Virtual `Life' Evolves in the Lab

CAN life exist inside a computer?

Before dismissing the question as idle speculation, it is necessary to come up with a definition of "life." As journalist Steven Levy shows in his newest book, "Artificial Life," it's a process scientists can now simulate with such fidelity that drawing distinctions between simulation and reality may no longer be a fruitful pursuit.

Life, argues Levy, has the ability to create something out of nothing. With enough energy and raw materials, organic life reproduces itself, as anybody who has grown a potted English ivy knows firsthand. The second quality of organic life is its ability to change - to custom-tailor itself to its environment through trial and error and survival of the fittest.

These days, the existence of self-replicating computer programs should come as no surprise to most computer users. Many have heard of computer "viruses." These programs, written by a few antisocial programmers, have hopped from floppy disk to hard-drive and back to floppy disk, and literally encircled the world. Levy argues that viruses like "Brain" and "Friday the 13th" are quasi-life forms, occupying a murky existence on the cusp between the living and the inanimate, just as biological viruses do. The difference between a biological virus and a computer virus is that one is based on molecules made of carbon, hydrogen, and nitrogen, and the other is based on information - a particular alignment of ones and zeros.

The real action in artificial life today isn't in the wild, but in laboratories at places like the Massachusetts Institute of Technology, the University of California/Los Angeles, and the University of Delaware. And the most exciting advances are not discovering new ways of propagating these artificial vermin, but developing new ones through a radical new computer-science technique: evolution.

The idea is simple: Build a world inside a computer that is populated by organisms that are, themselves, computer programs. Give the programs tasks to complete. At the end of each generation, pair up the most successful computer programs and have them "mate."

How do computer programs mate? They create progeny programs that contain instructions randomly derived from each "parent." In practice, 99 percent of the programs created this way fail to work. But some survive. And after thousands of generations have passed with millions of individual programs created, some of them work better than those devised by humans.

One exciting development Levy describes is at the University of Delaware, where an evolutionary biologist named Thomas Ray built an artificial ecology inside his computer's memory bank. Within hours after inoculating the system with a program named "Ancestor," the environment had evolved different programs, predators on those programs, and an assortment of other digital flora and fauna.

Like his previous bestseller, "Hackers: Heroes of the Computer Revolution," Levy's new book is at its best when he is looking into the professional lives of computer scientists and engineers. He has a flair for bringing these stories to life, describing, for example, how Dr. Ray inoculated the soup of his computer simulation late one night, working on a table in the bedroom of his apartment while his wife slept.

Sadly, as with "Hackers," in his quest to make the book approachable to those without degrees in computer science or higher math, Levy eliminates many details that would make the book more enjoyable to those expert audiences. That's a shame, because the technically inclined are the main readership for this volume.

Obsessed with the fate of the space program and the superconducting supercollider, the nation's popular-science press has largely ignored the amazing creation happening inside the laboratories that are researching artificial life. Artificial life often arises inside a computer without the experimenter explicitly coding in the rules for its existence. These artificial organisms learn to exploit the nooks and crannies of their universe in ways never dreamed of by their human programmers.

"The conclusion I draw from it," Levy quotes Ray as saying, "is that virtual life is out there, waiting for us to create environments for it to evolve into."