Some population-genetic
arguments used by
creationists
This is material originally assembled for the PopGenLunch weekly seminar at the University
of Washington, Seattle, for a
talk by Joe Felsenstein on 7 June
2016, and susequently supplemented and edited. It lists different arguments
based on population
genetics that creationists and Intelligent Design advocates have tried to use
to refute evolutionary theory (these are usually not different people, but
some of the same people using different arguments).
Hardy-Weinberg prevents change in gene frequencies.
I
refute this the next day at The Skeptical Zone, pointing out that
Hardy-Weinberg "equilibrium" describes the genotype frequencies that result
from given gene frequencies, not a force that holds gene frequencies constant.
Sal Cordova's demonstration that natural selection has little effect on
fixation probabilities
My
rebuttal two days later at Panda's Thumb. The figure Cordova cited about probability of
loss of the allele in the next few generations did not accurately reflect
the ultimate probability of fixation. (The Panda's Thumb blog has
recently moved to a new server, and the comments on archived threads are not
yet available -- they will be gradually restored over the next few months).
William Dembski's 2002 conservation-law argument that Complex Specified
Information cannot be achieved by natural evolutionary forces, owing to a Law
of Conservation of Complex Specified Information
Dembski's
book in its revised edition at Amazon. The book sketches his Law of Conservation of Complex
Specified Information, and the chapter describing it is described by Dembski as "the climax of
the book. Here I examine evolutionary algorithms, which constitute the mathematical underpinnings
of Darwinism. I show that evolutionary algorithms are in principle incapable of generating
specified complexity."
Wesley Elsberry's and Jeffrey Shallit's detailed negative
review of it, in a preprint version. See
especially their discovery that Dembski's Law of Conservation of Complex
Specified Information violates one of his own stated conditions.
My argument in Reports of the National Center for Science Education
in 2007 that points out that Dembski's "Conservation Law" does not keep the
Specification unchanged before and after evolutionary forces operate, as they
ought to in order to refute evolution. I think that this is the fundamental reason why the
Law of Conservation of Complex Specified Information fails to do what it is intended to do. It does not show that evolution cannot achieve
high levels of adaptation.
In the Discovery Institute's science blog Evolution News and Science
Today on April 10, 2018, Andrew Jones argues that this is a misunderstanding of Dembski's
argument, and states that Winston Ewert answered my criticism "several years
ago" (No, Ewert did not answer it). Then he hedges and ends up saying "why not
address the common sense argument?" He ends up equating Dembski's Conservation
Law argument with Dembski, Ewert, and Marks's "active information" argument,
which is different.
William Dembski's revision of his argument to define Specified
Complexity (essentially CSI) as that which cannot be achieved by natural
evolutionary forces.
My 2013 argument at Panda's Thumb that this now makes Specified Complexity a useless add-on to his
argument, with the users having to do all the heavy lifting and prove themselves
that natural selection cannot explain the adaptation, before trying to apply
Dembski's concept (which you do in order to show that natural selection cannot explain the adaptation). See the problem
there?
Walter Remine's "cost of natural selection" argument -- he actually
independently reinvented Ewens's and Crow's version of the cost of natural selection.
His self-published book The
Biotic Message: Evolution Versus Message Theory in which
he argues that the Creator designed into life a message to us, a message
that evolution
didn't happen, and that he, Remine, happens to be the first person to read it.
God's Messenger On Earth, I guess. The cost argument in the book is described
also in the paper mentioned above.
There are numerous threads on the newsgroup
sci.bio.evolution
in which I debate Remine on this issue. They include cogent rebuttals of
Remine by others such as "Perplexed in Peoria" and John Harshman.
William Dembski and Robert Marks published a peer-reviewed
article in IEEE Transactions on Systems, Man and Cybernetics A, Systems & Humans in 2009.
It defined a quantity called "active information" which reflects how much information is provided to
the evolutionary process by the smoothness of the fitness surface. This was by way of
arguing that information needed to be "front-loaded" for an evolutionary process to work in
improving fitness. At Panda's Thumb, I noted briefly that their argument did not establish that
there needed to be any Design Intervention in the evolutionary process, and that
the smoothness of the fitness surface does not need to result from previous design,
but can simply be the result of the weakness of long-range forces in ordinary physics.
William Dembski replied
to this at Uncommon Descent, which was then his blog,
and I replied to that at Panda's Thumb.
Dembski, Ewert and Marks's argument that most "evolutionary searches" can
do no better than random wandering among genotypes.
A
rebuttal by me and Tom English at Panda's Thumb in 2015 showing
that when we confine attention to evolving systems that actually reproduce and
have fitnesses, just that fact supplies much of the needed information. For
that, no intelligent designer is needed. So the broader collection of
processes that they call "evolutionary searches" is a huge straw-man (or
straw-evolutionary-processes). Only by including a great many processes that either
have genotypes whose success is not affected by their fitness, or even ones
that have change favor lower fitness, are they
able to show that "evolutionary search" does as badly as random
sampling.
A magnificent 2018 post by commenter "DiEb" (Dietmar Eben)
at the blog The Skeptical Zone surveys many possible search and optimization
algorithms with a very thorough and informed listing, concluding that Dembski, Ewert and
Marks, by decoupling the function evaluated from the function being
optimized, loosen the definition of searches unreasonably.
