On Mon, 24 Oct 2016 08:36:51 -0400, jillery <
69jp...@gmail.com>
wrote:
I have the paper and it is indeed something that would be of great
interest here. I am responding with this summary of its contents as a
response to jillery's original posting of it in order to eliminate
some of the later irrelevancies in other posts.
For the record, the paper is
On Mathematical Anti-Evolutionism
Jason Rosenhouse
Science & Education
March 2016, Volume 25, Issue 1, pp 95–114
http://link.springer.com/article/10.1007/s11191-015-9801-7
Fair use interpretation of copyright law might allow me to distribute
copies to my colleagues and students for academic study but certainly
does not allow copying it here so I will describe/review it in some
detail. If you have questions about the original, please email me
directly by removing the two underscores in my listed email address.
The full abstract was made available above by jillery (that is legal).
The paper opens with a brief history of mathematical approaches to
"disproving" evolution. This goes back to a 1925 book by WA Williams:
"The Evolution of Man Scientifically Disproved." However most earlier
works before the early 2000's were based on scientific or
philosophical objections to evolution. These include Phillip Johnson,
Michael Behe, William Dembski, and Jonathan Wells. Although Dembski's
works included mathematics, that was not the major subject.
Since 2005 it seems that mathematical objectins have come to the fore.
The major recent anti-evolution works based on mathematical arguments
are cited and a number of references given to debunking the works. It
is possible that mathematic proofs that evolution can't work carries
great weight among the uninitated because of the sophistication and
precision attributed to math. Indeed we scientists often claim that
science "can't actually prove anything" but mathematicians regularly
produce proofs that are incontrovertible and undeniable. So math
carries a special cachet for those who can't follow the details.
It should be noted that this paper does not pretend to be a thorough
cataloging of all the ills of the anti-evolutionists nor a complete
review of all the debunking arguments. However it does do an
excellent job of reviewing many of the more important anti-evolution
works and describes the major flaws on those arguments.
There is a general pattern to the anti-evolution argument based on the
notion of evolution as a "search" through a set or space of possible
alternatives. Existing complex biological structures are specific
points in this space. Mathematics demonstrates that it is extremely
implausible (impossibly small probability) that any naturalistic
process or algorithmic procedure could ever find those target points.
The seach space may be composed of genes, proteins, whole genomes or
phenotypes. It does matter. The same argument is applied: how can
you reach a complex tiny target in the vastness of the space of
unsuccessful or unlivable alternatives? This whole concept suffers
from the problem that human problems such as those in computer science
or its applications has predifined targets that serve our own goals.
Biological evolution, on the other hand, is undirected and explores
genetic space where the only criterion is producing something that
works.
As an alternative, the target may not be an existing biological
structure/organism/species but rather an optimum fitness value. Here
the mathematical or computational model fails because biological
fitness is an incredibly complex result of a large number of
variables, not a simple function defined across the space. Further,
fitness is not predefined but itself "evolves" along with the changing
environment which includes all the living things in it.
There are two major ways of developing the mathematics to disprove
evolution: compute a probability value of reaching the target that is
incredibly small or produce a general principle or theorem to show
that the target is unreachable.
The first, calculation of a probability, is based on combinatorics;
that is, counting the number of possibilities. Four example there are
four bases so a DNA segment n characters long has 4^n possibilities.
Similarly a protein n amino acids long has 20^n possibilities. For
even rather small values of n, these are enormous numbers. Evolution
manages to find a small target in this enormous space, either by
random assembly or by some search strategy (algorithm) that threads
its way through the space until finding the target. Both of these are
easily "proven" to be essentially impossible.
To develop the model with any mathematical rigor, though, you need to
understand the topology and the metric of the space of possibilities.
That means knowing what points are in the "neighborhood" of eacdh
point and also how to measure how far apart two points are. You also
need to know the probability distribution over the space. The simple
fact is that in real life "protein space is vast, but its probability
distribution is hi8ghly non-uniform. Natural selection has the
consequence that enormous swaths of the space receive a probability
very close to zero, while the probability of the regions of
functionality will be relatively large." (quoted from the paper). A
second simple fact is that even though "every modern endpoint of
evolution is highly improbable, but that this by itself does not count
against the theory." This quote is cited as being developed at length
by Sober.)
The algorithmic search strategy, finding a pathway through the space
to the target, But the topology of the search space is unspecified. To
demonstrate a failure in random walks through the space (hill climbing
algorithms, for example) requires that functional points (proteins,
genomes) are highly isolated so that evolution would have to cross
vast non-functional regions to move. We do know that neutral amino
acid substitutions and neutral mutations are common and therefore
around each functional point there is a rather large swatch of
functionality. Search arguments must address these issues properly or
else are meaningless.
