17 Comments

Of course, for theism to be an explanation of any of this, we need the laws connecting God's volition to what actually happens in the world ("if God wills that X, then X, and if God wills that Y, then Y, and...") to hang together in a similar way. So your argument once again reduces to the thesis that this is a simpler set of ultimate laws than one the naturalist proposes. But it isn't; it's much more complicated, and in fact almost ineffably complicated since it's going to resist any computable description that would precisely pin down what God would or wouldn't be able to do.

Expand full comment

Why does it have to be computable?

What would be wrong with saying God is just what you get when you have an mind unlimited in power and knowledge, for instance? That's pretty darn simple--one thing, two properties, 0 limits.

Expand full comment

Well, as I was trying to argue in the comments section of that other post, "unlimited power and knowledge" highly underspecify God, because those two descriptions probably don't pick out a unique maximal set of abilities/beliefs. Once you try to actually spell out the specific maximal sets in question that God has, you run into immense difficulties that yield complexity.

You could try to respond: "Okay, but I don't have to specify exactly which maximal set of abilities God has. My theory merely says there is some such set and leaves it at that. An existential quantification over X's can be simple, even if each individual X is complicated." Now two can play this game. Maybe any specific naturalistic explanation of the facts you enumerate in your post is complicated, but the existential quantification "there is some naturalistic theory T such that if T, all those other facts are consequences" seems not significantly more complicated than God's having a maximal set of abilities. And this statement *does* successfully "explain" all those facts under naturalism, but I don't think you'd find it very satisfying. Clearly there's some extent to which existential quantification over highly complex domains really does involve immense complexity, right?

>Why does it have to be computable?

It doesn't have to be computable, in the sense that it's perfectly logically coherent/epistemically possible to suggest that the ultimate laws are uncomputable in some fashion. But you'll pay an absurdly high price in terms of the complexity of your theory.

Expand full comment

Why does uncomputability mean it’s super complicated?

This is an interesting point, though I guess specifying all powerful seems pretty simple: god can bring about any metaphysically possible world. But I’m uncertain enough about it that I think theism still has a good claim to a high prior. It seems like if you have specified the properties of an entity to within some degree of specification, and that description is simple, then the theory is simple even if it’s not clear exactly how to figure out whether it has some more specific trait.

Expand full comment

>Why does uncomputability mean it’s super complicated?

The best formalization of simplicity/complexity we have so far has the simplicity/complexity of a theory T as something like "a measure of the shortest description that completely characterizes T." If that description is really short, it's simple, and if it's really long, it's complex. Individual uncomputable hypotheses, by this metric, will then be infinitely complex!

Now, this is just one formalization, and I don't think that makes it the end-all-be-all of the matter. Maybe it leaves something important out, and a better one will come along tomorrow; this is the sort of thing that happens a lot in formal epistemology. But it should be taken as being at least suggestive, until that better analysis comes along.

> This is an interesting point, though I guess specifying all powerful seems pretty simple: god can bring about any metaphysically possible world

Taken at face value, this definition runs into the standard paradoxes of (naive?) omnipotence. They'll actually be more difficult for you than for most theists, since you think God is contingent, and that leads to additional paradoxes. But admittedly maybe you'll decide later that God is necessary.

> It seems like if you have specified the properties of an entity to within some degree of specification, and that description is simple, then the theory is simple even if it’s not clear exactly how to figure out whether it has some more specific trait.

It's going to be hard to cash out what exactly it means to specify a property within a certain degree.

Anyway, it's true that many simple theories have consequences that are hard to figure out. For example, in Newtonian physics (a paradigmatically simple theory!), it's going to be extraordinarily difficult in practice to figure out the solution of an arbitrary three-body problem, say, a trillion years out in advance. But the theory at least has the virtue that you *could* figure it out in principle, if you had enough time, pencil and paper, and information about the precise initial conditions. But if Newtonian physics instead said that the three-body problem eventually adopted the dynamics of some non-computable system and you could *never* describe its behavior with all the thinking time in the world, that would make it complicated indeed.

