From: Tim Ikeda (email@example.com)
Date: Fri Sep 13 2002 - 00:57:24 EDT
Hello again Josh.
Isn't this getting long? ;^)
>>"Yes. I remember that this system was used for illustrating
>>the concept of irreducible complexity. But as a test case for
>>determining whether IC systems can evolve it has some problems."
>--I agree that the problems you have defined are worth consideration,
>however I don't believe that the flagella is a poor test case (eveb
>in light of your considerations) because of the primary fact that
>it is such a clear example of what can be considered to be
>irreducible complexity. It is a well-defined system with a well-
>defined role. Things like cascades and metabolic pathways may not be
>so easily defined as is a mechanical machine devoted to swimming
>(and the ability to perform protein trafficking in some systems.)
I'm not saying that flagella aren't IC. They clearly are. But whether
a system is a great example of a concept doesn't mean that it makes
a great model system. A blue whale is a clear example of a mammal
with a developmental pathway-- how could you miss beached whale?,
they weigh tons -- but you don't study mammalian developmental
biology in whales. IC isn't the issue; the question of evolvability
*is*. The problems I've defined explain *why* the bacterial flagellum
is a *terrible* system for determining whether IC systems can evolve
or require an intervention by a designer.
>>"The first is that the organism or class of organisms which later gave
>>rise to bacteria with flagella is not well defined. So knowing what
>>components were around in the immediate pre-flagellar progenitor is
>>extremely difficult to determine (actually it's currently impossible).
>>Second, the acquisition of the flagellum likely happened roughly two
>>billion years ago, which seriously reduces the odds that any sequence
>>homologies will survive (It can happen but it's probably extremely
>>rare). The effects of time in conjunction with strong, optimizing
>>selection adds to the difficulty. Basically, as a means of directly
>>addressing the question of whether IC systems are accessible to
>>evolution, the bacterial flagellum does not strike me as an
>>experimentally resolvable system."
>--This discussion assumes that the flagella evolved, and that the only
>way to test it truly is to uncover the way it evolved. If we can
>establish that the current structure that exists today could not have
>evolved for some reason, we need not worry about evidence for its
>evolution. Are you suggesting that ID purposely chooses examples that
>have no evidence to deny their claims? I would tend to think that
>they have chosen a clearly defined system that exemplifies the
>qualities of IC as a beginning. Later their efforts can be
>diversified to other systems. In the end, if they hypothetically
>proved beyond doubt that the flagella could not evolve, then the
>evidence you want has never existed for the flagella and your
>considerations are a diversion. Of course this may not necessarily
>happen without further data, however I think that would be their goal.
I don't agree that the only *hypothetical* way to test whether the
flagella evolved is to uncover they way it evolved. While doing this
would certainly work, it may be "hypothetically" possible to
demonstrate the negative, that no natural mechanisms could have
produced flagella. But as they say, the difference between theory
and practice is that in theory, theory and practice are the same,
but in practice, they're not. My considerations are anything *but*
a "diversion". Instead, they a practical recognition of the extreme
technical difficulties it making such negative determinations.
In reference to Dembski's probability calculations re:the bacterial
flagellum, please note that William has only presented data for a
complete, de novo creation of such structure through random assembly.
Most observers suspect he still has a tiny bit more explaining to do.
>>"Cilia could be interesting from the perspective that they do exist
>>in different forms in different species and therefore it may be
>>possible to examine a relatively "new" structure with regard to
>>possible pre- and post-emergent states. That is, in some cases it
>>may be possible to determine what components were available in the
>>ancestral state and compare this to an organism carrying a newer
>>version of the system. Components of cilia (and their closely
>>related proteins) and differences in how they operate in other
>>cell functions might also be an interesting study."
>--Which seems like a more interesting study for someone trying to
>negate the notion of IC and ID. If ID wants to get off the ground
>they have to choose an example and make their case with it. Then
>they can move on to others systems and on to the difficult task
>of determining what constitutes minimal complexity and added
>complexity built off of a minimal- functioning system, systems
>built through evolution, systems not able to be built by
>evolutionary mechanisms, etc.
