I've got one more post coming out of the conversation at Dispatches from the Culture Wars.
One common complaint I hear about design arguments that involve laws at the outset of the universe being designed is that they simultaneously make two things true. Design arguments generally take some surprising fact in nature, and then they appeal to a designer to explain why the surprising fact makes more sense than it would otherwise. The fine-tuning argument, for instance, takes the extremely small range of cosmological constants that would allow rational life as a reason for thinking there must have been a designer who intended rational life to be possibile and formed the laws accordingly.
The objection then comes in. The anti-ID move is to say that there's a contradiction between two things. (1) The natural laws are such that the origin of rational life is unlikely (otherwise there would be no reason to appeal to a designer). (2) The natural laws are such that the origin of rational life is highly likely (otherwise there designer hypothesis has done no work, and we're right back where we started).
This argument is a classic equivocation. It says something that is true when you use your terms in one sense, and then it considers something else alongside it that is true if you use your terms in a different sense. When it puts them together, it gets a false conclusion because it doesn't account for the fact that these two things are not true in the same sense and thus can't combine in this way. A classic example of equivocation is saying (1) you put your money in the bank and (2) that the flooding in the river is overflowing all the banks, concluding (3) you better take your money out of the bank for fear that it will get waterlogged. It isn't the financial institutions that are being covered with water, and that's where your money is.
To see how this problem comes up in this anti-ID argument, I think it will be helpful to look to the general form of a design argument. Design arguments are inductive arguments that take one surprising piece of evidence and then try to explain why such a surprising thing would be true. Other examples of this form of argument abound. We should be surprised to see the coincidental murder, in a very distinctive manner, of five different people simultaneously in five different places. We shouldn't expect five people to do something so similar all at the same time. What are the chances of that? At least we shouldn't expect that unlikely result if there wasn't some sort of collusion between them. This is actually a very good inductive argument for a theory that takes these people to be involved in some sort of conspiracy together. If the collusion theory is true, then the result we've observed becomes very likely. When people conspire to do (at least) five similar acts of murder all at the same time, it's fairly likely that those acts of murder could happen all at once in a very similar manner. This is how this kind of inductive explanation works.
Now let's apply the anti-ID argument. This inductive argument for a conspiracy theory behind these criminal acts leads to an interesting contradiction. It implies both that the murders are very unlikely (otherwise there's nothing to be surprised at) and that the murders are very likely (otherwise the conspiracy hypothesis doesn't help us at all). So if this is a good objection to ID, then it applies to any inductive argument that seeks to explain a surprising event by hypothesizing something that would lower the surprise factor considerably. But obviously these are not bad arguments. It really is reasonable to conclude that there's a conspiracy behind such an amazing coincidence. So something must have gone wrong in the criticism.
The problem is that the objection slips between different senses of probability. It's very unlikely, given the information we have at the outset, that five such murders would take place simultaneously and with such great similarity in five different places. That's the sense in which it's unlikely, the sense in which we're calculating probability in terms of what we should expect given our initial information. But there's also a sense in which the five murders become very probable. Given the hypothesis that there is a conspiracy to commit these five murders, the fact that they all occur becomes highly probably in comparison to its unlikeliness without assuming the hypothesis. But these are just very different probability claims. They're perfectly consistent with each other, and thus there's no contradiction.
Now apply this to intelligent design. We see a surprising result, something that (according to the ID argument) appears to be irreducibly complex, the sort of thing that possibly could arise with unguided natural causes but shouldn't be expected. It's highly unlikely. It cries out for explanation, and the explanation is that an intelligent mind is behind it. So something unlikely according to one set of data (our original information) is made likely by the hypothesis of a designer. The hypothesis makes it very likely, but that doesn't invalidate its unlikelihood given the assumption that all causes are unguided. There's no contradiction at all here.
Postscript on front-loading ID: In the discussion at Dispatches from the Culture Wars, this general argument was applied to a particular view I was sketching. I was arguing that someone could hold to the legitimacy of ID arguments while believing the universe to be a closed system, so long as the designer designed the universe at the outset to exhibit signs of design. Out of all the possible systems of natural laws, the designer chose one that would lead to the sorts of signs of design that ID people generally believe there to be. This view is sometimes called the front-loading ID hypothesis. All the design is worked into the closed system of the universe from the outset rather than the more standard miraculous intervention model of ID that most ID advocates accept. It's not the usual ID view, but it's perfectly consistent with accepting ID arguments as pretty good arguments.
The argument then resurfaces. Before we posit a designer, it seems really unlikely that these features of irreducible complexity would occur in nature. So we appeal to a designer. But if we were to take that designer to have front-loaded all the design into the very laws of nature, don't we then have a system of natural laws that would guarantee the results we have? But if the laws of nature would guarantee these examples of so-called irreducible complexity, then those examples were inevitable. If they were inevitable, then there's really no need to posit a designer, because they would have happened without the designer.
The objection sounds a little better in this case, but it really is just the same mistake. The event looks very unlikely given unguided natural causes, and that's why you appeal to a designer. Once you have the front-loading ID hypothesis with design, then you do have an inevitable event that appears designed. But that doesn't invalidate the argument, because the argument isn't claiming that the event isn't inevitable. It isn't claiming that the event really is unlikely either. It's simply claiming that the event is unlikely given just unguided natural causes. If there aren't just unguided natural causes, then the event is likely after all. But the argument doesn't rely on whether it's actually likely or unlikely. The argument is against there being only unguided natural causes, and just having those makes this result seem very unlikely, so it must be that something besides just unguided natural laws must be behind this event. Once it's clear what the argument is saying, the objection just makes no sense.