I had not intended to do another post on this topic so soon after the last one. But I have just read
an astonishingly bad post over at Uncommon Descent that discusses this issue, and I cannot resist responding.
The post is called, “Where Do We Get the Probabilities?” It was written by Winston Ewert, and it opens like this:
What is the probability of a structure like the bacterial flagellum evolving under Darwinian processes? This is the question on which the entire debate over Darwinian evolution turns. If the bacterial flagellum’s evolution is absurdly improbable, than Darwinism is false. On the other hand, if the flagellum is reasonably probable than Darwinism looks like a perfectly plausible explanation for life.
This is poorly phrased, for reasons I explained in my earlier post. An absurdly low probability by itself would tell us nothing one way or the other about the plausibility of Darwinian evolution. As has often been noted, improbable things happen all the time. It could well be that any particular outcome of billions of years of evolution occurs with low probability. To go from a low probability of flagellar evolution to the conclusion that it is effectively impossible requires an additional argument, to show that this is not the sort of improbability that can be dismissed. The ID folks have been entirely unsuccessful in this regard. Ewert, of course, claims otherwise:
Dembski’s development of specified complexity depends on having established that the probability of structures like the bacterial flagellum is absurdly low under Darwinian mechanisms. Specified complexity provides the justification for rejecting Darwinian evolution on the basis of the absurdly low probability. It does nothing to help establish the low probability. Anyone arguing the Darwinian evolution has a low probability of success because of CSI has put the cart before the horse. You have to show that the probability of the bacterial flagellum is low before applying CSI to show that Darwinism is a bad explanation.
Sadly, there are two big problems with this. The first is that you cannot hope to do a relevant probability calculation without first clearly specifying the event in question. It is not that you first carry out the calculation, and then move on to the next step of showing that the structure is “specified.” The specification and the calculation have to go hand in hand, as otherwise you have no idea what you’re calculating the probability of. Picking out the relevant event, though, is no small problem, as I pointed out in my earlier post.
The even bigger problem, though, is that there is no way to give any meaning to Dembski’s notion of “specification” that is useful for drawing conclusions about the plausibility of evolution. That is why, in his writing, he nearly always only applies it to tinker toy examples and then argues by vague analogy to draw biological conclusions. The one place where he actually tried to do a calculation for the flagellum (in his book No Free Lunch), the whole effort was so riddled with errors it could hardly be taken seriously.
To see the problem, let’s use a standard example. The specific sequence of H’s and T’s that arise in five hundred flips of a coin is exceedingly unlikely, but that is not necessarily suspicious since something had to happen. But five hundred heads would be suspicious, roughly because it conforms to an easily describable pattern. ID folks try to extrapolate this sort of reasoning to structures like the flagellum. It’s not only improbable, they say, but it’s “specified” by virtue of its functionality.
But this argument doesn’t work at all, and for a simple reason. When we reason about what is, and is not, likely to occur when flipping a coin, we have a good grasp on the relevant probability space. We generally take it for granted that we are talking about a fair coin tossed in a fair way, meaning we are justified in equipping the space with a uniform probability distribution. That is precisely what we can not do when we are discussing genetic evolution. In this case we have no way of determining the correct distribution to apply to the space.
The prolonged action of natural selection ensures that most gene sequences have a probability close to zero of ever occurring (or persisting for long if they do occur) while the small percentage of functional sequences have a relatively high probability. That is why simplistic analogies with coin tossing (or any of the other stock examples ID folks use) do not work. We know enough about the relevant space so that five hundred heads immediately jumps out at us as requiring a special explanation. We have no such justification for arguing comparably with respect to the flagellum.
So what is the probability of a bacterial flagellum under Darwinian mechanisms? Obviously, we can’t expect to know the exact probability, but can we at least determine whether or not its absurdly improbable? That’s the question on which the whole debate rests. It seems that any arguments over Darwinism should be focused on arguments about this probability. It is the key to the whole discussion.
This is just nonsense. The inability to define the relevant probability space in assessing the outcomes of prolonged evolution scuppers the whole attempt to apply probability in the way ID folks prefer. We have no good way even of estimating the probability, and we would have no helpful way of interpreting the probability even if we could somehow estimate it.
These points are obvious to anyone who understands probability theory, which is why actual biologists do not try to carry out these sorts of calculations. To make probability the question on which the entire debate revolves betrays a complete incomprehension of the subject. Probability theory is very useful for understanding many aspects of evolution, but it is not useful for drawing grand conclusions about what is possible (or plausible) and what is not.
Intelligent design proponents have long offered a number of arguments attempting to show that Darwinian evolution accords a low probability to structures such as the bacterial flagellum. Darwin’s Black Box argues that irreducible complexity is highly improbable to evolve. The Edge of Evolution argues that non-trivial constructive mutations are too improbable for Darwinian evolution. Doug Axe’s protein work argues that protein evolution is too improbable. The fact is, almost every work by intelligent design proponents has been directed towards arguing that Darwinian evolution is too improbable to work. There is no mystery about why we intelligent design proponents think that evolution is improbable.
Intelligent design critics are going to dispute all of these arguments I mention. That’s fine. But dispute those arguments. Don’t act as though we’ve never given explanations for why we think that Darwinism is an improbable account of the complexity of life. Don’t attack specified complexity for not showing that Darwinism is improbable. That was never the intent of specified complexity. It is the intent of a host of other arguments put forward by intelligent design proponents.
This is a joke, surely. Who is Ewert talking about when he says ID’s critics need to “dispute those arguments”? Who is acting like ID folks have never given explanations for what they believe? We have replied at length to all of the points Ewert raises. We do sometimes express incredulity that the ID folks could possibly believe what they are saying, but that is a different matter.
Behe’s book Darwin’s Black Box was one of the most thoroughly reviewed scientific books in recent history, and it was not difficult for scientists to explain what was wrong with his argument. They pointed out that “irreducible complexity” as Behe defined it was a silly notion that was entirely unhelpful in assessing the plausibility of evolution. At a theoretical level it is trivial to devise scenarios through which such systems could evolve gradually, and actual research scientists have been very successful at showing how these scenarios have played out in specific cases.
Likewise for Behe’s and Axe’s bloviations about the nature of sequence space. In addition to the conceptual errors especially in Behe’s book The Edge of Evolution, their whole approach plainly cannot justify the bold conclusions they are trying to draw. You’re never going to be able to draw grand conclusions about the plausibility of protein evolution from a few small experiments undertaken on modern organisms. Moreover, actual research in protein evolution does not at all suggest that the relevant space is structured in the way they need to make their case. We know, for example, that most mutations are neutral, which already suggests that individual proteins are surrounded by groups of other functional proteins.
Actual researchers see not just that ID is putting forward bad arguments, but that its whole approach to the subject is just a conceptual dead end. So the issue is not that ID’s critics have simply ignored the arguments ID folks have put forward. It is that they have replied at length, but people like Ewert don’t like what they have said. So they heckle from the sidelines, demand a standard of proof under which anything less than a videotape of natural history is casually dismissed, and then just repeat ad nauseum the same points they have always made. Helpful when preaching to the choir, but not so helpful for saying anything interesting about natural history.