Swann, S. Andrew. Zimmerman's Algorithm
DAW Books, New York 2000
Topics: algorithm, artificial intelligence, computer security, terrorism, virus
Themes: (artificial) intelligence, life (artificial), religion
Motifs: sentient ai
"The genetic algorithm has been used for decades in the computer sciences," Nolan said. "Putting the
idea in layman's terms—you start with a large pool of computer programs with random instructions. The
person running the experiment grades each program proportionally on the extent each is able to complete
some task. Those that grade in the top ten percent are allowed to 'breed' to create a new pool of computer
programs—" p 153
"Christ, I want to
know how we allowed a single individual to be responsible for critical security measures in so many
systems. I don't understand this math crap, but that much information in one head was a security risk from
D'Arcy nodded. "That's obvious in retrospect. But given the algorithms she developed, she was the only
individual qualified to develop security measures against them. Apparently the psych profiles—or the people
who interpreted them—confused Zimmerman's dedication with loyalty."
Slowly, as if he misunderstood what D'Arcy had just said, he asked, "Are you saying that these
programs developed some sort of intelligence?"
"Collectively, yes. It may not be on a par with the human brain, but it could easily be equivalent to some
of the higher vertebrates. A true artificial intelligence. A quantum leap in the ability to process information.
A system that could learn, deal with unforeseen circumstances. It also would be robust in the face of attack
like no other software. A distributed, modular system— like the Internet—resilient in the face of hostile
action, resistant to viruses, damage to servers . . . It could even reprogram its security to respond to threats
its operators would never see. It would be the ultimate operating system."