Tuesday, August 30, 2005

On Intelligent Design and Evolutionary Computation

Atrios, at Eschaton, has only this to say:
This Washington Post column is, perhaps, the stupidest thing I have *ever* read.

I went through it, and wow ... I agree.

Columnist Sally Jenkins' rumination on Intelligent Design wanders from topic to topic. She seems to like the idea that athletes are "transcendental," in the sense that they can do things the rest of us can't. Then she gets into the idea that athletes do stupid stuff. She lists a couple of dumb things that jocks have done ... well, you don't need Google for that. Then she gets to her point. She wants to talk about Intelligent Design:
But athletes also are explorers of the boundaries of physiology and neuroscience, and some intelligent design proponents therefore suggest they can be walking human laboratories for their theories.

First, let's get rid of the idea that ID (intelligent design) is a form of sly creationism. It isn't. ID is unfairly confused with the movement to teach creationism in public schools. The most serious ID proponents are complexity theorists, legitimate scientists among them, who believe that strict Darwinism and especially neo-Darwinism (the notion that all of our qualities are the product of random mutation) is inadequate to explain the high level of organization at work in the world. Creationists are attracted to ID, and one of its founding fathers, University of California law professor Phillip Johnson, is a devout Presbyterian. But you don't have to be a creationist to think there might be something to it, or to agree with Johnson when he says, "The human body is packed with marvels, eyes and lungs and cells, and evolutionary gradualism can't account for that." Just Check the ID

But of course that's wrong. Evolutionary theory has no problem explaining "marvels" like these. Remember, life has had five billion years to work these things out. Eyes and lungs and cells exist in many types in various species, and the evolutionary descent of modern forms is not really mysterious at all. It may be amazing to us as humans, but the process of evolution is clearly sufficiently powerful to produce these complex forms.

Oh, and let's get rid of the idea that ID (intelligent design) is anything besides a form of sly creationism. That's all it is. None of its adherents are "legitimate scientists," as she asserts, at least they are not legitimate scientists who publish research on ID. Simply stated, it is not science.

As she mentions later in the story:
Crackpot speculation? Maybe -- maybe not. ID certainly lacks a body of scientific data, and opponents are right to argue that the idea isn't developed enough to be taught as equivalent to evolution.

Look, there's no data. There are no publications. There is no theory.

It is crackpot speculation, and nothing else. There's no maybe about it. If it "lacks a body of scientific data," it's just plain not science, it is exactly crackpot speculation -- did an editor look at this? It is simply irresponsible for a paper like The Post to lead uneducated readers down this path.

One thing Ms. Jenkins seems to want to say is, if this is intelligent design, how come there're so many things wrong? Why do body parts wear down and break, and not work right sometimes? But then, she wants to think that maybe, even though there's no evidence for it, it just might be true, there just might be an intelligent designer behind the complexity of life.

That's absurd. There is no evidence to support the idea, and no valid inferential chain that concludes that there is an intelligence behind life. If you want to believe in deity, you will have to take it on faith, because the empirical world does not provide any evidence one way or the other. There are no phenomena that can only be understood through reference to a deity. On the other hand, if you prefer to think that deity is behind and inside all of it, there's no evidence to prove you wrong. Science and religion don't need to be in conflict, unless they both try to explain the same phenomena. And that's where different ones of us prefer to accept one kind of answer or another. Me, I like the explanation with the evidence.

I'm going to mention something real quick here and get off it, because I don't think people will find it very interesting. But I have never seen anyone else make this case.

There are some mathematical and engineering problems that are so hard, nobody knows how to solve them. There might be a lot of variables, and maybe every time you change the value of one, it changes the effects of all the others. There may be combinations of values that produce a pretty good solution, but better solutions exist somewhere else, in a set of values that are entirely different, and you want to find those.

The best way to solve a problem like that, these days, is through the use of something called Evolutionary Computation (EC). This is a kind of computer program that starts with random guesses at the solution to the problem, and then uses Darwinian processes -- typically recombination, mutation, and selection or "survival of the fittest" -- to evolve problem solutions.

Every year I go to a couple of conferences on this topic, and know something about it (actually, I publish several papers every year on the subject), but it's a little nerdy for this blog. The reason I bring it up here is that it needs to be noted that evolutionary processes are more powerful for optimization, for instance, for tuning a species' characteristics to an environmental niche, than any known methods that rely on human "intelligent design." These random programs can do things no heuristic program can do.

PS This is a strange coincidence. Just as I wrote "I have never seen anyone else make this case," just as I was about to submit this post to the blog, I got an email from somebody over at GMU, pointing out this article in yesterday's Boston Globe: And Now, Digital Evolution.
Recent developments in computer science provide new perspective on "intelligent design," the view that life's complexity could only have arisen through the hand of an intelligent designer. These developments show that complex and useful designs can indeed emerge from random Darwinian processes.
A growing sub-field of computer science is devoted to "evolutionary computation." The user of such a system specifies the ingredients that can be used and how the "goodness" of any particular design can be measured. The system then creates and tests thousands or millions of random combinations of the ingredients. The better combinations are allowed to produce "children" by mutation (random changes) and recombination (random part-swapping). This often produces, after many generations, genuinely novel and useful designs and inventions.

Evolutionary computation has proven to be useful for solving practical problems. It has been adopted by researchers and engineers, and it is the focus of scholarly journals and international conferences.

Go read the article, it is better than my little description.


Post a Comment

<< Home