There is no doubt that flagella are IC. Likewise, there is no doubt
that there are hundreds more IC systems in cells. Any two connected
steps in the blood clotting cascade are IC. The sigma-45/RNA pol
interaction in bacterial transcription is IC. The interaction of
many protein tyrosine kinases and their targets are IC. Irreducible
complexity is a pretty simple concept: Removal or mutation of at
least one component of a system leads to impairment or loss of a
previously existing function. There are many interactions in the
cell, particularly in regulation and signaling pathways, where this
What I am suggesting is that some ID proponents are making their case
with a system (i.e. flagella) for which making any case is difficult.
If I was trying to avoid negation of IC=ID, that's a system I would
focus upon. Were I interested in addressing the core issue, I'd
choose something more recent in an organism with well-represented,
living relatives available for comparative study.
>>"I believe it is best to study a system in which the pre- and post-IC
>>emergence data is most readily available."
>--I don't know that such a system is readily identified, I feel
>that you would have rattled off a few if that were the case.
Actually, for me, much of the difficulty arises because I don't know
the proposed divergence dates for a particular systems with which I am
familiar. However, you may recall that I did describe some areas where
I thought one would likely find such examples.
Since Tuesday I ran across an older article by Howard Ochman titled
"Molecular archaeology of the Escherichia coli genome" (J.G. Lawrence &
H. Ochman PNAS 95:9413-9417, Aug 1998), that I thought might provide
a start. In this paper, Howard and Jeff compared the entire E.coli and
Salmonella typhimurium genomes (est. divergence time 100 Myr) to
determine what parts appear to have been acquired by horizontal gene
transfer. They estimate that about 18% of E.coli ORFs (open reading
frames: each presumably encoding a protein or RNA product) were
introduced into the genome in at least 234 transfer events. They
also estimate that up to 1.6 megabases of DNA may have been "sampled"
and lost by E.coli since its divergence from S.typhimurium.
Now, the horizontally transferred genes do not address the issue
of IC directly, but the process of absorbing and adapting these
new components might involve "new" IC systems. For example, many of
the regulatory components controlling the expression of the new genes
either may not have transferred into E.coli or may not have been
entirely compatible with the bacterium's previous mechanisms. So,
I suggest someone, such as Mike Behe or yourself, go through the list
of "new" E.coli genes (entire operons in many cases), pick out one
which is missing in closely related bacteria such as S.typhimurium
and other sister species, and see if there are any new IC regulatory
systems involved (most regulatory systems will be IC). Another
angle would be to find which new components interact with the original
E.coli proteins to determine whether the points of interaction have an
IC relationship. Then examine what makes the systems IC and see whether
they likely emerged from pre-existing components or are de novo
The benefit of doing this in E.coli is incredible. Its genome and
many of its relatives are already sequenced. Also, it's easily the
most studied and best understood organism in the world. And there
are good molecular & biochemical tools to manipulate the organism.
>Do you think the blood-clotting system, despite its "more recent
>appearance in evolution", fits this category? I think that within
>the next ten years, this type of data will be much more available
>as more and more genomes are sequenced, until then I don't think
>any system IC or not has enough pre- and post- emergence data to
>carefully argue as to its evolution.
Parts of the blood-clotting system could actually fit in this category.
Although blood clotting in other organisms is not nearly as well
studied as it is in humans and say, rats, there is data from reptiles,
amphibians fish and invertebrates. Thus, there might be a chance to
compile an inventory of proteins that existed before the modern
mammalian system (serine proteases, for example).
>This is why the entire controversy exists in the first place,
>because the data you want them to have before they carry out their
>analysis is largely unavailable for any system. Most research
>focuses on trying to discover the entire repertoire involved
>with system X in the first place, not to mention discovering
>every interaction and function and then analyzing those in
>systems other organisms and how they are related in function
>and components, etc. You have jumped several decades or more
>ahead of what science has discovered in many areas.
