Why I’m Staying Out of Responding to Scott Alexander’s Recent Piece

In a recent article titled Why I’m Staying Out Of The Substack Religion Debate, Kabbalistic fiction author1 Scott Alexander wades into the Substack religion debate. In it, Scott explains at a high level why he thinks it unlikely that God exists, setting aside all the little details of all the arguments. And he replied to your favorite Substacker, who I’ll leave unnamed out of modesty. In my view, carefully considering Scott’s case makes clear precisely why God’s existence is so plausible.

Scott opens his argument with the following parable: imagine a group of people who thought aliens assisted Napoleon in his rise. Now, there is actually a lot of evidence for this hypothesis. A bunch of stuff about Napoleon is weird and can be explained by alien assistance. So you can imagine an alt-world where Substackers make complicated Bayesian arguments for the hypothesis that aliens assisted Napoleon and Lydia McGrew claims that the evidence for alien interference is good for a Bayes factor of 10^44 or whatever.

But this is a terrible hypothesis, obviously! Why is it terrible? Scott’s answer: same reason theism is a terrible hypothesis. Specifically, his diagnosis is that the hypothesis is:

1. Very complex and ad hoc. Why would the aliens intervene secretly to rescue Napoleon?

2. Doesn’t even explain the data that well. Why do the aliens remain hidden? Why did Napoleon lose sometimes? Anytime you posit some super ad hoc hypothesis involving powerful hidden entities that don’t even do exactly what you’d expect, your theory pays a big cost.

As Scott puts it:

The very strongest version of the overfitting penalty should go to theories of the form:

Why did this anomalous thing happen? Because there was a powerful entity trying to make that specific thing happen. No, you can’t see it, it hid all evidence of its own existence. Yeah, the outcome doesn’t match any sane entity’s plausible goals, but I can sort of tell a story for why they kind of sort of vaguely resemble those goals. No, you can’t apply human limitations to it, the entity is an exotic form of intelligence for which we have no other evidence.

I agree that these theories are generally bad. But I don’t think this applies to theism for two basic reasons. First, it isn’t true that we have “no other evidence” for theism. Second, I don’t think theism is ad hoc.

It would be one thing if God were just being trotted out to explain one or two random anomalies. Instead, God provides a good explanation of about 30 features of the world. Now, when you pad your list with 30 arguments, they won’t all be winners. But there are a bunch of good arguments on the list—at least the first nine strike me as very forceful, even considered on their own.

These arguments don’t require positing weird eccentric motivations of God—they posit more general facts like “a perfect being would prefer life to random floating clouds of hydrogen gas,” “if God’s going to create at all, it’s not terribly unlikely he’d do a lot of creating,” and “God prefers agents with minds to random mental static.” None of these require that we guess God’s favorite toothpaste flavor!

And sometimes we do posit that there’s an invisible entity with weird properties to explain observable evidence. Case in point: dark matter. It is literally invisible mostly undetectable matter that you can’t see or measure directly. And you need very finely-tuned invisible matter to explain the data—not just any invisible matter will do. Nonetheless, we’re pretty sure it exists because it explains a lot of otherwise mysterious features of the world. So sometimes inferences of the form “positing this invisible thing and adding some auxiliary assumptions explains the world way better than alternatives, so we should believe in it,” are correct.

Note that any time you use an agent to explain anything, you’ll posit an agent who doesn’t do exactly what you expect. The prior probability of any agent doing any particular thing is very low. If I see a crude drawing on your desk, though my prior in you making a crude drawing is low, I’d be pretty sure you drew it.

Suppose I come back to my home and see that many of my valuables have been stolen. However, surprisingly, a good number of valuables weren’t taken. Positing that a burglar broke in involves thinking:

Why did this anomalous thing happen? Because there was a powerful entity trying to make that specific thing happen. No, you can’t see it, it hid all evidence of its own existence. Yeah, the outcome doesn’t match any sane entity’s plausible goals, but I can sort of tell a story for why they kind of sort of vaguely resemble those goals.

Now, neither this example or the dark matter example are exactly like theism. We have independent evidence for burglars. But if we can have reason to think that invisible entities lacking outside evidence caused anomalies, and we can have reason to think that particular unevidenced agents caused anomalies, then it’s not clear why we can’t think that invisible agents caused anomalies.

