I’ve been reading some of Branden Fitelson’s bayesian papers recently, (one in particular on the issues between likelihoodists and bayesians is fantastic), and came across a review by him, Stephens, and Sober of Dembski’s book on design theories (F/S/S, for short). The review is long as far as reviews ordinarily go–15 pages or so–and is an excellent source for the inadequacies of Dembski’s metaphilosophical position on when design explanations should be accepted.
The piece has a really interesting passage, however, one that contains a common mistake that philosophers tend to make. The mistake is this: Philosopher X says p is true; philosopher Y isn’t convinced that p is true, but has no direct argument to show that p is false; so, Y attacks some generalization that implies p instead.
My favorite example of this tendency among philosophers occurs in conversations with my former colleague Michael about where to go to lunch. “Where should we eat?” “I don’t know, how about Shakespeare’s?” “Why do you want to go there?” “No special reason, we just haven’t been there in awhile.” “Do you always want to go to lunch at places where you haven’t been in awhile?”
Here’s the F/S/S version, referring to an argument that has 3 premises, the relevant one being one concerning Dembski’s conditional independence requirement (CINDE):
(2) If CINDE is true and S is warranted in accepting H (i.e., that E is due to chance), then S should assign Pr(E | I) = Pr(E). . . .
We grant premiss (1) for the sake of argument. We’ve already explained why (3) is false. So is premiss (2); it seems to rely on something like the following principle:
(*) If S should assign Pr(E|H&I) = p and S is warranted in accepting H, then S should
assign Pr(E|I) = p.
If (*) were true, (2) would be true. However, (*) is false.
First, note that premise (2), as stated, is far from obvious (and would be even if I took the time to explain it fully). So there’s a problem here. But F/S/S want to go further; they claim that (2) is false. But notice that their argument cannot demonstrate this claim. What is going on is a display of the philosopher’s predilection noted above. Notice the sequence in the passage. First, it is claimed that (2) is false. Why? Because it “seems to rely on” a certain claim that entails it, a claim that is demonstrably false. The language of “seeming” here is instructive, I think. F/S/S don’t simply say that a defense of (2) is needed, and the best they can think of is (*); they make a much stronger claim. The connection between (2) and (*) is so strong that a phenomenalistic connection is claimed to be present. That’s the predilection: we don’t just ask what generalization might lie behind a claim, the claim itself leads us so quickly to the issue of the underlying generalization that we often don’t acknowledge the move, sometimes describing the move in terms of “seeming” language.
None of this helps Dembski, of course. (2) is far from obvious, and (*) is a good example of the kind of principle he’ll have to cite to defend (2), and it won’t help; so it’s a mystery how (2) could be true. Besides, as F/S/S show, the argument has other flaws as well.