ASA Listserv response AAAS review of No Free Lunch (2002)

15 August

There is an interesting review by H. Allen Orr of Bill Dembski's new book (No
Free Lunch) in the Boston Review. The URL is:

http://bostonreview.mit.edu/BR27.3/orr.html

I haven't read the book so I can't comment on the accuracy of the review. At
face value, he carves up Dembski's ideas pretty well. (Actually, if Orr is
correct, Dembski carves himself up with no need for help from non-IDers.)

One quote from the review: "Dembski may not have much of an argument, but
you've got to admit he's got chutzpah."

Karl

19 August

Karl-

I'd like to comment breifly on the Orr review. The following paragraph
supposedly disproves Dembski's efforts:

"The problem with all this is so simple that I hate to bring it up. But here
goes: Darwinism isn't trying to reach a prespecified target. Darwinism, I
regret to report, is sheer cold demographics. Darwinism says that my
sequence has more kids than your sequence and so my sequence gets common and
yours gets rare. If there's another sequence out there that has more kids
than mine, it'll displace me. But there's no pre-set target in this game.
(Why would evolution care about a pre-set place? Are we to believe that
evolution is just inordinately fond of ATGGCAGGCAGT÷?) Dembski can pick a
prespecified target, average over all fitness functions, and show that no
algorithm beats blind search until he's blue in the face. The calculation is
irrelevant. Evolution isn't searching for anything and Darwinism is not
therefore a search algorithm."

There is a key simple flaw with Orr's analysis. His simple idea that my
sequence has more kids than yours is filled with hidden goals or targets.
To have any sequence generate "more" kids you must have met the following
requirements

1. Stable, Biologically active sequence.
2. Sequence able to replicate.
3. Sequence capable of being improved.

In Dawkins book he mentions histone 3, a protein of about 100 residues which
change in only two or three positions throughout all the diversity of life
and throughout "evolution." This sequence meets the first two requirements
but has never been improved since it was first happened upon. This data
tells us that precursors to this sequence must not have been active because
the sequence is NOT malleable- all residues must be fixed for the protein to
perform the job it currently holds. In the end Orr is assuming that we can
start with criteria 1 and 2 firmly in place and then hypothesize about 3.
Assuming that an inactive sequence can generate any kids at all as an answer
to the derivation of novel functional genetic sequence completely misses the
question.

Josh Bembenek

19 August

>There is a key simple flaw with Orr's analysis. His simple idea
>that my sequence has more kids than yours is filled with hidden
>goals or targets.<
>To have any sequence generate "more" kids you must have met the
>following requirements<
>1. Stable, Biologically active sequence.
>2. Sequence able to replicate.
>3. Sequence capable of being improved.

However, these targets are enormous. Dembski's error is to specify
the goal much more than nature does. Only pre-specified complexity
is improbable. Post-hoc, you can get as specific as you like for any
random sequence. To assert that particular complexity is
sufficiently specified to be improbable requires proving that the
number of possible ways to achieve a similar goal is small. However,
we have no idea how many different ways hypothetical living organisms
might be able to do things.

Furthermore, 1 and 3 are not required. A sequence needs to be
moderately stable, but it does not have to be biologically active to
increase in frequency in a population, nor must it be biologically
active to be mutated into something useful. 3 is not a requirement in
two ways. First, neutral or even detrimental variants may end up
generating more kids than a better variant, either through genetic
drift or through association with other, beneficial sequences.
Secondly, all sequences are capable of being improved upon, depending
upon the function that you chose to look at. Histone 3 is a terrible
digestive enzyme, for example. Even for the current function of a
gene, usually slight improvements are possible.

