Casey Luskin's self-flagellation
A few weeks back, I posted a rebuttal to the Casey Luskin/Michael Behe interview, which itself rebutted the Pallen/Matzke Nature Reviews Microbiology paper on flagellum evolution.
As you may recall, my key beefs were that:
(a) Numerous IDers have claimed that the cooption explanation for the origin of the flagellum doesn’t work because, they said, 30 of the 40 flagellum proteins don’t have homologs (Luskin had said 2/3) and thus were being coopted from nowhere. This claim was disproven in the NRM article, but the clear errors of Scott Minnich, Stephen Meyer, Thomas Woodward, William Dembski, Hugh Ross, and Casey Luskin himself were not admited.
(b) Behe wrote in Darwin’s Black Box that the flagellum had 40 required parts (a widely repeated claim); the NRM article showed this was wrong, and yet neither Behe nor Luskin conceded the error.
From this I reiterated my original conclusion, which is that the IDists don’t even know the facts on their flagship “designed” system, and furthermore mostly just copy each other’s talking points, they don’t correct each other, and they have therefore been misinforming the public for years.
Well, Casey Luskin posted a response in a comment to the interview. In the comments to my interview rebuttal post, I pointed out the basic reasons his response was ludricrous. But I guess Luskin didn’t see that, because he just posted his final response in the interview page, indignantly claiming that we didn’t acknowledge his concession:
about 2 hours ago, Casey Luskin said:
I find it incredibly telling that Nick and many other Pandas Thumb people have accused me of all kinds of wrongdoing at /archives/2006/10/luskin-intervie.html but I can find no one, including Nick, have noted that I retracted any claims I made previously about the number of unique flagellar parts. That says a lot about them: even when I retract a claim Nick asked me to, he remains silent on the thread, and that’s my last post on this thread.
Well, let’s have a look at Luskin’s “concession”:
Firstly, I’d like to note that I have now retracted any claims I personally made about the number of “unique” flagellar proteins. The claim from Jason about my claims came from footnote 14 from an article (the “Lugnuts article”) I wrote regarding the Kitzmiller ruling and the flagellum. Having seen the Pallen & Matzke article, I now realize that different authorities are making different claims. But this claim only came from a small footnote and it really isn’t crucial to my argument: in my opinion, you can make all the homology arguments you want, as far as my essay is concerned, even there were clear homologies determined for every flagellar gene, this would only solve the “availability problem” described in my article. So determining the number of unique proteins in the flagellum was not even a necessary part of my argument.
So far, this is somewhat decent – Luskin made a mistake, admits it, and then says why it wasn’t important (obviously I think it is important, but let’s not get sidetracked).
But what did Luskin write in his revised footnote? It appears that he takes it all back:
With regards to the flagellum, there is dispute as to whether many flagellar genes can even be said to be homologous to genes outside of the flagellum. This dispute is witnessed in the conclusion of pro-ID biologist Mike Gene who writes when critiquing Nick Matzke’s model attempting to evolve a flagellum: “The various dissimilarities (some very profound) listed above, along with the weakness of the criteria for inferring homology, is only rendered more problematic by the seemingly arbitrary nature of the chosen matches.” Matzke’s model may be found at http://www.talkdesign.org/faqs/flagellum.html Mike Gene’s critique may be found at http://www.idthink.net/biot/flag6/index.html Obviously there are different types of flagella, but Matzke admits that least 1/4 of flagellar proteins have no known homology, and also acknowledges that “the flagellar research community has scarcely begun to consider how these systems have evolved.” See Mark J. Pallen and Nicholas J. Matzke, “From The Origin of Species to the origin of bacterial flagella,” Nature Reviews Microbiology, AOP, published online 5 September 2006; doi:10.1038/nrmicro1493
Unfortunately for Luskin, the Mike Gene quote he uses is critiquing a specific sub-hypothesis I made in 2003, the hypothesis that core T3SS proteins were homologous to ATP synthase proteins. This concerns only one of the 20+ homologies listed in the NRM table. And, doubly unfortunately for Luskin, it turns out that for the one homology in the NRM table that Mike Gene actually did criticize – the hypothesis that flagellar protein FliH was homologous to ATP synthase protein Fo-b – I have been proven right in the peer-reviewed literature.* I’m not making this up – I am cited and everything. See the update to the BFE for a summary; someday I will blog this part of the story in proper detail, but it takes a lot of background to explain well.
