what's already there. Much faster than conventional brute-force graphics. Much more efficient."
Realtime Interrupt Page 14