>In Dawkins book he mentions histone 3, a protein of about 100
>residues which change in only two or three positions throughout all
>the diversity of life and throughout "evolution." This sequence
>meets the first two requirements but has never been improved since
>it was first happened upon. This data tells us that precursors to
>this sequence must not have been active because the sequence is NOT
>malleable- all residues must be fixed for the protein to perform the
>job it currently holds. <

No; these data suggest that a precursor would have had a different
function, at least in detail, if it were a functional sequence. I
would guess that histone 3 may have common ancestry with other
histones and thus presumably descend from a more generic DNA-binding
protein. However, DawkinÌs data are outdated or incorrect, as a
search on GenBank for histone 3 yields not only several Ïhistone
3-like genesÓ but also several genes identified as histone 3. I
looked at ten sequences (see below), trying for a wide taxonomic
spread, and found variation in about 42 or so positions (depending on
how you count indels). One reference in particular seems noteworthy:
Bernhard,D. 1999. Several highly divergent histone H3 genes are
present in the hypotrichous ciliate Stylonychia lemnae. FEMS
Microbiol. Lett. 175 (1), 45-50. Apart from a sequence titled
Ïpossible H3 geneÓ from an archaeobacterium, I did not spot
prokaryote genes, so the claim that it occurs throughout all life may
be in!
accurate.

>Assuming that an inactive sequence can generate any kids at all as
>an answer to the derivation of novel functional genetic sequence
>completely misses the question.<

Not necessarily inactive. Also, there is a problem here in the
definition of novel. Is histone 3 versus histone 2 novel? How
different do sequences have to be to count as novel? Important
structural proteins in advanced eyes (vertebrate and cephalopod) show
minimal or no sequence divergence from certain enzymes-the exact same
sequences was capable of a novel function (albeit one probably not
too closely constrained by protein structure). However, inactive and
active sequences alike are inherited, and, in a finite population,
will change in frequency in the population.

Histone 3 sequences (note that not all are complete):
martkqtarkstggkaprkqlatkaarksapatggvkkphryrpgtvalreirryqkstellirklpfqrlvreiaqdfktdlrfqssavmalqeaceaylvglfedtnlcaihakrvtimpkdiqlarrirgera
martkqtarkstggkaprkqlaskaarksapvaggikkphryrpgtvalreirryqkstdllirklpfqrlvreiaqdfktdlrfqssavlalqeaaeaylvglfedtnlcaihakrvtimpkdiqlarrirgers
artkqtarkstggkaprkqlatkaarksapatggvkkphrfrpgtvalreirkyqkstellirklpfqrlvreiaqdfktdlrfqssavsalqeaaeaylvglfedtnlcaihakrvtimpkdiqlarrirgera
martkqtarkstggkaprkqlatkaarksapatggvkkphryrpgtvalreirryqkstellirklpfqrlvreiaqdfktdlrfqssavmalqeaseaylvglfedtnlcaihakritimpkdiqlarrirgera
tggkaprkqlatkaarksapatggvkkphryrpgtvalreirryqkstellirklpfqrlvreiaqdfktdlrfqssavmalqeaseaylvglfedtnlcaihakrvt
qqlaskaarksapstggvkkphrykpgtvalreirryqkstellirklpfqrlvreiaqdfksdlrfqssaigalqesvesylvslfedtnlcaihakrvtiqsvsll
ntggkaprkhiahkqakklsaaaatggvkkphrfrpgtvalreirrfqkstellirklpfqrlvreiasefkndlrfqssavlalqeaseaylvglfedtnlaaihakrvtimp
mertkqterkstggkaprkqlaskeerksapvstgikkphryrpgtvalreirksqkstdllirklpfqrlvreiaqeyktdlrfqsqavlalqeaaeaylvglfedtnlcaihakrvtiilkdiqlarrirgers
pgtcaireirkfqkstslliqcapfqrlvrgverqkeglrfqssaimalqeateayivslmadtnlacihakrvtiqpkdiqlalrlrgerh
ggvkkphryrpgtvalreirryqkstellirklpfqrlvreiaqdfktdlrfqsaaigalqeaseaylvglfedtnlcaihak

     Dr. David Campbell
     Old Seashells
     University of Alabama
     Biodiversity & Systematics
     Dept. Biological Sciences
     Box 870345
     Tuscaloosa, AL 35487 USA
     bivalve@mail.davidson.alumlink.com

20 August

Dr. Campbell

Thank you for the response, here are my comments:

>There is a key simple flaw with Orr's analysis. His simple idea
>that my sequence has more kids than yours is filled with hidden
>goals or targets.To have any sequence generate "more" kids you must have
met the
>following requirements
>1. Stable, Biologically active sequence.
>2. Sequence able to replicate.
>3. Sequence capable of being improved.

However, these targets are enormous.

