Reflections on Dembski's "Vise Document"

Note: This page was written some time ago, when I was still a Christian. I'm not any more, so it's likely that some material in here no longer reflects my current opinions.


On 2005-11-08, "Intelligent Design" proponent William Dembski proposed what he calls (tongue, presumably, in cheek) a "Vise Strategy" as a successor to the "Wedge", the aim being to "squeeze the truth out of Darwinists". You can read the whole thing at (pdf, 23 pages).

Dembski suggests that "Darwinists" (a term used incessantly by anti-evolutionists, and scarcely ever by their opponents) can, and should,be shown up as "bigoted extremists", "condescending elitists" or "closet ID theorists", by asking them a series of questions intended to reveal their hidden agendas and presuppositions. Well, I happen to be what Dembski calls a "Darwinist" and I'm not aware of being any of those terrible things, so let's try the obvious experiment: the following are my answers to all of Dembski's questions. We'll see whether I end up in a morass of contradictions, or showing my embarrassing true colours, or what.

I don't propose to offer much in the way of support for my answers, except where the questions actually ask for it. Dembski contends that just by answering the questions I should find myself in "a steel trap that leave[s] the Darwinists no room to escape". So, after answering the questions, I'll look back and see whether that's happened, and make some other remarks on the questions and on the comments that Dembski scatters among them.

Questions and answers

For the sake of realism, I'm answering these questions in order, without looking ahead for traps or anything. I'm taking them in the same order as Dembski gives them in, and not skipping any unless they're obviously completely inapplicable. Dembski's questions are in italics; my answers are in roman except for occasional bits of emphasis. All headings are mine.

The nature of science

Is it fair to say that you regard intelligent design as not a part of science? Would you agree that proponents of intelligent design who characterize it as a "scientific discipline" or as a "scientific theory" are mistaken? Yes. It isn't altogether impossible a priori that something along the lines of "intelligent design theory" might some day be part of science, but for now there is no scientific theory, and no scientific discipline, of "intelligent design".

One aside that seems worth making: the term "intelligent design" is ambiguous in various ways; it can mean what the ID folks say has happened, or the belief that it's happened, or the belief that there's compelling scientific evidence that it's happened, or a putative scientific theory showing that it's happened, or a putative scientific theory exploring its details. It would be a useful exercise to distinguish clearly all the possible meanings of "intelligent design" and disambiguate every use of that phrase, for the avoidance of equivocation; but I'm not going to try to do it now.

Would you characterize intelligent design as a "pseudoscience"? I don't think it's as pseudo as, say, astrology. For some purposes it would be better characterized as really, really bad science than as outright pseudoscience. But, on the whole, yes.

Would it be fair to say that, in your view, what makes intelligent design a pseudoscience is that it is religion masquerading as science? If ID is something other than science, what exactly is it? No, it wouldn't be quite right to say that the reason why ID isn't science is that it's "religion masquerading as science" -- although there's some truth in that accusation. There are a number of reasons why ID is not science. (As there ought to be, if it isn't. "Science" is a fuzzily defined notion, whose boundaries are loosely defined by several different criteria. For something to be firmly outside the realm of science, it should typically be found to miss on several of those criteria.) So: ID is not science, because ...

Are you a scientist? Only in a weak sense: my professional qualifications are in a somewhat scientific subject, and my work is in a somewhat scientific field. (So I'm a "scientist" in the same sense as William Dembski.)

Do you feel qualified to assess whether something is or is not properly a part of science? What are your qualifications in this regard? I feel qualified to have an opinion on the subject, just as I feel qualified to have an opinion about whether there's a God, whether taxation is a good thing, and so on. I don't claim that my opinion is authoritative, or that anyone else should be obliged to agree with it.

Do you think that simply by your being a scientist, you are qualified to assess whether or not something is or is not properly a part of science? In the only sense of "qualified" that I claimed above, I think I'm so "qualified" quite apart from whether I'm a scientist or not. I don't think this is very interesting.

Have you read any books on the history and philosophy of science? Yes, a few, but not very recently.

Would you agree that in the history of science, ideas that started out as "pseudoscientific" may eventually become properly scientific, for example, the transformation of alchemy into chemistry? As I understand it, the "transformation of alchemy into chemistry" is an example not of ideas changing from pseudoscience to science, but of a community of practice making that change by changing their ideas. I don't think many of the ideas of alchemy would be regarded as scientific. But I do think such transformations can happen, in principle. For instance, maybe Democritus's "atoms" were pseudoscience in his time (though the term is a bit unfair, as science-as-we-now-understand-it didn't exist yet) but something a bit like them was science in Dalton's time. (The modern conceptions nearest to Democritean atoms are different enough from them that I don't think the ideas can reasonably be considered the same.) This sort of transformation generally happens, I think, by "coincidence": the original practitioners aren't doing science and don't understand why their ideas will eventually turn out to be right. Hence, in such instances the initial label of "pseudoscience" is fair.

Is it possible that ID could fall into this category, as the transformation into a rigorous science of something that in the past was not regarded as properly scientific? That would probably depend on how, and how broadly, "ID" is defined. As I said right at the start, something along those lines could conceivably happen. I'd guess that the resulting scientific discipline would have to be very different from what ID advocates are currently doing.

Are there precise criteria that tell you what belongs to science and what doesn't? No; the boundaries of science are fuzzy.

Then on what basis do you preclude ID from being science? In that case, isn't your ruling out ID as belonging to science purely a subjective judgement? How do you rule it out as not being science if you have no criteria for judging what's in and what's outside of science? The previous question was about precise criteria, and suddenly this question assumes that a "no" answer means that I have no criteria at all! Bah. Compare the famous example of "games". There doesn't seem to be a useful definition of "game" in terms of precise criteria; rather, we have some fuzzy notions of what makes something a game, and various prototypical examples and non-examples, and so on. So, indeed, for some things it's a judgement call whether they should be called "games" or not. But the answer to "Is the manufacture and testing of nuclear warheads a game?" isn't in any useful sense "purely subjective", because there are so many ways in which that activity fails to be game-like. Similarly (though to a lesser extent) there are so many ways in which "intelligent design" fails to be science-like that I'm comfortable saying outright that it isn't science. Let me reiterate that I don't claim to be an authority on the subject; but you asked my opinion, and I'm giving it.