Expand full comment

I think I reject formulations like that. Modal realism seems simple, but hard to compute, for instance. I'm sort of sympathetic to a view on which there isn't a precise formula for specifying simplicity.

//Taken at face value, this definition runs into the standard paradoxes of (naive?) omnipotence.//

Those arise from supposing he can bring about impossible things. Not sure why thinking God is contingent is key for that.

//Anyway, it's true that many simple theories have consequences that are hard to figure out. For example, in Newtonian physics (a paradigmatically simple theory!), it's going to be extraordinarily difficult in practice to figure out the solution of an arbitrary three-body problem, say, a trillion years out in advance. But the theory at least has the virtue that you *could* figure it out in principle, if you had enough time, pencil and paper, and information about the precise initial conditions.//

But same thing with perfection--you'd just need to figure out the sequence of properties that is best.

Just want to say, I've found your blog comments really thought provoking. One of the great blessings of having a blog like this is that a lot of smart and interesting commentors turn up and say interesting things! So thanks!

Expand full comment

> I think I reject formulations like that.

Then it seems that your position collapses to

"I intuitively find God to be simple and natural laws to be complex on priors which I refuse to update, therefore theism is more likely than naturalism"

Which isn't really a strong argument. The standard philosophical replies here are "Too bad that your intuitions are so uncorrelated with reality" and "Have fun with your unfalsifiable beliefs".

Expand full comment

>I think I reject formulations like that. Modal realism seems simple, but hard to compute, for instance. I'm sort of sympathetic to a view on which there isn't a precise formula for specifying simplicity.

I don't see how modal realism is uncomputable. It does posit the existence of uncomputable objects/processes, of course, but the point is that there is a computable procedure to decide whether an arbitrary metaphysically possible object/process exists somewhere out there. (The procedure, under modal realism, is "always answer yes!") In other words, there's a difference between the theory itself being uncomputable versus the theory's posits being uncomputable. Although perhaps both have an impact, to differing extents: as a non-modal realist, I wouldn't have too much of an issue with faulting it at least a little bit for positing uncomputable things.

>Those arise from supposing he can bring about impossible things. Not sure why thinking God is contingent is key for that.

Well, for example, if God is contingent, he can't bring about the metaphysically possible world where he never existed. But the problems I had in mind are broader than that. For example, the famous "McEar" counterexample of a guy named "McEar" who metaphysically necessarily can only scratch his ear. Even if we found out that McEar really exists, we wouldn't say he's omnipotent, even though he can do anything it's metaphysically possible for him to be able to do, or bring about any state of affairs it's metaphysically possible for him to bring about (i.e., the ones in which he does or doesn't scratch his ear).

>But same thing with perfection--you'd just need to figure out the sequence of properties that is best.

Right, and I don't see the reason to believe there's any computable procedure you could use to identify that sequence, even in principle, as long as you started from only a finite number of assumptions. You could maybe figure out the status of individual properties vis-a-vis an optimal agent's abilities, but to make the theory simple, I argue you'd need to be able to do it with just about any property. By contrast, in Newtonian physics, a large-enough-but-still-finite amount of information will determine the state of the system at any other time, at least up to the desired degree of precision.

>Just want to say, I've found your blog comments really thought provoking. One of the great blessings of having a blog like this is that a lot of smart and interesting commentors turn up and say interesting things! So thanks!

Thank you! It's appreciated, and I love your blog. Even though I disagree with a large swathe of your metaphysics, your writing always comes off as a refreshing island of sanity in an otherwise depressingly insane internet!

Expand full comment

I love to read arguments for the existence of God. I deeply admire the work of eons of philosophers who grapple with this difficult work.

You were very candid about the limitations of this particular argument, and I really appreciate that too.

I’m not personally wired to do this kind of work, but I benefit from those of you who are.

Expand full comment

Oh I wasn’t saying my argument had limitations I was saying Feser’s did

Expand full comment