 I recently attended the Discovery Institute event at Southern Methodist University in Dallas, Texas. It was titled Four Nails in Darwin's Coffin. What follows is the audio only of the question I asked of the panel and the answers they gave. If you're hoping to see blood or a shouting match or a witty retort, you're out of luck. I actually had some questions prepared specifically about the evolution of Wales, but since they'd addressed that point already, I ad-libbed a question that had really been bothering me. It concerns the limitation they placed on evolutionary change in development. Before I play the clip, let me set the scene. For those of you unfamiliar with SMU's reputation, it's affiliated with the United Methodist Church, but not overtly religious. It's a very prestigious business school, very conservative, and the students have a reputation for coming from wealthy families. The event was planned to start at 7 p.m., but they were 45 minutes late. We watched a one-hour movie of DI fellows saying incredibly stupid things about the Cambrian radiation. Nothing new was presented. The most humorous comment was John Wells saying that we have no idea how long the explosion took. It might have happened overnight. Such comments are obvious pandering to the young Earth creationists who fund the Discovery Institute. No mention was made of the fact that the Cambrian radiation did not give rise to a single vertebrate vascular plant or insect. So it's almost 9 p.m. when they start the sessions, which lasted for about an hour. Stephen Myers was the emcee, Jonathan Wells was there, along with Doug Axe from the Biologic Institute, a wholly-owned subsidiary of DI, Paul Nelson and Richard Sternberg. Myers is a philosopher. Wells has PhDs in biology and theology, but has never published a research paper. Paul Nelson is a philosopher of biology. Sternberg is a theoretical biologist, which leaves only Doug Axe as an actual research biologist. Doug has nine papers in Medline, and Sternberg has exactly one. So the total number of peer-reviewed science papers published by the entire panel is 10. That's about half of what Thunderfoot has published, and a little less than I managed before I went to industry. I'm not talking to equals, in other words. The impression I got from the crowd is that many of them were part of youth groups or campus groups of local churches. They moved in large crowds and knew the sponsor well enough to call them by first name. Here's the clip. I'll pause it at several points to explain the questions and responses. So this relates, I think, to several of the areas you talked about today. But one concept I didn't hear discuss was gene duplication. Specifically, how that relates to mutations resulting in lethality early in development. If there are multiple copies of a gene and they're less constrained, does that change your assessment of the possibility of mutations during development? Or the probabilities. Good question. The topic of gene duplication relates to the model of the resumption model that I presented. Certainly, if you have extra copies of genes within the genotype, within the genome, in theory, this provides the substrate for all kinds of innovative changes. And in theory, you should be able to relate gene duplications to innovations that have occurred either within laboratory populations or along an evolutionary tree. The problem is that whereas you find a number of gene duplicates that are around, many of them are sub-functional. Those that we know in plants, those that we know in animals. So it's still another question of whether or not they do provide that substratum. But there's another aspect to it, and it relates to a figure that I showed. After a rather small number of changes, you've got this extra gene out there. Presumably, it's not being... changes aren't being said by natural selection. After four, five, six substitutions, it drifts into non-functionality. More often than not, it becomes a pseudo-gene. So it remains a tantalizing hypothesis that's constantly presented as a... covers, if you will, a multitude of theoretical reasons. But it's yet to be demonstrated that you really can tie an innovation to it. I want to pause here because I want to summarize what he just said. Gene duplications have never been shown to generate new functions, and pseudo-genes are irrelevant for evolution. Remember that he just said that. He's also said that we aren't able to place gene duplications on a phylogenetic tree, and I've been showing you papers published just in the last three months that put the light of that statement. It's well known that some plants are polyploid, which means they have more than the standard number of chromosomes. These duplications change the way the plant grows and what soil it can grow in. So that's a gain of function from gene duplication in a laboratory model. For example, the human tumor necrosis factor receptor is part of a superfamily of over 34 genes in our genome, each with different function, but only relatively minor differences in protein sequence. This is a recurring theme in biology. A useful gene is co-opted for many different functions. The way this happens is that the gene is duplicated, drifts, and is activated with a new purpose. Watch in the next clip as Richard Sternberg, Discovery Institute fellow, confirms that this is the case in contrast to where he just said that gene duplication is not a source of new functions. Well, specifically on what you just said, pseudogenes can also be reactivated. That's true. Well, pseudogenes, it depends. Often when they're reactivated, I do not know of an instance. I mean, there may be one. The pseudogenes that have supposedly adopted functionality are often performing a role in the cell in development that is different from the putative ancestral sequence. So if you have, like, a protein-coding gene, and you've got an extra copy, and it became a pseudogene, then the new role is often something else. It does not go back to what it was doing. But, nevertheless, it's very hard to pin down a specific innovation where you say, aha, at this point, give you an example of bony fishes. You have anywhere from 22,000 plus species of bony fishes. A compelling idea is that what happened is the genome was duplicated. I mean, the whole genome was duplicated. They became effectively tetraploids for a while, or at least part of their genome became, instead of two copies of each gene, you have four copies, and that you can explain, you know, puffer fishes, sea horses, flounders, etc., from this. But the problem is that when you actually look at the many innovations you have, it's very difficult to tie them to any of these extra copies that are lying around. I'm going to pause again and spare Rick his misery. Steven Meyer is about to step in to stop the travesty. Did you follow where we've gone from? We started at gene duplications don't produce innovation. We passed through pseudo genes producing new functions, and now we're at the point where Rick admits he's observed developmental gene duplication in the evolutionary history of his bony fish. But it's just really hard to tie the gene duplication to the innovation, except in flounder and sea horses. Feel free to go back and see where he realizes his error and starts correcting. No wonder Steven is going to stop this and pass the ball to Doug Axe. Oh, what fun I was having at this point. Rick, could I get Doug to speak to that as well in just a sec? Because I'd like to point out that whereas the gene duplication, operating as Rick is in his critique, still within the neo-Darwinian framework, accepting all the assumptions that neo-Darwinism asked you to assume, you could entertain the gene duplication hypothesis as a way of solving some of these problems. But when you get to the kind of thing that Jonathan and Paul are talking about, the problem of developing innovations at the body plan level, simply having extra copies of DNA is going to help you. But even at the level that Doug is addressing, I don't think it solves the problem, because you still have this huge combinatorial space to search. I just wondered if you could speak to that, Doug? Can I restate the question for Dr. Axe? Is the combinatorial space really all possible proteins that could be formed, or is it the current complement of proteins that exist? In other words, do cells try to invent new proteins by randomly jamming the amino acids together, or do they modify existing ones? So you're talking about new protein structures. New protein functions, I thought was the issue. Well, the fact is there are 2,000 fundamentally different protein structures, so a Darwinist has to explain how you get to different structures. So often new functions seem to require new structures, because that's what we see. When we see new structures, they're often performing new functions. There are 55,000 proteins in the Protein Data Bank. There are only 80 fold families known, usually called domains. Three of those folds, SH2, TIM, and B3, B4 account for a high percent of all functional fold domains. We call these the super families, or the super folds. In other words, most proteins reuse what already works. They don't go searching for new ways to fold proteins very often. So Doug made a terrible analogy of optimal fitness searches in protein domains to Google queries. I'd like to modify it. Suppose you wanted to generate functional web pages that would draw visitors. Would you have to invent fantastical animals and clever stories? Or would you just cut and paste porn, fail jokes and pictures of cute cats into a random generator, and spit out 90% of the internet? Here on YouTube, you just need a steady stream of makeup advice, licensed music videos, and cute things sneezing. Proteins are more like the internet than a well-written novel. The same motifs are reused over and over, and only sometimes do we see the emergence of some new successful meme, which is quickly seized upon and adapted into thousands of other crossover innovations. An example is the ABC transporter, which is shorter for ATP binding cassette transporter. It's a modular unit. Recombination can plop down into any gene and turn it into something that moves molecules in and out of the cell against a gradient. It's part of everything, from antibiotic resistance in bacteria to human cystic fibrosis. We call it a cassette because it's more or less self-contained information about how to do something. We know it's ancient because it's highly conserved. I'll deal with where domains come from in just a minute after Doug has a chance to be interrupted by Steve Meyer yet again. But a space that has to be searched isn't just 2,000 possibilities. Those are the things that we see out there doing the job. How that happens, nobody has a clue, but presumably it would have to involve extensive cutting and pasting and mutating. And effectively, if you look at the sequence difference between a novel structure and the closest sequence you can find, the sequences are so wildly different that you're effectively, as I tried to give in my simple example, having to search the whole space. You basically have to be able to pull things out that are so remote in the space that it makes no difference what you started with. So effectively, in that case, having a second copy of the same gene isn't going to help you solve the problem? Let me clarify that. When I say search the whole space, I do not mean you have to have tried all possibilities. What I mean is you have to wander so far in the space that the starting point is irrelevant. New folds and therefore new domains arise very rarely, but they aren't impossible, as Doug thinks. That's because to get from A to B, as I show in the diagram on the right, you have to have a version of A that is pretty bad at its job. That's only possible if there's a duplicate copy that's good. Or if A isn't a critical fold, then the bad form of A finds that it can do the job of fold B, but only very poorly. But that poor version that can do both A and B has a selective advantage. Further evolution will result in stepwise movements up that shallow hill to the fitness peak for fold B. What Doug has done is to imagine a fitness landscape where there can be no overlap between A and B, no gentle slope to climb between them. The good news, at least for me, is that several papers in the last few years have shown that the model on the right is more accurate. My message to Doug is that whenever you and the rest of the scientific world disagree, sometimes it's healthy to question your own conclusions rather than everyone else's. So I don't mean to monopolize time, but I know that there are literally hundreds of thousands of different sequences that all do the same function in different organisms. Right, for example, a jellyfish, adenine, whatever, reductase. Homologous. Homologous, right, but doing the same function. So there must be hundreds of thousands of possible functional proteins accomplishing the same purpose in different organisms. Is that true? Sure, and you could go a lot higher than that. But if the JNB 2004 paper, which I can give you the reference for, carefully examined the prevalence of those functional variants, and there's trillions of them, but the thing is you have to divide by 20 raised to the 150th power, which is stalking the enlarge. So it doesn't matter that you have trillions of functional sequences, the denominator is so large it becomes impossible. I'd say of the three people I interacted with, Doug gets a small measure of grudging respect. Sternberg was a buffoon, and Meyer was a slimy weasel who could only recite mantras. To all of the panelists, I want to say thank you for being such good sports and providing us with such great entertainment. Thanks for watching.