LETTERS TO THE EDITOR

HEADLINE: EVOLUTION AND ERROR

BYLINE: ALIK WIDGE; SQUIRREL HILL

    I wish to reply to a significant fallacy in Dennis Wert's Jan. 23
letter, "Evolution vs. Creation." Mr. Wert, chairman of Creation Science
Fellowship Inc., asks if it would be a good idea for a software designer
to "[refuse] intelligent design of their software in favor of random errors
to produce a good program," and then goes on to assume that the answer is
"no." In fact, it is, "yes."

    There is a whole field of computer science that falls under the rough
label of "evolutionary programming"; it includes well-known technologies
such as neural networks and genetic algorithms. The simple principle of
evolutionary programming is that if one does not understand a problem well
enough to produce a solution by mathematical design, one can have the
computer take random guesses and tell it "right" or "wrong" until (through
some clever "learning" algorithms based on the idea of natural selection)
it learns the pattern.

      Would the idea "fly on Wall Street"? There have been plenty of
technology startups devoted to these technologies. Moreover, an engine
designed by genetic algorithm is currently being prototyped and studied by
Caterpillar Inc. after simulations suggested it could cut diesel engine
emissions by 50 percent.

    These technologies are hardly perfect, and they are not suitable for
all situations because we still need to improve the mathematics behind
them. However, human beings are also imperfect; that's why we get sick and
die. The fact remains that "random error" can produce results that work
well enough to be packaged and sold.

    And for the record, I am neither a strict creationist nor a strict
evolutionary atheist. People want to believe in God; it makes our lives a
bit brighter. I, for one, think evolution would be a fine way for God to
have populated the Earth.

    ALIK WIDGE

POWRÓT