Yet another unfortunate fact for Luskin (these are quite common in this post) is that, while he mentions that “Matzke admits that least 1/4 of flagellar proteins have no known homology”, he fails to mention to his readers that all but 2 of those proteins are not even universally required in functioning bacterial flagella. They are reducible parts, not irreducible parts (and this is probably why their homologs are hard to find; if the parts aren’t even evolutionarily conserved within flagella, the probabilities are low they will be conserved for even more distantly-related proteins outside of flagella). Some of these “parts” don’t even give a detectable mutant phenotype when they are experimentally knocked out of the genome.
OK, so we have established that Luskin was clueless in citing Mike Gene against the Pallen/Matzke homology table, and he was clueless in making something of the 1/4 figure, when the truly important figure is that only 2/42 proteins (5%) are both required and unique on current data.
Luskin fails to admit the error of Minnich/Meyer/Dembski/Woodward/Ross
Even with all of the above, it might be worth giving Luskin a little credit for admitting that his 2/3 figure was an error, except for the fact that Luskin takes it all back in a desperate attempt to preserve the credibility of his authority figures:
Pallen and Matzke may be right, or maybe Mike Gene, Scott Minnich, and Stephen Meyer are right. It all depends on how you define “unique” and “homologous” and given that Matzke’s prior homology arguments have been strongly challenged by ID proponents (see the Mike Gene reference), I’m preferring to wait and see, but I personally now take no position on the matter.
Again we see the uncomprehending citation of Mike Gene, but your main reaction here should be, What? So Luskin admits that his claim that 2/3 of flagellum proteins were unique (lacked homologs) was wrong, but he still thinks that Minnich/Meyer/Dembski/Woodward/Ross’s claim that 3/4 of flagellum proteins were unique might be right?
It is crystal clear that Luskin was talking about the same issue as everyone else. They all said, basically, “Sure, 10 of the 40 flagellar proteins have homologs in the T3SS, but where do evolutionists gets the other 30 parts? Those are unique! You’re coopting them from nowhere! Gotcha!”
There is really nothing else to do here except to stammer. Forthwith, I stammer: “But- you said–and then you said- Oh, I give up.”
But it gets better
Amazingly, this isn’t the end of the story. Luskin’s next move is to try and defend Behe’s claim from Darwin’s Black Box that the flagellum “requires about forty other proteins for function” (in addition to several he discusses). In the NRM paper, we showed that only 20 proteins are actually universally required in all functional flagella.
Luskin tries to bump this number up to save Behe:
According to Behe, there are over 40 necessary flagellar parts. Matzke and Pallen identify 20 parts which are “indispensable”. So we’re almost halfway there.
Halfway! How hopeful!
Actually, if you think about it, you basically can’t bump this number up, because the list of 20 indispensable proteins was created by taking the standard parts list of the Salmonella flagellum and then removing proteins that are observed to be unecessary by experiment or in natural systems. The number can go down by new observations, but it can only go up if a currently-published observation of inessentiality turns out to be plain wrong, which is unlikely.**
Luskin continues:
Macnab (1987) studied E. Coli and S. typhimurium and found between 33 and 34 “necessary” genes. From what I can tell, there isn’t a lot of crossover between the two lists of required genes, but I acknowledge that the terminology may have changed and so it’s hard for me to tell. I’m not sure if gene names have changed (I’m happily open to clarification here), but combining the two lists I see over 50 unique flagellar parts:
…and then Luskin lists the 20 required proteins from the NRM paper, 33 apparently different E. coli proteins from a 1987 paper by Macnab, and another 33 proteins from Salmonella in Macnab 1987.
Luskin then adds 20+33 and gets…over 50! Behe is vindicated!
combining the two lists I see over 50 unique flagellar parts
Combining Matzke / Pallen’s, and one of MacNab’s lists minus the clear crossovers (motA and motB) leads to over 50 required flagellar parts.
There is only one problem here. Anyone who really knows their flagella knows that the protein naming scheme for flagella changed in the late 1980s.
The problem back then was that different labs were using different naming schemes, and this meant that what was basically the same protein in 5 different bacterial species could have 5 different names. Sometimes the same name could refer to two entirely dissimilar proteins. The whole thing was a mess, and so in 1988 the leading flagellum researchers got together and started from scratch so that the same homologous protein in different species would get the same name. Luskin is sort of vaguely aware that this kind of thing might be a possibility and might mess up his calculation, but for some reason he didn’t bother to do the simple thing and get on PubMed to double-check.
In fact, in about 30 seconds on the web I was able to find the paper that unified the flagellum terminology: Iino et al. (1988). “New unified nomenclature for the flagellar genes of Escherichia coli and Salmonella typhimurium.” Microbiol Rev. 52(4): 533–535. It is free online at PubMed.
If you look at page 2, you can see the table that converts between the old scheme and the new…and you will see that all of the new names proteins are listed in the NRM table.
To sum up: Luskin made A Big Mistake.