Rosenhouse then turns his attention to the now infamous 1966 Wistar
Conference as serving as the foundation for the framework of
subsequent developments. There were two main presentations, Murray
Eden, a EE from MIT, and Marcel-Paul Schutzenberger, a mathematician
from U. Paris.
Eden's argument was based on naive combinatorics, counting
possibilities and calculating probabilities. The space of possible
proteins is vast: there are some 10^325 possible proteins of length
250 or less but the total number of protein molecules that could have
ever existed is only about 10^52. So how could evolution have
"explored" any significant portion of that space to arrive at the
successes we see? Eden could see no mechanistic, naturalistic
constraint to would produce the result. Eden has left out the biology
in his notion of random construction. Once you have abiogenesis, some
form of initial life, then you only need search the protein space in
the vicinity of the starting point. The seach branches out like a
spider-web pattern and "natural selection has the consequence that we
stay on the threads of the web and do not stray into the regions of
non-functional proteins." Denton's book "Evolution: A Theory in
Crisins" and "Meyer's book "Darwin's Doubt" rebroadcast Eden's
argument.
The problem with abiogenesis is somewhat different because it is still
an unsolved problem. Still an awful lot more is known now that was in
1966. In any event, the works under consideration attacking evolution
usually deal with the impossibility of biological evolution following
the initial formation of life. Abiogenesis is a separate topic, as we
try to point out here.
Schutzenberger bases his objections to evolution by comparison to
computer code or formal language (in the mathematical logic sense).
The code is a string of characters that results in a phenotype. Random
typographical changes in computer code destroy the program completely.
The chances of random change producing functional results is
impossibly small. Here, too, the argument ignores the biology.
Changes in a functional protein usually do not "halt the program" but
rather produce a protein with a different or altered function
(including a non-function). And we have decades of subsequent work in
"genetic algorithms" that produce very successful computer code based
on exactly the procedures that are claimed to make evolution
impossible.
The legacy of the Wistar conference, as Rosenhouse describes it, has
three components. First, evolution is modeled as a combinatorial
process. Second, following Eden, a calculation is made not based on
biological constraints, producing ridiculously small probabilities for
evolutionary success. Third, following Schutzenberger, a general
principle is searched for that denies evolution's plausibility, again
failing to recognize biological constraints.
I have only covered the first eight pages of Rosenhouse's paper. He
continues with five pages on the probability arguments including long
sections on Dembski's "Complex Specified Information" and Behe's "Edge
of Evolution". Then follows three pages on the general principles
argument including a discussion of the Dembski's "No Free Lunch" book
based on the "no free lunch" arguments about search strategy in the
computer science literature by Wolpert ande MacReady. It also
considers the "conservation of information" argument.
I have no idea whether anybody is at all interested in those details
(or even in any of my current summary). So I will leave that to
another day. Sorry, Dr. Dr. Kleinman, you will have to wait. Your
argument is essentially Behe's in "Edge of Evolution" about evolution
of chloroquine resistance in malaria.
The conclusion is actually rather simple and simple minded. The
so-called mathematical disproofs of evolution are based entirely on
unfounded assumptions about how evolution works. These assumptions
are unstated, yet implicit in the models. Given the assumptions, the
models might well be derived correctly in the mathematical sense. But
if the assumptins are false then no proper conclusion can be drawn
from the correct mathematics. And examining and testing the
assumptions can well be done without all the mathematical formalism
piled on top. However the mathematical formalism is so "impressive"
as to completely overshadow the assumptions which are only implicit
and hidden.
As an example I quote: "Behe's 'edge of evolution' calculations were
based on the assumption that certain protein complexes could only
evolve via the occurrence of multiple, simultaneous mutations. This
assumption does all the work in his argument. If the assumption were
true, the challenge to evolution would be clear: a probability
calculation would be unnecessary. Since the assumption is not true, or
is at least ungrounded, the calculation has no value."
In the Wistar conference, one participant wrote "I do really think
that if they [the mathematicians and engineers and computer
scientists] want us to take these things seriously, they have to
present them in a way in which not only do we understand them but in
which they make biological sense." Rosenhouse ends saying "That is a
perfect description of modern mathematical anti-evolutionism."
[This is more than enough for one go. Also I am rather pressed for
time and so forego effective proof-reading and correction of egregious
misphrasings. I will leave that for you.]