--According to what calculations? For example what proportion of all
possible protein sequences are even soluble in a water based environment?
What percent fold in solution in a repeatable and predictable fashion? On
what basis do you make this claim?

Dembski's error is to specify the goal much more than nature does. Only
pre-specified complexity is improbable.

--I believe that the specification for a water soluble protein is quite
stringent, not to mention one that folds properly, and has biological
activity that enhances the adaptive capability of an organism or cell or
protobiont. Here is a relevant section from "Biochemistry" by Lubert Stryer
(p. 418) that I posted previously in a massive post:

"The way out of this dilemma is to recognize the power of cumulative
selection. Richard Dawkins, in The Blind Watchmaker, asked how long it would
take a monkey poking randomly at a typewriter to reproduce Hamlet's remark
to Polonius, "Methinks it is like a weasel." An astronomically large number
of keystrokes, of the order of 10^40, would be required. However, suppose
that we preserved each correct character and allowed the monkey to retype
only the wrong ones. In this case, only a few thousand keystrokes on
average would be needed. The crucial difference between these cases is that
the first employs a completely random search [the hemoglobin number] whereas
in the second, partially correct intermediates are retained. The essence of
protein folding is the retention of partially correct intermediates.
However, the protein-folding problem is much more difficult that the one
presented to our simian Shakespeare. First, proteins are only marginally
stable. The free-energy difference between the folded and unfolded states of
a typical 100-residue protein is 10kcal/mol. The average stabilization per
residue is only 0.1kcal/mol, which is less than random thermal energy
(RT=0.6kcal/mol at room temperature). This means that correct
intermediates, especially those formed early in folding, can be lost. The
analogy is that the monkey would be quite free to undo its correct
keystrokes. Second, the criterion of correctness is not a residue-by-residue
scrutiny of conformation by an omniscient observer [as Dawkins analogy
provides with the computer program] but rather the total free energy of the
transient species. Intermediates can be scored only by their free energies.
Third, some intermediates, called kinetic traps, have a favorable free
energy but are not on the path to final folded protein form. No wonder then
that protein folding is such an intriguing problem for both theoriticians
and experimentalists."

Not only is this an intriguing problem for experimentalists, but it must be
incorporated into whatever theoretical issues we imagine when discussing
useful, productive, protein sequences capable of conferring natural
selective advantage to an organism. Your statement that the target is huge
for nature's search of useful sequence doesn't seem to match Stryer's
comments on the protein folding problem and thus protein structure/function
in general. In addition, we aren't discussing intermediates here in this
section, we are discussing the properties of fully functional, fully
adapted, highly tailored sequences perfectly matched to do their job. Even
these protein sequences are "marginally stable."

To assert that particular complexity is sufficiently specified to be
improbable requires proving that the number of possible ways to achieve a
similar goal is small.

--Conversely what proves that the number of ways to achieve a similar goal
is large? Consider all possible protein sequences available at an average
of 150 amino acids, an astronomical number. Just sorting through the
sequence space to arrive at molecules that can fold in solution in a
predictable three dimensional structure that can eventually do something is
probably quite rare incorporating -- try obtaining soluble proteins in the
lab from molecules that are supposed to fold. Regardless of whether you
find the numbers large or small, what data backs up your claim??

However, we have no idea how many different ways hypothetical living
organisms might be able to do things.

--So why make the statement that there are many ways? The data we have
about organisms we know about don't support this idea-- it tells us just how
difficult it is to do what currently gets done. An overall theme is that a
very limited class of proteins perform a very limited subset of goals. You
do not find histones in chimps and completely different molecules in fish
doing the same job. If it were so easy to perform the job of a histone,
every organism could have its own special one perhaps unrecognizable in
sequence relation to other organisms. The origin of the molecule within the
scheme of evolution doesn't help either, if one sequence is just as good as
the next, eventually the next should be found at an equal rate of
opportunity with the first. The fact that no other sequences have been
found to perform histone's job indicate that this is the best one at its
job.

Furthermore, 1 and 3 are not required. A sequence needs to be moderately
stable....

--Even the best are only "marginally stable." What do you mean by moderate,
is your comparison with something other than random thermal energy?

....but it does not have to be biologically active to increase in frequency
in a population, nor must it be biologically active to be mutated into
something useful.

--For evolution to act, it must start with a template that has some function
that confers advantage to natural selection. If something has no function
to do anything, it cannot be naturally-selected to do nothing better. What
you are talking about is the spontaneous generation of function, not the
evolution of function. If mutation acts in the absence of any function and
thus in the absence of any selective advantage of the organism, then this is
an uncoupling of the evolutionary paradigm and is equivalent to miracle
occurrences at each arrival of novel genetic sequence.

3 is not a requirement in two ways. First, neutral or even detrimental
variants may end up generating more kids than a better variant, either
through genetic drift or through association with other, beneficial
sequences.

--This is still "improvement" in the sense that the gene is becoming more
abundant. Given enough time evolution will select out detrimental genes
regardless of their association with better genes-- they will either be
mutated back into more fit genes, or they will be breeded out due to their
disadvantages by evolutionary processes. Unless the detrimental gene itself
confers selective advantage like sickle-cell anemia, will that gene persist.
   What is more important is that if you completely lose the genes' function
(easily done with stop codons and frame-shifts,) it cannot be selected to be
improved because it isn't doing anything. Therefore if you destroy a gene,
mutations will simply accumulate at random mostly destroying the gene
completely unless you mutate the gene back to a place where it can function.

Secondly, all sequences are capable of being improved upon, depending upon
the function that you chose to look at. Histone 3 is a terrible digestive
enzyme, for example. Even for the current function of a gene, usually
slight improvements are possible.

--Do you care to comment on the minimal number of mutations required to
allow histone 3 act in a natural-selectiion-advantageous manner as a
protease? It will look nothing like histone 3 when your done! I find this
concept to completely disregard the relationship between protein sequence,
protein structure and protein function. They are tightly related, and
suggesting that any given sequence can perform any given function seems to
ignore about this particular relationship.

No; these data suggest that a precursor would have had a different
function, at least in detail, if it were a functional sequence.

--Allen Orr remarks in his criticism of Behe's work, that (Boston Review @
http://bostonreview.mit.edu/bostonreview/br21.6/orr.html ): "First it will
do no good to suggest that all the required parts of some biochemical
pathway popped up simultaneously by mutation. Although this "solution"
yields a functioning system in one fell swoop, it's so hopelessly unlikely
that no Darwinian takes it seriously. As Behe rightly says, we gain nothing
by replacing a problem with a miracle. Second, we might think that some of
the parts of an irreducibly complex system evolved step by step for some
other purpose and were then recruited wholesale to a new function. But this
is also unlikely. You may as well hope that half your car's transmission
will suddenly help out in the airbag department. Such things might happen
very, very rarely, but they surely do not offer a general solution to
irreducible complexity."

I would guess that histone 3 may have common ancestry with other histones
and thus presumably descend from a more generic DNA-binding protein.
However, DawkinÃs data are outdated or incorrect, as a search on GenBank
for histone 3 yields not only several ¦histone 3-like genes² but also
several genes identified as histone 3.

--All the histone genes you posted have very a related sequence, very
conservative residue changes (like threonine substituted for serine, one
carbon atom plus hydrogen change difference....) I would bet that they all
have the same three-dimensional structure and that critical residues are
invariant throughout all sequences. This analysis does not change the point
I was trying to make nor does it widen the relationship between sequence,
structure, and function for proteins. It does show that histone 3 is
tolerates more substitutions than hemoglobin or cytochrome C, but it only
means my example wasn't perfect. The point is still there.

Not necessarily inactive.

-Why not?

Also, there is a problem here in the definition of novel. Is histone 3
versus histone 2 novel? How different do sequences have to be to count as
novel? Important structural proteins in advanced eyes (vertebrate and
cephalopod) show minimal or no sequence divergence from certain enzymes-the
exact same sequences was capable of a novel function (albeit one probably
not too closely constrained by protein structure).

--And as indicated by Orr, occurs very very rarely. There are such examples
which occur like antifreeze proteins and the like, but generally the trend
is as stated above. Definition of novel would be a class of sequences
sharing the same three-dimensional structure and serve similar functions.

Josh Bembenek

1 September

Follow the continuing discussion for much more on this subject!