Creation Myths: A chat with Michael Behe on Irreducible Complexity

Then his claim is a contradiction. If his argument is that he has a challenge to science, his claim must be definable. If his claim isn’t definable, he doesn’t have a challenge to anything.


Except in this case he did state it like that quite explicitly.


To be fair, I believe he does clarify quite a few times that some very simple IC systems can evolve, but then that’s the point of contention how complex is too complex to be accessible? Behe seems unable to give a really objective criteria of the dividing line, which is obviously a problem.


In which case–your video?

AFAIK, the only statement that comes close to being explicit was the one extracted under oath in the Dover case.

We may see it as a hypothesis, but Behe’s followers definitely don’t. Behe is very talented in hoodwinking them (and maybe himself) in that way.

Never up front with laypeople, though. And then it becomes just another argument between experts, which helps Behe sell books while never testing his hypothesis. That’s the real “ID” here. Idea: what if we framed it more accurately by calling it “intelligent avoidance of doing science” (IADS)?

I wouldn’t say that he “seems unable.” I would say that he strenuously avoids giving criteria because on some level, he knows that his hypothesis is globally false. It’s also not obvious at all to his acolytes.

1 Like

This is where he does two things.

First is his “Edge” of evolution-type arguments, where he will reference his “waiting longer for two mutations” paper and speak about chloroquine resistance.

Second is he invokes the assumption of an exponential decline in the probability of adding more and more components to a system.

1 Like

No. Not to me at least.

I’m open to correction here, but I have a hard time seeing this as a “two or three-component IC system.” This is my understanding:

We have an organism that already has a pre-existing anaerobic citrate metabolism trait. Through a gene duplication, and a couple of mutations related to acetate metabolism (what I gathered from this video, and this point) , it can now metabolize citrate aerobically. It seems like a stretch to me to call this a “two the three component IC system”

I also read from this comment from @dsterncardinale (I might not have linked that properly):

My favorite thing about this experiment, other than the simple fact that it exists, is that the aerobic citrate metabolism trait is a directly observed experimental refutation of irreducible complexity [emphasis mine]

And so as I’m reading this, this ‘refutation’ can be extrapolated to any IC challenge such as blood clotting cascade, or ATP synthase.

Again, that seems like a stretch, an over-extrapolation. It makes me question the phrase I hear repeated: “IC has been refuted”. Just being honest…

1 Like

Climbing Mt. Everest is a powerful challenge for human beings, and yet they do it all of the time.

It depends on how you frame the IC hypothesis. If it is framed, as Behe did during out conversation, as “any IC system cannot evolve via undirected processes”, then it has been refuted. Now he then went on to attach a bunch of other conditions to that over the next 45 minutes, such as the claim that the non-duplication parts of the cit+ trait were not part of the same “system” so it doesn’t count, a qualification which, honestly, does not seem reasonable.


Why not?

Even Behe concedes that a 2-3 component system can evolve in a matter of decades. If he’s right about this, then why does it not follow that more components just take more time? Is there some sort of magical barrier that comes down and says “nope, that’s enough here, no more!”?

There’s a function: Citrate transport under aerobic conditions.

To implement this function you need at least two components:

  1. A promoter that ensures the the citrate transporter gene is actively transcribed when oxygen is present.
  2. The gene encoding the citrate transporter.

If you remove the promoter, the function stops working. If you remove the gene encoding the transporter, the function stops working. So both components are required for the function aerobic citrate transport.

Irreducible complexity: If you remove a component from a system that contributes to it’s function and it stops working then it is irreducibly complex.

The function evolved. Thus an irreducibly complex function requiring at least 2 components evolved.


I don’t really see what the problem is here. What is being argued is that the “irreducibility” of some extant system does not constitute a barrier to it’s evolution.

The argument is that if we can show that systems composed of multiple interacting parts that contribute to it’s overall function, which will stop functioning if certain parts are removed(so they are irreducible), can still evolve, then we have shown that at least this aspect of the argument from irreducible complexity is not a challenge to evolution.

So it does not suffice to merely point out that ATP synthase(for example) has a function which, if you remove parts of it, it stops being able to perform. Because it has already been shown that this kind of mutual dependence relationship between individual components can evolve, by cooption(exaptation) of components that themselves performed other functions (and which they still some times do), or used to be part of other systems.

This is why Behe went on to move the goalposts and insist that there must be some sort of degree of complexity(number of parts of the system becomes the new crucial aspect) beyond which irreducibly complex systems cannot evolve.

Behe originally argued his point about the irreducibility being unevolvable by focusing his argument on a Darwinian emergence of the system, whereby he argued against a gradual increase in the basic function as the system elaborated. He said the system’s irreducibility prevented this type of gradual “Darwinian”(cumulative selection-dominated) emergence where the function came early with a few components, and then gradually improved (with each step being selected for cumulatively because the system functioned better and better) as more and more components eere added. His point of course being that if you remove a component and the system stops working, then there is no such step-wise Darwinian pathway to the extant complex system’s function.

Biologists have responded in two ways. First by pointing out there are neutral pathways to complexity, and by pointing out that parts can be coopted from other functions, and the system as a whole can be exapted to change function along the way.