Please list all the criteria you can think of that demarcate science from non-science. Are you sure these are all of them? If you are not sure these are all of them, how can you be sure that your criteria are the right ones? There are a great many. I've listed, or at least alluded to, several of them above. Of course I'm not sure that my set is complete. I strongly doubt that "all of them" really has meaning; there will surely be many possible sets with substantially the same consequences. Likewise for "the right ones"; there are probably plenty of sets of criteria that could be considered "the right ones".

Do these criteria work in all cases? Do they tell you in every instance what's in and what's outside of science? Are there no exceptions? Of course they don't give a clear answer in every case, because some things simply aren't either clearly in or clearly out. I'd put string theory on the boundary, for instance. Maybe cold fusion, too. Maybe "hydrino theory". Again, that doesn't stop there being instances in which the answer is clear. General relativity is clearly in. Astrology is clearly out. I think ID is fairly clearly out, though (as I said before) not as clearly out as astrology. One especially fuzzy boundary is the one between non-science and really bad science. I'm not entirely certain which side ID is on.

Let's consider one very commonly accepted criterion for what's in and what's outside of science, namely testability. Would you say that testability is a criterion for demarcating science? In other words, if a claim isn't testable, then it's not scientific? Would you agree with this? I'm not sure that I'd agree at the level of individual claims. A claim needs to be assessed in the context of a surrounding theory, and one can always adjust the theory instead of the claim if experimental consequences come out unfavourably. Still, if a theory doesn't make testable predictions then that's a heavy argument against its being "science"; and if proponents of a theory show themselves inclined to revise anything and everything other than their theory when unfavourable evidence comes in, that's a heavy argument against the claim that they're doing "science".

Would you give as one of the reasons that ID is not science that it is untestable? Yes, and indeed I already did give that as one of the reasons. It may be better considered as a reason why ID advocates aren't doing science than as a reason why ID, in the abstract, isn't science.

Let's stay with testability for a bit. You've agreed that if something is not testable, then it does not properly belong to science. Is that right? Not quite; my actual position is stated above.

Testability of methodological naturalism

Have you heard of the term "methodological materialism" (also sometimes called "methodological naturalism")? Yes, and I consider the latter -- much more usual -- term for it a better one; some of the "natural" entities postulated by science, such as wave functions, could with some plausibility be called immaterial.

Do you regard methodological naturalism as a regulative principle for science? In other words, do you believe that science should be limited to offering only materialistic explanations of natural phenomena? ... This is not a trick question. By materialistic explanations I simply mean explanations that appeal only to matter, energy, and their interactions as governed by the laws of physics and chemistry. Do you regard methodological naturalism in this sense as a regulative principle for science? There are a few ways in which I'm uncomfortable with the principle as stated, though there's clearly something right about it. Firstly, it seems to say that only the entities currently known to physics are admissible in explanations. That's obviously silly; new discoveries in physics, some of them involving whole new kinds of stuff, happen every now and then. Secondly, "should be limited to" is somewhat ambiguous. Thirdly, you can do science (perhaps in a slightly attenuated sense of the term) taking as foundation some set of concepts that don't go all the way down to fundamental physics, as in the "social sciences".

I do believe in a principle of methodological naturalism, but I really don't like your statement of it. I think the key idea is this: Explaining the behaviour of A in terms of that of B is a scientific activity only in so far as the behaviour of B is clearly specified. So you can offer explanations in terms of electrons and call that science, because we have a pretty detailed account of how electrons are supposed to behave. If you want to offer explanations in terms of, say, human reactions to incentives, then for what you're doing to be science you need to say what those reactions are supposed to be; that's what economists do, sometimes. Explanations in terms of God are only scientific to the extent that you have an underlying theory of what God is supposed to do. Which, on the whole, we don't, and we aren't supposed to, what with God being ineffable and sovereign and all that.

(As an answer to, approximately, one of the things I said above, Dembski proposes that the questioner should say ...)

I see. You think there are higher-order phenomena that cannot be accounted for in terms of any sort of reductive materialism. Would you then admit that intelligence is a higher-order process that's fundamental to nature and that can be invoked in scientific explanations? Would you be comfortable in claiming that intelligence constitutes a legitimate category of explanation within the natural sciences? If so, then how can you say that intelligent design is non-scientific? If not, then how can you deny holding to methodological materialism? You can use intelligence in scientific explanations if you've got a specification of how intelligence behaves. So a psychologist who explains some results in terms of differences in intelligence could be offering a somewhat-scientific explanation. I don't regard intelligence as a process that's "fundamental to nature", at least not if I understand that phrase correctly. I don't deny holding to methodological naturalism; see above.

Could you explain the scientific status of methodological materialism? For instance, you stated that testability is a criterion for true science. Is there any scientific experiment that tests methodological materialism? Could you describe such an experiment? Methodological naturalism is not a scientific claim; indeed, it isn't a claim at all, but a technique. If there's an underlying claim, it would be something like "Methodological naturalism is an effective strategy"; we could test it, for instance, by having some people do biology using methodological naturalism and some people (let's call them "the Discovery Institute") do biology without it, and see which group produces more and better scientific research. Well, it turns out that this has been done and methodological naturalism is doing pretty well.

Are there theoretical reasons from science for accepting methodological materialism? For instance, we know on the basis of the second law of thermodynamics that the search for perpetual motion machines cannot succeed. Are there any theoretical reasons for thinking that scientific inquiries that veer outside the strictures of methodological materialism cannot succeed? Can you think of any such reasons? No, of course there's no reason why someone who rejects methodological naturalism can't succeed; one can get good results using suboptimal procedures. Kepler made some good astronomical discoveries despite having strange ideas about the planets and nested Platonic solids. There is empirical evidence that methodological naturalism works well in practice; see above. There is a theoretical reason for expecting this (though it's doubtless informed by hindsight): without it, it's too easy to give up too early.

A compelling reason for holding to methodological materialism would be if it could be demonstrated conclusively that all natural phenomena invariably submit to materialistic explanations. Is there any such demonstration? No, of course not.

(Conditional on the appeal -- as in my answers above -- to the empirical success of methodological naturalism so far, Dembski suggests asking ...)

