Home » Library » Modern Library » Naturalism, Fine-Tuning, and Flies

Naturalism, Fine-Tuning, and Flies



Discoveries in cosmology over the past several decades have revealed that if the values of the physical constants and the initial conditions of our universe were just slightly different, our universe would be uninhabitable to life. For example, Newton’s law of gravitation is expressed by the equation F=G(m1·m2/r2). G represents the gravitational constant, which takes the value 6.673 x 10-11 N · (m/kg)2. If G were just slightly larger or smaller, the universe would either collapse in on itself or blow apart, making life impossible.[1] G is therefore said to be “fine-tuned.” Cosmologists now think that several constants, as well as the universe’s initial conditions, are similarly fine-tuned.[2] In the words of Stephen Hawking and Leonard Mlodinow:

The laws of nature form a system that is extremely fine-tuned, and very little in physical law can be altered without destroying the possibility of the development of life as we know it. Were it not for a series of startling coincidences in the precise details of physical law, it seems, humans and similar life-forms would never have come into being.[3]

Many theists have argued that these findings reveal divine design in the cosmos. For example, apologist Hugh Ross writes: “Does the fine-tuning imply purposeful design? So many parameters must be fine-tuned and the degree of fine-tuning is so high, no other conclusion seems possible.”[4] And it’s not just apologists who see teleological implications; cosmologist Paul Davies writes: “It seems as though somebody has fine-tuned nature’s numbers to make the universe…. The impression of design is overwhelming.”[5]

The fine-tuning argument (FTA) for theism is typically presented as a Bayesian argument—one that compares the likelihood of actualizing a life-permitting universe (LPU) under theism (T) to that under naturalism (N).[6] A standard presentation of the argument goes something like this:

  1. P(LPU/N) << 1.
  2. ~P(LPU/T) << 1
  3. Therefore, LPU favors T over N[7]

In other words, according to the FTA, the existence of a life-permitting universe is very improbable if naturalism is true, but not very improbable if theism is true, and therefore the fact that our universe supports life increases the probability of theism relative to naturalism. In this paper I aim to challenge the first premise of the FTA.

Is the Existence of a Life-Permitting Universe Very Improbable on Naturalism?

Is P(LPU/N) low? Several philosophers have expressed skepticism that it is. Timothy and Lydia McGrew, for example,
have argued that the notion of probability is meaningless in this context because the conceptual space of possible universes cannot be normalized; that is, the probabilities of each conceivable universe do not add up to 1.[8] For the sake of argument, however, I will assume that this problem can be solved and instead focus on another problem with the claim that P(LPU/N) is low.

At the outset, it is important to note that the need for fine-tuning is relative to our specific laws of physics. As Richard Swinburne writes: “Given the actual laws of nature or laws at all similar thereto, boundary conditions [and physical constants] will have to lie within a narrow range of the present [values] if intelligent life is to evolve.”[9] Consider the gravitational constant again. The FTA takes the equation for the force of gravity—F=G(m1·m2/r2)—and asks what would happen if we altered G. But why not alter the rest of the equation as well? Why not consider alternative universes with different laws altogether? If we really want to play “what if?” and compare our universe against all theoretically possible universes, then we should also consider alternative universes with different laws entirely. After all, our laws are no less contingent than the values of their constants.[10]

Life-permitting universes may be rare among the limited subset of possible universes that share our laws, but who knows whether they are rare among the total set of possible universes. In other words, even if this ratio is small:

possible life-permitting universes with our laws / total possible universes with our laws

…it doesn’t follow that the following ratio is also small:

possible life-permitting universes / total possible universes

But to justify the claim that P(LPU/N) is low, we need to establish that the second ratio is small.
For if life-permitting universes took up a large portion of the space of total possible universes, then the fact that our universe supports life wouldn’t be all that surprising. This would be true regardless of whether life-permitting universes were rare within the small subset of possible universes that share our laws.

FTA proponents recognize both that the argument only considers a small subset of the total number of possible universes, and that they don’t know what would happen under different laws. For instance, William Lane Craig writes:

Maybe in a universe governed by different equations, the gravitational constant G could have a greatly different value and yet life still exist… [but] the correct formulation [of the FTA] concerns universes governed by the same laws of nature as ours, but with different values of the constants and quantities. Because the equations remain the same, we can predict what the world would be like, if, say, the gravitational constant were doubled.[11]

Or consider this comment by Robin Collins:

Our physics does not tell us what would happen if we increased the strong nuclear force by a factor of 101,000. If we naively applied current physics to that situation, we should conclude that no complex life would be possible because atomic nuclei would be crushed. If a new physics applies, however, entirely new and almost inconceivable effects could occur that make complex life possible.[12]

What, then, justifies limiting the FTA to the subset of possible universes that share our laws? As Collins explains, doing so is simply a matter of convenience:

[T]he justification for varying a constant instead of varying the mathematical form of a law in the fine-tuning argument is that, in the reference class of law structure picked out by varying a constant, we can make some estimate of the proportion of life-permitting law structures. This is something we probably could not do if our reference class involved variations of mathematical form.[13]

Here Collins is basically conceding that we have no idea whether or not life-permitting universes are rare among the total set of possible universes. We can only say that they are rare among a small subset. An atheist would naturally object that while it might be pragmatically convenient for the FTA advocate to focus on this subset, doing so robs the FTA of its significance. How can we attach any weight to the conclusion of the argument if it ignores so much relevant information?

Leslie’s Fly on the Wall Analogy

Theists have typically tried to answer this question by appealing to an analogy introduced by John Leslie.[14] Imagine that you’re in a dark room and you illuminate a limited portion of the wall with a flashlight. In the middle of the illuminated portion is a fly. Suddenly a shot rings out and a bullet kills the fly. Because the fly took up such a small portion of the illuminated space, you assume that this bullet wasn’t fired at random. It must have been aimed at the fly. Of course, it’s possible that the portion of the wall that you can’t see is absolutely covered with flies, such that it’s not unlikely that a bullet fired at random would hit a fly. Nonetheless, it is still rational to think the bullet was aimed, even though we cannot see the whole wall, given that the fly was such a small target. Collins concludes: “it seems intuitively clear that the [bullet] hitting the [fly] in this case does confirm the aimed hypothesis over the chance hypothesis.”[15]

Just as we cannot see the whole wall, the argument goes, we cannot “see” the whole set of possible universes. However, just as we can infer design when a target is hit in the illuminated portion of the wall, we can similarly infer design when a “target” (i.e., the life-permitting range of values) is “hit” in the subset of possible universes that we can examine.

I do not find this argument to be particularly persuasive. I will admit that we are justified in inferring design in the fly analogy. But why are we justified in inferring it? The reason is that we have good cause to believe that the illuminated portion of the wall is a representative sample of the wall as a whole. But then why can’t we also say that the subset of possible universes with our laws is a representative sample of the set of all possible universes? Because we don’t think that the illuminated portion of the wall is a representative sample simply as the result of an inductive generalization. That is, our reasoning is not:

  1. This part of the wall isn’t covered in flies
  2. Therefore, the rest of the wall probably isn’t covered in flies either.

Instead, we think that the wall probably isn’t covered in flies because our background knowledge includes extensive experience with walls and flies. We know that it is rare for an entire wall to be covered in flies. We also know that flies don’t travel in packs of thousands. Even if we can only see a small portion of the wall, we can safely guess what the rest of the wall looks like. Thus, quite apart from the observation that the illuminated portion of the wall is not heavily populated with flies, we have independent reasons to think the rest of the wall is similar. If we didn’t know anything about how many flies there are in the world, or how flies behave, or if we had never seen any walls before, then we would have very little reason to think that the illuminated portion of the wall was a representative sample of the wall as a whole.

