Ev Ever Again -- Eying an Evolutionary Simulation
A writer at The Skeptical Zone, Patrick, recently contributed a post on the computer simulation ev. He takes aim at William Dembski, Robert Marks, and the Evolutionary Informatics Lab's analysis of that simulation. However, the events he discusses actually show a history of Darwinists repeatedly misunderstanding or misrepresenting arguments for intelligent design.
Patrick fundamentally mistakes the claim we are making about ev (and evolutionary simulations in general). Regarding a response to Schneider from the Evolutionary Informatics Lab, he says:
He admits again that evolution does work in certain environments.
Patrick treats this as an admission that undermines our argument, but it is what we have been saying over and over again. We have argued that evolution only works in certain environments. In order for Darwinian evolution to work, you must have an appropriate fitness landscape. This is what Dembski said in "America's Obsession with Design," the first time I'm aware of that he wrote about ev. It is also what Dembski wrote in No Free Lunch. It is also what we said in "A Vivisection of the ev Computer Organism: Identifying Sources of Active Information,"as well as our response to Schneider's criticisms.
We are not claiming and have not claimed that evolution simply doesn't work in any environment. We have claimed that it only works in certain environments. This is the entire point behind the concept of active information and conservation of information: if you have an environment where evolution is successful it must be due to the configuration of the environment.
The real world is one of [the environments where evolution does work].
This is the point that we would dispute. Our contention is not that evolution cannot work in any environment, but that the real world is not one of the environments where evolution works. The question is not whether evolution can navigate an ideal fitness landscape, but whether the real world's fitness landscape is sufficiently ideal for evolution to navigate it. It is not a question of whether or not evolution can follow the signals natural selection provides, but whether natural selection provides the right signals.
Where does that leave us with ev? Well, ev can tell us that given its particular environment, evolution works. But that is not under dispute. What can ev tell us about the real world environment, and how suitable it is for evolution? Does ev suggest that reality would be amenable to evolution, or is it simply an irrelevant model that has little to do with the real world?
We have argued that ev is, at best, irrelevant to the real world. In the real world, natural selection favors organisms which perform well in the current environment. On the other hand, ev'smodel of natural selection measures the difference between an organism and an ideal target. As Richard Dawkins said about his own target-based model, "life isn't like that." You cannot evaluate fitness as distance to a target, and present that as a model of natural selection.
That's not a target. It provides no details about what a solution would look like or how to reduce the distance measured, it simply indicates how far away a genome is from being a solution. In fact, it does less than that because it doesn't provide any information about the difference between an existing recognizer and an ideal recognizer.
It is unclear what Patrick thinks the word "target" means. Nothing stated in the paragraph above has anything to do with whether or not a target exists. In fact, he talks about the target, calling it the solution. Patrick seems to think that it's not a target unless detailed information is provided about it to the simulation. But that is simply not what target means.
Even so, providing the distance to the target is detailed information. Distance to a solution is information about what a solution would look like. A distance that decreases when the genome moves towards a solution is information about how to reduce the distance. Furthermore, distance between the ideal recognizer and the existing recognizer is clearly information about the difference. Every sort of information that Patrick mentions is in fact provided by the distance to the target.
Patrick claims that Dembski made an incorrect claim about ev, which, he says, was thoroughly refuted by Schneider, and which Dembski has never admitted.
In 2001, William Dembski claimed that ev does not demonstrate an information increase and further claimed that Schneider "smuggled in" information via his rule for handling ties in fitness.
What Schneider shows is that without this special rule, ev still succeeds, although it does take somewhat longer. Thus he and Patrick claim that they've demonstrated that this special rule was not a source of information.
However, they are simply incorrect. Dembski pointed to multiple sources of information. He wrote in "America's Obsession with Design":
Schneider's choice of fitness function is the most obvious place where he smuggles in complex specified information. But there are others.
Dembski goes on to discuss the special rule as an additional example of a source of information. Dembski identified multiple source of information, and as such the removal of only one of them would not necessarily be expected to bring evolution to a halt. It would be expected to hinder or slow evolution, and this is exactly what is observed.
Imagine coming across a dubious salesman with an allegedly magic swimming pool. He claims that the swimming pool fills itself with water without needing any external source of water. This seems unlikely, and upon investigation you point out that the swimming pool is filling with water because it is pouring rain and there is also a hose dispensing a trickle of water into it. In response, the salesperson pulls out the hose, and says, "Look! See? The water is still rising. The swimming pool is magic!"
