|
Author
|
Topic: The Bayesian Probability of Intelligent Design
|
Doug Wedel
Member
Member # 1901
|
posted 09. March 2006 10:41
Bayesian Probability measures the degree of belief an individual has in an uncertain proposition. I wish very briefly to discuss the Bayesian probability of Intelligent Design.
In the 1990s two mathematicians, Wolpert and Macready, published a series of papers focused on what have come to be known as the "no free lunch theorems". William Dembski, in 2002, published a book in which he argued that these theorems proved that evolutionary algorithms could not produce information (the inference being that life was designed by a designer who operated outside the domains of evolution). Wolpert, at al, strongly challenged Dembski's claims, charging that his use of this argument was fatally imprecise.
Those of us who are nonmathematicians find ourselves in a bit of a pickle here. How are we to evaluate these arguments full of strange symbols and curlicues? How are we to adjust our Bayesian probabilities in light of these conflicting claims of mathematicians?
In December, 2005, Wolpert and Macready published a formal refutation of Dembski's view in a reputable peer-reviewed IEEE publication. In this proof, entitled "Coevolutionary Free Lunches", Wolpert and Macready extend their discussion beyond the territory covered in their original No Free Lunch Theorems to discuss what they call "self-play" problems which they characterize as follows:
quote:
In these problems, the set of players work together to produce a champion, who then engages one or more antagonists in a subsequent multiplayer game. In contrast to the traditional optimization case where the NFL results hold, we show that in self-play there are free lunches: in coevolution some algorithms have better performance than other algorithms, averaged across all possible problems. However, in the typical coevolutionary scenarios encountered in biology, where there is no champion, the NFL theorems still hold.
Without in any way criticizing or even characterizing Dembski's earlier work, it now appears that until Professor Dembski or someone else refutes this refutation by Wolpert and Macready in a proof which can make it through rigorous peer review, that a thinking nonmathematician is FORCED to lower his/her Bayesian probability of Dembski's formal proof of Intelligent Design to below the threshold of likeliness. To the extent that we are scientific in our approach to this question, it seems to me that we must now conclude that at present the proper inference is that MATHEMATICAL PROOF EXISTS THAT COEVOLUTION CAN GENERATE INFORMATION.
IP: Logged
|
|
Micah Sparacio
Member
Member # 6
|
posted 09. March 2006 11:26
So being scientific in one's approach to this problem, basically means giving one's ascent to the latest published material in a respected peer-review journal? Is that what you are saying?
This is way too simplistic and a poor heuristic. Anyone who does that is making a mistake. Given the recent exposure of shoddy or bogus peer-reviewed literature, any thinking person should refrain from judgement until she has struggled through the material herself.
Not to mention the fact that you make way too many assumptions throughout your post (e.g. that thinking nonmathematicians apply Bayesian probabilities when determining the force or likelihood of proofs).
Further, the title of your post here indicates that you are considering the Bayesian likelihood of intelligent design, not Dembski's formal proof of CSI using the No Free Lunch theorems. There are many thinking people whose "inference to the best explanation" leads them to accept some version of intelligent design and who would be completely unfazed by a refutation of Dembski's formulation. So, even if there were such a refutation, in many people's eyes, it would only lessen the likelihood of Dembski's formal proof being relevant, but not the likelihood of intelligent design being true (so the title of your post fails on its promises).
Finally, no one ever doubted that coevolution could produce information. The question is whether coevolution can produce information of a certain high-complexity which stands in a certain relation (conformation) to a distinct pattern of information that is independent of the process that produced it.
IP: Logged
|
|
Doug Wedel
Member
Member # 1901
|
posted 09. March 2006 12:13
Thanks for your response. I'll just let my original post stand on its own and go back to the other thread unless you wanted to pursue any of this any further, which I'd be happy to do.
IP: Logged
|
|
|
|
|
Printer-friendly view of this topic