Craig on the teleological argument

This post is a continuation of the discussion of Craig’s arguments for the existence of God that I began here.

William Lane Craig, Dembski’s explanatory eilter, and anti-evolutionism

I titled the post as a whole “Craig on the Teleological Argument,” but partly this has to be a discussion of Craig and the Intelligent Design movement, since Craig has made a decision to build his version of the teleological argument on the ideas of prominent anti-evolutionist/“Intelligent Design theorist” William Dembski. Craig’s version of the argument appeals to the alleged “fine-tuning” of various features of the physical universe (such as the strength of the force of gravity), and comes in this logical form:

1) The fine-tuning of the universe is due to either physical necessity, chance, or design.
2)It is not due to physical necessity or chance.
3)Therefore, it is due to design.

The inspiration for this way of stating the argument Dembski’s idea of the explanatory filter. Dembski claims that the main way we infer something is designed by ruling out the alternative explanations of physical necessity and chance.

In Reasonable Faith, Craig de-emphasizes his reliance on Dembski, but in other contexts he had leaned heavily on Dembski’s work. For example, Craig has claimed that some criticisms of the teleological argument fail because they don’t appreciate the power Dembski’s ideas. Also, Craig has long had an awkward relationship to evolution, and Reasonable Faith provides some clear examples of this. So it’s worth saying something about Craig’s use of Dembski’s ideas and his relationship to anti-evolutionism in general

To get a better sense of Craig’s position on intelligent design, here’s how Craig began his opening statement in his debate with Francisco Ayala on the validity of intelligent design (from the 26:50 mark on):

In any debate it’s critical that we begin by clearly defining our terms, and this is especially important with respect to tonight’s topic, because there is such widespread misunderstanding of what Intelligent Design is. Taken in its broadest sense, ID is the study of justifiable design inferences… ID theory is applicable in a broad variety of fields, for example cryptography, forensic science, intellectual property protection, the search for extraterrestrial intelligence, and so forth… Undoubtedly one of the most sophisticated accounts of design inferences comes from the mathematician William Dembski, in his book The Design Inference published by Cambridge University Press… Dembski and other ID theorists have made the controversial claim that a design inference is warranted in the field of biology…

Here, Craig tries to counteract negative perceptions of the Intelligent Design movement by defining it the way that casts the movement in the best light. He also does his best to link ID to mainstream intellectual endeavors like forensics and SETI, and hints that ID is only controversial when applied to biology. But any honest look at the ID movement reveals that anti-evolutionism is not one controversial consequence of ID, but the main driving force behind the movement.

Furthermore, the claim that ID captures how mainstream scientists work has been rejected by those scientists. Archaeologist and forensic scientist Gary S. Hurd has described how Dembski incorrectly dismisses as secondary considerations that are central to inferring intentional agency in Hurd’s areas of expertise. For example, to determine if something is a man-made tool, archaeologists rely on studies of present-day tool use and studies of the materials the objects are made out of. One interesting example that Hurd gives is that of primitive hammers called “hammer stones.” When first created, there is no way to identify a particular rock as a hammer stone; archaeologists identify them mainly based on known patterns of wear-and-tear found on used hammer stones.[1] Similarly, SETI researchers have pointed out that they look for signs of intelligence not based on Dembski’s criterion, but based on what kinds of signals are known to be produced by various kinds of sources. Notably, both archaeologists and SETI researchers are using information that is unavailable in Dembski’s inference to God.

Furthermore, Dembski’s status as a mathematician is not so great as Craig would probably like his audience to believe. While Dembski’s first Ph.D. was indeed in mathematics, one attempt to discover what mathematical papers Dembski has published found only a single publication.[2] Worse, on close examination, Dembski’s use of mathematical ideas tends to be inconsistent or flat-out wrong, as in Dembski’s equivocations between different senses of “information” and “complexity.”[3]

I have to say I’m a bit surprised at Craig’s frequent endorsements Dembski in recent years, given his usual desire to position himself as a mainstream scholar and distance himself from creationism. In the previous edition of Reasonable Faith, he recommended to his readers that they actively distance themselves from creationism. Similarly, in the third edition he commends the fine-tuning argument to his readers because it avoids “the emotionally loaded question of biological evolution” (p. 193).

