“For the most part?”
“There are other things as well, but not so many. Mostly they seem to be dragons, but of an odd sort.”
“Odd in what way?”
“Like the rest of this thing’s magic—cold.” He looked up at Bal-Simba. “Lord, I have never seen anything like it. Nor have any of the other Watchers.”
“What do you think they are doing?”
“I cannot say with certainty, but it appears they are scouting, perhaps testing our defenses. At their present rate they will reach our lands ere long.”
Bal-Simba considered. “Then best we seek these things out to see what they are. Order our patrols south again, but cautiously. And try to steer them to a small group they can meet in overwhelming strength.”
“Jerry tells me you have developed a weapon against our enemy,” Bal-Simba said without preamble as he walked into the programmers’ work room.
“Yep,” Taj said proudly. “It’s a lysing virus. Or maybe a self-reproducing restriction enzyme would be a better way to describe it”
Jerry squinted at the code hanging above the desk Taj was using. “Describing it in English would be better yet”
“Okay,” Taj said. “Basically the problem is that this virus of the enemy’s glues spells together, with some transcription errors. Then those new spells compete against each other in what amounts to a Core Wars tournament where only the fittest survive. Eventually the winners get big and nasty.”
He gestured to the code. “What this virus does is exactly the opposite. It breaks spells into pieces at certain specific points, sort of makes them come unglued.”
“What’s going to prevent this thing from running wild and reducing every piece of code to rubble?”
Tajikawa smiled, looking more satanic than ever. “It won’t affect a piece of code smaller than a certain size.”
“Wait a minute. How do you keep the anti-virus from mutating?”
Again the satanic smile. “You can’t. It has to mutate if it’s going to do its job because the sticky virus is going to mutate. But we can make sure it won’t attack anything smaller than the limit. Here, take a look.”
Jerry scanned the indicated portion of the code.
Taj reached past him and pointed to several sections of the listing. You will note that there is not a test in there for code size. Nor is it localized to one part of the program. It’s more subtle than that.”
Jerry nodded. “Clever.”
“As far as we know there are no programs that big. None of yours anyway. It won’t prevent things from forming, but it will limit their size and that will probably limit their power.”
“Probably?”
Taj shrugged. “Theoretically these things could become efficient enough to be pretty potent within that limit, but with the smaller code sizes the global minima tend to be in pretty steep wells on the state surface. Plus there are a lot of local minima to act as traps. A genetic algorithm might reach a minimum but it would be pretty much a random event. Like the monkeys at the typewriters trying to produce Shakespeare.” He frowned. “Of course there is a question of how many monkeys and typewriters we’ve got here.” He got a faraway look as he considered the problem.
“Will this thing leave us worse off?” asked Bal-Simba, who had understood perhaps a quarter of what Tajikawa had just said.
“No.”
“Then we will do it.” He paused. “How long will it take for this thing to work?”
“It starts as soon as we tell it to execute,” Taj said. It will start here and then spread like the original virus did.”
“Wait a minute,” Jerry said, “how long will it take to affect what’s in the City of Night?”
“That’s a ways from here right?”
“And it’s protected by some kind of magic barrier.”
“Oh, the barrier shouldn’t be a problem. Eventually it will diffuse through or be carried through by an infected spell.”
“How long,” Jerry asked slowly, “is eventually?”
“Fermi numbers, around ten years.”
Bal-Simba looked at him. “What kind of numbers?”
“Fermi numbers. You know, within an order of magnitude.”
“In other words,” Jerry added, “it could happen in anywhere from one year to a century.” He shook his head, “But even a year is way too long.”