But wouldn't you agree that there are many natural phenomena for which we haven't a clue how they can be accounted for in terms of materialistic explanation? Take the origin of life? Isn't the origin of life a wide open problem for biology, one which gives no indication of submitting to materialistic explanation? It's certainly wide open, which is hardly surprising considering how far in the past it is. "No indication of submitting to materialistic explanation" is ambiguous. There are various candidate (but sketchy) materialistic explanations; it certainly isn't obvious to me that they're no good. So, sure, we don't know what happened, just as we don't know what happened to Ambrose Bierce. That doesn't, in itself, mean that there's any sort of crisis for naturalistic accounts of the world, any more than our ignorance about Bierce means we should think it likely that he was abducted by aliens.

Would you agree, then, that methodological naturalism is not scientifically testable, that there is no way to confirm it scientifically, and therefore that it is not a scientific claim? As mentioned above, it's not a claim at all. The claim "methodological naturalism is effective" is a claim, and it's somewhat testable, by somewhat scientific means, and it emerges quite well from that testing. Stronger claims like "Everything we'll ever care about can be usefully explained using no notions not found in fundamental physics" are harder to test and may be false. The claim "Methodological naturalism is characteristic of science" is a claim about science rather than within it (though the sociology of science is, to some extent, itself a science).

Is there any way to show scientifically that materialistic explanations provide a true account for all natural phenomena? Is it possible that the best materialistic explanation of a natural phenomenon is not the true explanation? If this is not possible, please explain why not. Of course there isn't; and, of course it is.

Since methodological materialism is not a scientific claim, what is its force as a rule for science? Firstly, that methodological naturalism has been found to be a very effective technique. Secondly, that without it it's too easy to give up and say "at this point, a miracle occurs". Much of scientific technique is about getting in the way of human laziness, bias, and bad thinking. The tendency to give up looking for explanations when the going gets heavy is one of the things scientific technique tries to get in the way of.

But if methodological materialism's authority as a rule for science derives from its success in guiding scientific inquiry, wouldn't it be safe to say that it is merely a working hypothesis for science? And as a working hypothesis, aren't scientists free to discard it when they find it "no longer works"? It's far from clear to me what "works" means here. But of course it might happen that something isn't explicable in scientific terms, and it certainly often happens that we don't know how to explain something in scientific terms. In such cases, we can stop doing science and start doing something else, and sometimes that gives what seem to be good results. (We do this all the time in our interactions with one another.) In general, what we then do isn't considered "science"; it typically lacks some useful attributes we associate with science: precision, testability, predictive force. (These consequences are clearest when methodological naturalism is understood in terms similar to the ones I described above: An explanation conforms to MN in so far as it eschews appeals to things we don't understand and can't model.)

If methodological materialism is not a scientific claim, how can it be unscientific for ID theorists to discard it as a working hypothesis for science? In the absence of methodological materialism as a regulative principle for science, what else is there that might prevent ID from being developed into a full-fledged science? You claimed earlier that ID is not testable. Is that the reason you think ID cannot be developed into a full-fledged science? Methodological naturalism is not a "claim" at all. It's "not scientific" in the sense that it isn't a claim within science, not in the sense that it doesn't describe what science is or what science should be. So I don't see what the obstacle is to declaring that those who drop it are being unscientific. As discussed above, they might in some instances be right; but "right" is not the same thing as "scientific". Someone who says that Bach is a better composer than Mozart, or that there is exactly one god, may be right, but those aren't scientific statements either.

I mentioned earlier some other things that argue against regarding ID "theory" as science. Whether there's any prospect of a genuinely scientific theory emerging that we culd rightly regard as continuous with todays' ID, I don't know -- though I don't see any sign of it happening. Untestability certainly seems like a problem for ID; Dembski's purported tests for design are rubbish.

But how can you say that ID is not testable? Over and over again, Darwin in his Origin of Species compared the ability of his theory to explain biological data with the ability of a design hypothesis to explain those same data. Moreover, Darwin stressed in the Origin that "a fair result can be obtained only by fully stating and balancing the facts and arguments on both sides of the question." How, then, can you say that ID is not testable when Darwin clearly claimed to be simultaneously testing a design hypothesis against his own theory? As I mentioned way back, the "testability" of a theory is often as much a matter of the attitude of its proponents as of the theory itself. ID, as practiced by its actual proponents, is not testable because they offer no way to test it; no experimental outcome that would lead them to admit defeat. That doesn't mean that it's impossible for any design hypothesis to be testable, and perhaps Darwin considered some testable ones. Since what's now called "ID theory" wasn't around in Darwin's time, you can't claim that it was ID that Darwin was proposing to test. (If you insist on doing so, then you'd better also say that ID is over a hundred years old. If it hasn't become a "full-fledged science" yet after more than a century, its prospects are looking pretty dim.)

Okay, you are still not convinced that ID is testable. Consider the following possibility: Darwinian biologists provide detailed testable scenarios for how the bacterial flagellum and other irreducibly complex molecular machines that Michael Behe has identified could have been produced by, as Darwin put it, "numerous successive slight modifications". In that case, wouldn't you agree that ID would be tested and found wanting? Some claims made by ID advocates in support of ID are certainly testable. Concrete claims that such-and-such a thing couldn't have evolved naturally are probably testable. The existence of some testable fragments within a body of claims doesn't make that body as a whole testable. If Behe's concrete claims about irreducible complexity were refuted, would you give up on ID? If not, then their testability doesn't mean that (for you) ID itself is testable.

Is ID creationism?

Let's talk about creation and creationism a bit. Is it fair to say that you think ID is a form of creationism? On the whole I'd say no; ID has been described as "creationism in a cheap tuxedo", but in fact I'd say that ID is the cheap tuxedo in which creationists have tried to disguise themselves and their views.

Does ID try to harmonize its scientific claims, like those about specified complexity and irreducible complexity, with the Bible? If so, please indicate. It's not clear what ID "does", not least because the term "ID" is so ambiguous. ID advocates certainly do that, but there's nothing particularly wrong with that. (Some evolutionists also try to harmonize their scientific claims with their religious beliefs and unbeliefs. Nothing wrong with that, either.)

