Re: "The Blind Watchmaker"

From: Stevan Harnad (harnad@coglit.soton.ac.uk)
Date: Sun Nov 02 1997 - 17:49:46 GMT


> From: Katherine Lyne <kml295@soton.ac.uk>
>
> Dawkins claims that if there is a watchmaker then he is blind
> as there is not ultimate purpose in evolution.
>
> " Natural selection is the Blind watchmaker, blind because it does not
> see ahead, does not plan consequences, has no purpose in view" Pg 21

All it "does" is to "select" the successful reproducers rather than
the unsuccessful ones. But of course no one needs to do that, because
that's what successful reproduction is!

> The question "How could an organ as complex as an eye evolve?" is
> stupid. It has because of necessity and evolutionary time. This is not
> evidence for a creator.

That question is certainly not stupid! It's quite natural for everyone to
wonder about it. But it has an answer. You need to look at the
illustrations of a good book on the evolution of the eye, showing how
the eye has been shaped by the Blind Watchmaker:

Evolution of the eye and visual system / edited by John R. Cronly-Dillon
and Richard L. Gregory. Boca Raton : CRC Press, 1991.

An evolutionary link on the Web:

http://www.gen.emory.edu/MEDWEB/keyword/electronic_publications/evolution.html

> Dawkins illustrates that everything is possible in time. Even a monkey
> typing "methinks it is like a weasel" is possible given enough time.

It's not quite that simple. Some things would take an infinite amount
of time. "Combinatorial Explosion" is what happens when you have to try
all possible combinations of a large enough number of things: All
possible 6-word combinations in English is an enormous number, but it
grows disproportionately more enormous as you lengthen the sequence
from 6, to 7, 8 or more words.

In general, searches that grow longer in proportion to their number of
elements are said to require "linear time." If each time you added an
element it would make the search 3 times as long (or 1000 times, or
any finite number of times as long), then the search is still said to be
do-able in "linear time." (A growth process in which all you do is add to
the number of steps, or multiply them by a constant number, the growth is
linear.) Evolution is feasible in linear time.

But if the search grows, not as a multiple of the number of steps, but
as their square or a higher power (cube, etc.), then we are heading in
the direction of combinatorial explosion, and not even steps taken at the
speed of light for the entire history of the universe leave long enough
to go through even a minuscule fraction of them. This is what happens
as the length of the sentences you are waiting for the monkey to type by
chance grows.

> If certain errors occur and are chosen, a computer can "generate" that
> sentence through random sequences of words and cumulative selection, in
> 64 generations. Darwinian selection is cumulative and therefore not
> random as choices are made according to the environment.

Trying many combinations by paper and pencil or by computer requires
trying each possibility, one after the other. If there were a way to try
more at one time, the search would become shorter (but would require more
space, for the growing tree of combinations.

Changes and recombinations are random, but they are changes in the
structure that was already there; a mutation does not get you from
being a chihuahua to being a Great Dane in one step. Centuries of dog
breeders have imitated natural selection (unknowingly) with artificial
selection: Here the Watchmaker is not Blind, but knows what he wants in
advance. The genome of the wolf contains the traits of all the breeds
of dogs you have ever seen, but under natural selection it would have
required changes in climate and in predator/prey ecology to favour any
one of them over any other. (Indeed, some of them -- such as a
foreshortened nose and the breathing difficulties that go with it --
would have handicapped them so much that they could not have survived
in the wild.)

> Dawkins then talks about a computer programme which he created which
> writes a programme simulating splitting cells.

It is interesting to watch how such evolution-simulations generate
structure out of simple elements without any human guidance. Dennett,
in his book, will call this Darwin's dangerous idea: That simple
algorithms (rules that are followed over and over) plus chance can
generate structure is what Dennett refers to as "Darwin's Dangerous
Idea."



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:08 GMT