In addition, it’s not true that God is some totally bizarro hidden entity that leaves no evidence behind. A giant portion of the world has had a direct experience as of God. Most people throughout human history have believed in him. If a giant chunk of the world claimed to have directly seen the aliens who had helped Napoleon, and the theory also explained many different anomalies, I’d start taking the theory more seriously!

Scott complains that the world isn’t what you’d expect given theism. But neither is it what you’d expect given atheism. Given atheism, I would expect a world with nothing of value, because most conceivable sets of laws don’t give rise to anything interesting. The theist has no monopoly on surprise.

Scott then gives the standard god of the gaps argument: in ancient times, primitive people used gods to explain everything. You’d see thunder and blame the gods. You’d have a headache every time you drank hot chocolate, so they’d say the gods cursed you. As we’ve learned more, we’ve seen otherwise powerful theistic arguments blown to smithereens by the steady march of science. Through science, we’ve learned that lightning is caused by electric discharge and the pain you get whenever you drank hot chocolate is caused by you not taking the spoon out. The argument for Zeus from lightning, while once a powerful argument beloved by the natural theologians, has fallen seriously out of favor.

Now, it isn’t actually true that throughout history people cited random unexplained bits of the natural world as evidence of a creator. Pagans thought that lowercase g gods were behind these phenomena, but monotheists didn’t cite random natural phenomena as evidence for God. As Aron Wall put it:

This type of reasoning (which is quite common, although I phrase it in my own words) tends to glide imperceptibly from popular pagan polytheists (who thought there was a divinity for every major or minor phenomenon) to the Hebrew monotheists (who resisted this trend as superstitious and wrong).

It was perfectly obvious to any pagan philosopher or early Christian that Nature proceeds according to orderly laws, and natural processes. Modern Science can take credit for unifying the description of many phenomena into common mathematical frameworks, but to act as though the existence of order in Nature is a modern discovery is simply absurd. It is true that this fact is in considerable tension with certain forms of Animism or Nature Polytheism. But certainly almost any astute monotheist living in the last two thousand years, is going to admit that God causes most things to happen, not through whim but through the operation of certain natural processes, which can be understood to some extent by human reason.

But even if they had, this wouldn’t be evidentially relevant. If people frequently make ridiculous inferences, that doesn’t cast doubt on any vaguely similar inferences. The argument for God from lightning is a terrible argument. In order for a fact to be evidence for God, it has to be likelier given theism than given atheism. There’s no obvious special reason why God would directly make thunder, beyond his general reasons for making an orderly world.

The argument for God from fine-tuning is not a ridiculous argument! And sensible theists aren’t just looking to random unexplained natural phenomena—nobody thought that God must be directly boring holes in Swiss cheese before we had an explanation of Swiss cheese. Instead, they cite facts as evidence for theism when they involve the deepest underpinnings of our world, are extremely implausible, and are extremely valuable.

So you shouldn’t cite God as explanations of mundane natural events absent extremely strong evidence. You should only cite God as the explanation of features of the world that are plausibly fundamental—not explained by anything deeper. By induction, we learn that God doesn’t monkey around in the random daily workings of the natural world—we learn little about what explains the deepest features of the world.

With that methodological clarification out of the way, there is only one really good theistic argument that has ever been defeated by scientific advance: the argument from life. Before we discovered evolution, God seemed like the only reasonable way to explain life. Now that we know about evolution, we have a better account. In its place, we have gotten many new good theistic arguments in just the last few decades. The psychophysical harmony argument is younger than I am—the fine-tuning argument is younger than my parents, the anthropic argument hasn’t even had its third birthday. So overall, the case for theism has gotten better with time.

One final point: all these methodological complaints pretty clearly give way in the face of really strong evidence. If the stars spelled out the first three chapters of the Gospel of John, that would obviously confirm theism or something like it. So if the case for theism is really strong, that’s enough to overcome Scott’s methodological worries.

The intrinsic probability of a hypothesis is how likely it is before you look at any evidence. The main reason that the aliens assisted Napoleon hypothesis isn’t any good is that it has a very low intrinsic probability. Before considering specific evidence, you should think that it’s ridiculously unlikely. Scott thinks that God’s existence is the same way.

I disagree. I think God is very simple and non-arbitrary, so his existence starts with a not unreasonable prior. Here’s my preferred story about how it is that God is simple: God is simply a mind without limits. Having no limits, nothing bounds either his powers or abilities, so he can do anything and knows everything. Because he knows everything, he knows all the moral facts. If you know that something is good, and you don’t have any contrary psychological inclinations, it’s plausible that you’ll have a default motivation to follow it. So once God knows everything, he’ll always do what’s best.