If what I'm asking is several decades ahead of available information
and technical knowledge in examining evolutionary histories, how
far out there is ID theorizing that evolution couldn't happen? If
it's so difficult to determine how things work in extant organisms,
how can one model all evolutionary possibilities in organisms that
existed billions of years and hundreds of billions of generations
ago? When Michael Behe et al. demand precise technical details,
such as clotting rates, pressure differentials, kinetic models of
protein interactions & etc. in the proposed evolution models of
the clotting system to demonstrate that ID didn't have an obvious
role in its origin, do you think this is reasonable? What if you
knew that we didn't have sufficient detail to theoretically
demonstrate that blood can clot *today* in *existing* organisms?
>(I work on the cell cycle and the systems controlling mitotic
>exit between s. cerevisiae and s. pombe appear to be quite
>different between extremely related organisms. Imagining how
>they relate evolutionarily at this point is quite premature in
>my book because we haven't clearly determined exactly how the
>system works in the first place, in terms of all components involved
>and function of each and relationship to other proteins in
If one considers humans and jellyfish very, very, very closely
related, then yes, with that caveat, S.cerevisiae (SC) and S.pombe
(SP) might be considered distant relations. It is estimated that
S.cerevisiae (SC) and S.pombe (SP) diverged somewhere in the
neighborhood of a billion years ago. The genetic distance between
SC and SP is roughly the same as the distance between humans and
SC. (FWIW: We are, however, probably more closely related to SC than
Bacillus subtilis is to E. coli). Yeasts are old. In addition to the
huge ancestral distance between these yeasts, other researchers have
considered additional factors, such as the relative amount of time
each of these yeasts spends in different parts of the cell cycle
as possibly contributing to the differences in their regulatory
systems. I do not think I would use S.cerevisiae and S.pombe for
comparison if I were looking to test a recently emerged IC system.
> I think it is unfair to ask for the problem to be resolved
>empirically before we theorize about what happened with biology
>through evolution or design.
I don't think it unfair to ask that people make the effort to test
their ideas once in a while, particularly before they declare
>>"If any mathematical evaluation of evolutionary probabilities is
>>likely to succeed, it can only be done on the most recently emerged
>>systems, not ones billions of years old."
>--Unless the mathematical argument can be conclusive enough to
>inform us that IC systems could not have been derived through
>evolution, and thus waiting for systems with sufficient precursors
>is waiting for nothing.
I think it is far easier to find systems with sufficient precursors
than to develop a mathematical proof that IC = ID.
>I would guess that within the next ten to twenty years we will
>have a much better answer for these questions, however with
>current data I see no great problem with looking at the flagellum.
If we are ten to twenty years away from better answers, why does
examining an ancient system that has slight chance of retaining
the necessary information to calculate evolutionary probabilities
present no great problem?
There are many reasons why your model systems for studying mitotic exit
controls are S.cerevisiae & S.pombe, and not African elephants. You can
thank people like Paul Nurse (Fun fact: Like Francis Collins, Paul is
another biologist who motorcycles) and Lee Hartwell for providing you
with a wonderful selection of cdc mutations and many others for
developing experimentally tractable systems for cloning, expression
and mapping -- Things that just aren't available yet in the pachyderm
system. While I see no great problem studying CHK and other protein
kinases in elephants (as long as we could spend a thousand years or
so in a post-doc and could convert the entire habitat of the planet
to maintain an elephant culture collection), I strongly suspect
there are other model systems out there that give faster, more
certain results. Interestingly, we could take the knowledge we
gain studying the better model systems and apply it to learn more
about the pachyderm system than we could if we had stuck with looking
at elephants alone.
As I mentioned before, where one spends one's limited resources of time
as scientist tends to have a great impact on one's ultimate success.
>In the end, even if they chose a more recent example of IC, they
>would be looking at the core IC structure, not all the variants
>or how the system has been modified. This would mean that they
>would perform the same calculations and get the same answers for
>the core of the IC structure despite however it is changed or
>adapted in various organisms.
"Core IC structure"? I think that term is redundant. With respect to
a *particular* function, a system is either IC or not. Like an onion,
one may be able strip layers of different functions off progressively
smaller subsets of an IC system, but each encapsulated unit nonetheless
represents an IC system.
