From: Wesley R. Elsberry <wre@cx33978-a.dt1.sdca.home.com>
Newsgroups: talk.origins
Date: 7 Jun 2001 19:10:29 -0400
Message-ID: <200106080021.RAA13020@CX33978-A.DT1.SDCA.HOME.COM>
Subject: Re: No free lunch theorems


In article <42YT6.14243$FS6.994617@NEWS6-WIN.SERVER.NTLWORLD.COM>,
Dene Bebbington wrote:

DB>Does anyone know whether the no free lunch theorems that
DB>Dembski refers are applicable to evolutionary algorithms?

Yes, they are.

DB>Apparently evolutionary algorithms have been successfully
DB>used to generate computer programs that solve real
DB>problems.

That would be genetic programming or evolutionary programming.
There are a variety of categories of evolutionary computation;
the production of programs is the topic of the above two
mentioned classes, and also is prominent in "artificial life"
simulations. Producing programs via evolutionary computation
is a tough field, but some success (defined here as "working
programs") has been had there. Other evolutionary computation
has been applied to various optimization and search problems,
with several successful applications (success defined here as
"it found a reasonably good solution").

DB>It seems to me that if the result of such algorithms are CSI
DB>then Dembski must be wrong.

NFL says that when you average the performance of an algorithm
over all "cost functions" of a problem, it performs no better
on average than blind search. That is for *any* algorithm,
not just evolutionary computation (which Dembski likes to
imply). This goes to early claims that certain forms of
evolutionary computation could be considered as general
problem-solvers that could be deployed without much domain
knowledge of a problem. NFL says that if you are concerned
about the relative efficiency of getting a solution, you have
to apply domain knowledge of the problem and cost function to
select an algorithm with good performance on that problem and
cost function. NFL isn't about essential capacity of an
algorithm to produce a solution; it is about comparative
efficiency of algorithms in producing solutions.

It's my opinion that Dembski misconstrues or misunderstands
what the NFL theorems say. I've passed word along that
Dembski's choice of "No Free Lunch" for the title of a book
that is due out this fall sets him up for embarrassment.
That's still the title, so far as I know. It will be
interesting to see how the reviews turn out. The introduction
to the book is online at
<http://www.idurc.org/nofreelunchintro.html>.

Dembski's latest foray into invoking NFL can be seen at
<http://www.metanexus.net/archives/message_fs.asp?ARCHIVEID=3294>.
It already has a couple of responses from Tom Schneider, which
can be seen at
<http://www.lecb.ncifcrf.gov/~toms/paper/ev/dembski/rebuttal.html>
and
<http://www.lecb.ncifcrf.gov/~toms/paper/ev/dembski/claimtest.html>.

And don't forget my Dembski page at
<http://inia.cls.org/~welsberr/ae/dembski_wa.html#sub_ea>.

If anybody has a question that they think should be posed to
Dr. Dembski, I'll be sharing the stage with him for a 90
minute workshop session on June 17th. Drop me email by the
13th, and maybe I'll get a chance to bring up your favorite
topic.

Time to get back to preparing figures for the dissertation...

--
Wesley R. Elsberry, Student in Wildlife & Fisheries Sciences, Tx A&M U.
Visit the Online Zoologists page (http://www.rtis.com/nat/user/elsberry)
Co-recipient of "Most Obnoxious Critic" status awarded by William A. Dembski
"It's not luck if you can do it consistently."-NB

[Other on-line articles]

[ID-Commentaries]