Now, is this account right? Certainly there are a few moving parts to it. You have to think each of the following:

1. The mind is non-physical.

2. If you knew everything and were fully rational, you’d always be motivated to do what’s best.

3. The idea of a limitless mind is simple, makes sense, and entails omnipotence and omniscience.

I accepted 1-2 even when I was an atheist. 3 seems a bit shakier, but still not crazy. None of the premises are trivial, but they’re not obviously false either. If they hold, then theism has a very high prior probability. So I think theism starts with an okay prior—not extremely high, not extremely low. Certainly it starts out much more plausible than that aliens helped Napoleon.

Scott has criticisms of this hypothesis. I’ll talk about them, but first, let me give three reasons at a high level why I think you shouldn’t assign theism an outrageously low prior probability.

First, if these moves work, theism is really really simple. It’s the kind of theory that physicists dream of: one fundamental parameter set to infinity. If you have a theory of debatable internal coherence, but extreme simplicity if it works out, then a lot of the probability space will be occupied by the theory.

Analogy: suppose you’re considering between two theories of physics. One might just be totally horribly broken and incoherent. But if it works, it’s ridiculously simple. The other is definitely very complicated. Overall, you should prefer theory 1 to theory 2. Because theories that are a bit simpler have a much higher prior, even with only a low prior on theism being coherent, it will still gobble up a lot more of the probability space than naturalism. I talk about this more here, if you’re interested.

Second, much of fundamental reality ends up being pretty weird when probed. Look at quantum mechanics or the foundations of math. As Alexander Pruss puts it, discussing various puzzling Catholic doctrines:

But to reject the doctrines for this reason alone would be mistaken. For similar points can be made about Relativity Theory and Quantum Mechanics. To say that simultaneity is relative or that a physical object has no position but rather a probability distribution over positions borders on contradiction, and the philosophical moves needed to defend these seem ad hoc designed to save the theories. If we’ve learned one thing from physics in the 20th century, it is that the true physics of the world is very strange indeed.

Nor are theology and science the only places where things are strange. Similar things can be said about the mathematics of infinity, or even just common sense claims such as that there is change (think of Zeno’s paradoxes) or that material objects persist over time (think of the Ship of Theseus and the paradoxes of material composition).

We can, thus, be very confident that created reality is very strange indeed. And hence, shouldn’t we expect similar strangeness—indeed, mystery—in the Creator and his relationship to us?

Thus, it wouldn’t be that surprising if we couldn’t see how a simple ultimate reality was simple. We’re just monkeys in shoes.

You shouldn’t take this too far. You shouldn’t think that the aliens helped Napoleon hypothesis has a decent prior of being super intrinsically likely for reasons unknown. But I think it’s pretty clear when you consider God’s essentially limitless properties and mysterious nature that it isn’t outrageously improbable that he’d be a simple sort of entity. Simple things can have many properties and effects—a simple electron exerts effects all throughout the universe.

Third, there are a lot of other potential ways of God being simple. Other views of how God is simple:

1. The divine simplicity view according to which God doesn’t have any properties or attributes—being literally the simplest conceivable thing.

2. God is simply a perfect being: possessing all perfections. I’m suspicious of this account because I don’t think perfections are real and joint-carving elements of reality, but it’s not obviously wrong.

3. God is a limitless agent. An agent is a mind that pairs with the world in some way: this without limits=a being whose mind matches the world (omniscience) and for whom the world matches its mind (omnipotence). We’ll come back to this idea later.

4. God is a mind of limitless power. If power is joint-carving, because power entails the power to bring facts to mind, it also entails omniscience—and then you get omnibenevolence for free.

5. God is a being of infinite goodness. Set a thing’s goodness score to the maximum degree it will go, and you get God. I have my gripes with this account, but it’s not obviously ludicrous.

6. God is pure act. I don’t really know what this means, but my Thomist friends think it means something and entails divine properties.

God’s existence might also follow from certain necessary truths. Godelian ontological arguments try to show this. Now, I’m pretty sure they’re all unsound, but the odds one of them works aren’t, like, one in a million.

One of these accounts succeeding doesn’t seem outrageously unlikely. It seems that you should think there’s at least a 1% chance that one of these accounts succeeds on which God has at least a 1% intrinsic probability. Combined, that gets you at least an intrinsic probability of 1/10,000.

