Steve Meyer recently gave a lecture summarizing the arguments put forward in his book Signature in the Cell  to an audience of 1,400 (including me) at Biola University.  After Steve sat down, two of his critics, Steve Matheson and Arthur Hunt, were invited to put their questions to him.
Matheson and Hunt both referred to my work and to Meyer’s use of it, Matheson having since posted his points on his blog.  As is often the case when complex subjects are debated in front of an audience, things got a bit muddled. I stood up at one point with the intent of commenting but wasn’t able to get the attention of the moderator, so I’ll comment here instead.
The specific work to which Meyer, Matheson and Hunt referred  has added to the scientific case for functional protein sequences being extraordinarily rare within the whole space of possibilities. Matheson started off by arguing not that this deduction of extraordinary rarity is incorrect, but rather that it is irrelevant to the debate between Darwinism and Design. According to him, “What is relevant is whether the protein’s place in sequence space is linked through achievable steps to other points in sequence space” in a manner traditionally represented by Darwin’s branching tree.  His reasoning seems simple enough:
I pointed out that a standard evolutionary account of that tree, whether it’s a tree of species or a tree of people or a tree of proteins, makes no prediction about the rarity (or commonness) of function or adaptation within the space that the tree inhabits. In the case of proteins, the branches of the tree are particular proteins, and the proteins are linked to each other by common ancestry. Whether each branch represents a fantastically rare structure that has a function, or just represents one choice among zillions of [comparably functional] alternatives, is really not relevant to the question of how the protein’s structure came to be. 
But this is a naive understanding of prediction. Theories have both consequences and assumptions. If a theory is correct, then its consequences will prove true when tested, and so will the assumptions on which it is predicated. The claim that a theory is correct therefore amounts to a prediction both that its consequences will prove true and that its assumptions will prove true.
Tree-like relationships are what Darwinian evolution produces—they are the consequence of its operation. But if Darwinism is to work as a theory of origins, it must explain not just trees but trees with remarkable transformations of form and function scattered throughout their branches. For a century and a half this has been the major point in dispute—whether such remarkable transformations can possibly happen through small adaptive steps. If as a rule they can, then Darwinism works. If as a rule they can’t, then Darwinism flops.
Matheson recognizes the immediate assumption on which the Darwinian account of proteins rests—that the whole set of biological proteins must be “linked through achievable steps”—but he doesn’t seem to see what would have to be true in order for that to be true. Assumptions rarely travel alone.
Part of the difficulty is that the degree of rarity we’re talking about here is so far beyond our everyday experience that our intuitions tend to be unreliable. When we think of extraordinarily rare events, we think of winning the lottery or being struck by lightening, both of which are actually very common events on the scale relevant to protein origins.
Picture this instead. Suppose a secretive organization has a large network of computers, each secured with a unique 39-character password composed from the full 94-charater set of ASCII printable characters. Unless serious mistakes have been made, these passwords would be much uglier than any you or I normally use (and much more secure as a result). Try memorizing this:
Now, if someone were to tell you that these computers can be hacked by the thousands through a trial-and-error process of guessing passwords, you ought to doubt their claim instinctively. But you would need to do some math to become fully confident in your skepticism. Most importantly, you would want to know how many trials a successful hack is expected to require, on average. Regardless of how the trials are performed, the answer ends up being at least half of the total number of password possibilities, which is the staggering figure of 10 raised to the power 77 (written out as 100,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000). Armed with this calculation, you should be very confident in your skepticism, because a 1 in 1077 chance of success is, for all practical purposes, no chance of success.
My experimentally based estimate of the rarity of functional proteins produced that same figure, making these likewise apparently beyond the reach of chance. So, with due caution, let’s transfer Matheson’s reasoning from the problem of protein origins to this hypothetical hacking problem. It is as though Steve Meyer has said that computers with passwords of the strength described above cannot be hacked by trial and error, and Steve Matheson has responded that password strength has nothing to do with it.
That’s a peculiar response. Reading between the lines, I suspect the train of thought is something like this: We know that there are millions of computers on the organization’s network, not just the thousands that are to be hacked, and we know that the network is arranged in a branching pattern with neighboring machines having passwords that differ by only one character, so hacking the first machine will make it easy to hack the rest.
Ummm… but we don’t really know these things. I can understand why Darwinists presume the equivalent things to be true for proteins (and even want them to be true), but Darwinism is itself the thing in question here, so all its presumptions need to be set aside.
Certainly an IT manager could configure a network in such a highly hacker-friendly way, if that were the objective. But absent any reason to think this was the objective, it would be a mistake to presume so. All we really know is that there are thousands of machines to be hacked and that they all use 39-character passwords. The only sensible deduction under these circumstances is that every attempt to hack one of these machines by sampling passwords must fail.
It seems to me that the default assumption for proteins ought to follow the same generalization—that fantastically rare points in vast spaces don’t line up like stepping stones unless something forces them to. Might there be such a force for proteins—even a non-teleological one? Conceivably. So, it would be perfectly reasonable to ask whether something might possibly force functional protein sequences to align in this way. But to dismiss their fantastic rarity as irrelevant, as Matheson has done, is to misunderstand the problem entirely.
In fact, although Matheson is right that my prior paper focused on the rarity of functional protein sequences rather than on their isolation, a recent paper examines directly the implications of their extreme rarity for protein evolution.  Rarity is by no means the only aspect of the problem that has to be considered, but it certainly is a key aspect, and in the final analysis it appears to be the decisive one.