Author’s note: Theoretical physicist Sean Carroll was recently interviewed by podcaster Alex O’Connor and asked to defend his stance that one of the most thought-provoking scientific arguments for God’s existence, the argument from cosmological fine-tuning, “is the best argument for God, but it’s still a terrible argument.” I am responding to Carroll and other critics of the fine-tuning argument in a series of posts.
Find the full series so far here.
In my last post, I catalogued Carroll’s main contentions against the fine-tuning argument, noted that he was going to be our stand-in for fine-tuning skeptics in general, and discussed why his objections (and other related ones) warrant scientific and philosophical engagement. Let me now outline the first part of a systematic response that will unfold across multiple subsequent posts.
Methodological Foundations and Refuting a Reversal of Likelihood
Using Carroll’s criticisms of the fine-tuning argument as a general guide, I propose to address objections to that argument systematically in six stages, each of which will subdivide into further parts. I describe the first two stages here.
Stage 1: The Methodological Foundation
We need to have a good grasp of what we’re doing when we assign probabilities to hypotheses, particularly broad metaphysical hypotheses like theism and naturalism. Carroll’s approach uses Bayesian probability theory, named after the English mathematician, philosopher, and Presbyterian minister, Thomas Bayes (1702-1761). Carroll is a subjective Bayesian, which means that he takes probabilities to be subjective degrees of belief, that is, psychological states that vary from person to person.1 When the rules of Bayesian reasoning are understood this way, they function to ensure that personal beliefs are consistent (non-contradictory), but any disputes about priors (what is believed to be true before the evidence in question is brought to bear) reduce to unresolvable disagreements over subjective inclinations.
Probabilities as Degrees of Support
This is the wrong way to think about these things. Rather than taking the subjectivist line, Nevin Climenhaga (2024) is right to argue that epistemic probability should be understood as measuring objective (mind-independent) degrees of support among propositions. This captures how the fine-tuning evidence functions quite well. As physicist Luke Barnes (2020: 1223n2) has observed, “Bayesian degrees of plausibility/support aren’t about what any individual knows or believes; they are about what one proposition implies about the plausibility of another.”
As we will see in later posts, this realization pulls the rug out beneath some of Carroll’s objections before they can get traction. Whatever degree of support exists between the fine-tuning evidence and theism is a timeless logical relation among propositions that has nothing to do with the order in which we learned things. This objectivity also significantly mitigates the so-called “problem of the priors” because, even though likelihood analysis already dampens this issue, we’re not dealing with conflicts over subjective beliefs before seeing the evidence, but factual disputes about explanatory relations.
Stage 2: The Likelihood Reversal Refuted
Carroll’s head-scratching claim that fine-tuning supports naturalism more than theism rests on an obfuscation. He argues that since theists believe that God could have created an immaterial reality, God has no need of any particular physical conditions to make things alive. Naturalism, however, requires a physical reality that is compatible with embodied life, so it’s much less surprising on naturalism that this is what we see.
Two Fatal Mistakes
But there are at least two mistakes here. First, there is a conflation of what God could do with what God would likely do given his purposes. The monotheistic tradition holds that God is a necessarily existent transcendent being who desires there to be rational creatures capable of intellectual discovery and moral development who will actively seek to know him, be in a loving relationship with him, and learn to be stewards of the reality he has made for them. Genuine love like this cannot be forced; it must be unconstrained and originate with the one who chooses to love. God first loved us and brought us into being in an environment where such choices were possible and where evidence of his existence is available to those who would look for it. The existence of a universe describable by elegant mathematical laws that are fine-tuned to provide the necessary conditions for life, along with our actual existence as creatures capable of discovering this truth about reality, is unsurprising in this context. Carroll’s objection therefore trades on a caricature of theistic commitments, not their actual content.
What is more, we can show that the probability of fine-tuning on the monotheistic tradition, that is, P(fine-tuning | theism), is quite high — not because God couldn’t do otherwise, but because fine-tuning coheres with God’s purposes as monotheism understands them. Meanwhile, P(fine-tuning | naturalism) is vanishingly low. Naturalism is completely indifferent and uninformative on this question, so the actual laws and constants of nature are brute facts with no tendency toward life-permitting values. As Robin Collins (2009: 207) frames the reasoning in his extended Bayesian likelihood argument for God’s existence on the basis of the fine-tuning evidence, when we consider a life-permitting universe on the naturalistic-single-universe hypothesis versus on the theistic hypothesis, the ratio of likelihoods is such that if a life-permitting universe is very, very epistemically unlikely on the naturalistic-single-universe hypothesis but not unlikely on theism, then the existence of a life-permitting universe strongly supports theism over the naturalistic-single-universe hypothesis.2 In such case, the likelihood ratio overwhelmingly favors theism. Indeed, we shall see that it does.
Carroll’s second mistake is even more serious. Just because naturalism requires a physical reality compatible with embodied life doesn’t mean that the existence of such a reality is objectively likely on naturalistic grounds. In fact, on naturalistic grounds, not only are we confronted with the brute, unexplained, and unexplainable fact that something exists rather than nothing at all, and that the something that exists has the specific structure and properties it does, we are also confronted with the fact that naturalism gives us no expectation that anything that just happens to exist would be compatible with embodied life. Naturalism is completely indifferent and uninformative as to why anything exists at all as opposed to absolute nothingness, and even given the inexplicable existence of something, it is completely indifferent and uninformative with respect to why this something would make embodied life possible.
Saying that naturalism supports fine-tuning more than theism, and doing so for the reasons that Carroll ostensibly does, is therefore risible. The generous interpretation is that he knows this, but he’s being disingenuous because he thinks no one will call him on it. That’s one bet he just lost. Realizing all of this does, however, make the seriousness of his commitment to multiverse explanations more understandable. Naturalism has no hope of shedding light on why there is something rather than absolutely nothing, but maybe if the something that does exist is rich enough with possibilities, it could shed light on why it makes embodied life likely or even inevitable.
Halvorson’s Misjudgments
While I am focused on probabilistic arguments with respect to the fine-tuning evidence, I would be remiss if I didn’t mention Hans Halvorson’s (2018) dismissal of cosmological fine-tuning arguments as self-undermining. He contends that if the laws of physics assign low probabilities to life-permitting conditions, then even theists should take these probabilities as definitive because God’s existence doesn’t provide any basis for rejecting them. Halvorson’s whole critique, however, rests on David Lewis’s philosophical account of “objective chance” as a mind-independent feature of reality that constrains belief regardless of anything else we might know (Lewis 1980), including, of course, God’s existence.
Technical, philosophical, and theological problems abound for Halvorson’s critique. Lewis himself acknowledged that his framework had severe internal difficulties, including that his account of present chances, which depends on the total pattern of events and states of affairs that actually obtain throughout spacetime, makes them hostage to what happens in the future as well as the past (Lewis 1994). This creates paradoxes when they are used as guides for belief. Also, Lewis’s opaque conception of “admissible information,” which Halvorson uses to screen off God’s intentions in evaluating the fine-tuning evidence under theism, has been shown by Ned Hall (Hall 1994, 2004) to be incoherent and lacking any principled criterion for its application.
Finally, as Luke Barnes (2018, 2020) points out, Halvorson’s analysis fundamentally mischaracterizes what creation is by treating God’s act of creating the universe as though it were a random process governed by fixed probabilities, when theologically it is nothing of the sort. It obviously is not. It’s an intentional divine act where God directly specifies both laws and initial conditions, just like a computer programmer simply sets the values of program parameters rather than generating them by lottery. Barnes’s rigorous formulation of the fine-tuning argument avoids these difficulties altogether. As noted earlier, he uses objective epistemic probabilities derived from the probability distributions that physicists routinely employ for parameter estimation and theory testing, not from obscure Lewisian “objective chances.” His formulation of the FTA sidesteps the quagmire of Lewis’s Humean probabilities and is entirely untouched by Halvorson’s misguided critique.
To Infinity and Beyond
Having introduced the relevant methodological considerations regarding the use of probability in evaluating the fine-tuning evidence, and having indicated how both Carroll and Halvorson get aspects of it seriously wrong, I’m now prepared to introduce the multiverse, which is Carroll’s go-to explanation for fine-tuning phenomena. The central questions that must be faced here are whether invoking essentially infinite explanatory resources is a wise move for the naturalist to make and, even if he does, has he gotten rid of fine-tuning issues or just displaced them to a meta-level? Beyond this, I need to introduce how I intend to handle Carroll’s (and other’s) objections that the FTA for theism neglects relevant negating evidence, indulges in post hoc fabrications, and ignores the undermining consideration of divine hiddenness. This is all before sketching the cumulative case toward which this series is building.
Next up: “The Multiverse, Evidential Escapades, and the Hiddenness of God.”
Notes
- See Sean Carroll, The Big Picture: On the Origins of Life, Meaning, and the Universe Itself (New York: Dutton, 2016), Chapter 14 “Planets of Belief”, and Chapter 18, “Abducting God.” Carroll paints a picture in which subjective Bayesian “priors” are updated with objective evidence and uses this framework to argue against certain religious and metaphysical — especially supernatural — claims. We will discuss how Bayesianism should function in this context and show how Carroll misapplies it in subsequent posts.
- See Robin Collins, (2009: 203ff). Collins also goes on to argue that the theistic hypothesis is better supported than the naturalistic hypothesis, even when considering the multiverse hypothesis.