By only removing the relatively minor source of water, the salesperson has misrepresented what you are saying. In the same way, by removing only the minor source of information and acting as though this undermined Dembski's argument, Schneider misrepresented what Dembski said. Dembski merely said it was a source of some information, and Schneider's experiment confirmed that. Dembski never implied that evolution would completely stop working without this additional source of information.
Patrick, following Schneider, objects that we did not measure the particular form of information that Schneider used in his paper.
They do not compute the information in the binding sites. So they didn't evaluate the relevant information (Rsequence) at all.
This is rather a case of the pot calling the kettle black. Schneider claimed to be addressing an objection against Darwinian theory:
The ev model quantitatively addresses the question of how life gains information, a valid issue recently raised by creationists.
But one of the sources that he references, Royal Truman's "The Problem of Information for the Theory of Evolution: Has Dawkins Really Solved It?," has this to say about Shannon information, the model of information Schneider used:
Shannon's theory of information, while useful in the context of telecommunications, does not seem to help anyone much in the evolution/creation debate.
If Schneider wished to address this question, he should have used a model of information as proposed by those asking the question. Schneider cannot simply adopt another model of information and pretend to have resolved the question.
Of course, Schneider's inappropriately switching information models does not give us the right to do so. However, we did not dispute Schneider's claims about Rsequence and Rfrequency. Rsequence is a measure of how similar the binding sites are to each other. When Schneider points out that Rsequence increases, this means the binding sites get more similar to each other over the course of evolution. We do not wish the dispute the fact that under ev's rules, the binding sites get more similar to each other, and thus calculating Rsequence remains irrelevant.
What we do dispute is Schneider's claim that he has answered the problem of information in evolution, and his assertion that his results are applicable to real-world biological situations. The Evolutionary Informatics Lab developed active information specifically to study these situations. Our paper was a case study in applying active information to this model. For the argument that we were making, active information was the relevant information measure.
Patrick continues, objecting to my paper, "Digital Irreducible Complexity," which discusses ev and irreducible complexity:
It appears that Ewert is the one with the misunderstanding here. If there is a destructive mutation in the genes that code for the recognizer, none of the binding sites will be recognized and, in the biological systems that ev models, the protein will not bind and the resulting capability will not be provided. It will "immediately and necessarily" cease to function. This makes the system irreducibly complex by Behe's definition.
Patrick observes, correctly, that destroying the recognizer will prevent any of the binding sites from being recognized, and thus the protein will immediately and necessary cease to function. Then he claims that this fits the definition of irreducible complexity. But that is incorrect, since he has to show that this is true for all of the parts, not just one of them. Later, he demonstrates that he is aware of this criterion, leaving it somewhat mysterious why he asserts that this makes the system irreducible complex.
Patrick does go on to discuss the other parts:
Binding sites are somewhat less brittle, simply because there are more of them. However, if there is a destructive mutation in one or more of the binding sites, the organism with that mutation will be less fit than others in the same population. In a real biological system, the function provided by the protein binding will be degraded at best and eliminated at worst. The organism will have effectively ceased to function.
Patrick begins by stating that the organism with a destructive mutation will be less fit than one without that mutation. This is true. He observes that in a real biological system, the function provided would be degraded at best. This is also true. But then he claims that the organism will have effectively ceased to function. Having functionality degraded is not the same thing as effectively ceasing to function. For functionality to be degraded means that some functionality remains, but it is not as good as the original. For it effectively to cease functioning means that the functionality is completely gone. In ev, destroying binding sites merely degrades functionality, it does not eliminate it. As such, it simply does not fit the definition of irreducible complexity.
Patrick accuses me of not understanding irreducible complexity, but he merely demonstrates his own confusion. He tries to show that ev evolves irreducible complexity by using two different definitions, both incorrect. Furthermore, Patrick did not even attempt to address the discussion of this point in my paper or other objections I raised against the claim that ev has evolved irreducible complexity.
Again and again, we have seen that our critics misunderstand and misrepresent our arguments. Dembski is represented as claiming that without a special rule, ev wouldn't work, but he only stated that it was one source of information amongst others. The Evolutionary Informatics Lab is represented as claiming that evolution cannot work in any environment, when we have repeatedly argued that it only works in some environments. Patrick has misconstrued the definition of irreducible complexity in two different ways. Not for the first time, our critics claim an easy and hollow victory against straw men.
Ev demonstrates evolution in an intelligently designed environment, not one that appropriately models nature. It does not evolve an irreducible complex system. The system it evolves fails the knockout test -- parts can be removed without the system effectively ceasing to function. The truth remains that critics like Patrick are unable or unwilling to engage our actual arguments. Our conclusion stands.
Image credit: © Kurhan / Dollar Photo Club.