Yet more
Luskin tries yet another argument:
Finally, I note that in the Kitzmiller trial, Scott Minnich testified about his own knockout experiments which determined that 35 flagellar parts were required (see here for a discussion).
The link there refers to Luskin’s discussion of Minnich’s testimony, where Minnich said “We put, knock out one part, put a good copy of the gene back in, and they can swim. By definition the system is irreducibly complex. We’ve done that with all 35 components of the flagellum, and we get the same effect.” I actually think Minnich was referring to the royal we here, as in “we scientists”, not his own lab, since to my knowledge he hasn’t personally done all of those knockouts (which were all originally done in the 1970s and 1980s in diverse labs).
I realize that Nick might respond that some flagella lack some of the parts found in Macnab’s list so the number of required parts is actually more like his estimate based upon his chosen samples for study. But the lack of some of these parts in other flagella does not imply that the flagella studied by Macnab and Minnich are not themselves irreducibly complex and require the parts they require. You have to take a system as it comes. If you can explain how to evolve step-by-step-mutation-wise some of the simpler flagella Nick discusses into the ones Macnab or Minnich studied, then you could justify that those parts aren’t necessary. But I haven’t seen that explanation–I’ve seen a coherent, complete system with a few dozen required parts in both cases.
OK, let’s stop and think about what Luskin is admitting here:
- The flagellum couldn’t evolve because all of its required parts had to come together at once in order to function.
- Some flagella function without some of those “required” parts.
Unfortunately for Luskin, the only inference we can draw here is:
- The claim that those missing parts were absolutely required is wrong, and therefore there is a way to have a “reduced” flagellum that still functions.
It gets even worse for ID when you learn that some flagellum parts are (a) missing in some species, (b) helpful, but not absolutely required in other species, and (c} required in other species. This exactly matches a longstanding pathway for the evolution of systems with multiple-required parts: first a part is selected because it is helpful, then co-evolution between the helper part and the helpee system makes the linkage tighter and tighter until they are codependent.
Finally, if Luskin is serious in his claim that all flagella are irreducible, but some flagellar are more irreducible than others, then he is basically admitting that he is proposing thousands or millions of instances of miraculous special creation of flagella, all in such a way that phylogenetic trees of conserved flagellar proteins just happen to match each other*** and match the phylogenetic trees of the bacteria in question. That is quite a coincidence if all of these different flagella were poofed into existence.
Summing up
Now, anyone can make mistakes, even people reasonably well-educated in the relevant area. But this wasn’t just any mistake. I think this litany of simple, purely amateurish errors gives us a real window into the heart of the ID movement and its promoters. Basically, Luskin – undoubtedly one of the most committed ID promoters anywhere – doesn’t know what he is talking about when it comes to the “icon of ID”, the bacterial flagellum. And yet, he talks about it anyway, in a quite unhesistating way. He has blogged about the flagellum, he has talked to the press about the flagellum at great length, he has written essays making use of the flagellum. But basically, he just unblinkingly repeats what his “trusted authorities” have said – the small group of academics on his side. He hasn’t checked to see if they actually know what they are talking about. He just assumes they do, despite the fact that pretty much the whole scientific community disagrees with them, which would induce some skepticism in most people. And so, when the errors of this small group are demonstrated, as in the NRM paper, he just assumes they are right, and scrambles around to try and patch up the holes, because, after all, they must know what they are talking about. This may be a PR strategy or a good way to deal with cognitive dissonance, but it’s not science.
But there is still a chance to do the right thing. Luskin could still take the brave step of admitting that Behe was wrong about 40+ required parts, and that Minnich & company were wrong about 30 unique proteins. The facts that indicate that this is the truth are simple. I know it is embarrassing to admit an error; but as I said before, in science, there is no other way forward. In the end, the shame is much greater if errors are stubbornly defended.
Notes
* There are other flagellum-ATPase homologies that I proposed but which I stated were more speculative. Mike Gene’s criticisms have more merit here; see the Update for which homologies I have surrendured and which I still propose. In any event, none of the speculative proposed flagellum-ATPase homologies are in the NRM table, only FliI/F1-alpha-beta (long accepted) and FliH/Fo-b (demonstrated by Pallen et al.), so Luskin’s citation of Mike Gene is still irrelevant.
** As I have mentioned before, some proteins were determined to be inessential because they are not found in genome searches. It is conceivable that a few of these will turn out to be universally present (FliH perhaps), but not all that likely, and even if this did happen, the protein in question would have to have very low sequence conservation and thus “low specifity/information” according to IDists.
*** In a statistically significant way, if not exactly. I believe both Nguyen et al. 2000 and Gophna et al. 2003 reached this conclusion although I don’t have the papers handy at the moment.