Short Reponse: What are you smoking?
First some general remarks.
I don't know much biology. Certainly Michael Behe knows a lot more biochemistry than I do, and fortunately he shares a great deal of his knowledge in this book. The story of the evolutionary arms race between humans and malaria makes for an interesting read, even if he goes into a bit more detail than one might like in a popular science book.
It's also refreshing that the young earth creationist will find no comfort in this book. No doubt the ID crowd wants to separate itself from its own historical evolution from YEC, but having read too much YEC material, it's pleasurable to read something that doesn't immediately discard practically every scientific discovery since Galileo. ID may not quite be science, but it's at least a much more scientific kind of non-science.
Behe spends most of his time directly on biology, but there is some other matter later in the book that ties it in with the anthropic principle and philosophy. On the whole, Behe's presentation of everything -- biology, physics, philosophy -- is similar to, but much better than Antony Flew's.
OK, let's get down to details. Behe points out that evolution has two main parts. There is the idea of common descent (which he does not deny) and the idea that mutation and selection can account for the observed changes in species (which he does not completely deny). To this end, he wants to show what the "edge" of evolution is. How much can Darwinian evolution do? Does evolutionary history show that more evolution has happened than Darwinian evolution can account for? He contends that the evidence goes beyond (far beyond) the edge of evolution. So some other mechanism has to be responsible for the rest of the observed changes.
It's sensible, but it amounts to an argument from ignorance. He looks at observed mutations in malaria and draws a line in the sand. "Okay, random evolutionary forces can do that much, but no more. Beyond that intelligence is required." But if and when an experiment showed that evolutionary processes can go beyond his line, he can just relocate his edge a little further out, and say that intelligence is required to go beyond that. This is not just an academic objection; since publication, Behe has already conceded that the edge needs to be drawn a bit further out:
Yes, I’m perfectly willing to concede that this does appear to be the development of a new viral protein-viral protein binding site, one which I overlooked when writing about HIV. So the square point in Figure 7.4 representing HIV should be placed on the Y axis at a value of one, instead of zero, and Table 7.1 should list one protein-binding site developed by HIV instead of zero.
In absolute terms, a step from 0 to 1 is not very far, but when considering whether something has never been observed or not, it's clearly of the utmost importance.
Behe also lets the perfect get in the way of the good. He discusses fitness landscapes and correctly notes that evolution may force a species on to a local maximum rather than the absolute maximum. Evolving away from the local maximum may be practically impossible, and critters will never reach the promised land of the absolute maximum. But no one ever thought evolution produced optimal designs. Evolution has no telos, an important point that Behe seems to forget from time to time. Evolution is blind to the absolute maximum, which is why we're stuck with blindspots and a vagus nerve that runs down our necks, loops around the aorta, and then goes back up to the larynx.
Behe criticizes Lenski's E. Coli experiment for showing that "it's easier for evolution to break things than make things". But one of the interesting things to come out of that lab just one year after the publication of Behe's book was a strain that evolved the ability to metabolize citrate. The edge moves out a little bit. Also interesting is the way the mutation evolved. It looks like a neutral mutation occurred halfway along the experiment that didn't have any obvious effect on the bacteria, but when a second mutation happened, citrate metabolism was activated. So it shows that a beneficial effect that requires 2 mutations can occur stepwise.
Behe discusses malaria in depth. The variant gene that causes sickle-cell (if you have two copies) provides protection against malaria (if you have one copy of the gene). There are a number of similar mutations in the hemoglobin gene. He brings up a rare variant, C-Harlem, and says it "has the advantages but not the drawbacks of sickle." And yet in his summary, he says that random mutation and selection in the war against malaria "is losing function, not gaining." It sounds like C-Harlem is a win-win.
Later he has a table of hemoglobin variants that protect against malaria and their other effects on the body. He says "Here's the bottom line: They are all damaging." And yet in the table, three entries have "none apparent" listed as the adverse effects. And C-Harlem is not in the table, as far as I can tell. Behe's trying to tell us that the edge of evolution is here when his own evidence says it's further out than that.
Behe doesn't like the arms-race analogy for evolution, but that's because he's not properly understanding it. He says, quite correctly, that for a gazelle to escape a lion it has more options than to just be faster than its compatriots. It could get more agile, "Or tougher skin. Or grow bigger. Or develop camouflage." Exactly, evolution has many many options, not just one. And obviously the real arms race (as Behe explicitly mentions) involved more than just making bigger and bigger bombs. One side makes bigger bombs, the other makes more accurate missiles. One side puts weapons in Cuba, another side puts weapons on submarines.
Imagining a telos toward which evolution must aim reduces all its options to one, and this of course reduces its likelihood. If we don't specify that it's Mrs. Krploknik who wins the lottery, and we don't specify that she must win with a particular set of 6 numbers, the odds of lottery-winning improve enormously.
Behe also discusses "coherent" mutations that add on to each other. But since evolution keeps a tiny improvement and moves on, a species can gain improvements in 10 different areas in many different ways. This is obviously much more likely than requiring that a species get 10 improvements in the same area.
Flipping things around, I could ask what is the edge of design? Chloroquine is a designed pharmaceutical that combats malaria. It beats the shit out of everything that evolution ever gave us to fight malaria. So the designer seems to be quite lazy in his work.
