“Eve, I can’t do anything.”
Corrigan was also preoccupied at this time with the problem of efficiently representing and storing the enormous amount of detail implicit in any realistic depiction of the world—which the F & F demonstration had highlighted. To represent everything absolutely faithfully was impracticable but also unnecessary, since a reality was real enough if it looked real enough. The problem, then, was to find good ways of being able to cheat and get away with it.
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.