Thursday, February 22, 2007

'Ole Occam and His Razor Strike Again!

In this post at Underwhelming Evidence, called "Argument by Technobabble," ID proponents try to simply ignore evidence that Behe's claim that mammalian coagulation (clotting) system is Irreducibly Complex (IC). Which demonstrates why ID is not science, despite the claims of its proponents: denigrating evidence by calling it "technobabble" is not the same as refuting or the same as actually presenting evidence of your own.

I especially liked this little bit of craziness:

One of the great things about Intelligent Design is that it simplifies things.
Unlike the theory of evolution it passes the test of Occam's razor the
universally true scentific [sic] axiom that given a choice between a complex theory
and a simple one, the simple one has the greatest probability of being right.
It's another way of saying that tall-tales and just-so stories are usually
false. ID is so simple that even people who are un-trained in biological
sciences can make great and astounding progress. What could be a greater
indicator of it's [sic] truth than that?

First off, Occam's Razor is not a "universally true scientific axiom." It is more like a guideline or, as stated in the Wikipedia entry on the subject, "a basic tool" for use in applying the scientific method. But it's not a perfect tool, or else scientists could simply look at all the theories that fit a particular set of data, pick the "simplest" one, and bingo!

It doesn't work that way. For one thing, Occam's Razor actually says:

All things being equal, the simplest solution tends to be the best one.

So, the simplest solution isn't always the correct solution, as the author would have us believe, but rather it tends to be. Further, determining what theory is actually the simplest is no easy task either, since all things are often not equal.

For instance, here the author wants us to believe that since, "an unknown intelligence did it!" sounds like a simpler concept than "gradual speciation through natural selection and other factors," it must, in fact, be simpler.


The underlying assumptions in the ID position are enormous. First off, though they deny the identity of the "designer" is important, and claim that it could be aliens, that's not what they think. It's God, specifically the Judeo-Christian God. Well, except for a few Moonies in the ID movement, but otherwise. So, in order to accept the "simpler" theory of ID, we have to accept a mass of other ideas that aren't simple at all. In fact, Christian theology is full of complicated ideas that have caused gallons of ink to be spewed over the past 2000 years, like the Trinity, Original Sin, the problem of free will, the problem of evil, and others just as unlikely and incoherent.

In fact, it's pretty ironic that one of the criticisms of Dawkins from Christians and ID proponents is that he doesn't understand the complexities of theology well enough to critique them, but then claim that ID is the "simpler" explanation, though it rests on those very theological complexities.

And then there's the classic bit of ID irrationality:

And the theory they use to "disprove" IC - it's the most absurdly complex idea
of all: They claim that irreducable things are not irreducable because the
components have other uses in other possibly unrelated structures in life. Of
course the fatal and obvious flaw in this argument is that the other structure
must by definition be also IC. Instead of having just one IC structure the
darwinists have actually proved that there are two! There is the precursor IC
structure and also the originally observed newer IC structure. If only they
taught that in schools!

Let's see how many logic flaws we can spot here. For one thing, we've got the classic moving the goalposts: they just move the IC claim back one step when something is shown not to be IC. Very nice. It's the exact same logic they use when they ask for "transitional" fossils: when you show them one, they just move the goalposts and claim that now that you have put a fossil in the middle of a gap, there are now two gaps! This way, they just keep setting the bar higher and higher, hoping no one will notice that the bar will never stop moving and can never be met.

In simpler terms, ID proponents are simply saying, "Show me X and I'll accept the theory of evolution." But, when you show them X, they just say, "Well, now I want to see Y." If someone did that to you in regular life, you'd get pissed pretty quickly.

For instance, imagine that you go pick up a package, and they ask you to prove your identity. They say, "We just need to see some ID." (Pun unintended). You show them your driver's license. They say, "Well, that's not enough. Driver's licenses can be faked. If you had a Social Security card, we'd believe it was you." So you show them your Social Security card. They say, "Well, we're still not sure."

Naturally, you'd say, "You said you'd believe it was me if I showed you my Social Security card!" Then, they say, "Well, we're just not sure. We'd believe you if you had a passport." When you then whip out your passport and they still won't give you the package, you'd realize they were just stalling and that no amount of identification is going to get it for you. You'd realize that they weren't requesting further evidence in good faith: they never cared about or intended to consider the evidence anyway.

And you can tell the author of this post is acting in bad faith because he pulls a little switcheroo when he says: "Of course the fatal and obvious flaw in this argument is that the other structure must by definition be also IC." Did you notice it? How do we know these other structures are IC? Because they're defined as IC. Why are the defined as IC? Because they are IC. It's a tautology. IC doesn't actually mean anything. Things don't become true just because you define them as true. I can't just define "my property" as anything I want, can I? Would you give me your car just because I said, "By definition, your car is my property, so hand it over?" I doubt it.

"If only the taught that in schools," indeed, except the author doesn't understand why what he just wrote should be taught in schools: to demonstrate errors in reasoning and how to avoid them.

What a total crock.


At 11:45 AM, Blogger R. Paul Wiegand said...

My problem with IC is that the definition is not at all clear, and thus the assertion that systems demonstrating that property cannot be evolved is essentially not falsifiable.

Believe it or not, I actually emailed Behe many years ago. I was just getting started in my field (Evolutionary Computation) and had read his book, and I thought I'd an interesting idea: Generate a clear, mathematical definition of IC in the abstract, then investigate a variety of algorithmic processes on example problems from that class.

Behe responded!

His first response was that the biochemical systems which he discusses were too complex to describe faithfully with abstract mathematics. I replied that this was not my intention, but only to capture the essential idea of IC as clearly as possible. He responded that I was free to investigate this idea, but that he was not interested.

Of course I can posit my own definition. In my day, I've taken a stab at several. Most of my ideas center around notions involving combinatorial spaces where the underlying components are non-linearnly connected in such a way that certain underlying pieces present contradictory evidence regarding the quality of the solution. The trick is to keep the definition general enough to encompass a wide variety of systems so that it is clear that the biochemical systems meet such criteria from the outset.

As it turns out, one can easily show that, in certain cases, such problems can be solved by simple evolutionary search methods.

But Behe and others can always claim after any analysis that this is not really what IC means. So it is essential for him to agree to the mathematical form beforehand, or his view cannot be assailed.

I can also posit the definition I think he really believes, but were it true it would be exceedingly difficult to show that his systems meet that definition.

Here's what I think he means: What people in my field call the Needle-in-a-Haystack problem.

f(x) = { n if (x==y)
{ 0 otherwise

If true, it is certainly a good foil for evolution (or any other search algorithm). In fact, no search algorithm can do better than random search on such a problem.

Here's the rub: One cannot prove that any of the systems discussed in Darwin's Black Box is, in fact, a Needle-in-a-Haystack problem (mathematically speaking).

In fact, the honus would be on Behe to justify that the systems he discusses meet that mathematical definition. To do so, one would have to demonstrate that there is absolutely no gradient information available in the search real space whatsoever ... highly implausible in the general case.

I am suspicious of any scientist that is not willing to participate in the falsification of their own work. Science, by definition, should be falsifiable. My problem with IC is that I really do not understand precisely what it is ... and until it can be described unambiguously, it cannot be falsified. If it cannot be falsified, it is not (good) science.

At 12:24 PM, Blogger mooglar said...

Exactly. If you come up with a testable hypothesis, it could be shown to be wrong. That's why Behe, and the rest of the ID crowd, avoid ever actually giving any kind of positive definition for what ID, or its components, are.

This is the sort of thinking that comes from being convinced at the outset that your beliefs are true: you'll do anything to protect them from critique.


Post a Comment

<< Home