Thursday, July 17, 2008

Blind Watchmaker Applet

Dr. Dawkins' Blind Watchmaker computer program is circulating again. When I tried to access it, my computer froze up solid, and had to be restarted in order to regain control. That notwithstanding, the site containing the applet (to which I will not return) did not seem to reveal the algorithm being used, other than the admonition that the selection process was "not" random.

Dawkins' first routine, "Methinks it is like a weasel", (or some such) also was not random. It preserved the "mutations" that were proper fits, until all the characters were filled out. This is directly contrary to biological understandings of the day, which claim that mutations are not preserved unless they are immediately beneficial to the reproduction capability of the organism.

Without access to the code or algorithm, Dawkins' next generation of 'evolutionary' programming is just as suspect as his first. For one thing, how are single mutations judged to be beneficial to the propagation capabilities of the child? Are mutations saved up in the advanced knowledge that they will be useful later? What "non-randomness" is considered useful in the selection process? Without clear access to the internal workings of the program, evidence is not available to support the reliability of the algorithm in representing evolutionary theory; so without such evidence, it cannot be accepted, especially in light of Dawkins' first program.

Dawkins' use of such programs can only be described as misleading, since they cannot be known to reflect the current state of evolutionary thinking. In most quarters, this would be considered fraudulent. Skeptics certainly would deride any such program purporting to show a 'first cause'. But Skeptics today seem selective in their skepticism.

I could be better convinced if evidence were produced of the accuracy of the program with respect to theory. Should anyone stumble on to such evidence, please point me to it.

4 comments:

Scott Hatfield . . . . said...

This is directly contrary to biological understandings of the day, which claim that mutations are not preserved unless they are immediately beneficial to the reproduction capability of the organism.

Stan, Stan, Stan. I have never met any biologist who makes the preceding claim. Mutations can be conserved simply through random chance (genetic drift). Cavalli-Sforza demonstrated this with human populations in the 1950's, and Kimura (among others) developed the math to describe mutations which are selectively neutral about four decades ago. The trick for conservation of a genetic element is not so much to immediately offer a given population some edge in survival and reproduction (which is not all that likely), to avoid immediately placing that population at a selective disadvantage.

As for algorithms for natural selection, these are easy to find. Here's one:

Choose the number of children, n, per generation
Choose a goal string g
Choose an initial parent p randomly
while (closeness(p,g) > 0)
Choose n children c[0], ... c[n-1] as mutations of p
Choose new p as the closest child to g
end while

I got that here.

Try it out for yourself!

Stan said...

Ah the goal string. Predetermined end point. Don't you see the difference? The goal is pre-selected, and the "beneficial mutations" are measured as closeness to the goal. This is not realistic! The algorithm forces the goal to be achieved. It is the effect of a directed, goal oriented, forced process. This is the exact issue with bad algorithms being used to simulate evolution, which is undirected, by definition!!

A few months ago you were arguing against mutation as a necessary condition for evolutionary movement outside the bounds of the group genome. I showed you what every reference offered, which was "evolution is natural selection on mutations". I hope we are beyond that.

Now I will have to research the source(s) where I found the references to mutations dying out in two or three generations if they are not useful. I'll get back to you on that.

In the meantime, the phrase "avoid placing the population at a selective disadvantage" doesn't make sense except in the case where the mutation is useless and benign.

Perhaps that is what you are saying, coupled with the idea that the mutation will be carried along and not die out; that a number of these benign mutations will occur; and that added together, these mutations will produce new features. Is that a fair summary?

Stan said...

Two references come easily to hand:

1. Cambridge Encyclopedia of Human Evolution, 1992, reprint 1999; pg 287, 288:

"Stabilising Selection. Life is characterised by stability rather than by change."


and

"Stabilising selection is a powerful agent that leads to genetic homogeneity. To explain how genetic variation is maintained in the face of such widespread censorship by natural selection is one of the main problems of population genetics"

and again, under "Neutral Theory of Evolution" p287:

"Most molecular polymorphism (which means most genetic polymorphism) is probably little influenced by selection. Instead it arises by mutation and is lost by chance."

2. "Evolutionary Dynamics of a Natural Population", Grant and Grant, 1989, reprint 1998, pg283:

Concluding their lengthy study of Darwin's finches on the Galapagos,

"...it is possible that a reduction is genetic variation had occurred as well, as a result of selection."

and after imaginary speciation (by song type, not beak length) was observed,

"These initial signs of population subdivision subsequently disappeared".

In fact, although the beak lengths oscillated during their observations, the Grants concluded:

"This variation [beak length oscillation] is maintained by the opposing processes of introgression and selection. They were not in balance during the decade. By the end of the study, a significant reduction in the phenotypic variance of beak length and beak width had taken place among the females, and nonsignificant differences in the same direction were apparent in the males. Given the high heritability, it is possible that a reduction is genetic variation had occurred as well, as a result of selection."
(emph added by me).

Please consider these and let me know of any more recent findings that show that mutations tend to be accumulated until they are useful.

Stan

Stan said...

Typo:
"reduction IN genetic variation had occurred...."
Sorry