Collins’ Case for Cosmic Design (2008)
Robin Collins offers three design arguments, one appealing to the existence of intelligent life and the fine-tuning upon which that life depends, one appealing to the beauty of the laws of physics, and one appealing to the intelligibility of the universe. It’s not completely clear what the conclusions of these arguments are. Sometimes Collins seems to be arguing for theism—for the hypothesis that an omnipotent, omniscient and morally perfect person created our universe. Other times he seems to be arguing for a much less specific hypothesis, one stating simply that a transcendent intelligent being (of some sort) designed our universe. I will call this second hypothesis the “generic design hypothesis.” Both hypotheses are mentioned multiple times in his opening case.
Although these two hypotheses are obviously compatible, a rigorous evaluation of Collins’ arguments is impossible if he is allowed to vacillate between them, precisely because the generic design hypothesis is much less specific (i.e., much smaller in scope) than theism. On the one hand, this lack of specificity makes the generic design hypothesis more “plausible” than theism—i.e., more probable than theism independent of the evidence. Thus, it is less vulnerable than theism to the objection that, even if some supernaturalist design hypothesis explains or predicts our total evidence better than naturalism, naturalism is still more probable all things considered because of its greater plausibility. On the other hand, the lack of specificity of the generic design hypothesis is also a handicap. Theism posits a morally perfect designer and thus implicitly attributes certain prima facie preferences to that designer. This makes it possible to construct a plausible argument for the claim that the existence of intelligent life is expected or probable given theism: since intelligent life has special value, a perfectly good God would have reason to create a universe containing it. The generic design hypothesis, however, says nothing about the goals or intentions or preferences of the designer or designers it posits. Thus, it hard to see why the assumption that the generic design hypothesis is true leads to any expectations at all about what the world would be like. In particular, it would seem that, given the fine-tuning data, intelligent life is just as unlikely given design with unspecified motives as it is given “chance.”
Clearly Collins can’t have it both ways. He can’t use the less specific generic design hypothesis when the issue is plausibility and then use the more specific theistic hypothesis when the issue is predictive or explanatory power. Since I believe that the disadvantages of employing the generic design hypothesis in Collins’ arguments are greater than those of employing theism, I will assume in the remainder of this chapter that Collins wants to formulate his three design arguments in terms of theism.
Another worry I have about Collins’ selection of hypotheses concerns the hypothesis to which Collins compares theism. That hypothesis is naturalism in the case of two of his three arguments, but it is “single-universe naturalism” in the case of his main argument: the so-called fine-tuning design argument. By “naturalism” Collins means the hypothesis that physical reality did not result from transcendent (nonnatural) intelligent design. By “single-universe naturalism,” Collins means naturalism conjoined with the view that physical reality consists of nothing more than our own (expanding) cosmos. According to Collins, single-universe naturalism has greater difficulty accounting for the fine-tuning data than (generic) naturalism because the latter hypothesis is compatible with the existence of multiple universes. Notice, however, that in addition to this disadvantage, single-universe naturalism is also intrinsically less probable than naturalism because it is more specific—single-universe naturalism says everything (generic) naturalism says and more; so it is more likely to say something false. Why on earth, then, would any atheist choose to defend single-universe naturalism instead of naturalism? Why saddle oneself with a more specific (and thus intrinsically less probable) hypothesis that actually explains the data worse than a less specific one?
Before I examine the details of Collins’ three arguments, I would like to respond briefly to his claim that his case for cosmic design is as strong as the scientific case for common descent. Some would call this a “bold” claim; others outrageous. But whatever one calls it, I suspect Collins himself would admit that his opening case falls far short of establishing it. It is important, however, to realize just how very far short it falls. For even assuming that his three core arguments are flawless, the best case scenario for Collins is that the various pieces of evidence to which those arguments appeal together raise the ratio of the probability of theism to the probability of single-universe naturalism many-fold. That’s all he can get from an application of the expectation principle or the likelihood principle. To get from there to the more significant conclusion that theism is more probable than single-universe naturalism, all things considered, he would have to perform three other very difficult tasks. First, he would have to show that his evidence isn’t offset or outweighed by the fact that single-universe naturalism is much more plausible than theism. (Of course, scientists rarely explicitly assess plausibility, but that doesn’t mean that plausibility considerations are not crucial in scientific reasoning or that they make no difference to how probable the evidence makes hypotheses like common descent or naturalism.) Second, he would have to show that his evidence favoring theism is not offset or outweighed by other evidence favoring single-universe naturalism, such as the existence of horrific undeserved suffering. And third, he would have to show that his evidence is not offset or outweighed by a combination of plausibility considerations and other evidence. He doesn’t even attempt any of these three tasks in his opening case. So he has not shown that theism is more probable than single-universe naturalism—not even close.
