>Dawkins is "plain wrong", as has been pointed out by Milton:
>
>"Dawkins' argument is a modern rendition of the traditional Darwinist
>approach and the error it falls into is that dubbed the 'Statistical
>Fallacy' by Francis Crick...Suppose we have a highly improbable event
>such as a perfect deal in bridge, where each of the four players
>receives a complete suit of cards. The odds against this happening
>are billions of billions of billions to one. Let us assume that since
>being manufactured the cards have been used for 99 deals and on the
>100th time the pack was shuffled, the perfect deal arose. Can we say
>that each of these previous shuffles, deals and plays of hands (number
>1 for instance) was a cumulative event that ultimately contributed to
>the perfect deal? Can we reduce the ultimate odds against the perfect
>deal by attempting to spread them around more thinly between the
>intermediate steps? Not afterwards, note, when we know the result,
>but at the time each step is occurring?
>
>The answer is no, we cannot. Like the supposedly evolving DNA, the
>cards have a memory in that the previous deals have contributed to
>their current order and the ultimate perfect deal. But being part way
>towards a perfect deal does not alter the odds on the ultimate deal,
>because some of the key random events determining the ultimate outcome
>have not yet taken place."
>
>(Milton R., "The Facts of Life: Shattering the Myth of Darwinism",
>Fourth Estate, London, 1992, p143)
>
>With this fallacy Dawkins' whole argument fails. And with it his whole
>Blind Watchmaker thesis.
>
Milton's argument misses the point: he throws out cumulative selection.
Suppose we consider a different situation that is more like what
evolutionists claim occurs in nature. Suppose we consider a large number,
N, of foursomes, each with their own deck of cards. The cards are dealt.
After the deal, the hands are ranked according to the number of one
particular suit (say clubs) that appears in them and the top x percent are
selected. These hands are reproduced and given to the other players who
had smaller numbers of clubs. Then each player is allowed to combine hands
with one other player to produce a single hand with the maximum number of
clubs. All cards given to "offspring" are replaced and the selection
process is repeated. In a scenario like this you can show that the
exponential allocation proven by Goldberg for genetic algorithms will
occur: the number of hands with large numbers of clubs will increase
exponentially. Eventually, in a finite number of hands, you will obtain a
hand with all clubs. Of course you can repeat the above four times and
obtain all four perfect hands in 4 times the time with probability one.
(statisticians please forgive any mangling of terminology in the above.
Likewise, for those of you who use genetic algorithms regularly, you may
not find my arguments to be rigorous, but I believe they are basically
correct with perhaps minor modifications)
Question: Did Milton misunderstand Dawkins? Or is he attempting to mislead
his readers?
Bill Hamilton | Vehicle Systems Research
GM R&D Center | Warren, MI 48090-9055
810 986 1474 (voice) | 810 986 3003 (FAX)