I’ve complained about this before, more than one time. Once again, Eric Holloway has declared that William Dembski’s method of inferring Design is obviously true and has not been refuted. Now I’ve complained of this argument before (here at PT) and Holloway’s only response has been to repeat himself. I pointed out specific refutations of Dembski’s use of Complex Specified Information to infer Design, gave links to them, and explained them.
Now Holloway has simply done it again. In this comment at The Skeptical Zone he says that
Briefly, Dembski’s innovation is that you can identify a pattern after the fact that allows you to reject the chance hypothesis in favor of the design hypothesis. Before, statistics used the Fisherian approach where you always have to pick the pattern before you observe the data.
This way we can infer design by just looking at an artifact without any knowledge of the artifact’s history.
Now that is surprising, because Dembski very specifically cited Fisher’s approach to statistical testing as the justification for his Design Inference. Far from replacing Fisher’s method, he was using it. His use of CSI required us to make a specification of a target set T and then compute the probability of getting a genome that was in that target set under “chance and necessity”. The target set was all genomes of high enough fitness, in effect. If this probability was small enough that such an event would occur fewer than one time in the history of our universe (say less than 10-150) then one inferred that seeing such an event allowed us to conclude in favor of Design, just as seeing a hand in poker that is improbably favorable to the dealer allows us to conclude that the dealer is cheating.
I’ve said this many times before, but the flaws in Dembski’s arguments were (1) in his 2002 argument he made use of a Law of Conservation of Complex Specified Information, while both using a form of it that could not do the job, and also violating an important condition of his own argument, and (2) in 2005-2006 he quietly abandoned that Law and reformulated his argument so that the probability that we had to compute was the probability under all relevant natural evolutionary processes. So, far from “just looking at an artifact without any knowledge of the artifact’s history”, we have to understand all the relevant possible historical processes and use them to calculate whether this artifact (or genome) is improbably well-adapted.
To see the specifics of my argument, follow the link above to my 2018 PT article. Holloway has refuted not a word of it. But he in his TSZ comment he has, once again, tried to dismiss it with a wave of his hand:
Anyways, this is all quite clear if you read the primary sources: Dembski’s papers and books. Don’t spend your time reading the rebuttals first. Most of them are quite confusingly worded and they’ll provide more fog than clarity. The only somewhat decent “rebuttals” are by Devine, and they are not so much rebuttals as they are identifying what I mentioned above, that Dembski’s mathematical work is already preempted in much of standard information theory and statistical literature.
(I will try to pa-troll this thread and delete any off-topic arguments, both the comments by off-topic trolls and by troll-chasers)