Suppose, however, that Collins somehow managed to perform these three tasks and thus justify the conclusion that theism is more probable all things considered than single-universe naturalism. It still doesn’t follow from that conclusion that theism is more likely to be true than naturalism, since, as mentioned above, naturalism is more probable than singe-universe naturalism. And even if he were able to show that the theism is more probable than naturalism, it doesn’t follow from that conclusion that theism is probably true since theism and naturalism are not contradictories and thus might both be improbable even if one is (many times) more probable than the other. Finally, it would be one thing to show that theism is probably true and quite another to show that it is highly probable like common descent is. The lesson here is that establishing the truth or even probable truth of a substantive positive metaphysical or scientific hypothesis like theism or common descent is an extraordinarily difficult task, much more difficult than showing the probable falsehood of such a hypothesis. Collins is nowhere near to accomplishing such a task.
All of my remarks so far assume that the core reasoning in Collins’ opening case is sound. This reasoning, however, has been challenged in a variety of ways. Some of these challenges are highly technical. I will restrict myself, however, to a single nontechnical challenge that so far as I know is defended here for the first time. I will contend that all three of his arguments commit what I call “the fallacy of understated evidence.” This fallacy (i.e., mistake in reasoning) is committed when one uses some relatively general known fact about X to support a hypothesis when a more specific fact about X (that is also known to obtain) fails to support that hypothesis. For example, a prosecutor might try to mislead a jury by pointing out that the defendant bought a knife just days before the victim was stabbed, neglecting to mention that the knife that was purchased is a butter knife. Of course, in Collins’ case, there is no intention to mislead; he doesn’t realize that he is understating the evidence. But the mistake is the same, whether intentional or not.
Consider first his main argument, the fine-tuning design argument. Collins summarizes this argument as follows:
These cases of fine-tuning presented above have often been cited as providing significant evidence that the cosmos is designed. The reason is that, because of the exceedingly special conditions required for the existence of life, it seems very improbable or surprising that the initial conditions, laws, and constants would be adjusted just right for highly complex life under what I call the naturalistic single-universe hypothesis, but not surprising under theism. Thus, the fine-tuning provides significant evidence for theism over the naturalistic single-universe hypothesis.
This argument commits the fallacy of understated evidence. Specifically, by understating what we know about life, Collins makes the fine-tuning data appear to support theism more than it really does. Notice that his evidence statement does include one detail beyond the bare fact that (complex) life exists, namely, that some of this life is intelligent. This serves to strengthen his case because, given that life exists, one has more reason on theism than on naturalism to expect some of that life to be intelligent. But he ignores other more specific facts about life that we know to obtain, facts that I contend favor single-universe naturalism. For example, while he bases his argument on the fact that complex intelligent life of some sort exists in some universe or another, he ignores the fact that the only sort of intelligent life we know to exist is specifically human and exists specifically in this universe.
This is important. To see why, suppose that Collins is right that the existence of intelligent life of some sort in some universe is very many times more probable on theism than on single-universe naturalism. This will be of no significance whatsoever if it is also true that
D: given that intelligent life of some sort exists in some universe, the fact that the only intelligent life we know of is human and that it inhabits this universe is very many times more probable on single-universe naturalism than it is on theism.
