In a column written about a year ago, I raised the question, How do we know when something is alive? Just what is life, and is there an easy way to decide that something is living or not living?

The answer I gave, that living things must metabolize, while nothing else can, was the best I could offer. Today I want to approach the question from a different direction, one which suggests that my original answer was too restrictive.

Previously, my focus was on what we could term "natural" forms, the plants, animals, fungi and bacteria that we find on Earth. But now let us generalize. Let us suppose that the living things that we see all around us comprise just a small part of all possible forms of life. In particular, the set of all living forms must include what today is known as "artificial life," or just "a-life."

The subject has a long and confusing pedigree. A good place to begin is with John von Neumann. He was one of the greatest mathematicians of the twentieth century, and when the first electronic computer was built in 1946 he at once saw its long- term potential. His work on what he called "automata" defined the limits on any machine that operated by performing a sequence of logical steps. One of the questions that most engaged his extraordinary mind was, Could a machine be made that was able to create more copies of itself? In other words, could automata reproduce?

Von Neumann died of cancer in 1957 at the age of 53, with the definitive book that he intended to write on automata theory still little more than a set of preliminary notes. However, he had already introduced a class of devices that another scientist, Arthur Burks, named "cellular automata." A cellular automaton consisted of a large number of identical units laid out on a huge checkerboard arrangement of squares. Each unit was in a certain state and had certain rules that it must follow, and the action taken in each cell depended on the condition of its nearest neighbor cells. Von Neumann was convinced that a form of cellular automaton could be defined which would reproduce itself, but he did not live long enough to provide a proof.

Von Neumann's cellular automata were complex. There were 29 possible states for any unit. Ten years after von Neumann's death, another extraordinary mathematician decided that a much simpler version might exist. John Horton Conway whittled away at the von Neumann model and its rules, and by the late 1960's he had come up with a cellular automaton of staggering simplicity. There were just two internal states, on or off, which Conway called "living" or "dead." A handful of simple rules decided what happened next, based on the condition of a cell's nearest neighbors. However, the consequences were far from simple. Using Conway's model of a cellular automaton produced stable units that moved systematically across the checkerboard, units that "died," units that swallowed up others, and units which gave rise to new units both similar to and very different from themselves. Not surprisingly, Conway called his invention "Life." Much more surprisingly, he was able to show that the game of Life could be used to simulate any general-purpose computer.

I am going to call the units of Conway's game "organisms," even though they exist only on sheets of squared paper or within a computer. One thing that those organisms could not do was evolve, since there was no equivalent in Conway's rules to the random mutations that allow natural living organisms to evolve. However, introducing the equivalent of mutation into cellular automata is no great trick. All that a programmer has to do is make an occasional random change to a rule or a unit's internal state. Mostly the organism affected will vanish from the game (die) but occasionally one will live, propagate and flourish.

Today there exists a wide variety of computer programs that simulate the ways that living creatures live, feed, reproduce, compete, evolve and die. Moreover, experiments using computer software have a huge advantage over experiments done with living "wetware." We consider that a bacterium reproduces rapidly if it can make a copy of itself in 20 minutes. With today's fastest computers, an a-life form can pass through a million generations in one second. This speed-up factor of a billion allows enormous numbers of variations to be evaluated. And because the rules are allowed to mutate as well as the condition of the units, we have no idea in advance where the evolution will take us.

A-life can teach us a great deal about the ways that natural life may evolve. It has not, so far, given us much insight into the way that natural life began. On the other hand, these are early days for a-life, and the "world" in which a-life can flourish - inside computers - is far more limited than the real world even though it is growing fast.

I will end with a comment and a prediction.

The comment: Anyone who has dealt with computers has run into the problem of computer viruses. These are sections of code, quite short, that are able to copy themselves and pass from one computer to another. Usually, a computer virus is like a disease virus, in that it does damage to its host. It will wipe out memory, or cause a hard disk failure. However, in nature the fraction of viruses causing damage is minute. Most of them are content to invade a host organism, reproduce, and quietly leave. A computer virus can do exactly the same thing, and you will never be aware of it unless you make a deliberate search using anti-virus software routines.

And now the prediction: Sometime during the next half century, some form of innocuous computer virus will be discovered. Software engineers will do their best to trace it back to its creator, but they will be unable to find a source. The virus will have arisen from random mutations in computer memory, induced by the same cosmic rays and other perturbing forces that produce mutations in living tissue.

Although the origins of natural life are hidden from us by four billion years of history, the origins of spontaneously arising a-life will be available for immediate study. The effects on our theories as to how life arose on Earth are unimaginable.

"Borderlands of Science" is syndicated by: