Thursday, July 31, 2008

Irreducible Complexity

This is the second in a series of posts regarding themes in Darwin’s Black Box and The Design Matrix.

Behe Starts a Furor

“By irreducibly complex I mean a single system composed of several well-matched, interacting parts that contribute to the basic function, wherein the removal of any one of the parts causes the system to effectively cease function.”
Michael Behe, Chapter 2, Darwin’s Black Box

In 1996, Dr. Michael Behe provided in his book, Darwin’s Black Box (DBB), this simple definition for a simple, yet potentially powerful concept – irreducible complexity (IC). In the subsequent years, an academic war of words broke out which beckoned the question: why the furor? Surely the above definition alone couldn’t have started the maelstrom of criticisms that began soon after it was published?

Let me propose that much of the hullaballoo (did I really just write hullaballoo???) is due to Behe’s application of IC, which amounts to IC = evolution impossible. To clarify Behe’s argument, while it is impossible for a direct evolutionary pathway to produce an IC system, it is possible for indirect evolutionary pathways to do so. However, when he scanned the scientific literature, Behe noted that there was no paper describing the indirect routes in significant detail.

The Traditional Template Invoked

The concept IC as defined by Michael Behe is simple, brilliant and stands as a potential marker of design. To provide backup, Behe eloquently presented several cases of IC systems (cilia, flagellum, blood clotting, etc.). He also anticipated most of the criticisms directed towards his thesis and answered them fairly adequately. However, Behe became entangled within the Tradition Template of the debate the moment he presented a negative argument (IC = evolution impossible), and even though Behe attempts to make a positive argument for design in Chapters 8 through 11, the negative argument dominates DBB. In my opinion, this tactic has halted the concept of IC in its tracks before teleologists could take it for a proper test drive. By arguing an impossibility, Behe unwittingly assumed the “traditional role” of the dissenter.

Accordingly, Behe’s critics were more than willing to assume their “traditional role” to demonstrate that it is possible for DE to produce IC systems. Kenneth Miller, a biologist at Brown University, is generally credited with proposing the best argument against Behe’s application of IC: cooption* - the parts of an IC system were coopted from parts of other precursor systems. With cooption, Miller showed it was possible for evolutionary mechanisms to develop IC systems (Note I said possible, not plausible nor probable). Since Behe is arguing it is impossible for evolution to produce an IC system, all Miller had to do was show it was merely possible. Thus it would appear that the Traditional Template has given a seemingly crushing blow to IC**.

Hopping Down the Bunny Trail

Enter Mike Gene. In his book, The Design Matrix (DM), Gene takes IC for a test drive within the Explanatory Continuum.

First, Gene pointed out that cooption was “really the only evolutionary explanation that has the potential to explain the origin of an [IC] system.” Second, he recognised a flaw in the cooption argument:

“The most basic problem with the conventional use of [cooption] is its complete reliance on chance.”
Mike Gene, Chapter 8, The Design Matrix

Third, Gene made the cooption explanation plausible by incorporating his working front-loaded evolution*** (FLE) hypothesis.

Mike Gene then applied the brakes and headed back to the starting line. He granted that cooption is possible, thus avoids getting entangled in the Traditional. Gene then investigated what independent evidence is needed to progress it to plausible.

If cooption was to be a viable explanation, it must be gradual. Then Mike Gene pointed out that to construct an IC system through gradual cooption, “the previous existence of simpler precursors and multiplied functions” should be abundant. If these precursors are missing, then the IC system can be said to consist of “system-dependent parts”.

“A system-dependent part would be something that does not exist or function apart from the context of the machine.”
Mike Gene, Chapter 8, The Design Matrix

With this, Mike Gene laid the framework for one of four criteria in his Design Matrix (more on this in the next post). This, coupled with FLE, has advanced IC from Behe’s simple yet powerful concept to a possible marker of design. Thanks to Mike Gene, IC has new life.

Behe, M., Darwin’s Black Box: The Biochemical Challenge to Evolution, 2006
Gene, M., The Design Matrix: A Consilience of Clues, 2007

*Even if I am wrong and Miller did not originally come up with cooption, he is at least credited with being the “front-man” for the argument against IC = evolution impossible.
**There are many other critiques of Behe and IC; one of the more extensive (and honest) ones comes from Thornhill and Ussery, “A Classification of Possible Routs in Darwinian Evolution.” from Journal of Theoretical Biology in 2000. A summary of their findings can be found in Chapter 8 of The Design Matrix
*** “Front-loading is the investment of a significant amount of information at the initial stage of evolution (the first life forms) whereby this information shapes and constrains subsequent evolution through its dissipation. This is not to say that every aspect of evolution is pre-programmed and determined. It merely means that life was built to evolve with tendencies as a consequence of carefully chosen initial states in combination with the way evolution works.” Mike Gene, Chapter 7, The Design Matrix


  1. The concept IC as defined by Michael Behe is simple, brilliant and stands as a potential marker of MAGIC.

  2. The word design in intelligent design means magically create. The designer is the Magic Man, also known as God. These are facts. Anyone who denies these facts is a liar trying to make magic look scientific. Intelligent design magic is a childish idiotic idea. It's not science. No matter how many times the Discovery Institute and other compulsive liars say it's science, it's still not science. Real scientists don't invoke intelligent design, also known as magic.

    Magical creation was renamed to Intelligent Design after the Supreme Court ruled in 1987 that Creation Science is not science.

    In 2005 a Federal Court ruled intelligent design is not science. That hasn't stopped Liars for Jesus, like Michael Behe, from trying to make intelligent design magical creation look scientific. He is fooling nobody. Real scientists laugh at Behe, while even creationists know Behe is trying to stick a religious belief into science.

    Michael Behe is an international joke. Lehigh University, where Behe has tenure, is ashamed to have Behe there for obvious reasons. Who knows how many intelligent students have avoided Lehigh University because they employ the lying moron Behe.

    Lehigh University published the following statement on their website. Read between the lines. They are calling Behe dishonest.

    Department Position on Evolution and "Intelligent Design"

    The faculty in the Department of Biological Sciences is committed to the highest standards of scientific integrity and academic function. This commitment carries with it unwavering support for academic freedom and the free exchange of ideas. It also demands the utmost respect for the scientific method, integrity in the conduct of research, and recognition that the validity of any scientific model comes only as a result of rational hypothesis testing, sound experimentation, and findings that can be replicated by others.

    The department faculty, then, are unequivocal in their support of evolutionary theory, which has its roots in the seminal work of Charles Darwin and has been supported by findings accumulated over 140 years. The sole dissenter from this position, Prof. Michael Behe, is a well-known proponent of "intelligent design." While we respect Prof. Behe's right to express his views, they are his alone and are in no way endorsed by the department. It is our collective position that intelligent design has no basis in science, has not been tested experimentally, and should not be regarded as scientific.

  3. Well, first, at least I finally learned a new word, teleologist.

    Now, I once tried but got bored reading DBB. So, I do not recall seeing "IC = evolution impossible". My recollection is that IC == macro evolution improbable.

    Do you, EE, have a quote to back up your assertion that Behe argues evolution is impossible?

    Bobxxxx, it is science. E.g., Behe has a whole chapter on the "two binding site" rule, which is falsifiable. The formation and testing of falsifiable hypotheses is science. Evolution is metaphysics.

  4. Welcome to EE, bob.
    "The concept IC as defined by Michael Behe is simple, brilliant and stands as a potential marker of MAGIC." (link added)

    Stay tuned to this 5-part series, bob. I think you'll enjoy the post where I list research themes found in The Design Matrix.

    I also see your heavy dependence on the U.S. court system. That's a double-edged sword. Do you really want the courts to determine what is and isn't science?

    "Lehigh University published the following statement on their website. Read between the lines. They are calling Behe dishonest."

    "Read between the lines" or read what you want to see? I don't see any implication of dishonesty in that statement. In fact, I agree with most of the statement, keeping in mind that evolutionary theory (more specifically macroevolution) is supported largely by circumstantial evidence.

    "I do not recall seeing "IC = evolution impossible...Do you, EE, have a quote to back up your assertion that Behe argues evolution is impossible?"

    What, I had a name change and no one told me? ;) j/k

    IC = evolution impossible is implied throughout DBB and through subsequent arguments Behe has made outside the book. I believe I qualified that equation in the second paragraph of my post:

    "To clarify Behe’s argument, while it is impossible for a direct evolutionary pathway to produce an IC system, it is possible for indirect evolutionary pathways to do so. However, when he scanned the scientific literature, Behe noted that there was no paper describing the indirect routes in significant detail."

    So I'll grant you that Behe is arguing that IC = indirect evolutionary pathways improbable, but that still opened the door for the likes of Miller et al. to argue that it is possible. All Behe had to do to avoid getting ensnared in the Traditional Template was to grant the possibility and then ask Miller for evidence to progress cooption from possibility to plausibility. It sounds like a game of semantics, but I've learned it makes a big difference (and I'm not a fan of semantic games).

  5. Hi JSS,

    I'm also an engineer who is interested in ID. The plausibility of coopton as a means for producing complex machines is what I would like to critique. At first blush it seems plausible that various components could be coopted to produce complex machines like the flagellum. However if one looks at this from a design engineering perspective, I think it begins to fall apart rather quickly.

    The problem I have with its plausibility can be found in the combinatorial dependencies within these systems. You can't just grab any ole components and slap them together and explect something like the flagellum to work. There are too many dependencies that have to be just right. This particulary applies to biological structures that developed gradually by any and all means. Each component shrinks the design space for what can come after. I've designed machines and systems for over 30 years. I can painfully attest to the fact that if you don't plan ahead correctly from your starting components you can quickly find "you can't get there from here".

    Here's a more detailed post on this where I critique Matzke's hypothetical gradualistic scenario for the flagellum. It would apply equally well for one that included cooption.

    Combinatorial Dependencies

  6. Steve, welcome to EE!

    Thank you for the link to one of your posts at TT. I have a previous post where I referenced another TT post of yours. It was a post that generated lots of discussion! (23 comments - the most at EE thus far! This ain't TT or Pharyngula, so I consider 23 good!) :)

    I am still reading through the comments of your linked post, but I would like to invite you to post here any time.
    Addressing your comment:
    1. I am well aware of the "combinatorial dependencies" of all protein machines. Mike Gene even touched on this in The Design Matrix. That is a point that cooption advocates need to address to advance it through the Explanatory Continuum.

    2. I don't even think cooption is plausible without sufficient circumstantial evidence (biological precursors). Even if this is accomplished, way more evidence would be required to advance cooption to the probable stage. Until then, evolution is a "just-so" story with regards to IC systems and protein machines.

    3. I have left the Traditional Template (as described by Mike Gene) behind and all my future arguments will be made within the Explanatory Continuum; otherwise we just return to spinning our wheels. I would encourage everyone to do the same.

  7. One problem I see is that biological and biochemical systems are not constrained by engineering sensibilities. For example, I explained in two essays on The Panda's Thumb just how IC can arise in one fell swoop, de novo, and via natural and random mechanisms. I'm fairly sure that the engineer in all of us would say "huh??!!". But facts are facts. And they tell us that there is much more to life than the engineer's logic.

    The essays:


  8. Art, thanks for the links. I'm going through them now.

    But to address one point:

    "One problem I see is that biological and biochemical systems are not constrained by engineering sensibilities."

    I'll give you one from my perspective: Newtonian mechanics, something that most engineers use everyday, don't dominate at the molecular level where I would surmise that quantum mechanics and the principles of nanotechnology would dominate.

  9. Alrighty. I've (sort of) made it through the first essay. Art, it is clear you are a smart guy and have written a good essay on the T-urf13 protein. Based on my (very) limited knowledge of biology, let me offer two points:

    1. If I've learned anything from evolution debates is plants are weird. I have accepted plant polyploidy as evidence of plant macroevolution in past debates. What I dispute is animal macroevolution. And I am sure people like yourself, Dave, rbh, Freelurker, and others will continue to attempt to show me the error of my ways. (by the way, I have no problem with the guys I've listed. They're always welcome at EE).

    2. It seems to me that the thrust of the IC argument stems from the assembly of proteins to produce a molecular machine and not necessarily concentrating on a single protein. Not to take anything away from your essay. It's good. I'm just throwing that out there.

    On to the second essay...

  10. Just finished scanning the second. I haven't read EoE yet, so I can't comment on much of your points. It also seems to be more of an extension of your first essay, so I really have nothing more to add at this time.

  11. About the comment on single proteins:

    1. Shouldn't we be focusing on function? Does it matter if multiple functions are brought to the table in one polypeptide or several?

    2. Jed Macosko has affirmed that single proteins can be considered as IC.

    3. The contrived and fallacious nature of the distinction between single vs multiple polypeptides is illustrated in a most excellent way by recalling the structures of bacterial and mammalian fatty acid synthesizing complexes.

  12. Hi JJS

    You say that Behe uses IC to make a positive argument for ID, as a marker for design. This strikes my as analogous to a detective saying he has no idea who did the murder, but at least he knows where to look for clues, which is to say it may well tell you where to look for evidence, but it is some distance from actually being evidence for IC.

    Behe published in 1996, so we must then be wondering what research he (or anyone else) has done that was prompted by IC. You mention cilia, flagellum, blood clotting; what work have teleologists done to study these systems and build an argument? As far as I know, none whatsoever.

    Mike Gene deascribes ID as a nascent protoscience. It looks doomed to stay that way.


  13. Art said:
    "1. Shouldn't we be focusing on function? Does it matter if multiple functions are brought to the table in one polypeptide or several?

    2. Jed Macosko has affirmed that single proteins can be considered as IC."

    Fair enough. It seems to me a sort of oversimplification to state a "one-component" IC system. However, I think the Modern Synthesis is not as able to explain IC systems as more components are added. I expect that the difficulties would increase exponentially rather than linearly. IMO, this is probably biology's version of what engineer's know as "scale effect": what works at smaller sizes doesn't necessarily work at larger sizes.

  14. Welcome to EE, Pixie. Man, I get one review of IC posted on TT, and all the "big leaguers" come to play at the asylum! ;)

    "Mike Gene deascribes ID as a nascent protoscience. It looks doomed to stay that way."

    And yet, later in The Design Matrix, Mike listed eight working hypotheses as potential starting points. Like I've always said, ID needs to shut up and get in the lab.

  15. Hi Art,

    I read your links. I don't have a problem, per se, with the idea that certain structures can come together (unguided) in one fell swoop. The issue for me, as a design engineer, is one of complexity and combinatorial probabilities. I think Behe has overreached in his telic conclusions for small IC structures but examples like the chloroquine resistance do highlight some important issues. One, for tightly specific mutation sets it can take a lot of random tries to reach them. Second, the more tightly specific mutation sets that are needed for a system, the more exponential the number of tries that are needed.

    Whereas chloroquine resistance only took two or three tightly specific mutations when I look at something like the bacterial flagellum there probably is more than a hundred tightly specific mutations when the assembly instructions are included.

    Now as an engineer I would expect that prexisting modules to be used when possible (cooption) but here again not just any ole modules will work. Not only must these modules be conformally compatible with each other, but they must also have the other engineering characteristics (tensile and torsional strength, fits and tolerances, have the right rpm and torque compatibility, toughness and wear resistance, and on an on. Many of these are interrelated with other components. Since the use of coopted components would more than likely be used for, at least, a somewhat different application there is no guarantee they will fit the new function.

    I think one reason why design engineers become skeptical of unguided emergence of these complex systems is that they are painfully aware of their own failures due to poor planning and how there are sometimes thousands of critical decisions required to get something that works. The more decisions that have to be made, the exponentially more ways to screw up.

    Now granted this is mostly an untuitive sense from engineering perspective, it still does reflect the real world situations for complex functional systems. So far I have not seen anyone address the combinatorial problem for atelic creation of these machines and systems, just a general statement that given enough time it will all work out. Something more than this will be necessary to dissuade, at least me, and perhaps many real-world engineers from their skepticism.

  16. Hi Steve,

    I think your engineering sense misleads when it comes to biochemistry. For example, the caricature that you weave, that protein-protein interactions are matters of exquisite, precise, rigidly-defined and ultimately highly-improbable combinations has been shown experimentally to be wrong. The ways by which networks can and do "assemble" around a primordial core are many and varied. Interaction interfaces are largely defined by small numbers of amino acids (fewer than Behe asserts), and the interfaces themselves may occur, not within rigid and highly-structured surfaces or domains, but in relatively floppy and plastic regions, domains that seem to be able to "mold" themselves in real and evolutionary time to take on new interacting partners.

    Of course, there's much more than a blog comment, or even a whole blog, can get into. But the place where the field of protein biochemistry takes us is the realization that there are no appreciable probabilistic hurdles when it comes to the evolution of interactions, or of multicomponent, even IC, complexes.

  17. Art,

    "I think your engineering sense misleads when it comes to biochemistry."

    In engineering materials are materials whether they be metals, plastics, or proteins. You seem to be discounting an engineering sensibility when it comes to organic systems but I think that if one looked at bioengineering technologies most of the basic engineering principals would be in play. Are you saying that the physics of machine design and the physical interaction of components doesn't apply to biotic systems? I think it would be informative if biologists sat down with a design engineer for a few weeks and they would see that no matter what materials were used the basic dynamics of creating a machine would be very similar.

    You speak of the plasticity of proteins. In engineering this refers to tolerances. I remember well calculating the accumulated tolerance of interrelated parts to see if the accumulation would work. For instance if the size and torque of the motor in the flagellum is outside certain specific tolerance the filament will either be uneffective or destroy itself from dynamic forces. In the long run, no matter what the materials it all comes down to physics.

    An interesting test would be to bioengineer a bacterial flagellum from scratch. My guess, from engineering similar systems, that even with proteins as the materials, very complex models would have to be build to run the "what if" scenarios on features of the interrelated components. If bearing was too small, crash. Shape of the filament components aren't the right shape or torque capacilities, crash. I could go on and on. In other words, engineering principals apply.

  18. Steve Petermann,
    "An interesting test would be to bioengineer a bacterial flagellum from scratch."

    But that's not a good analogy. Putting the evolutionary model in engineering terms, the requirement is not to produce a bacterial flagellum; the requirement is only to survive to reproduce.

    Many of your comments appear to carry this assumption that what has been produced is something that was previously specified.

  19. Nobody here seems to have caught the point that, in evolution, the potential function of a structure can change as environmental conditions change. This makes it impossible to even tell whether or not a given structure is irreducibly complex. A mousetrap may not be able to trap mice if it's missing the arm, but it still makes a good tie-clip. And the bacterial flagellum isn't irreducibly complex either.