Dembski's Free Lunch

[QED]

In the topic titled Human Evolution

Now that we have probability statisticians working on the principles of evolution, we'll all have to study the intelligently designed higher mathematics of William Dembski also.

http://en.wikipedia.org/wiki/Universal_ ... lity_bound

http://www.designinference.com/

Dembski's probability bound estimates have been studied and even I can see straight away that they are not in the least bit applicable to proving the necessity of intelligent design or disproving evolution by natural selection. In order to use a simple combinatorial expansion of all the possible states of matter in the universe to "prove" that evolution can't be possible assumes that evolution operates by chance alone which it does not. So here Dembski is merely attacking a strawman. What he leaves out is the all important selection criteria.

Interestingly Dembski acknowledges that Genetic Algorithms can be used demonstrate practical examples of evolution at work admitting that there is something "oddly compelling and almost magical" (No Free Lunch p.221) about the ability of Genetic Algorithms to find solutions unlike anything produced by human designers. However he still believes them to be no better than blind chance:

Dembski stated in his book No Free Lunch (page 212):

The No Free Lunch theorems show that evolutionary algorithms, apart from careful fine-tuning by a programmer, are no better than blind search and thus no better than pure chance.

The ev program is an evolutionary algorithm that shows that without fine tuning the selection process leads to an information gain...

...This control experiment shows that when the ev program is run without selection there is no information increase. Therefore we can attribute the information increases observed with selection on entirely to that selection. In other words, an evolutionary algorithm does far better (almost 13 standard deviations!) than 'pure chance' which is the situation when there is no selection. This falsifies Dembski's statement about No Free Lunch Theorems.

Once selection is added the above result is absolutely to be expected given that the selection criteria becomes a "steering" agency guiding the subject though its genetic state space. Genetic Programming and Genetic Algorithms are "kids stuff" these days so it's very careless for people like Dembski to come out with easily refuted statements like the one above.

Richard Wein has also written a comprehensive rebuttal to Dembski, entitled Not a Free Lunch But a Box of Chocolates. Of course I can offer a long list of critiques of Dembski's work but seeing as how we have touched on Genetic Algorithms and Genetic Programming and their demonstration that "design" can come from sources other than intelligent operators like humans, it would be interesting to debate on the ramifications for proponents of ID. If it can be shown that it does not take intelligence to produce design is it not fair to say that they haven't got a leg to stand on?

[Jose]

But, isn't this already giving Dembski a leg up in the debate? The simple analogy, which has been discussed here before, is shuffling a deck of cards. An outcome will always happen. Is it fair to say that "a bunch of math" shows that outcome to require intelligent design?

The fundamental premise of Dembski's whole approach is that we can work backwards from the end product of one round of evolution, and figure out what was designed. If we can't prove that one round of card shuffling requires a god to create the final pattern, why should we think that the outcome any other one-time process has to be special? His entire logic seems to be what he calls "specified complex information," where the key is "specified." That's the only situation in which any of the statistical calculations make any sense.

What this means is that he is basing his entire thesis on the idea that evolution planned all along to create humans. It didn't. It can't.

He can't distinguish genes from siRNA sequences from spacer DNA. The way we do that in practice is by evolutionary comparisons. Spacer is free to mutate, and evolves far faster than sequences that have selection pressure maintaining them.

However, wrong though he may be, I bet he gets a lot of free lunches by going around giving talks about it.