But anyway, Scott has an objection to my account according to which God is a limitless mind. He even calls my argument “especially bad.” Here’s the first bit of his objection.

This relies on a sleight of hand where you pretend to be subtracting something to make your target simpler, but the thing you’re subtracting is “limits” or “incapacity”. But these are negatives (ie the positive is capabilities). When you claim at the end that all of this subtraction has produced a very simple thing, you’re wrong, because you’ve been subtracting negatives, ie adding.

If this trick worked, it should be easier to build a starship than a biplane. You would start with aircraft-without-limits, and then you’d only need to add a few limits to get the starship, but you would need to add many more before you got the biplane.

Suppose we have two theories. Theory 1 says the universe is infinite. Theory 2 says the universe is 10 quintillion meters across exactly. Should we say theory 2 has a higher prior because theory 1 is “subtracting negatives, ie adding?” No! Sometimes theories that posit more total things are simpler. The theory of modal realism—that every possible world exists—is a lot simpler than the theory that every possible world exists except one, even though it has one more world.

When does a theory pay a cost for having a limit? Dustin Crummett has a great post on this, but two contributing factors:

1. When the simplest way to describe it involves describing the thing without limits and then adding a limit. The simplest way to describe a being that knows every fact but one is to describe a being that knows every fact and then add a proviso about it not knowing one fact. It’s not to list every fact that it knows in order.

2. When the limit is in some way arbitrary. If God doesn’t know some fact, why that fact rather than another one?

This is why the theory that nature is uniform is much better than the theory that nature breaks down in one random patch somewhere. The simplest way to describe the patch theory is to describe uniform laws and then add a proviso about them not being uniform in some region. The theory is also arbitrary. Why do laws break down in that patch rather than some other one?

So if we can characterize all of God’s properties in terms of a single unlimited thing set to infinity, then theism is a very good theory. I claim we can: God is a mind set to infinity. And minds are fundamental bits of the universe, not reducing to any non-mental thing.

My guess at Adelstein’s response: he is a nominalist about aircraft, but a realist about mind. You can’t improve airplanes by removing their limits, because “airplane” is a semi-fake concept that we apply to a particular object which must be laboriously constructed component by component. But Mind is just a thing that floats out there in the void, so it’s fair to ask what it would look like if allowed to wax to its full power.

Yes, that’s correct. Simplicity is (something like) how long it takes to describe something in an ideal language. The ideal language has to be able to include talk about any fundamental things, otherwise you could never describe them because they wouldn’t reduce. Aircrafts are complicated physical things. Minds are irreducible. So just as it’s fine to invoke math when describing the physical world, because math is fundamental, it’s fine to invoke minds. You get to build a symbol for minds into your language. You can’t build a symbol for airplanes.

Can we reject this assumption? A common view is that the physical language has to describe everything in terms of math. But if minds are fundamental things and they don’t reduce to math, then this is just to declare minds infinitely complex by fiat. It doesn’t seem that laws that give rise to conscious experience are automatically infinitely complex.

But most applications of dualism to this debate on another sleight-of-hand. Consciousness is complicated and mysterious and hard to explain. Therefore, maybe it happens on some exotic substrate. A reasonable name for that exotic substrate would be Mind. And we all know that the Mind is where computation happens. Therefore, maybe computation happens on an exotic substrate.

I agree that consciousness is complicated and mysterious and hard to explain. I sort of suspect it won’t turn out to have an exotic substrate, but maybe I’m wrong about that, and I don’t want to speak with confidence here.

But computation isn’t complicated and mysterious and hard to explain. Normal neural nets explain it fine. We have great Himalayan mountains worth of evidence that computation happens on the ordinary physical substrate of the brain:

[proceeds to list a lot of that evidence]

But the dualist portion of the argument for divine simplicity (and therefore for God) relies on an exotic substrate for computation in particular. The argument isn’t that God might be maximally conscious (although if He exists, this is likely true). It’s that “mind without limits” is enough to buy unlimited computation (eg know everything, design the universe). If we unbundle Mind into the explicable parts that we know happen for natural reasons, and the inexplicable parts which might have exotic substrate, computation is in the explicable-natural part, and so any exotic substrate that consciousness forces us to posit is useless for getting us infinite computational capacity.