When it comes to the FTA, we don’t have the requisite background knowledge to say that the “unseeable” universes in the set of possible universes are similar to the ones that we can see. We know that flies don’t usually cover entire walls, so we have antecedent reasons to be very confident about what the unseeable parts of the wall look like. We don’t have any such antecedent reasons to build expectations about what the “unseeable” universes look like. We also know that people and guns exist, and that people don’t particularly like flies. By contrast, we don’t already know that there is God who would want to create a life-permitting universe. As Neil Manson writes in his critique of the fly analogy:

This is a fool’s game. We are drawn into it only if we are persuaded by Leslie’s story of the fly on the wall. That story works, however, only because Leslie illicitly imports a perspective. We know how big we are, we know how big flies are relative to us, and we know what it is for an empty area surrounding a fly to be ‘largish’ relative to a fly. There is no correspondingly natural perspective when it comes to surveying the space of sets of possible parameter values.[16]

Keith Parsons makes a similar point about our lack of background knowledge in this context:

The alleged improbability of the cosmic “coincidences” is often illustrated by comparing it to extremely improbable mundane situations—such as every gun jamming at once in a firing squad. But such analogies are not apt. We have prior experience of rifles and their performance and, on the basis of that experience, we know how unlikely it is that, say, ten of them would simultaneously jam. We have no similar experiences that would justify such an inference about the cosmic “coincidences.”… Once this disanalogy is recognized, the “fine tuning” argument loses all of its intuitive appeal.[17]

Collins seems to hint at the idea that his calculations may not be highly significant after all, for he writes that “because we are considering only one reference class of possible law structures… it is unclear how much weight to attach to the values of epistemic probabilities one obtains using this reference class. Hence, one cannot simply map the epistemic probabilities obtained in this way onto the degrees of belief we should have at the end of the day.”[18] This is quite a concession on Collins’ part!

Sample Sizes and Representativeness

Conceivably, a theist might ask: “Why can’t we simply infer that life-permitting universes are rare in the set of all possible universes because they are rare in the subset that we can examine?” To be sure, scientists routinely use subsets to draw conclusions about a set as a whole. For example, a criminologist may study gang members in a California prison and then generalize his conclusions to gang members in California as a whole. The usefulness of this type of method depends on the external validity of the data—its representativeness. A study that uses a large, random sample of the U.S. population is likely to be reliable because the sample is representative of the country as a whole. Unfortunately, practical considerations often limit researchers’ ability to use such an ideal sample. For instance, a psychologist at an Ivy League college may limit her experimental subjects to students in the cafeteria because it’s convenient for her, but Ivy League college students are unlikely to be representative of the population at large.

Thus, practical concerns—such as financial limitations, lack of time, or lack of access—often get in the way of obtaining representative samples. This is precisely our situation when surveying possible universes. It is an unfortunate fact that scientists are severely limited in their ability to determine the habitability of universes with alternative laws of physics. They can only make predictions about universes that share our physics, but vary in the values of their constants. And that does not provide a random sample of possible universes. Rather, it provides a small (indeed, infinitesimally small), homogenous sample, which is exactly what researchers want to avoid. Collins himself says that the significance of the FTA will depend on how representative this subset of possible universes is of the set of total possible universes, yet admits that “there is no completely objective procedure for addressing this [question].”[19]

However, when scientists are selecting a sample class, Collins argues, they do not need to have positive reasons to believe that the class is representative. All that matters is that there is no reason to think that the class is relevantly biased. He writes:

Tests of the long-term efficacy of certain vitamins, for instance, are often restricted to a reference class of randomly selected doctors and nurses in certain participating hospitals, since these are the only individuals that they can reliably trace for extended periods of time. The assumption of such tests is that we have no reason to think that the doctors and nurses are relevantly different than people who are neither doctors nor nurses, and thus that the reference class is not biased.[20]

Collins thinks that using doctors and nurses to represent all people is no different than using possible universes without varying their laws to represent all possible universes. But the two are in fact quite different. First, we don’t trust these studies simply because there is no reason to think that doctors and nurses are a relevantly biased sample. Collins ignores that our background knowledge is filled with facts indicating that doctors and nurses are relevantly similar to the total population. Doctors and nurses are like all humans in that they all evolved from a common ancestor and have the same physiological makeup. This background knowledge about the shared origins and biology of all humans provides positive evidence that a vitamin that has a certain effect on doctors and nurses will have similar effects on other humans. We have no similar background knowledge regarding possible universes that would provide positive evidence that Collin’s sample is representative.