But there are a couple of things that go beyond just "trying to harmonize" and that seem to me to give more cause for concern. Firstly, some ID folks have said that ID is basically theology in disguise, as when Dembski said that "intelligent design is just the Logos theology of John's gospel restated in the idiom of information theory". Secondly, some ID folks seem to put ID forward as the result of impartial scientific investigation, when in fact they started from the assumption that things are the product of "intelligent design" (on the basis of their religious views) and went searching for reasons. The converse sometimes happens among evolutionists -- Dawkins probably offers examples -- and it's a bad sign there, too; but what's worse about ID is that this seems to be the history of just about every ID proponent, whereas it doesn't seem that every evolutionist begins from a position of atheism or naturalism.

Is it fair to say that ID is not in the business of matching up its scientific claims with the Genesis record of creation or any other system of religious belief? If otherwise, please indicate. Again, it's not clear how to determine what "ID" does, as distinct from what its proponents do. Some of those proponent do try to match up their allegedly scientific claims with their theology, but when they do so they say they aren't doing ID theory. I'm willing to answer "yes" to this one, at least for the purposes of argument.

Is it fair to say that ID is not young earth creationism, also known as scientific creationism or creation science? It's not the same thing as any of those (though "creation science" is also not the same thing as "young earth creationism", so there's something fishy here).

Is it possible to hold to ID and not be a Christian, Jew or Muslim? Is it possible to be a Buddhist and hold to ID? Is it possible to be a Hindu and hold to ID? Of course. It's probably even possible to be an atheist and hold to ID, though I don't know of any examples. In practice, though, proponents of ID almost always seem to start off by rejecting the usual scientific account of the history of life for religious reasons, and then turn to ID for support.

Is it possible to hold to ID for philosophical reasons that have nothing to do with any sort of conventional belief in God? In other words, can one hold to ID and not believe in God, much less a creator God? In principle yes; in practice, it's very rare. See above.

Would you agree that Aristotle, who held to an eternal universe and an inherent purposiveness within nature (i.e., not imposed on nature from the outside), did not have a conventional belief in God but would today properly be regarded as an ID advocate? No. Aristotle argued for the existence of God as, for instance, the "First Mover". He wrote, "We say therefore that God is a living being, eternal, most good, so that life and duration continuous and eternal belong to God; for this is God" [Metaphysics, 1072b]. As to whether he would rightly be called an ID proponent today, I don't know. If it's true that he attributed the nature of living things to an "inherent purposiveness" without an intelligence behind it, then I'm not sure why he should be called an intelligent design theorist. If he thought there was an intelligence behind it, then it's not clear to me that his beliefs on this score are much different from those of typical theists.

Are you familiar with Antony Flew's recent embracing of intelligent design despite his rejection of conventional belief in God (for instance, he explicitly rejects personal immortality)? Aware of, rather than familiar with. I'm also aware that along with that change of view he also embraced deism, and that he has said "I now realize that I have made a fool of myself by believing that there were no presentable theories of the development of inanimate matter up to the first living creature capable of reproduction" -- that being the issue on which he found ID arguments convincing. So Flew isn't a great example either of a non-religious person or an ID believer, let alone of a non-religious ID believer.

Let's now turn to someone like Kenneth Miller, who has remarked "I'm an orthodox Catholic and an orthodox Darwinian". Miller, as a Catholic believer, holds to a doctrine of creation. Is Miller a creationist? Not according to my definition. Possibly according to some very loose definitions, like one adopted by William Dembski when he wrote about "Mere Creation".

(The following question, Dembski observes, would need some tailoring when addressed to anyone other than Miller himself. I'll quote it as it stands and answer it as if suitably tailored.)

Prof. Miller, as an orthodox Catholic, is it fair to say that you subscribe to orthodox views of divine action? In particular, do you believe that God has acted miraculously in salvation history, parting the Red Sea, performing miracles in the life of Jesus, notably his miracles of healing, transforming water into wine, and above all the Virgin Birth and Christ's Resurrection? Were these miracles plain to see? For instance, when Jesus changed the water into wine, was it evident that a miracle had taken place? I expect Miller's answer would resemble mine: yes, God has sometimes acted miraculously, and some but not all of these miracles would have been plain to see for those with their eyes open.

So you agree that God is able to act miraculously and that God has indeed acted miraculously and discernibly in salvation history. What then prevents God from acting miraculously and discernibly in natural history? Nothing. Obviously.

Okay, let me get this straight. Miller is an orthodox Catholic. He holds to a creator God who has acted miraculously in history. And yet he is not a creationist. On the other hand, there are ID proponents (like David Berlinski) who have no religious belief and who, simply in virtue of supporting ID, are, according to you, creationists. Wouldn't it be fair to say that it is simply an abuse of language to identify ID with creationism? I don't myself identify ID with creationism; see above. I wouldn't say it's "simply an abuse of language" to do so, since making the identification points up an important fact, namely that the great majority of ID proponents are using ID to buttress creationism. But yes, I consider it incorrect to identify the two.

Your main beef with ID seems to be that it claims that material causation is an incomplete category for scientific explanation. Is that true or is there any other criticism that you think is more significant? If it is true, how can you claim that ID is creationism? Creationism suggests some positive account of an intelligence creating the world. But your problem with ID seems to be in its denial that a certain category of causation can account for everything in nature. I have several other disagreements with ID, and reasons for considering it unscientific, besides the one mentioned; see above. And I don't claim that ID "is" creationism.

Methodological naturalism again

Are you merely a methodological materialist or are you also a metaphysical/philosophical materialist? In other words, do you pretend that everything happens by material causation merely for the sake of science but then bracket that assumption in other areas of your life? Or do you really hold that everything happens by material causation -- period? If the latter, on what grounds do you hold to metaphysical materialism? Can that position be scientifically justified? How so? If you claim merely to be a methodological materialist, then whence the confidence that material causation is adequate for science? I am a methodological naturalist only. I don't "pretend" that everything happens by material causation, for the sake of science or otherwise. I'm not even sure that "material causation" can be defined in such a way as to be a useful term, but let's suppose it can. I simply consider that non-naturalistic explanations are not part of science, because (see above for what I mean by "non-naturalistic" in this context) they are inevitably not checkable empirically and short on useful predictions.

What is the nature of nature? Does nature operate purely by material causation? If not, how could we know it? I'm not sure what the scope of "nature" is here. I don't, on the whole, think that everything happens by material causation. There might be some mileage in restricting the term "nature" to things that do (following the widespread but fuzzy distinction between "natural" and "supernatural", in which case the answer would be "yes, by definition".