On the other hand, Craig has shown a willingness to repeat creationist canards when it suits him. I mentioned a couple examples of this in my review of the second edition of Craig’s book, though these seem to have been removed from the third edition. The third edition, though, has examples of this tendency–really the line about “the emotionally loaded question of biological evolution” is one of them, since it tries to give the impression that scientists who reject creationism do so for emotional reasons and hopes readers will miss the fact Craig is ignoring the massive amounts of evidence for evolution. While evolution is an emotional issue for many scientists, the reason for this isn’t what Craig wants his readers to think. Richard Dawkins (a prime example of the kind of scientist Craig likes to take shots at) put it nicely:

Imagine yourself a classical scholar who has spent a lifetime studying Roman history in all its rich detail. Now somebody comes along, with a degree in marine engineering or mediaeval musicology, and tries to argue that the Romans never existed. Wouldn’t you find it hard to suppress your impatience?

I’d add that the frustration goes beyond the simple fact of being contradicted by people with no expertise in your field. The fact of the matter is that the creationist literature is full of factual mistakes obvious to anyone with a basic knowledge of the relevant science. Worse, some falsehoods continue to reappear no matter how many times scientists publicly correct them. Under such circumstances, it’s hard for scientists not to question the honesty of prominent creationists.

Similarly, the beginning of Craig’s discussion of the teleological argument says:

The explanatory adequacy of the neo-Darwinian mechanisms of random mutation and natural selection with respect to observed biological complexity has been sharply challenged, as advances in microbiology have served to disclose the breathtaking complexity of the mirco-machinery of a single cell, not to speak of higher level organisms. The field of origin of life studies is in turmoil, as the old scenarios of the chemical origin of life in the primordial soup have collapsed, and no new, better theory is on the horizon…

…Due to sociological factors surrounding the neo-Darwinian theory of biological evolution… biologists have been for the most part extremely loath so much as to even contemplate a design hypothesis, lest they let a creationist foot in the door…

Based on what actual scientists have had to say on the matter, I have to say their reluctance to take design seriously seems based on the flimsiness of the reasons given for the design hypothesis. Most scientists regard Michael Behe’s irreducible complexity claims (which Craig’s line about “advances in microbiology” appears to be a reference a reference to) as based simply on ignorance of current science, which Behe has revealed by his demonstrably false statements on the matter. Similarly, nothing I’ve read by scientists who actually know something about origin of life studies suggests the field is in turmoil or lacks any plausible hypotheses to investigate. If Craig is right about the state of the field the experts have failed to notice.

A non-Dembskian fine tuning argument?

For these reasons, Craig’s endorsements of the ideas of the anti-evolution movement are a disaster, but as mentioned above, Craig de-emphasizes his reliance on Dembski. In fact, he claims his first premise (“The fine-tuning of the universe is due to either physical necessity, chance, or design”) should be uncontroversial, because physical necessity and chance exhaust the alternatives to the design hypothesis. However, it is hard to see how this is so unless both “physical necessity” and “chance” are taken in a much broader way than they are normally understood.

For example, while there is an explanation of why the Earth orbits the Sun the way it does in terms of physical laws, it would sound odd to say that it is physically necessary that the Earth orbits the Sun in a particular way. So maybe when Craig says “physical necessity,” what he really means is “any explanation in some way involving physical laws. Again, in his book with Walter Sinnott-Armstrong, Craig says that “chance takes the values of the variables to be brute facts” (p. 63) but “brute facts” are often defined as facts having no further explanation, so if “chance” is understood as a kind of explanation, the “chance” options and the “brute facts” options are incompatible. Finally, Craig’s premise ignores the possibility of something’s being explained by a combination of chance and design. Many criticisms of the theory of natural selection rest on a failure to realize that natural selection relies on both random processes (mutation) and fairly predictable processes (selection).

Perhaps Craig’s first premise should be restated as follows: “fine tuning is either a brute fact or is explained by design or is explained by some hypothesis somehow involving physical necessity and/or chance.” However, if this is what Craig is really trying to say, it is unclear how he proposes to rule out alternatives to design, because “chance” and “physical necessity” both end up being names for a vast array of distinct options, and it is not clear how Craig thinks he can rule out so many different options via a few brief arguments.

Before looking at Craig’s attempts to rule out those alternatives, I want to make one other point about the strangeness of how the fine tuning argument is usually framed (and not just by Craig). Fine tuning is always framed in terms of a few specific features of the universe, which, allegedly, could not have been otherwise if the universe is to be life-permitting. But at most they are necessary for life if the laws of physics have the basic structure which physicists think they have. Presumably, if God exists, then being all-knowing he could have thought up radically different laws of physics which would have allowed life to exist without fine tuning, and being all-powerful he could have willed those laws to be the laws governing our universe. Thus, if God exists, we might easily have not lived in a fine-tuned universe, so the God hypothesis doesn’t seem to really explain fine-tuning.

The issue of fine-tuning would be more interesting if we had some reason to suspect that the universe were under the control of some extraordinarily powerful, but not omnipotent, beings and we had further reasons to suspect it would be easier for those beings to influence the allegedly fine-tuned features of the universe than it would be for them to influence things like the basic structure of our physical laws. The inference that someone really did monkey with the laws of physics would be compelling if that were our situation, but unfortunately it isn’t our situation.

Physical explanation

On the question of physical necessity, his discussion is mostly dismissive. He claims that “on the face of it, this alternative seems extraordinarily implausible.” But scientists are a long track record of discovering physical explanations for previously unexplained phenomena, so what basis is there for dismissing the possibility of a physical explanation as obviously implausible in this case? Craig has little to say beyond a brief discussion of current scientific theories, but the status of current scientific theories has little relevance to the question of whether there is likely to be a good physical explanation of alleged fine-tuning in the future.

Similarly, Craig claims, “The person who maintains that the universe must be life-permitting is taking a radical line which requires strong proof.” Aside from assuming a too-narrow view of physical explanation (as we saw with the example of Earth’s orbit), this claim invites the question of why the God hypothesis isn’t “a radical line which requires strong proof.” Why be so harsh on naturalistic explanations and give theism a pass? Here, Craig seems to be using the same sort of double standard he uses in his appeal to the Big Bang to prove God (discussed in my review).

Craig’s casual dismissal of physical explanations seems especially ill-founded in the case of the initial low entropy of the universe. Since life relies on chemical processes involving transitions from low entropy to high entropy, life could not have existed unless the initial entropy of the universe had been very low. Now, it is one thing to be suspicious of the idea of a physical explanation for a quantity being such-and-such a value, no more, no less, but what’s so odd about the idea that the of a physical reason for a quantity just being very low? The entropy case is interesting because, unlike some of Craig’s probabilities, the probability of the universe having a low initial entropy has a clear meaning: it means the probability on the condition of the matter of the early universe being arranged at random. But what’s strange about the matter of the early universe not being randomly arranged?

It is worth noting that if “physical necessity” just means something like “being entailed by the laws of physics,” then the laws of physics are physically necessary because they entail themselves. This means that if thinks like the strength of the gravitational force are thought of as part of the laws of physics (rather than, as Craig seems to want to think of them independent entities on par with particles and celestial bodies) then the strength of gravity is physically necessary. This trivial truth may not sound like an explanation of anything, but it does highlight the strangeness of trying to explain some features of the universe while ignoring others.

Finally, Craig claims that physical explanations of apparent fine tuning would likely introduce new features of the universe which would, themselves, have to be fine-tuned. Here, Craig is falling into one of the crudest sorts of bad theistic arguments: no matter how well various features of the universe are explained, the theist simply demands another explanation for the explanation, assuming that we’ll have to use God as an explanation sooner or later. But if we are to accept an infinite chain of explanations, why think God need be included in that chain? Schopenhauer famously said that the causal principle is not a hired cab which we may dismiss when we reach our destination; he might have added that the need for explanation is not a hired cab which we can order wherever we want.

Chance and multiple universes.

As for chance, what argument does Craig think he has that rules out both chance and some “brute facts” approach? Craig apparently thinks it’s just obvious that citing long odds against the universe being the way it is is sufficient, but in most cases it is not initially clear what the cited numbers mean. Craig explains that, when he says a life-permitting universe is improbable, he is simple defining probability so that “There are simply a vastly greater proportion of more life-prohibiting universes in our local area of possible universes than there are life-permitting universes.”

Aside from the vagueness of the phrase “local area of possible universes,” I simply fail to see the relevance of probability, so defined, to the problem of understanding why our universe is the way it is. Proponents of the fine tuning argument like to claim that the fine tuning is analogous to some well-known chancy process (a lottery, firing a gun). But being improbable in the sense that Craig defines “probability” just doesn’t entail being improbable in the more familiar sense that picking the correct lottery number is improbable. Craig seems to come very close to asserting “either the universe is designed or the universe’s properties were chosen in a cosmic lottery”–but there’s no reason to think that.

Here, the discussion ends up centering on one kind of chance-involving hypothesis, the many universes hypothesis, which is willing to concede that it is unlikely that any given universe would be life-supporting, but our universe is one of a large number of universes within a larger “multiverse,” and only a minority of these universes are life-supporting. The many universes hypothesis is analogous to the widely-accepted view that it is unlikely that any star system would have a planet of just the right kind, at the right distance from the star, to support life, but our star system is just one of a whole universe of star systems, only a minority of which are life supporting.

Craig’s discussion of the multiverse hypothesis is one of the relatively few places where his argument appear to be based on straightforward scientific ignorance. His primary objection seems to be that “there is no evidence for the existence of a world ensemble apart from fine-tuning itself.” Many scientists disagree. Physicist Taner Edis has been especially blunt on this point:

This is bloody nonsense. Physicists take multiple universes seriously, but not because we necessarily like the idea (I, for one, would be happier avoiding it) or because it would get the intelligent design people off our back. We take it seriously as a possibility, mainly because it’s damn difficult to avoid if you play around with any kind of quantum cosmology. Starting with inflationary cosmologies, we have been playing around with scenarios involving either a vast number of universe-bubbles or huge numbers of at least metastable vacua for some time now. As I said, they’re difficult to avoid—that is, without invoking arbitrary principles for the sole purpose of sweeping multiple universes under the carpet.[4]

In the discussion in Reasonable Faith, Craig actually does discuss the work of one scientific defender of the multiverse hypothesis, Alexander Vilenkin, but persists in insinuating that atheism is the only motivation for the multiverse hypothesis by talking about “Vilenkin’s attempt to explain away the fine-tuning of the universe,” even IVilenkin’s book Many Worlds in One, which provides the basis for Craig’s discussion, shows few signs of a preoccupation with rebutting theism.

Craig’s main argument against Vilenkin’s view is that it requires that events can be simultaneous from one perspective in spite of being not simultaneous from another perspective, and Craig finds this problematic. However, relativity of simultaneity is a central consequences of widely-accepted understandings of Einstein’s theory of general relativity. In some of Craig’s lesser-known works, he has made clear that he rejects standard interpretations of Einstein’s theory in favor of a neo-Lorentzian interpretation, on which there is absolute simultaneity. Perhaps he has good reasons for those views, but in Reasonable Faith Craig simply avoids talking about the issues raised by Einstein’s theory. Thus it is hard to take his criticisms of Vilenkin seriously as written.

Craig’s other complaint about the multiverse hypothesis is that, if the matter in the universe were arranged at random, a deceptive universe would be much more likely than a universe that we could successfully investigate. But why assume that the matter in individual universes would have to be arranged at random if the multiverse hypothesis is true? At most, this point is only problematic if the multiverse hypothesis is supposed to explain the initial low entropy of the universe. It says nothing against the multiverse hypothesis in general.

