I used to think of proofs of the existence of God as basically attempts to compel assent to a particular religious doctrine through a sort of sleight of hand:
- Prove, based on reasonable-seeming general axioms, the existence of some sort of ultimate entity.
- Name this entity "God."
- Conflate this with the particular God-based model of the world and right action embedded in your own religion.
While in many cases this may actually be the motivation, I now see a totally different thing people might have been trying to do with such "proofs." In many cultures at many places and times, people have talked about right action, or the nature of reality, with reference to some sort of ultimate entity that seems to have some sort of human-like agency or preferences, an Ultimate Lawmaker or Ultimate Actor of some sort. These systems often have good reasons for particular claims they're making - they often constitute a workable strategy - but saying it with reference to an ultimate agent is frequently an embarrassment.
A natural criticism of such framings is that the ultimate entity seems like an extra hypothesis of which we have no need. Why not just describe everything literally? Why attribute moral principles to the will of an ultimate agent? Why attribute the existence of the universe to the same thing? There are plenty of perplexities associated with this. One of the most popular is that while we'd be happy to follow the advice of an omniscient omnibenevolent agent, it's hard to attribute the existence of the universe to the same omnipotent agent, given how obviously suboptimal many things are locally.
(Evidence: sometimes we improve things, and we're not even omniscient or omnibenevolent, much less omnipotent. Evidence: the need for such advice in the first place. The Bible mentions both these problems.)
The reason why God gets posited is that systems are simple not because they posit few objects, but because they have a short message length. A compact set of axioms and rules of inference is an example of this. For instance, Euclid's Elements allows the construction of an infinite variety of shapes and proof of a very large number of statements, but is based on a comparatively small set of assumptions, definitions, and rules of inference mostly stated at the beginning.
Given this, proofs for the existence of God can also be construed as an argument that, in some frameworks that are reasonably natural to work with, we can't help but posit a God if we carry them to their natural conclusion. The Prime Mover or Greatest Possible Existent occupies a position somewhat similar to the Axiom of Choice. So, not attributing things to God means either abandoning those frameworks entirely (which may make some important true things much more difficult to say), or refusing to follow them to their logical conclusion (epistemic motivated stopping, not an especially rational practice), or affirming a contradiction.
These proofs don't prove anything else about God - they don't, for instance, say anything about whether any particular religion is true - they only serve to identify the question "does God exist?" with the question of whether we find it helpful to use the particular axioms and modes of inference involved in the proof.
I think that most such proofs still fail, in part because the people engaging in them are somewhat confused about what they're doing, but they fail a bit more sympathetically than I used to think; they're not explanations for why everyone should buy into their system, they're explanations for why their system can't help but talk about God, despite the obvious problems this presents.
This post inspired by my recent reading of Yoram Hazony's The Philosophy of Hebrew Scripture.