Unpacking J. H. Sobel's take on fine-tuning
A remarkably poor showing from a respected philosopher
Jordan Howard Sobel (1929-2010) was a widely respected philosopher who made contributions in ethics, logic, decision theory, philosophy of religion, and value theory. Considered one of the best atheistic thinkers, his book Logic and Theism (Cambridge University Press, 2009) remains the pièce de résistance of the philosophical atheist’s toolkit. He was no Richard Dawkins: he engaged with the issues deeply, and wasn’t afraid to interact with theism’s best defenders.
A section of Logic and Theism addresses the fine-tuning argument. The fine-tuning argument, for those unaware, infers the likelihood of theism from what we shall call the “fine-tuning fact”—that many of the physical laws, constants, and initial conditions of the universe fall into an extremely small range of life-permitting values relative to the range of possible values they could have taken. Robin Collins and Luke Barnes are two of the foremost defenders of the argument today, and many atheists take it seriously.
Sobel’s comments on the fine-tuning argument are remarkably poor for a philosopher of his caliber. And though I am a theist with a vested interest in the success of the fine-tuning argument, I don’t think all his errors are in one direction: he makes an error that seriously handicaps his own view.
First, Sobel seems to think that the knowledge of physics required to verify the fine-tuning fact is a problem. Note that Sobel previously distinguished the “1776 version” of the argument from design (biological complexity) and the “millennial version” (the cosmological fine-tuning argument), and that Cleanthes is a character in one of David Hume’s Dialogues Concerning Natural Religion:
There is a remarkable difference between the 1776 version and the millennial versions of the argument from design. The evidence for design in 1776 was for all to see: “Look ’round,” Cleanthes could enjoin. The new evidence for design is for all but a very few to take on faith and authority…
[quotation of Peter van Inwagen on fine-tuning)
There you have a general indication of the new evidence, about fourth hand, from me, who got it as you can see from van Inwagen, who probably got it from teachers and reporters of science, who got it from some front-line researchers’ reflections on the significance of some features of contemporary physical theory.
(p. 278)
I don’t see the problem that Sobel does. Perhaps the relative obscurity of the fine-tuning fact makes the intuition less clear, but that says nothing about the strength of the argument. Further, though perhaps information on fine-tuning was harder to come by when Sobel published the book (2009), recent treatments of the fine-tuning argument (see Luke Barnes and Geraint Lewis, A Fortunate Universe) are relatively accessible for someone with one or two undergraduate semesters in physics or chemistry.
Further, one could argue that the evidence of cosmological fine-tuning emphatically is for all to see. Look around the room and notice that things are staying together and not collapsing in on themselves or flying apart. Note that things have stayed this way for a while. Note that the universe permits a bunch of different kinds of things to exist rather than just one kind of thing, and think about all the other ways it could have been. It’s not hard to perceive the strength of the cosmological fine-tuning argument. The technical details merely quantify and add force to what anyone can directly perceive.
Moving on, Sobel discusses the category of “many cosmoi hypotheses”, more commonly called multiverse hypotheses, as alternatives to the theistic explanation of fine-tuning. There are two kinds here: simple many cosmoi hypotheses and evolutionary many cosmoi hypotheses.
Simple many cosmoi hypotheses are those which hold that the values of the fundamental constants in each universe are independent from any other universe. That is, the values of the constants in some universe U are generated randomly without reference to the values in any other universe. Sobel thinks this theory is more intrinsically likely than theism. However, for him, it has very little predictive power. Why is this? Recall that predictive power is a measure of how well a theory predicts that which it’s supposed to explain. Sobel argues that while a simple many cosmoi hypothesis makes it very likely that there would exist, somewhere, life-permitting universes, it does not make it very likely that this particular universe would be life-permitting:
The likelihood of any given particular cosmos, including the one in which we find ourselves, randomly generating for itself the life-supporting parameter assignments (to simplify I am assuming that there is exactly one such assignment), or any other particular parameter assignments, assuming finitely many possible assignments, f, is 1/f, given that, to recall a ‘crucial feature’ of simple many cosmoi theory, assignments for each cosmos are determined by some random chance process operating on the ‘space’ of all possible assignments.
(p. 280)
For Sobel, the fact that needs to be explained is that the universe in which we find ourselves is life-permitting. To make the point a little clearer, let’s suppose that our universe was the 19,880,741st universe generated by some universe-generating mechanism in a simple many cosmoi hypothesis. Sobel claims that a simple many cosmoi hypothesis gives us no more reason to think that the 19,880,741st universe will be life-permitting instead of the 19,880,742nd, or the 19,880,743rd, or so on. The probability of #19,880,741 being the lucky one is still microsopic. Somewhere, in the vast ensemble of universes generated by this mechanism, we would expect there to be a life-permitting universe. But the odds of it being our universe—universe #19,880,741—are not great. Thus, simple many cosmoi hypotheses do not predict our observations, and so they are not a good response to the fine-tuning argument.
