Joined: Dec. 2003
I saw your and Shallit's paper referenced on the ARN forum, read it, started thinking about Dembski's use of the word "specification," and have a few comments about it. This isn't exactly a critique of your paper but it may be helpful.
If we could visit many earth-like planets in the universe, we would expect to see some things that were similar to things on Earth and other things that were different. For example, we would expect to see life, we wouldn't expect to see George Bush. We would expect sometimes to hear language, we wouldn't expect to hear English. The first kind of thing is what I suspect Dembski had in mind when he first defined his "specification." The second kind of thing is what he would call a chance event.
Dembski's says that his "specification" is the rejection region used in Fisherian statistics. However, there are usually many possible rejection regions that one might choose when assessing the null hypothesis for a questioned event. Dembski in The Design Inference says that one of the events in the specification must be the event at issue. This seems to be cheating since, if the event is in the rejection region, the null hypothesis (that nature did it) is rejected. However, Dembski claims that this is legitimate providing one can specify the rest of a rejection region using information that is independent of the null hypothesis. Dembski says he wants to avoid "cherry picking," meaning that he does not want to pick a rejection region merely because it contains the questioned event. But if many rejection regions are possible and he deliberately picks one that contains the questioned event, what else is he doing? Put another way, how does he demonstrate that he is not "cherry picking"?
If the null hypothesis is that nature did it and if the information that defines the specification is also derived from nature, would Dembski still conclude that rejecting the null hypothesis implies that the event must be a design event? Put another way, is it logical to conclude design without using a rejection region that is based on a competing design hypothesis? Dembski's examples of rejection regions typically postulate that a man or a man-like alien is responsible. Examples are Caputo and the prime numbers in the film "Contact." He also said that the bacterial flagellum was like a boat motor.
Having said all that, it is not obvious to me that Dembski actually needs the concept of a specification. Dembski's goal is to show that there is a designer who might be God. From the "Intelligent Design Coming Clean" paper on his web site: "... a designer, who for both Van Till and me is God...." Dembski doesn't need a general procedure to do this. All he needs is a good argument for one event. A specification is an event (usually, an event that is a collection of other events). If he can show that an event that is a specification must have been designed, then it is unimportant that the specification specifies other events. Dembski has the answer he wants. Put another way, a specification is nothing more than an event that would be interesting to working scientists anyway.
I don't understand "Appendix A.1 A different kind of specification." Some strings are random and cannot be compressed, some strings can be compressed using a known program, and still other strings could be compressed except that we don't know how. If there is a program to compress a string, it could be the invention of an intelligent designer or it could be a model of a natural process. So what does this have to do with specifications?
A suggestion that may be helpful in your quest to shorten your paper: Focus on issues that Dembski can't repair. Ignore issues such as the claim that telephone numbers are CSI or the error in the prime number sequence. Discussions about Dembski's gaffes tend to obscure the more significant problems in Dembski's writing.