Behe has responded to the point about neutral evolution by saying neutral evolution just makes it very unlikely because you’re missing selection to aid the fixation of each new step. And Behe has responded to the point about exaptation by arguing that (paraphrasing) the probability drops off with each added component. (A point which I have above tried to argue seems without basis).

Biologists have responded to Behe’s claims about neutral processes in evolution by saying Behe ignores constructive neutral evolution, which can be inherently biased towards complexity. I have not seen him respond to this.


Watch from this point until about 15 minutes it. Right around 14, 14 and a half, after a little bit of poking and prodding, we get to an actual clear statement of what IC means in terms of whether or not something can evolve. Behe says that if a system meets the criteria for IC, it cannot evolve by any “undirected” process - including all the stuff we’ve figured out since 1859.

And that, I think, was one of the two big takeaways, at least for me. Because that is a clear, testable prediction. And any number of directly observed examples refute it.


This is exactly what is said:
At about 13 minutes in Dan takes about one minute to list things he considers non-Darwinian mechanisms and causes of change. He mentions (among other things) neutral theory, exaptation, horizontal gene transfer, epigenetics, then says:

Dan: The reason I harp on this is I wanna make sure I have this crystal clear.
Dan: When we say a system has irreducible complexity - does it preclude evolution by all these mechanisms - just the ones from 1859 - just the ones from the 1940s…?
Behe: Well - I, I’m glad to say it precludes them by all those processes you just mentioned…
Dan: Okay.
Behe: …I just wanna say that, in my mind, I consider them all random changes, remember Darwin didn’t know about mutations or anything - he said random changes plus natural selection - and you got look at each of them carefully - I think neutral mutation fits happily into Darwinian evolution except that the mutation has a selection coefficient of 0 instead of a negative or a positive one - uh, so… so that doesn’t strike me as a big deal.
Behe: And gene duplication - meh that’s fine that’s one event though. Ahh so, yeah - any unguided processes.
Dan: Alright, any unguided pro…(?)
Behe: Nods
Dan: Okay, any unguided processes!
Behe: Right.


Wow, you were good at nailing him down!

Your typing in the PPT seemed to be very effective at limiting his wiggle room. It was sad that the first thing Behe went to was quote-mining Darwin. That he would demand that someone show him neutral evolution, instead of bothering to learn about it himself, was truly pathetic.

And his claim that Darwin talked about “random changes” is not consistent with my memory. Didn’t Darwin only talk about existing heritable variation?


First I believe it is helpful to think of ID as a method vs a hypothesis. In his discussion with @swamidass he described design detection as having a stronger signal depending on the arrangement of parts you are observing. This is discussed below in the debate they had about 15 minutes in.

1 Like

First, yes, as Behe indicates, it shows that Darwinism is capable of getting 2-3 changes to occur before selection. I certainly don’t have a hard with this. But my understanding is that these were 1) a gene duplication, and 2) a few point mutations (correct me if I’m wrong). Again, to me it’s a stretch to call those three events occurring before selection a “2-3 component system.”

Here are some observations:

  1. Applying the cit+ modification to IC is painting with a pretty broad brush.

  2. The LTEE is a package deal. You have to take the good with the bad: devolution. As Behe has pointed out, the majority of beneficial mutations were loss-of-function, and the overall size of the genome is smaller. If, as you implied, you extend that experiment out a million years, that devolution will have an adverse impact.

So that’s my answer to “Why is it not obvious?”

I feel like I followed your logic just fine. I just want to give a perspective from a creationist “in the stands.” Here’s how I ‘see it’:

You’ve become convinced that the cit+ modification allows you dismiss Behe’s challenges to things like blood clotting & flagellum, which fall under the category referred to as IC. With Behe, you went through a series of debates over definitions, dividing lines and falsifiability in order to get Behe to either agree to something, or perhaps say something that justifies your counter-hypothesis. To me I can’t help from thinking this sounds like quibbling over wordings in search of some “loop-hole” (again, my perspective). And not surprisingly, Behe wasn’t compliant to that.

You may win some quibble over the definition of IC, the lack of “dividing lines” or lack of falsifiability, but those do nothing to convince me that IC has been ‘refuted’. It just sounds like you really want to refute it, which is admirable. You certainly have the freedom to come to your own conclusion of ‘refutation’. But don’t be surprised that others aren’t convinced.

Additionally (again, my “view from the stands”), I feel like this phrase reveals what I’m saying:

BTW I’d have to disagree. And this is the really important part. For me it’s not about ‘framing’, it’s about the fact that large ‘leaps’ exist, call them what you want. Not just flagellum, but also things like single-celled to multiple-celled, asexual to sexual, various steps along the way to a working eye. Some of these are significant. And a chain is only as strong as its weakest link.

It only takes one link, so here’s one: blood clotting. It’s covered in Behe’s “A Mousetrap for Darwin”, which I recently read about. After reading that, I concluded that blood clotting cascade IS a serious challenge that no-one has refuted.