Consider the following riddle (posed by Robert Pennock): If you call a tail a leg, how many legs does a dog have? Wouldn't you agree that the answer is four: calling a tail a leg doesn't make it one? Accordingly, wouldn't it be prejudicial to define nature as a closed system of material entities in which everything happens by material causation? Wouldn't you agree that nature is what nature is, and it is not the business of scientists to prescribe what nature is like in advance of actually investigating nature? I think the riddle actually goes back to Abraham Lincoln. Calling a tail a leg doesn't make it one, but there might be contexts in which "leg-or-tail" was a useful concept. Similarly, there might be contexts in which "whatever happens by material causation" is a useful concept, and "nature" might be the nearest thing we have to a good term for it. But that's all speculation. I don't define "nature" in any such terms; I do agree that the world is whatever it is; and I agree that it's possible in principle that we might find compelling evidence that some things aren't susceptible to what we currently consider scientific explanation.

However: that doesn't mean that we should redefine "science". For one thing, such evidence as the ID movement claims to have for things that aren't explicable in what's-currently-considered-scientific terms is hopelessly weak. For another, even if there were such evidence that wouldn't make redefining science the right course of action. Scientific enquiry gains many advantages from its refusal to consider "unscientific" explanations (which, please note, I do not consider to be the same as "explanations that use notions other than the ones currently considered fundamental by physicists"; see above), and giving that up might not be worth it for the benefit it would bring. Every individual scientist who has any unscientific beliefs and admits it (that would be just about all of them) has made a similar decision in his or her own life.

Consider the following statement: "To make methodological materialism a defining feature of science commits the premodern sin of forcing nature into a priori categories rather than allowing nature to speak for itself." Do you consider this statement right or wrong? If wrong, why? I consider it wrong, because making methodological naturalism a defining feature of science is placing restrictions on science, not on nature.

Testability again

Let's return to the issue of testability in science. Do you agree that for a proposition to be scientific it must be testable? Good. Not so fast. I'm not sure that I do agree. Untestability is a very, very bad sign; but firstly, it applies to whole theories as well as to propositions, and secondly, testability isn't the only criterion for what's science and what isn't, and some untestable propositions might get in on their other "merits". For instance, "The behaviour of electrons is well described by certain partial differential equations" is probably too vague to refute even if false, but I don't think that stops it being a scientific statement.

Would you agree, further, that testability is not necessarily an all-or-none affair? In other words, would you agree that testability is concerned with confirmation and disconfirmation, and that these come in degrees, so that it makes sense to talk about the degree to which a proposition is tested? For instance, in testing whether a coin is fair, would finding that the coin landed heads twenty times in a row more strongly disconfirm the coin's fairness than finding that it landed only ten times in a row? Yes, of course.

Okay, so we're agreed that science is about testable propositions and that testability of these propositions can come in degrees. Now, let me ask you this: Is testability symmetric? In other words, if a proposition is testable, is its negation also testable? Not entirely. It is more important for a proposition to be refutable if false than for it to be provable if true. (Of course, "refutable" really means something like "able to be shown to be tremendously improbable" and "provable" something like "able to be shown to be tremendously unlikely to be wrong".) Negating a proposition interchanges refutability and provability, so it may change the extent to which a statement is testable.

AS a general rule, if a proposition is testable, isn't its negation also testable? Can you help me to understand how a proposition can be testable, but its negation not be testable? To say that a proposition is testable is to say that it can be placed in harm's way of empirical data -- that it might be wrong and that its wrongness may be confirmed through empirical data, wouldn't you agree? Testability means that the proposition can be put to a test and if it fails the test, then it loses credibility and its negation gains in credibility? Wouldn't you agree?

Doesn't it then follow that whenever a proposition is testable, so is its negation, with a test for one posing a test also for the other?

The explanation of what "testable" means is pretty good. But it doesn't remotely follow from it that testability is symmetrical, and I don't understand why Dembski thinks it does.

Let me therefore ask you, are the following propositions scientific and, as a consequence, testable? (1) Humans and other primates share a common ancestor. (2) All organisms on Earth share a common ancestor. (3) Life on Earth arose by material causes. Are the negations of these propositions therefore scientific and testable? If not, why not? Of these: #1 is somewhat testable, and so is its negation. #2 is somewhat testable, and so is its negation. In each of these cases, it's difficult to establish conclusively that two things do or don't share a common ancestor, which is why I've said only "somewhat testable" for each. #3 is less testable than either of those, and its negation if anything even less so. I am not inclined to call either #3 or its negation a scientific proposition.

Let's focus on the third of these propositions. How is it tested? If its negation is not testable, how can the original proposition be testable? Wouldn't it then simply be like arithmetic -- simply a necessary truth and not something in contact with empirical data? The third proposition seems very difficult to test, to me. It's possible to imagine circumstances in which (the usual understanding of) it would be refuted -- for instance, say we encounter aliens who claim to have put that first life on earth, having designed it carefully themselves, and they kept very detailed records that we can check. But most of the ways in which #3 could be false probably wouldn't lead to any way for us to see that it's false. Its negation is also not very testable; in fact, it seems just about entirely untestable. We could come up with a very plausible natural history of the origin of life and establish that it's compatible with everything we know (or, at least, maybe we could), but that wouldn't be anything close to proof that it actually happened that way. So the negation of #3 is even more untestable than #3 itself, and I'm even less inclined to consider it scientific. But I don't see how either #3 or its negation is much like arithmetic -- these aren't the sorts of things you can plausibly claim to know a priori, they aren't truths of pure logic, and they certainly aren't true by definition.

Let's now turn to evolution. Back in 1989 Richard Dawkins remarked that those who don't hold to evolution are "ignorant, stupid or insane (or wicked, but I'd rather not consider that)". Is Dawkins right? Quite possibly. I seem to recall that he pointed out that ignorance afflicts us all and that to be ignorant of something is not necessarily cause for shame; I would certainly want to point that out in connection with such statements. I don't think I can comment any more definitely without knowing more exactly what meaning of "evolution" is in play here; different people use that word in very different ways.

Common descent

Evolutionists distinguish between common descent (also known as universal common ancestry) and the mechanism of evolution. Common descent is a historical claim. It says that all organisms trace their lineage back to a last universal common ancestor (sometimes abbreviated LUCA). Do you hold to common descent? Please be as detailed as you can in describing the scientific evidence that leads you to that belief. I suspect that if you go far enough back, the mechanisms of reproduction and inheritance get a bit tangled up (as they still are, to some extent, with viruses), and certain technical difficulties may arise in talking about a "universal common ancestor", and most especially in talking about a last one. But I think the basic idea is very well supported. For instance, look at the extent to which living things on earth use the same mechanisms -- DNA and proteins, for two obvious examples at a very low level. Other explanations are, of course, possible: maybe there's some hitherto unobserved mechanism by which features of one kind of life get copied into another without common ancestry, or maybe someone or something created lots of lineages separately but chose to make them share all those features just as if they had common ancestry, or something. Common descent seems like the simplest explanation available, especially with the standard scientific restriction of avoiding "non-naturalistic" explanations in terms of ill-understood and unspecified entities like intelligent designers.

Are you familiar with the work of Carl Woese and Ford Doolittle? What is their view of the origin of life? Is it monophyletic or polyphyletic? Do you accept their conclusions? Why or why not? Would you agree that these are reputable scientists? Doesn't their work throw into question common descent? If not, why not? Do you accept that there were multiple origins of life but that the multi-cellular life that now exists traces its lineage back to a last universal common ancestor? I'm not familiar with their work, but let me just ask Google... So, it looks like this is about the prevalence of horizontal gene transfer among some simple living things, and about the idea that early living things might have emerged from a kind of gene soup. OK, maybe. This is just the kind of thing I mentioned above about "if you go far enough back"; it doesn't seem to me to throw into question anything that matters. (If something like this turns out to be right, then whether it amounts to "multiple origins of life" might depend on how you define "life", which is frankly neither interesting nor important.) In Woese's and Doolittle's accounts, as I understand it, we still have common ancestry, but maybe it goes back to before there were things you'd want to call "organisms".

No doubt you have heard of the Cambrian explosion. Isn't it the case that the fossil evidence suggests that many of the animal phyla first appear over a period of 5 to 10 million years in the Cambrian rocks without evident precursors? Well, kinda. There's a period of a few tens of millions of years during which early representatives of many things-that-are-now-phyla are first seen. They weren't as different then as their descendants are now, of course, so talking about different phyla can be misleading; and it seems pretty plausible that what actually emerged "quickly" (over mere tens of millions of years!) in the Cambrian was the "invention" of hard, fossilizable body parts.

What multicellular precursors are there to the Cambrian fauna? Why should we think that these are ancestral to the Cambrian fauna? We may not have found the ancestors of the known Cambrian fauna. Various pre-Cambrian fossils are known, notably the Ediacaran and Vendian ones. Are they ancestral to the known Cambrian fauna? I don't know.

Stephen Jay Gould and Simon Conway Morris have both cast doubt on whether the Ediacaran fauna are ancestral to the Cambrian fauna. Are you familiar with their arguments? Do you share their doubts? If not, why not? I'm not familiar with their arguments, so let me just go and ask Google again ... Well, I haven't had much luck; I turned up some things that look very much like both Gould and Morris saying (or at least assuming) that the Ediacaran fauna are ancestral to the Cambrian. So I can't comment on their alleged views to the contrary.

Consider an octopus, a starfish, an insect, and a fish. To what phyla do these belong? Is there solid fossil evidence that these share a common ancestor? If so, please provide the details. Mollusca, Echinodermata, Arthropoda, Chordata, respectively. Fossil evidence? I doubt it. The evidence for common ancestry between phyla is internal rather than external.

Do you regard the Cambrian explosion as providing a challenge to common descent? If not, why not? No; I can't see the least reason why I should.

Mechanisms of evolution

I want next to turn to the mechanisms of evolution. What are the mechanisms of evolution? Are these all of them? There are lots, and probably some of them aren't known. There's descent with modification, which in turn is a consequence of several submechanisms: reproduction, sex, mutation, crossover, and so on. There's natural selection, which comes in various forms such as differential survival and sexual selection. There's artificial selection, which you could consider a special case of coevolution. You might consider "neutral drift" a mechanism, though I'm more inclined to call it a consequence of what happens when you have variation that isn't strongly acted on by selection. There are some interesting processes like horizontal gene transfer and (perhaps) whatever turned separate organisms into organelles like mitochondria. These days there's even genetic engineering. I'm sure I haven't listed all of them; I doubt whether all of them are known.

So, you're not sure that these are all the mechanisms that drive the process of biological evolution. Is intelligence a mechanism? If you can't be sure that you've got all the relevant mechanisms of evolution, how can you rule out intelligence as a factor in biological evolution? Intelligence isn't a mechanism of evolution; it's the wrong kind of thing to be. But some things that intelligent beings do (such as genetic engineering and artificial selection) are mechanisms of evolution. Artificial selection has had considerable impact on the evolution of some kinds of organisms.

What, that isn't what you meant? Well, too bad you didn't ask the right question.

Okay, you're convinced that the neo-Darwinian mechanism of natural selection and random genetic change is the most important factor in biological evolution. Why is that? What is the evidence that it deserves this place in evolutionary theorizing? Please stop putting words into my mouth; I didn't say what you claim I'm "convinced" of. However: The evidence that this neo-Darwinian mechanism is a major factor in biological evolution (the most important? Who knows?) is that we know it happens, we can watch it, it has big observable effects on timescales we can observe, and it offers what seem like good prospects of explaining a whole lot of things about life on earth.

Are you familiar with the bacterial flagellum, a miniature bidirectional motor-driven propeller that moves certain bacteria through their watery environments? Are you familiar with the standard account told about its evolution, namely, that a microsyringe embedded into this system eventually evolved into it? Do you accept this explanation? I know that there is such a thing as a flagellum, and that if you're inclined to describe biological things as if they were products of engineering then you might call it a "miniature bidirectional motor-driven propeller". I know that one conjecture about its origin is that it is derived from a secretory system (which, if you're inclined etc., you might describe as "a microsyringe"). That seems pretty plausible to me, but I'm not expert enough in microbiology to claim that it's definitely the right answer.