As with the discussion of physical necessity, Craig’s discussion of the multiverse provides another example of an argument based mainly on a double standard. Craig insists that the multiverse hypothesis is not plausible unless there is a plausible mechanism for generating an ensemble of universes, but what plausible mechanism is there for generating a god?

Final comment

Though Craig’s presentation of the teleological argument lacks the obvious flaws found in his standard presentation of the moral argument, on closer inspection it still compares some of his better thought-out arguments. Craig’s presentation comes off as cobbled together from bits of other people’s presentations of design arguments, and Craig doesn’t seem to have thought carefully enough about how to turn these bits into a coherent argument. I strongly suspect that there are better presentations of the fine-tuning argument available, and I’ve especially heard good things about Robin Collins’ presentation. Since I’m currently reading the Blackwell Companion to Natural Theology, which contains an article by Collins, I expect I’ll have more to say about this in the near-future.

[1] Gary S. Hurd, “The Explanatory Filter, Archaeology, and Forensics.” in Why Intelligent Design Fails ed. Matt Young and Taner Edis. New Brunswick: Rutgers University Press, 2004. pp. 107-120
[2] Jeffery Shallit and Wesley Elsberry, “Playing Games with Probability: Dembski’s Complex Specified Information.” in Young and Edis 2004. p. 129
[3] Ibid., see also Taner Edis “Chance and Necessity—and Intelligent Design?” and Mark Perakh “There Is a Free Lunch After All: William Dembski’s Wrong Answers to Irrelevant Questions” in Young and Edis 2004, as well as various criticisms of Dembski at TalkOrigins.
[4] See also Edis’ books The Ghost in the Universe (Amherst: Prometheus Books, 2002) pp. 89-95 and Science and Nonbelief (Westport: Greenwood Press, 2006) pp. 54-61

Leave a comment


  1. It makes little sense to me to criticize, on the one hand, the multiple universe hypothesis on the grounds that it has no evidence other than fine-tuning, and on the other hand to argue that fine-tuning supports the theistic design hypothesis. This seems to presuppose that other theistic arguments are successful, which I don’t think Craig would want to do here. Besides, the multiple universe hypothesis is much more qualitatively parsimonious than the theistic design one.

    Collins has a rather exhaustive paper defending the teleological argument which I found at Common Sense Atheism: It’s indeed better by far than any other presentation of the argument I’ve encountered. (Nevertheless, I think the reasoning within falls out of some larger errors committed by Bayesians regarding modeling belief states as probabilities.)

  2. Chris Hallquist

    Wow. Luke posts everything on his site. Does he have any respect for copyright law? Not that I’m a saint in that area, but I would never be so blatant about it.

  3. Best article I have found on Dembski is this one by Elsberry and Shallit.

    Collins’ presentation is good but I always wonder whether Bayesian concepts of epistemic (or subjective) probability collapse once we have some data on which to base them.

    For example, is the fact that a friend you were just thinking about happens to ring you epistemically surprising? Maybe if you know nothing about the actual (or theoretical) probabilities involved; maybe not once you do.

    Fine-tuning arguments seem to work on the fact that we don’t really know what the actual or theoretical probabilities involved are.

  4. Chris Hallquist

    Yeah, I just took a look at Collins’ article, and his ideas about how to calculate the relevant probabilities are kinida weird. And he claims that saying “God did it” is automatically not ad hoc because God is an old idea.

  5. “Fine-tuning arguments seem to work on the fact that we don’t really know what the actual or theoretical probabilities involved are.”

    Well, in one sense you can view fine-tuning arguments as attempting to show precisely that we can have an idea of what they are. If there’s no deeper reason why the constants fall in the narrow life-permitting range, it would be “utterly astonishing;” therefore, probably there actually was a deeper reason out there which made the constants likely to have been life-permitting, and the proponents of FTA say this reason is probably a designer’s volition, etc.

  6. Very well written and informative. Seems like this guy is barking up the wrong tree – when it comes to whether physical neccessity, design or chance created a universe that supports life there’s literally nothing to choose between them, and all these arguments are a stretch or flat out wrong.

  7. Nice article. I have bookmarked it, and will return.

Trackbacks and Pingbacks: