In the Wake of the Fine-Tuning Argument

A scientific theory is fine-tuned when you must make a suspiciously precise assumption for it to fit the data.^[1] The pride and joy of modern physics, The Standard Model, is fine-tuned. Unless you make suspiciously precise assumptions about a number of dimensionless parameters,^[2] the standard model does not fit with observation. This is a defect. Scientists don’t like fine-tuned theories and generally take fine-tuning as motivation to go find a better theory that needs no suspiciously precise assumptions to fit the data.

Finding better theories than the standard model has proven challenging. Some opportunistic theists have taken these facts to construct an argument for theism. It goes roughly like this: given the assumption that the universe has no particular bent toward the flourishing of life (e.g. what you’d expect given atheism), fine-tuning is really very improbable. But given the assumption that the universe does have such a bent (e.g. what you’d expect given theism), fine-tuning is not that improbable. Assuming roughly fair priors for theism and atheism, this gives theism a big Bayesian win. That’s a hugely simplified version of the argument: go read a cutting edge version for the nuances.

This post is not actually about whether the fine-tuning argument “works.” Like most philosophical arguments, it’s a work in progress as very clever people come up with ongoing cycles of objection and reply. Maybe it will end up failing, e.g. because normalizability can’t be solved. It’s a hard problem and I don’t think any of the current solutions out there hold up very well. My goal in this post is not to adjudicate that kind of question. Instead, this post is about what I think the argument is likely to show/not show and what follows for philosophy of religion if I’m right.

Here’s what I think the fine-tuning argument will show. The king is dead. In 20^th century metaphysics, reductive single universe physicalism was the king. It was the “normal” view, against which all other views reacted or strove. Gods, spooks, multiverses, and the like required justification. Reductive physicalism did the justifying, of numerous research programs. And largely because of the colossal probabilistic hit single-universe reductive physicalism takes from things like the cosmological constant, I think we will end up concluding that it is not viable.

Here's what I think the fine-tuning argument will not show. It will not distinguish theism from a cloud of successor-hypotheses vying for the throne. All you need to get a big Bayesian boost from fine-tuning is to render physical life, or even stars, or even elements with atomic number >3, not horrendously improbable. This includes theism, sure. But it also includes multiverses. It includes small-god theisms as well as big-god theisms, despite big-god theisms taking up all the oxygen in the theism discussions. It includes non-reductive/non-physicalist variants of atheism, such as axiarchism and Nagel-style cosmic minds. It includes things like paganism and animism that don’t neatly fit even as small god-theisms or non-reductive atheisms. Partly because of considerations like Hans Halvorson’s theological critique of the fine-tuning argument, it is not at all clear to me that theism’s big Bayesian boost is going to be better or worse, or by how much, than these other options.

So, I think the portion of philosophy of religion that concerns itself with the existence of God debate will need to come to grips with these other views. The rest of philosophy of religion will presumably chug along as usual. But the kinds of things we need to concern ourselves with will shift in ways that may seem surprising. My focus will be on one particular such shift: the vulnerability of what we will christen “The New Natural Theology” to atheistic axiarchism.

Natural theology has always followed the same pattern. It begins some kind of theistic proof. Traditionally these have been ontological or cosmological arguments, but one of the hallmarks of New Natural Theology is its reliance on design arguments instead. Once the argument is finished, we usually don’t have a complete God. We might have a being who is pure act^[3] (Aquinas), the greatest conceivable being (Anselm), first in efficient causation, final causation, and excellence (Scotus), or perfectly simple (Maimonides). There is then a lengthy project of working out what other attributes this kind of being ought to have. This tends to be the majority of the work, and regularly takes many times more words than the proofs that these authors are most commonly read for.

New Natural Theology follows this same pattern. But the starting theistic arguments are different. These include the fine-tuning argument, but also arguments like the argument from consciousness, the moral knowledge argument, and the twin harmony arguments: nomological and psychophysical. Notably, all of these are design arguments of some sort or other. They start with some striking fact about the world and, in their most-common Bayesian form, argue that it is more to be expected given theism than given naturalism. The defense of this differential Bayes factor almost invariably invokes the twin observations that the striking fact is objectively good/valuable, and that theism involves an agent whose actions are more likely to produce more valuable things.

While these arguments position well against naturalism, with its background hypothesis of a universe fundamentally indifferent to the good, they position more poorly against something like atheistic axiarchism. You can think of the God that New Natural Theology delivers as something like Axiarchism+, or Axiarchism With A Personality. Since the ‘+’ or the personality don’t actually do any predictive work in new design arguments, bog-standard atheistic axiarchism (which posits a fundamental orientation toward the good in the workings of the universe) will predict the key evidence (things like fine-tuning, consciousness, moral knowledge, and various harmonies) about as well as bog-standard theism.

That leaves us with the priors. The prior of bare axiarchism will exceed the prior of either atheistic axiarchism or theistic axiarchism, since it is logically weaker than both. But which hypothesis about agency or personality is more intrinsically probable? My inclination is to say that they will favor the atheistic version. It’s more parsimonious. Its fundamental ontology will be things like fields, particles, or the cosmos (depending on whether we go priority monist or priority pluralist) that everyone already believes in, while the theistic version needs novel fundamental ontology (God). Its fundamental ideology will be things like goodness and natural properties that everyone already believes in, while the theistic version needs whatever conceptual elements are required to assemble the divine agent/person. Anything required for its fundamental nomology will be either come from fundamental physics or be duplicated in the theist’s axiology, which will have to be some variety of Platonism or Augustinianism. These issues are too complicated to resolve with a paragraph, so consider this more a speculative call for future work than a definitive argument.

Lastly, we need new sources of theistic predictions. What, besides an orientation toward the good, drives divine agency? Theists have to pay the theoretical price of a divine personality. What does it predict outside of axiarchic predictions? This is ultimately where the dialectic will have to go (I don’t have a ton of faith in fights over “the right priors” to end up being much more than autobiography for the various authors involved). And it’s something the focus on theism v. naturalism, which on some deep level (if you spend lots of time in both the evil and the new natural theology literatures) keeps boiling down to “the universe is indifferent to the good” vs. “no it isn’t,” hasn’t really produced. We’re going to need more of it.

---

^[1] That’s not my definition; it comes from Luke Barnes and Geraint Lewis in A Fortunate Universe.

^[2] That is, parameters where the number you use isn’t sensitive to your choice of units; the proton:neutron mass ratio is an example.

^[3] This is a scholastic technical term that roughly means a being who cannot change or be intrinsically different than it actually is.