Would you agree that this microsyringe, known as a type three secretory system (abbreviated TTSS), is much simpler than the flagellum (requiring only about 12 different proteins whereas a full flagellum requires about 40 different proteins)? How then does pointing to the TTSS as a precursor of the flagellum explain it? How is this different from pointing to a motor of a motorcycle and saying that the motor evolved into the motorcycle? How does pointing to the TTSS give us the "numerous successive slight modifications" that Darwin described as necessary in any evolutionary pathway? Sure, the TTSS is simpler than the eubacterial flagellum. It would have to be, to be any use in an evolutionary explanation. Merely pointing to the TTSS doesn't explain the evolution of flagella; as I understand it, evolutionary biologists don't merely point and say "behold!". In any case, consider the way in which the flagellum entered anti-evolutionist discourse: as an example of an "irreducibly complex" system. If it's true that a TTSS is somewhat like a flagellum but much simpler, but still useful, then the alleged irreducible complexity of the flagellum is dead. That's the main point here.

Have you read the work of Milton Saier at UCSD? Are you aware that Saier's work suggests that the TTSS evolved from the flagellum rather than into it? Wouldn't you agree that the challenge of evolution is to explain how you get complex systems from simpler ones and not vice versa? Thus, if Saier is right, wouldn't you agree that to explain the TTSS as evolving from the flagellum is only of limited evolutionary interest and that it leaves untouched the evolution of the bacterial flagellum in the first place? I'm not familiar with Saier's work. Google to the rescue again ... well, unfortunately it almost entirely turns up ID proponents' webpages saying that Saier's work has this consequence, but nothing useful from Saier himself or explanations detailed enough for me to have any chance of assessing how compelling the argument is. So I can't comment on this, beyond saying two things. Firstly, that if indeed the TTSS evolved from the flagellum and not vice versa then their similarity is not helpful in explaining how flagella came to be as they are; secondly, that the existence of one person who takes a particular view doesn't prove that view correct.

Are you familiar with the writings of James Shapiro (who is on faculty at the University of Chicago) and Franklin Harold (who is an emeritus professor at Colorado State University)? Shapiro is a molecular biologist, Harold a cell biologist. They both claim that there are no detailed Darwinian accounts for the evolution of systems like the flagellum. Do you agree with their assessment? Are there any other evolutionary mechanisms that yield a detailed, testable scenario for the origin of the bacterial flagellum? I am not familiar with their writings. Your claims about their claims are too ambiguous to make any useful reply to: what's meant by "detailed" or "systems like", for instance? It wouldn't greatly surprise me, though, to find that no one has so far produced an account of how a eubacterial flagellum might have evolved, with every step's details filled in. Is that supposed to be a problem?

Theodosius Dobzhansky, one of the founders of the neo-Darwinian synthesis, remarked toward the end of his life that nothing in biology makes sense except in the light of evolution. Do you accept this statement? I think it was an exaggeration.

But isn't it the case that for systems like the bacterial flagellum, evolutionary biology has no clue how they came about? So was Dobzhansky wrong? There are things for which we don't have a detailed evolutionary explanation. That seems to me to be entirely as we should expect even if present-day evolutionary theory is right in every particular. As I said above, I think Dobzhansky exaggerated; but I don't see how the existence of things for which we don't currently have an evolutionary explanation would refute his statement, so that second question appears to me to be a non sequitur.

Testability of evolutionary theory

Earlier you expressed reservation about ID being testable. Do you also share such reservations about the testability of evolutionary theory? No? Could you explain how evolutionary theory is testable? What sort of evidence would count against evolutionary theory? I expect some assertions about evolution that would count as part of "evolutionary theory" aren't very testable; see my remarks above about how testability is an attribute of systems and of people, and not only of individual propositions. However, I do think that evolutionary theory is on the whole much more testable than "intelligent design theory". Here are some imaginable observations that would, for me, be strong evidence against evolutionary theory as it stands at present.

The evolutionist J B S Haldane once remarked that what would convince him that evolutionary theory was wrong was finding a rabbit fossil in Precambrian rocks. Would such a finding convince you that evolutionary theory is wrong? And wrong in what sense? Would it show that common descent is wrong? If such a fossil were found in Precambrian rocks, why not simply explain it as an evolutionary convergence? If there were just a single finding, I'd probably think hoaxing or previously unknown geological processes a more likely explanation. But suppose there were multiple discoveries by different teams without any obvious axes to grind, always rabbits, always in the Precambrian. I'd regard that as a good sign that something was wrong with our present understanding, and I'd be wondering whether everything we currently think we know about evolution is rubbish somehow. Yes, it could conceivably be some sort of convergent evolution, but these rabbits would be the only organisms of their kind in the Precambrian, and (in the scenario I'm talking about, with multiple discoveries) it would be very strange to find all these Precambrian rabbits and no sign of their predecessors or successors. Yes, a failure of common descent would be one thing to consider in that situation.

Suppose we bracket the issue of common descent and accept, for sake of argument, that all organisms trace their lineage back to a last universal common ancestor. In that case, why should we believe that natural selection and random genetic change is the principal mechanism driving biological evolution? Is that claim testable? We should believe that, or at least something approximating to it (see above, where I answered a similar question already), because random genetic change and natural selection are known to happen; because they are clearly capable of explaining rather a lot of what we observe about the nature of life on earth; and because no other mechanism of similar explanatory power and consistent with the evidence is known. ("Neutral drift" might be another such mechanism; again, I've commented briefly on it above.) Of course, it might turn out that there are other natural mechanisms that we haven't found yet, and we should be prepared to accept that if we find them. And it might turn out that there are supernatural mechanisms too, such as divine intervention; but they are not something science can deal with, so when doing science we should work on the assumption that such mechanisms aren't operative.

Darwin in his Origin of Species remarked that if it could be demonstrated that some complex structure "could not possibly" have come about "by numerous successive slight modifications" then his theory would absolutely break down. But he hastened to add that he could think of no such case. But how is restricting evolutionary paths as proceeding by "numerous successive slight modifications" any restriction at all? How could the claim that some system did not evolve by numerous successive slight modifications ever be tested? Please describe in detail how this possibility could be tested. If it cannot be tested, then how can evolutionary theory be regarded as scientific? That's a curious question for an ID proponent to ask, since "ID theory" purports to show that certain things could not possibly have arisen by numerous successive slight modifications, or at least could only have done so by a succession of extremely low-probability events. I would think it would be very difficult to demonstrate that something couldn't have arisen in such a way. That doesn't mean that evolutionary theory isn't testable, it means that one imaginary way of testing it isn't available. A more likely mode of falsification would be showing that something didn't (rather than couldn't) emerge by successive slight modifications. There are others; see above.