First, one terminological note: the thing I’m arguing for isn’t actually what theists mean by divine simplicity. Divine simplicity is a technical term that means that God has no attributes or parts that aren’t identical to himself. It’s a pretty wonky doctrine and not one that I accept. It doesn’t just mean that God is very simple!

Second, Scott mischaracterizes the arguments for dualism (even just applied to this domain). The zombie argument, say, depends on the conceivability of zombies and the bridge from conceivability to possibility. Mary’s room depends on the idea that Mary knows every physical fact but doesn’t know what it’s like to see red. The master argument against physicalism goes that the physical facts are exhaustively about structure and function which isn’t even in principle the subject matter of consciousness.

Think what you want about these arguments. But none of them depend on the inference “Consciousness is complicated and mysterious and hard to explain. Therefore, maybe it happens on some exotic substrate.” There are lots of things that are mysterious and hard to explain: quantum mechanics, to give one example. No one posits an exotic substrate. The reason people are dualists is that there are arguments not just for consciousness being mysterious but for it being irreducible.

In an earlier draft, Scott said “most good arguments for dualism rely on another sleight-of-hand” instead of “most applications of dualism to this debate on another sleight-of-hand.” I do not get how the second thing changes the meaning. It’s not like we give different arguments for dualism when arguing for theism.

Now, I agree that you shouldn’t think of God’s essential property as being infinite computation. Infinite computational abilities are derived from his deeper properties, not things that we build in. Because God is omniscient, he can do infinitely complex calculations. But that’s not because we’re building into the concept of mind that it is identical to computing.

As an analogy, consider Scott’s favorite theory: Tegmark’s mathematical realism. On this theory, every single mathematical structure is instantiated. Every world that you can describe with math exists. Across the infinite worlds, there is infinite computation. But that’s not because you’re building in at the ground level infinite computational ability. Rather, you’re deriving it from the fundamental elements of the theory.

But why is a mind “allowed to wax to its full power” omniscient and omnipotent? Why not just infinitely intense conscious experience? I agree this step is the shakiest, but let me say two things about it.

One: just intuitively, a mind that doesn’t know something is limited in some essential respect. Something lies outside the bounds of its awareness. Imagine a mind wishing for something to be done but being unable to bring it about. Definitely that’s limited in some way! Now, there’s a deeper question about whether this kind of unlimitedness gets to be built into our fundamental theory, but it at least doesn’t seem like the answer is obviously negative.

Second, minds pair with the world in two distinct ways. One line of influence runs world to mind, and this refers to how the mind represents the world. The other runs mind to world, and this refers to the mind’s ability to bring about effects in the world. To know the essential facts about minds, you need to know both how the mind produces effects in the world and how the world produces effects in the mind.

Take both of these to infinity. Then the mind can produce anything in the world and anything in the world shows up in the mind. From there you get omniscience and omnipotence. If a mind’s match with the world is set to infinity, then from there, you get all the divine attributes. Infinite computation comes for free!

Again, I admit that this is a bit wobbly. My claim is modest: it is not outrageously improbable that some account along these lines will work. If that’s right, then theism doesn’t start with an outrageously low prior. I go back and forth about how high theism’s prior is—sometimes I think it’s higher than naturalism, while other times, I think it’s pretty low. But in any case, if the evidence is really good, you should be a theist.

Scott closes his piece:

If computational dualism is unsustainable, then the trick of increasing simplicity by removing limits works no better for minds than for airplanes, and we can no longer wiggle out of the penalty for pseudo-conspiratorial powerful-entity-based explanations by calling it a simplicity penalty and claiming that our hypothesis is infinitely simple; theism must face the penalty’s full weight. Its buckling under that intolerable burden produces the feeling of “although the evidence for and against this theory is hard to quantify on its own terms, it should have been ruled out preemptively and never come to our attention in the first place, the same as an alien-Napoleon theory”. And that feeling makes me less excited than I once was to participate in the weeds of religion-vs-atheism debates.

(and by “I’m not participating”, I mean “I’m participating by writing this post about how I won’t participate”. Sorry!)

But those are a lot of ifs. Insofar as your argument for theism being ridiculously implausible depends on a number of controversial claims, then you should think odds aren’t terrible that you’re mistaken. Thus, if the case for theism is really good, you should become a theist. So any argument of the form “theism should be ruled out of as a serious theory because of contentious assumptions X, Y, and Z,” involves playing the game, rather than getting a reason not to play it.