If D is true (or even if it just can’t be shown to be false), then the more fully stated evidence about intelligent life won’t (clearly) favor theism over single-universe naturalism. Similarly, suppose that the defendant’s recent purchase of a knife is more likely on the hypothesis (which I will call “GUILT”) that the defendant stabbed the victim than on the hypothesis (“INNOCENCE”) that he did not stab the victim. This is of no consequence if we also know that the knife he purchased is a butter knife and that, given that he did buy a knife, it is more likely on INNOCENCE than it is on GUILT that the knife is a butter knife. But is D true or at least not provably false? There is very good reason to believe so. To begin with, given the existence of intelligent life, it is certain on single-universe naturalism that it exists in this universe. This is not so on theism, which as Collins admits, makes the existence of many universes likely.
More importantly, if theism is true, then God is not only morally perfect, he is omnipotent and so could make many different sorts of intelligent life, probably infinitely many, including intelligent beings that are much more impressive than human beings. On single-universe naturalism, by contrast, one would expect that, if there is intelligent life, it will be relatively unimpressive. I want to emphasize the word “relatively” here, because I am not denying that human beings are impressive in many ways. But examined from the perspective of what is possible for an omnipotent being, we are, in terms of intelligence, a hair’s breadth away from monkeys. Again, one would expect this on single-universe naturalism because the more intelligent the life, the less likely it is that naturalistic processes would produce it. Of course, if one believes in God and, looking around, finds nothing more impressive than human beings, one will be forced to conclude that God wanted to make beings with very limited intelligence. But surely one would not have predicted this beforehand. There are indefinitely many different kinds of creatures that an omnipotent being would have the power to create and that, other things being equal, would be more valuable to create than humans. Antecedently, a God would be more likely to create these more impressive creatures than to create us.
One might object, however, that a good God would not be obligated to create the best and that a loving God might very well want to create and love inferior beings like us, especially since that doesn’t preclude his also creating other more impressive beings. I don’t deny that a God might create beings like us—that is certainly possible. Similarly, Collins admits that fine-tuning is possible even if single-universe naturalism is true. The issue, however, is what is antecedently likely—what a reasonable person would expect beforehand. And human beings have many features that make them an unlikely choice, no matter how many other sorts of beings God creates. This is especially true, if we take the term “human” not merely in the biological sense but in a fuller sense that implies some of our most notable and notorious characteristics. In this sense of the term “human,” the sense intended when someone says “I’m only human,” being human implies being naturally selfish (not to mention territorial and aggressive), which greatly limits our potential for developing morally, especially given our limited life span. It also includes the fact that we are profoundly ignorant beings, especially when it comes to moral and religious matters, as is obvious from the fact that we disagree or are uncertain about many important moral and religious issues. We also naturally identify with others that we perceive to be like ourselves, leading, if not always to prejudice and intolerance, at least to isolation for those different from the norm. (I could go on and on, but it’s too depressing.)
Now I don’t mean to claim that there is no good in humanity, that we are not wondrous simians in many respects. But again, when evaluated in the light of what is possible for a being that literally has no nonlogical limits to its power, we hardly belong on any list of “creatures a God would be most likely to create.” One might respond that God would be likely to create us precisely because of our inferiority, precisely because we would require a relationship with God in order to achieve our greatest good, in order to be “saved.” But we have no antecedent reason to believe that is the case. In fact, we have two antecedent reasons to believe or at least suspect that this is not the case. First, a morally perfect God would have no need to glorify herself by creating deeply flawed beings just so that she could play the role of savior. Second, it is doubtful that an omniscient (and morally perfect) God could have a meaningful personal relationship with human beings—not because of God’s limitations but because of ours. The “cognitive distance” between the mind of such a God and our minds is vastly greater than the distance between our minds and the minds of earthworms (assuming earthworms have minds). So who knows whether a relationship with such a being would be our “greatest good”? (It seems rather more likely that the greatest good of animals like us would be much more “down to earth.”) Granted, a God would be omnipotent as well, but not even an omnipotent being can bring about logically impossible states of affairs like my having a meaningful relationship with an earthworm. In any case, given our ignorance of what a mind like God’s would be like, there doesn’t seem to be any way to settle this issue, which suffices to rebut the objection that a God would be likely to create inferior intelligent beings like us.