MY CONCLUSION: Hearing “refuted” from evolutionists has been losing its meaning. It becomes a trust issue. I hear “IC has been refuted.” Oh really? Then I hear “I see lots of phylogeny signals.” Interesting. “Overwhelming evidence!” I see…

Behe is still making the same mistake you always make… He assumes a PURPOSEFUL arrangement when all we have evidence for is a FUNCTIONAL arrangement. Purposeful assumes conscious intent which has never been demonstrated. Conscious intent in creating a “design” is the thing Behe is supposed to be showing. On the other hand natural processes are empirically observed to produce function arrangements since such arrangements aid in reproductive success.


It’s not really clear what you mean by “before selection”? At no point during the LTEE was selection removed.
A second point is your use of the term “Darwinism”. I want to make sure I understand how that term functions(what it refers to) in a sentence you use.

Are we to take it you are referring to everything that is part of modern evolutionary theory with that term, including things like neutral theory, mutation bias, horizontal gene transfer, constructive neutral evolution, and so on? That similar to Behe, “Darwinism” is really just another word for everything mainstream science considers part of the modern theory of biological evolution?

That’s true yes.

But you’re not telling me why. The system has at least two components, does it not? And they both are necessary for the function, aren’t they?

So what is the “stretch” here?

You call them “events” as if that is somehow an argument against the evolution of irreducibly complex systems, but how else would something evolve but by a series of mutational events having phenotypical effects?

I can’t make sense of this point you’ve raised here.

That doesn’t appear to be an observation, but more like a subjective characterization. You appear to be suggesting(if I understand you correctly) that someone has gone beyond the definitions, but you don’t seem to have explained how exactly it was violated or overstepped.

It’s a stretch(why?). They’re events(what else would they be?). We’re painting with a broad brush. These are basically the essence of your complaints and I don’t understand them as they are.

Perhaps this can help clarify.
Pick one sentence here you think is factually incorrect and explain why:
A) If a system consists of multiple interacting parts that contribute to it’s essential function where if you remove any of these parts and the system stops functioning, then the system is irreducibly complex.
B) There is the function in question(aerobic citrate transport) that is performed by a system of multiple interacting parts.
C) If you remove any of these parts the system stops functioning.
D) The system is therefore irreducibly complex.
E) The system evolved by mutational events.

I’m happy to agree that I think this experiment will continue showing fitness increases because genes with functions that were adaptive in the ancestral environment of the mammalian gut are gradually purged by selection as they constitute a slight metabolic cost to carry around now that most of the challenges faced in nature are removed.

That said, I don’t think this is actually relevant to the subject matter of irreducible complexity (or the cit+ function specifically).

In a way I think this is a sort distraction that doesn’t make sense. To see what I mean, just dial up the knobs on both prongs of our respective arguments (more complexity in the system, but loss of much more ancestrally functional genome).

Suppose this experiment had run for a million years and during this span of time the total genome size had been reduced by 50% so we had ended up with a bacterium with a roughly 2.25 mb genome. Half as many total genes, half as much genome size is the net result after a million years.

But during this interval of time an irreducibly complex system consisting of 15 different parts (instead of 2 or 3) had evolved, with each of these parts deriving from duplications of genes that performed other functions in the genome. So here the complexity of this particular system had increased further, while the overall genome size had decreased by the loss of many other unnecessary genes. Could I still not then say that evolution can produce an irreducibly complex function, because it has occurred simultaneously with the loss of many other functions?

Where does the line cross for you? If a 4 component system evolves concomitant with a 1% loss in genome size, is that too much? What if it’s only a 0.5% loss of genome size and a 5 component IC system? Why must genome size expand or stay constant simultaneously with elaboration of the system in question? That was clearly not part of the original definition of irreducibly complexity, so why are you invoking it here as a sort of “packaged deal” argument against it?


Yes, it depends. Heh. I wish he would just come and hash this out on the forum, because to me, it seems like this all boils down to his understanding of the probabilities and mutation rates involved, with a heaping tablespoon of denying the principle of inference. IC just appears, to be his pet hypothesis that, like tissue paper, that must inevitably give way to the what his actual argument is.


But here’s the thing: you’re basically making the same argument as Behe: that there are systems SO complex that OF COURSE they have to be beyond some evolvability threshold.

Okay…and the criteria for that are…? Not IC, since we can find any number of such systems with directly observed evolutionary histories, many of them more complicated than Cit+.

The problem is that if you’re going to make the affirmative claim that some stuff can evolve but other stuff can’t, you better have a clear answer to the question “how can I tell the difference?”

“It’s really complicated” isn’t going to cut it. “Irreducible complexity” doesn’t do it because you get false positives. “Irreducible complexity but like really complex, like a flagellum” is just “we-haven’t-literally-watched-it-evolve whack-a-mole”.

So if you want to say it’s okay for cit+ to evolve, but not a flagellum, then we have a problem, because IC catches both. So you need a different, objective way of telling them apart in this context, and empirical justification for that standard.

(I’ll also note that blood clotting is one of the examples for which a literal stack of research was provided for Behe on the stand at Dover, so to say that is an unevolvable system has a whiff of argument from incredulity to it.)