Second, doctors and nurses make up a sample of a finite population. The larger the sample of doctors and nurses, the closer it is in size to the total population. This is why researchers prefer larger samples. The population of possible universes, however, is infinite. No matter how large the sample size of possible universes, it will always be infinitesimally small in comparison to the class of all possible universes.

Thus, even if we were able to determine that life-permitting universes were rare within this sample, we wouldn’t have sufficient reason to generalize from this sample to the set of all possible universes as a whole. For this reason, we are simply not in a position to assess P(LPU/N), and therefore cannot justify the first premise of the FTA.


[1] Robin Collins, “The Teleological Argument: An Exploration of the Fine-Tuning of the Cosmos” in The Blackwell Companion to Natural Theology ed. William Lane Craig and J. P. Moreland (Malden, MA: Blackwell, 2009): 202-281, pp. 214-215.

[2] Some scientists have expressed skepticism over the extent to which the universe is fine-tuned. For example, see: Victor J. Stenger, The Fallacy of Fine-Tuning: Why The Universe Is Not Designed for Us (Amherst, NY: Prometheus Books, 2011); and Fred C. Adams, “Stars in Other Universes: Stellar Structure with Different Fundamental Constants.” Journal of Cosmology and Astroparticle Physics: An IOP and SISSA Journal No. 8, Article 010 (August 7, 2008): 1-28.

[3] Stephen Hawking and Leonard Mlodinow, The Grand Design (New York, NY: Bantam Books, 2010), p. 161.

[4] Hugh Ross, The Creator and the Cosmos: How the Greatest Scientific Discoveries of the Century Reveal God (Colorado Springs, CO: NavPress, 2001).

[5] Paul Davies, The Cosmic Blueprint (New York, NY; Simon & Schuster, 1988), p. 203.

[6] For a discussion of the logic and application of Bayesianism, see Richard Swinburne, Bayes’s Theorem (Oxford, UK: Oxford University Press, 2002).

[7] For example, see Collins, “The Teleological Argument,” p. 207; and Richard Swinburne, The Existence of God (Oxford, UK: Oxford University Press, 2004), p. 189. For alternative formulations of the FTA, see Neil A. Manson, “Introduction” in God and Design: The Teleological Argument and Modern Science ed. Neil A. Manson (New York, NY: Routledge, 2003): 1-23.

[8] Timothy McGrew and Lydia McGrew, “A Response to Robin Collins and Alexander R. Pruss.” Philosophia Christi, Vol. 7, No. 2 (2005): 425-443; Timothy McGrew, Lydia McGrew, and Eric Vestrup, “Probabilities and the Fine-Tuning Argument: A Skeptical View.” Mind Vol. 110, No. 440 (2001): 1027-1038.

[9] Swinburne, The Existence of God, p. 175.

[10] This the crux of David Hume’s problem of induction.

[11] William Lane Craig, Reasonable Faith: Christian Truth and Apologetics (3rd ed.) (Wheaton, IL: Crossway Books, 2008), p. 159.

[12] Collins, “The Teleological Argument,” p. 248.

[13] Collins, “The Teleological Argument,” p. 246.

[14] John Leslie, Universes (New York, NY: Routledge, 1989), p. 17.

[15] Collins, “The Teleological Argument,” p. 245.

[16] Neil A. Manson, “There Is No Adequate Definition of ‘Fine-tuned for Life’.” Inquiry: An Interdisciplinary Journal of Philosophy, Vol. 43, Issue 3 (2000): 341-351, p. 350.

[17] Keith Parsons, “Is There a Case for Christian Theism?” in Does God Exist? The Great Debate ed. J. P. Moreland and Kai Nielsen (Nashville, TN: Thomas Nelson, 1990): 177-196, p. 182.

[18] Collins, “The Teleological Argument,” pp. 240-241.

[19] Collins, “The Teleological Argument,” p. 241.

[20] Collins, “The Teleological Argument,” pp. 245-246.

Copyright ©2015 Aron Lucas. The electronic version is copyright ©2015 by Internet Infidels, Inc. with the written permission of Aron Lucas. All rights reserved.