The Key Fallacy of Dembski's "Explanatory Filter" Method

by Chris Cogan (Copyright 2005)

Feedback, discussion, comments, questions: Chris Cogan, ccogan@ou.edu

 

(Note: This was to be an addendum to another essay, but I decided to make this into a small essay in its own right. It basically goes after Dembski's "Explanatory Filter" method right where it lives: In its basic "logic" (or, more accurately, as you will see, illogic). Let me know if you find any real flaws in it.)

William Dembski's "Explanatory Filter" depends on dividing all events or processes into three kinds:

1. Chance
(Dembski is ambiguous about this category. He claims at one point that it may include events that have a probability of either one or zero, events that are the product of the working of natural law. This would eliminate the need for a separate category for "Regularity," and, in any case, when he applies it to his straw man misrepresentation of evolution, he effectively treats evolution as chance in the ordinary sense. The fallacy of equivocation seems to be part of his normal way of arguing for his claims.)

2. Regularity, or necessity
(As indicated above, this category would seem to be excluded from the filter altogether by at least one of Dembski's different specifications of his "filter's" categories. But if we include it, then there should be no third category for intelligent design, since intelligent design, especially in the theistic sense that Dembski is really arguing for, would have to be subsumed under this category (since any design by God would be necessarily perfect and free of chance and rigidly determined by whatever constraints and rules of design God would use). Further, even in the realm of human design, it would count as a mixture of necessity and chance, except in the simplest cases (and those would all fall under regularity alone).)

3. Intelligent Design
(Dembski weasels on whether the design involved must
be intelligent or not -- but if it need not be, then all of
his books are worse than wrong: They are then pointless,
and irrelevant to the Intelligent Design debate.)

But, we already know that the world of things and/or events and processes cannot be rationally divided up into just these three categories. There are four more that this filter conveniently ignores, though only one of them is important here:

1. Combinations of design and chance.

2. Combinations of design and regularity.

3. Combinations of design, chance, and regularity.

4. Combinations of chance and regularity.

He effectively ignores these because the last one ruins his entire "Explanatory Filter" method. Evolution is (one form of) chance and regularity combined, and thus falls into the last of the excluded categories.

He claims not to exclude this fourth category, but, in application, he does in fact exclude it. By excluding this as a possibility, he begs the question. Since the question is whether evolution can produce the results claimed for it by evolutionists, it can hardly be valid to exclude it in the initial stage of defining the filter. Evolution, if it is to be filtered out, must be filtered out by the working of the filter as it is applied to evolution, not in the process of defining the filter itself. To make his filter valid for his purpose, he must include at least the this last category in his filter, and then show that the things he claims are design or are designed things are things that cannot fall into this fourth category. By starting out with a classification that does not allow for evolution at all, he automatically "proves" that anything that is not just chance and not just regularity must be designed. This is almost a perfect textbook case of circular reasoning, because he starts out with premise that only allow for his chosen conclusions to be reached, no matter what the facts are.

It is also a variation on the old false-alternative fallacy: Start with a number of alternatives that is actually smaller (at least by one) than the number that actually need to be considered, and then, by elimination, show that the desired conclusion is the only one left of the alternatives considered (while not even mentioning the others, or while dismissing them out-of-hand so as not to have them lying around pointedly undermining one's arguments).

Since Dembski has not admitted any other alternatives to be considered, he may appear to prove something when in fact he has not. In this case, Dembski starts with only three alternatives when there are obviously at least four to consider (four that are relevant to the evolution vs. design issue, that is). By arbitrarily excluding the evolutionary option from ever even being considered by his filtering process, he is guaranteed to produce only the results he wants to prove.

To make it absolutely clear why this is invalid, suppose we make another three-category explanatory filter, but defined with only the following categories:

1. Chance

2. Regularity

3. Evolution

Now, would it be so amazing if, when we applied this filter, we never found any intelligent design, even in computers, encyclopedias, cars, airplanes, and large automated factories?

Hardly. How could it possibly arrive at an intelligent design conclusion for such things, when we don't even permit that category to be considered? Once we examine a sophisticated computer and conclude that it is not the product of chance or of mere regularity, we only have one alternative left: Evolution. Therefore, if this explanatory filter were valid, we would have to conclude that everything that is not a matter of chance or of mere regularity is purely a product of evolutionary processes, without any intelligence being involved at all.

Both Dembski's filter and mine work by elimination of alternatives. Anything that is not in one of the first two categories is automatically put into the third category because we have not allowed for any other categories.

Since there are in fact four categories to be considered as far as the evolution vs. intelligent design issue is concerned, neither Dembski's nor my filter is a valid filter for the purpose -- unless the purpose is to appear to support conclusions that are not justified by the facts, of course.

The difference between Dembski and myself in this respect is that I don't expect anyone to take my "explanatory filter" seriously, except as an illustration of the fallacy of Dembski's filter, whereas Dembski insists that his be taken seriously despite the fact that it has the same basic flaws as mine does.

 

Feedback, discussion, comments, questions: Chris Cogan, ccogan@ou.edu