Do you see the problem here? Sobel is conflating two different ‘fine-tuning facts’ in a way that, ironically, complicates the atheist’s job. Robin Collins helpfully distinguishes between three fine-tuning facts that need to be explained (this is from his chapter in the Blackwell Companion to Natural Theology):
The observer-relative fine-tuning fact: we observe a life-permitting universe rather than a non-life-permitting universe.
The indexical fine-tuning fact: that this universe, the one we inhabit, is life-permitting (this here is functioning as an indexical that picks out the universe we inhabit).
The existential fine-tuning fact: that a life-permitting universe exists.
Fact 1 is logically entailed by the fact that we exist. Fact 2 is what Sobel is trying to explain. Fact 3 is what Sobel should be trying to explain.
Since it exponentially multiplies the number of universes, a simple many cosmoi hypothesis would prima facie provide a reasonable account of Fact 3. Sobel is right to point out that such a hypothesis doesn’t make any particular universe any more likely to be life-permitting. Where’s he’s wrong is in the significance that he attaches to that fact. We know that we exist, and consequently that our universe is life-permitting. It doesn’t matter “which universe,” indexically speaking, ended up being one of the life-permitting ones. What we’re trying to explain is why we exist at all, and our existence doesn’t depend on the constants in universe #19,880,741. It requires at least one life-permitting universe somewhere.
Suppose instead of universe #19,880,741 it were universe #382,007,227 that ended up being life-permitting. Then we would have found ourselves in that universe rather than the first. No matter. Nothing changes. The indexical fine-tuning fact is not very interesting. It’s the existential fine-tuning fact that requires explanation.
By trying to explain the indexical fact rather than the existential fact, Sobel ends up selling the multiverse hypothesis short, and goes on to relegate it to the same tier as theism (which, in his mind, is not a very good explanation of fine-tuning):
Their advantages of inherent plausibility and disadvantages in would-have-been predictive power to my mind balance to place simple many cosmoi and rational tuner theories about on a par, and equally probable on the evidence of fine-tuning.
(p. 280)
So much for simple many cosmoi hypotheses. Sobel goes on to discuss a different kind of theory, evolutionary many cosmoi hypothesis. In these, the values of the fundamental laws and constants in each universes is dependent on the values of such fundamental laws and constants in logically prior universes. Sobel places the most stock in Lee Smolin’s evolutionary cosmology. As I understand the theory, Smolin argues that the origin of every black hole spawns a new universe. The fundamental constants of a universe spawned from a black hole will randomly deviate from its parent universe within a narrow range of possibilities, leaving child universes which are fairly similar to their parents. Over many “generations” of universes, the universes which “reproduce” the most are those which produce large numbers of black holes, and so those eventually come to dominate the space of possible universes. However, since the constants which allow for black holes are also those which permit stars (the force of gravity, for example), by selecting for black holes, the evolutionary process favors universes with stars. And stars are needed for life. Life is a side effect of selecting for black holes:
Voilà!! Tuning for black holes that is incidentally sufficient for stars, and so for life. What a nice theory!
(p. 280)
Sobel thinks this theory is more likely than either theism or a simple many cosmoi hypothesis, but not quite more probable than not. It certainly sounds nice, though there are outstanding questions, not least of which is why our universe isn’t a wasteland of black holes. There’s also a dearth of empirical evidence for Smolin’s view. Other objections might be raised. For now, I want to jump to Sobel’s conclusions on the fine-tuning argument (here “FT” is theism and “MU” is a simple multiverse):
Were I persuaded that the apparent fine-tuning of which I hear talk cannot be grounded in a ‘deeper theory’, I would put down the appearances of fine-tuning to our dumb luck. Otherwise, supposing a worked-out deep theory of fine-tuning of this cosmos, say a theory along the lines of Smolin’s evolutionary many cosmoi theory, were not only more plausible than FT and MU (as in my view his present largely speculative theory already is by far) but actually acceptable, that is, more probable than not (as in my view his theory presently is not), I would accept it as an explanation of the facts of fine-tuning of this place and put that theory’s fundamental assumptions down to that luck. After all, not everything can be explained, not everything can have a reason (if, it is hardly necessary to qualify, anything is contingent), and a good ‘deep theory’ of ‘parameters for life’ could very well be a place at which reasons and explanations for them run out.
(pp. 284-5, emphasis added)
Let me just be blunt and say that dumb luck is a terrible explanation for fine-tuning. The probabilities in question are so small and the stakes so great that it’s hard to take this response very seriously. So far, I haven’t discussed Sobel’s reasons for discounting the theistic explanation. They center around the “how” and “why” of the theistic hypothesis—what does it really mean for a designer to fine-tune the laws and constants of nature? And why would a designer do so?
Without unpacking potential responses to these questions or Sobel’s elaborations on them, it suffices to say that even if we are unsure of the answers to these questions, we still ought to favor a vaguely-defined design hypothesis to a sheer dumb luck hypothesis.
Sobel’s take on fine-tuning is brilliantly written and fun to think about, but I’m afraid he’s fallen short here in multiple ways, including a significant one that disadvantages his position. I conclude that weighty critiques of fine-tuning will come from elsewhere.