To sum up, my main point is that, while it may be true that on single-universe naturalism the existence of anything as impressive as human beings is very unlikely, it is also true that on theism the existence of intelligent beings as unimpressive and flawed as humans is very unlikely. Further, given that human beings do exist, it is certain on single-universe naturalism, but not on theism, that they exist in this universe (i.e., in the one universe that we know to exist). Therefore, Collins’ fine-tuning design argument is unconvincing. To repair it, he would need to show that the availability to an omnipotent God of infinitely many apparently better alternatives to creating human beings doesn’t make the existence of human life at least as improbable given theism as it is given single-universe naturalism. Collins does not show this; he doesn’t even address the issue.
Collins’ other two design arguments are no more convincing than his fine-tuning argument. The first is an argument from the beauty or elegance of the laws of physics. Unfortunately, short on space, Collins doesn’t tell us what it means for a mathematical equation to be “beautiful” or “elegant.” If it just means simpler, then such beauty is not surprising on naturalism. Typically, the more uniform nature is, the simpler the mathematical equations that can describe it. And uniformity is intrinsically more probable than variety or change whether or not naturalism is true. But putting that aside, even if some aspects of the universe are quite beautiful, it certainly doesn’t appear to be as beautiful as one would expect if God made it. For example, while it has an abundance of visual beauty, it contains relatively little auditory and tactile beauty. Again, if one wants to construct a truly convincing “argument from beauty,” then one must not understate the evidence. One must not just focus on known facts about beauty that favor theism and ignore other things we know about beauty that favor naturalism.
Collins’ third argument also understates the evidence. I agree with him that a highly intelligible universe is somewhat more likely given theism than it is given naturalism. But here again this evidence is offset by other more specific evidence about the same topic. Specifically, I argued at the beginning of the general introduction to this e-book that, given that the universe is comprehensible, the fact that so much about it can be understood without any appeal to supernatural agency is much more likely on naturalism than on theism. So the fully stated evidence concerning the intelligibility of the universe doesn’t seem to support either theism or naturalism over the other.
I conclude that none of Collins’ three arguments is convincing. Each understates the evidence in a way that makes what we know about some topic appear to favor theism more than it really does. Further, as I argued in the first part of this essay, even if Collins’ arguments were convincing, it would not follow that there is “significant evidence that the cosmos is designed,” for whether the evidence in question is significant (as opposed to just strong or potentially significant) depends on whether that evidence makes theism or the generic design hypothesis credible or at least more credible than its denial, and, as I have shown, that in turn depends on a variety of issues that Collins does not address. Granted, he cannot be expected to cover all of these issues (e.g., plausibility, other evidence, alternatives to theism and naturalism, etc.) in a short paper, but then he should be willing to admit that whether or not his evidence is significant is an open question.
Continue the Debate
 For an explanation of how probability depends on specificity or scope, see section 2 of my opening case, “Natural Selection and the Problem of Evil.”
 One might object that we can sometimes detect human design without knowing what the designer’s goals were. My reply is that we can do this because we know by experience that humans often design objects composed of metal, stone, etc. that have gears or smooth surfaces or other distinctive features. Given this background knowledge, one can infer that an object found in an area currently or previously inhabited by human beings was made by human beings, and one can make this inference without knowing what the purpose of the object is. This background knowledge is of no use, however, when the designer in question is not only nonhuman, but transcendent.
 I don’t mean to imply here that Collins actually is trying to have it both ways. He completely ignores the issue of plausibility in his opening case.
 He says he believes this claim but does not say that he has established it.
 The argument has other problems as well. For example, who is to say that a more complicated mathematical structure would not be even more beautiful than a simple or “elegant” one? Also, a simpler universe seems more likely on naturalism than on theism; so if there is beauty in simplicity, it’s hard to see why that is evidence favoring theism.
 I am grateful to Robin Collins and William Hasker for helpful email discussion of some flawed ancestors of the points I make in this paper.
Copyright ©2008 Paul Draper. The electronic version is copyright ©2008 by Internet Infidels, Inc. with the written permission of Paul Draper. All rights reserved.