I'm also not sure you understood why I suggested we look at IC systems
with varied components. By definition, an IC system has necessary
components for a function or it is not IC. Two systems with a few
differences in their components that are necessary for function
represent *two IC systems*. If, as some suggest, IC systems are not
accessible via evolution then conversions *between* IC systems is
also not possible. By inference, this means that the two IC systems,
although they may have mostly the same complement of components,
*must* involve additional ID events. One can work this idea two
ways: Developing a catalog of common variations between similar IC
systems or locating the components that were most recently acquired.
Obviously, newer systems (or sub-components) are more likely to retain
the information necessary to evaluate evolutionary or creationary
scenarios. And the newer, the better. The entire system need not
be evaluated, only the portions that differ and are themselves IC.
Studying common but independent variations of an IC system could
enable one to determine what part of the system is most amenable to
accepting changes. One could then focus on what precursors and
mechanisms might allow such modifications. Note that making
comparisons and examining or exploiting difference between similar
systems is a classic research method in the biological sciences.
> I believe it is the job of those opposed to these ideas
>to provide a system that has evidence that you are looking for and
>provide a detailed explanation for the emergence of IC, not the
Myself, I think it is a shared responsibility among those who choose
to promote either idea. If someone wants to make the general claim that
IC systems are not accessible to evolution and, importantly, wants
to convince other scientists, I think it prudent to present a case
in which the best and most reliable data is available. In other words,
they should make the most convincing, iron-clad case they can.
Similarly, if one want to demonstrate that a particular set of
evolutionary mechanisms could have produced even a system as
complex as a flagellum then they should make their best case.
You can't make "I don't know" support either evolutionary or
>If such a system fitting your criteria were readily identifiable,
>I would think that someone would have already pointed it out to
>Dembski, Behe et al., and the argument would quickly pass away.
A quick poll of my coworkers in three biotech companies and
UC Berkeley revealed no awareness of Behe, Dembski or IC.
However, it is likely that someone (Ken Miller, Terry Gray?) has
presented possible examples. The common rebuttals from Behe
and others have been that the proposed systems weren't composed
of "tightly" interacting parts, weren't complex enough, didn't
represent new functions, weren't detailed enough, or weren't made
of enough new components. Part of the problem is how one defined
a system's "function" or determined how tightly components must
interact was never precisely described. Behe's original description
was one of a system of interacting parts in which removal of anyone
parts would destroy or seriously impair the function of the system.
Years ago, when Behe first discussed his ideas on talk.origins, I
worried that he would move the goalposts of what constituted a
"function" and how "tight" a level of interaction was required.
That concern was valid.
More recently, Dembski has been presenting the notion of IC called
"Version 2.0". This is:
1. Removal of one part destroys *original* function.
2. Removal of multiple parts kills system's *original* function
3. System has numerous complex interacting parts
4. System is minimally complex in relation to its minimal function for
Note that these points were added to eliminate the possibility of
co-opting components from other systems with different functions,
intermediate & temporary components that provide bridging to an IC
system final state and addition/removal or two or more components
at the same time (e.g. removing a positive & negative signal
simultaneously) -- Essentially, all the likely mechanisms which
people have proposed (which also tend to be supported by sequence
similarities in "newer IC systems) for the evolution of IC systems.
So now an IC v2.0 system is one of multiple interacting components
which were added in a serious of numerous successive modifications,
all the while retaining the same (if minimally selectable) function.
In short, an unevolvable system. While I'll agree that an unevolvable
system *is* an unevolvable system, I'm not so sure I'll agree that
IC systems must necessary arise via the route Dembski uses in his
revised definition. That's clearly reversing the logic of the original.
IC was originally formulated as an empirically testable property
of a system that could be determined independently of the system's
origin. Behe's goal was to demonstrate that IC systems were not
unlikely to arise via evolution. The new version runs backwards
by defining an unevolvable system. If Behe wants to switch to
the new version, that's fine: But now he must show that there exists
a system that couldn't have evolved. A system that was IC under
version 1.0 is not necessarily IC as defined under v2.0.
This archive was generated by hypermail 2.1.4 : Fri Sep 13 2002 - 01:31:22 EDT