Behe discusses the same kind of lottery analogy as I have, but he forces things into having only one possible winner. As a yardstick, he notes that about 1 in 10^20 malaria critters will have a resistance to chloroquine. This mutation required two nucleotide base changes in the same protein, which made it 'difficult' to evolve. There have been fewer than 10^20 humans and human ancestors on earth for the past few million years, Soooo.. human beings can't have evolved anything 'more complicated' than a single nucleotide switch.
But there seem to be several problems with this. Behe keeps insisting that the two nucleotides have to change at the same time, but as we saw with the E. Coli, that is not necessarily true.
And more importantly... our good friend C-Harlem is a mutation of the hemoglobin gene with two nucleotide base changes. So even within humans, Behe's argument is wrong. Why does he need to sweep this under the rug? Because it's the whole game right here. Going from 0 to 1 makes the Darwinian game possible, going from 1 to 2 makes the Darwinian game possible in the alotted time. If the mutations have to be simultaneous, then the time required is squared for two bases, cubed for three bases, and so on. There really would be a pretty rigid edge of evolution.
But if they can be sequential instead of simultaneous, then the time required doubles, triples, etc. If humans can get malarial resistance in 10,000 years, then we can get a lot more over the course of the last billion years.
Behe likes to talk about the rarity of mutations. "Any one particular nucleotide ... is freshly substituted about once every hundred million births." This makes it sound awfully rare -- if there are 6 billion people on earth, then only 60 have even a chance for improvement at that locus. But again it is because he is fixating on a particular nucleotide.
Every human on earth has about 100-200 mutations. So this means that each of us has 100-200 ways of improving.
Most of these mutations are neutral (or at least, not fatal), but this is also important, because it means that there's a huge pool of one-nucleotide misspellings out there that might have us halfway to a two nucleotide switch.
He also again talks about how destructive mutations are: "most mutations decrease an organism's overall functioning". If we each have 100-200 mutations, how destructive can it be?
He gets bogged down by telos again, complaining that the mutation that confers warfarin resistance decreases the effectiveness of the affected enzyme. So the mutation is "a net benefit only in desperate times." Yes, evolution can't see the telos. If there happens to be warfarin around, then evolution selects for the benefit. If there isn't any, it selects the other way. In both cases, evolution selects for the better gene given a particular environment. It's just a mistake to complain that the result is not well-adapted for the other environment.
Behe spooges about flagella and cilia. They look pretty darn complicated to me! But check out this chutzpah.
Behe complains that Ken Miller's "rejoinder [to Behe's argument] appeared in a trade book, not in the scientific literature."
But Behe's argument was made in a trade book. And this complaint is being made in a trade book. Why doesn't Behe get off his ass and publish in the scientific literature? I'm not competent to judge a battle between Miller and Behe, but I know Behe didn't fare well in the Kitzmiller trial.
Behe leaps directly for argument from ignorance: "Despite the amazing advance of molecular biology as a whole, despite the sequencing of hundreds of entire genomes and other leaps in knowledge ... the situation concerning missing Darwinian explanations for ... the cilium is utterly unchanged."
I'll also point out that hundreds of genomes provides oodles of data, but explanation may well take a bit longer to generate. The stupid human genome is less than a gig, but CASTOR at CERN has 19786.89 terabytes of physics data, so by Behe's logic, physics should have explained everything by now.
Anyway, the sequencing data is great, but it's just that, raw data. Explanations take time.
Proving Behe and Flew shop in the same analogy-store, Behe also tackles the finite monkey theorem.
Switching from odds of nucleotide replacement to odds of protein binding sites, Behe comes up with similar numbers and arguments. He finds that 1 in 10^20 is the odds for generating one protein binding site.
And humans have generated one protein binding site on hemoglobin with only 10^8 organisms.
So again hemoglobin disproves Behe's numbers, though he explains it away (in the real sense of explaining something away) by saying that hemoglobin is special since it is highly concentrated in red blood cells.
And of course, he goes directly to requiring simultaneous mutations to get two protein binding sites. Same shit, different day.
Behe mentions the protein FKBP. A change in one amino acid causes it to bind to itself with moderate strength. Rather than read this as a great success of Darwinian evolution - a single swap leads to protein binding, here's how he plays it: "it looked like the protein was pre-engineered to be complementary to itself, but was kept apart in the premutated version."
He's already made up his mind that any increase in function cannot be due to Darwinian mechanisms, so when an increase in function shows up, he immediately sees the telos. And if there was telos, then there had to have been preloaded design.
I'm really glad that Behe knows (and accepts) a lot more science than the YEC crowd, but he knows so much that bits of it chip away his own thesis, forcing it to become unfalsifiable: random mutation and selection can't produce something like this, therefore when something like this arises, it must have been due to previous design.
Behe addresses the objection that multiple mutations can happen sequentially rather than simultaneously. He makes some unconvincing telos-y statements.
Behe compares Darwinian evolution to the luminiferous ether. Poor analogy, since Darwinian evolution (at least at the level Behe is willing to admit) is observed.
Behe goes on into anthropic style arguments that I already discussed in Flew's book.
Behe considers theistic evolution to be incompatible with Darwinian evolution. That may be so 'philosophically', but they are experimentally indistinguishable. Scientists (of whichever persuasion) will agree as to the scientific facts, even if they have different philosophical stances on the nature of ultimate explanations. This again cuts to the heart of whether ID is science or not. I don't think it is. As for Behe, does he think ID is science, religion or philosophy?
"In a sense it hardly matters."
Wrong answer, bucko.