Do you accept that there are other mechanisms involved in biological evolution besides natural selection and random genetic change? If so, how do biologists known that the totality of these mechanisms account for all of biological complexity and diversity? Is the claim that these mechanisms account for all of biological complexity and diversity itself testable? Have you tested it? How so? How can it be tested? If it should be tested and disconfirmed (as can always happen to testable propositions), then what is the alternative hypothesis that correspondingly is confirmed? Wouldn't it have to be a design hypothesis? If not, why not? There are other such mechanisms, such as horizontal gene transfer and artificial selection. Biologists don't know, and generally don't pretend to know, that all the mechanisms of evolutionary change are known. The claim that they are is nether testable nor scientific, which would be a problem if evolutionary theory depended on its being scientific. But it doesn't. (We also don't know that there are no fundamental forces other than gravity, the electroweak force, and the strong nuclear force. That doesn't stop physics as currently practiced being scientific and testable, leaving aside the usual doubts about string theory.) A testable theory doesn't have to have a specific alternative hypothesis; something could be testable but also be the only coherent theory we have. An alternative to orthodox evolutionary theory wouldn't have to be a design hypothesis; for instance, some form of Lamarckism might be such an alternative without "design" being involved.


Perhaps I'm fooling myself, but I don't seem to have been tied in knots here. If Dembski's questions have revealed contradictions in my thinking, I'm unable to spot them; if they've shown me up as a bigot, an elitist or a closet ID theorist, then again I'm apparently too obtuse to see.

What am I?

Well, what is Dembski claiming I should turn out to be? He lists three kinds of "Darwinist", distinguished oddly enough not by their opinions about evolution and natural selection or even their opinions about ID, but by their opinions about religion. (But, of course, ID has nothing at all to do with religion; whyever would anyone think it had?)

Of Dembski's descriptions, that of the "Kenneth Miller Darwinist" is the nearest to describing me. So, I ought to be revealed as a "closet ID theorist" by Dembski's questions. It doesn't seem that way to me. Here, for definiteness, are my opinions on ID; I don't think they are the opinions of a closet ID theorist, and they seem entirely consistent with my answers above. "Intelligent design" is a slippery term, so I'll have to distinguish between several different things it might mean.

Dembski is much exercised about whether his victims consider ID to be "science" or not, and about whether they consider it to be "creationism" or not.

Is it science? It's hovering somewhere near the border between non-science and really bad science. When the question is "Could X have arisen through the known evolutionary mechanisms?", what's being done is generally science; unfortunately it's extremely bad science, and the arguments presented are woeful. When the question is "Does X in fact show that Y arose by the action of an intelligent being?", what's being done is generally not science. (And the arguments presented are just as woeful as before.)

Is it creationism? No, it isn't. But it's being done (almost exclusively) by creationists, and just about everyone who finds "ID theory" credible appear to do so because of their prior creationist convictions. There are a few exceptions, which might be a problem if I were claiming that ID simply is a form of creationism; I'm not.

I would say the same thing, by the way, about much of what's called "creation science". It isn't impossible to believe (say) that the second law of thermodynamics makes evolution impossible without being a creationist, and probably there are some non-creationists who believe just that. But the almost invariable motivation of those who present such arguments is a desire to bolster creationism. So it is with "intelligent design theory".

So, Dr Dembski, am I a "closet ID theorist"?

What's Dembski trying to do?

Although Dembski claims that ID is not fundamentally about religion, it's interesting that

The real aim of the "Vise strategy" is not to "squeeze the truth out of Darwinists"; it is to play to the jury and take advantage of the fact that juries in US courts usually consist mostly of Christians. So Dembski wants atheists to be shown up as such, and he wants to give the impression that to be anything other than an atheist is to be pro-ID.

Ruminations on Dembski's questions and comments

It surprised me slightly how often Dembski's proposed line of questioning assumes that the victim has answered in particular ways, and how consistently he is wrong when I'm the victim. I expect this is partly because he really believes that all opponents of ID think in a particular way, and partly because he wants to illustrate the sorts of things you can do and isn't bothered by the fact that his questions won't always be appropriate ... but I also suspect that part of this is mere wish-fulfilment fantasy: he likes to imagine having the poor idiot Darwinist on the ropes. Ah, well.

More than once, Dembski uses the following line of attack: (1) "Please list all the Xs you can think of." (2) "Are you absolutely certain that that's all of them?" (3) "If you aren't absolutely certain, how can you be sure that there isn't an ID-ish X?" This seems to me absolutely stupid.

More generally, again and again Dembski seems to confuse lack of perfect knowledge with weakness of theory. Do we know exactly how life began? No, so the idea that it might have happened naturally is bogus. Do we know exactly how bacterial flagella evolved? No, so we might as well take them to have been designed. Do we know exactly where the Cambrian fauna came from? No, so probably they just sprang out of nowhere. This may be a good debating tactic; perhaps it plays well with a jury; but, again, it's absolutely stupid.

Dembski consistently uses the term "materialism" rather than "naturalism" when talking about, e.g., "methodological naturalism/materialism". I think this is because he wants to portray naturalists (of whichever sort) as bigoted rationalists, and "materialism" sounds worse than "naturalism". But, unfortunately for Dembski, "naturalism" is a much better term for the position involved. I've used "naturalism" in my answers.

Dembski repeatedly tries to foist onto his victim the idea that non-naturalist explanations are not only unscientific but also necessarily wrong, and he represents (orthodox) science as adopting naturalism as an axiom and not merely a methodology. I can see that he has tactical reasons for doing this, but it's rubbish.

Dembski more than once shows that he thinks he's proved something useful if he gets his victim to admit that there are imaginable future circumstances in which something a bit like ID might be shown to be right. For instance, if abandoning methodological naturalism turned out to be more effective scientifically and ID were somehow "developed into a full-fledged science". (The implication being, of course, that what it is now is at least a fledgeling science. Sorry, Bill, but it ain't so.) It seems to me that this is entirely uninteresting; These circumstances are hypothetical, and the question is whether ID now is in any way scientific, not whether something in the future that resembles ID as chemistry resembles alchemy might be scientific.