A lively discussion with much technical information follows, including comments
by me, Tom English, DiEb, and Dembski and Marks's mathematically skilled colleague George Montañez.
By the way, the 2011 volume, Biological Information: New Perspectives is
available online for free, here.
It reprints papers from a 2011 conference arranged by John Sanford in a meeting room rented from
Cornell University's School of Hotel Management, and represented by its
organizers as a conference "held at Cornell University".
It is a useful compendium of the arguments by Dembski, Marks, their students,
and others associated with the Discovery Institute.
Eric Holloway and others have been arguing that
conservation theorems for Kolmogorov mutual
information justify William Dembski's argument that Complex Specified
Information cannot be achieved by natural
selection by showing that a quantity called Algorithmic Specified Complexity
is conserved. I
have argued that this is wrong. ASC,
put forward in 2014 by Ewert, Marks, and
Dembski, is a quantity that computes how unusually simple the description
of a bitstring is. Ewert, Marks and Dembski never connect ASC to anything
involving how well adapted an organism is, nor do they explain whether the
counterpart to the bitstring would be the genotype or the phenotype.
So this argument has nothing to do with
fitness or the achievement of adaptation. Dembski's CSI is at least defined in
terms of adaptedness, but the whole
Algorithmic Specified Complexity literature does not make any connection to
fitness. I argue its total irrelevancy to
what evolution can or cannot do.
This has nothing to do with
fitness or the achievement of adaptation. Dembski's CSI is at least defined in
terms of adaptedness, but the whole
Algorithmic Specified Complexity literature does not make any connection to
fitness. I argue its total irrelevancy to
what evolution can or cannot do.
Tom English has also argued at TSZ that, in any
case, there is
no conservation theorem for ASC.
Wolf-Ekkehard Lönnig's argument that natural selection will be
unable to favor the genotype of highest fitness at Evolution News and Views
in March, 2016
The argument that natural selection is unlikely to find the best genotype that
has ever appeared is cited by Lönnig to a book by Lucien Cuénot in 1951.
Cuénot was a great pioneer of Mendelian genetics early in the 20th century.
To what extent is Lönnig's (and Cuénot's) argument showing that natural selection cannot
greatly increase fitness, or effectively find adaptations? And to what extent
is it simply worrying that it might not?
William Basener and John Sanford's paper in
Journal of Mathematical Biology in 2017 which argues that R.A. Fisher's famous (or
notorious) Fundamental Theorem of Natural Selection needs to have additional terms added
to account for mutation, and that when this is done the mean fitness of populations
tends to steadily decline.
My argument in a
post at The Skeptical Zone on January 9, 2018 that they have misread the population
genetics literature, and that the incorporation of mutation effects into population genetics
theory largely happened in the 1920s, before R.A. Fisher's 1930 publication of his Fundamental
Theorem, and that it did not depend on the validity or invalidity of that Theorem.
Basener and Sanford replied here to the argument that they misunderstood the population genetics
literature, giving a lot of quotes by and about R.A. Fisher. I have argued
here that they fail to establish that R.A. Fisher's
Fundamental Theorem of Natural Selection was fundamental to the mathematics of
selection versus mutation. Rather, that mathematics was already
well established in the population genetics literature before 1930.
A more specific argument by me and Michael Lynch (of Arizona State University) in The Skeptical Zone on January 23, 2018, that Basener and Sanfords model of mutation and
selection ignored recombination between loci, and assumed a particular form for the
distribution of net fitness of the haploid genomes that implied interactions between genes
that led to mutation overwhelming the selection against deleterious mutants. It argued that
including recombination would result in a very different result.
A major Intelligent Design argument, the Irreducible Complexity
argument of Michael Behe, has a part which argues on population-genetic grounds
that an adaptation that requires that several individually-deleterious mutations be
occur that are, in combination, advantageous cannot occur in real populations.
This is summarized in his book The Edge of Evolution (which you will
find here, at Amazon)
Another argument, this one in defense of young-earth creationism, is by
John Sanford, a sometime faculty member at Cornell
University. He has published a book, Genetic Entropy and the Mystery of the
Genome on how mutation rates in humans are too high and too deleterious for
humans to have been around before 4004 BC. He and his co-thinkers believe that
continuation of that
deleterious mutation imply that The End
Is Nigh. (Really! I am not making this up).
Searches on the names of William Dembski, Robert Marks, and others mentioned
above will also disclose YouTube videos of talks by them putting forth their
arguments. For more on their views see Marks's website for his Evolutionary Informatics Lab.
Graduate courses on these arguments
There are, as yet, no courses on the above arguments and counterarguments.
Understanding these issues really should be part of the education of every
graduate student in evolutionary biology.
Why should we care? Why do you need to know these arguments? Because maybe they are correct
(if they were, wouldn't you want to at least know?). Also they get used to
persuade others, and many biologists who participate in the online debates
are a bit scared of technical
population genetics and unsure how to answer them. And you
will see that all this does really matter, if, under
a government friendly to the Religious Right,
creationists use them to justify cutting off your funding. You may not care now, but
if that happens, you will really care then!
