Why I'm Staying Out Of The Substack Religion Debate

I.

Matthew Adelstein (aka Bentham’s Bulldog) and Richard Chappell debate the existence of God. So do JP Andrew, Benjamin Tettu, Sangwon Lee, Ethan Muse, other Ethan, and “Atheistic Journals”, all within this month! I thought the omnipresence of the early ‘00s online atheism debate was a time-bound historical phenomenon. But now that the blogosphere has reincarnated as Substack, it’s back, nearly as strong as ever.

I enjoyed participated in the early 00's version, but the current incarnation leaves me cold. I can’t get as interested as I once did in the details of the Fine-Tuning Argument, the Argument From Miracles, et cetera. It’s not just that I’ve already served my time in the theology mines. My lack of belief in God intuitively feels like it precedes all of these things.

Consider by analogy a society where people regularly debated whether aliens assisted the rise of Napoleon. There are many reasons to believe in such assistance: Napoleon’s rise from Corsican nobody to Emperor and would-be world conqueror was a bizarre deviation from the usual pattern of history. He won battles outnumbered 4-to-1, when everyone knows victory usually goes to the larger army. He became monarch of France immediately after it had declared eternal enmity against the concept of monarchy and resolved to murder anyone who didn’t hate monarchy enough. And the arguments against such intervention are comparatively flimsy. True, nobody saw aliens, but a sufficiently advanced civilization could conduct its interventions in secret. True, our prior for alien intervention is generally low, but maybe this is unjustified: we have no idea how many historical events the aliens have intervened in; Alexander the Great was pretty strange too. True, Napoleon finally lost, but this is not decisive; maybe the aliens only wanted him to blaze brightly for a few years before his eventual defeat, or maybe stronger aliens were backing the Duke of Wellington.

In a world where this hypothesis somehow entered the space of serious debate, it would be hard to dislodge. Our own historical community is protected from it not because they can precisely add up the Bayesian evidence for and against each of these considerations and conclude its falsehood, but because common sense precludes entertaining it in the first place. This isn’t to say that Reason doesn’t work - if someone could precisely add up the Bayesian evidence for and against these considerations, they would get the right answer - but as a historical matter, this type of delicate reasoning wasn’t how our society resolved this question.

My atheism is more deeply rooted in whatever prevented us from considering this question in the first place than in the type of delicate reasoning that could get us out of it. I admit this is dangerous. It risks degenerating into insult-slinging - “your religion is just as stupid as believing aliens caused Napoleon!”. Even if it avoids that failure mode, it would at best come out as the absurdity heuristic, a dangerous fallacy where someone concludes that they can dismiss a matter without careful consideration because it feels absurd to them. This would have dismissed nuclear weapons in 1900, heavier-than-air flight in 1850, and atheism throughout most of history (surely it’s absurd that complex life could come to exist just by random chance). It’s also useless: religion obviously doesn’t feel absurd to religious people, so it can do no work convincing them. So rather than stop here, we should dig in further: why do we exclude the Napoleon scenario from consideration, and what can we learn from it?

The “aliens caused Napoleon” scenario fails because of overfitting. Anything at all can be explained by “there was a powerful hidden entity trying to make that specific thing happen”. Why does the location of the Great Pyramid encode the speed of light to seven decimal places? There was a powerful hidden entity involved, and it aimed at that specific outcome. How did Donald Trump come from behind to win the 2016 election? Well, there was a powerful hidden entity, it wanted Donald Trump to win, and so it arranged to make that happen. These pseudo-explanations aren’t merely as good as real explanations, they’re strictly better. Trump rode a wave of right-wing populism, but why didn’t that show up in earlier polls? How did Napoleon win all those battles? The “powerful hidden entity trying to make that specific thing happen” explanation plows through objections like these easily. Any reasoner who ever wants to get past that to a more useful model will need to give the “powerful hidden entity” story a preemptive penalty.

We usually apply this penalty under the term “conspiracy theories”. Since no individual human is powerful enough to unilaterally choreograph the course of history, it must be a cabal of elites; since we haven’t seen such a cabal, it must be hiding its own existence. We debunk such theories by observing that such conspiracies are hard; a worldwide plot would have to maintain secrecy for decades or centuries, even though the real government sometimes accidentally invites hostile journalists to their Signal chats. To avoid these counterarguments, the really intense conspiracy theorists add more epicycles: the conspiracy has memory-wipe technology that they can use to cover up leaks! But this only makes the problem more difficult; now the conspiracy needs to be able to invent and conceal technology decades ahead of the rest of the world. Eventually the demands on our credulity grow too great, and the edifice topples.

“Aliens caused Napoleon” demonstrates that “powerful hidden entity” explanations are a broader and more dangerous category than conspiracy theories. Aliens are invulnerable to the usual “people couldn’t conspire that well” heuristic: maybe there was only one alien, or the aliens are all loyal to their own species, or they don’t have Signal groupchats. They’re invulnerable to the argument that they couldn’t have implausible supertechnology - of course aliens could have implausible supertechnology! These “victories” don’t make the alien situation better than the usual conspiracies. If anything they make it worse, showing how perfectly it’s been optimized to remove all of the usual safeguards.

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.

This could certainly explain everything. Don’t take my word for it. Take the word of every culture throughout history: whenever they had something their primitive knowledge couldn’t explain, they imagined a powerful entity built on an exotic substrate who intended that outcome. Why is there thunder and lighting? Because the Storm God is very powerful and, uh, is trying to kill enemies who appear suspiciously regularly every hot summer afternoon. Why does the sun shine? Because the Sun God is very powerful, and decided it should do that, in order to scare off the Dragon of Darkness, who we definitely didn’t make up just for this one myth.

As science progressed, anomalies got fewer and fewer. When you understand some laws but not others, the things you don’t understand seem to violate natural law. If you don’t understand electricity, then lightning - sudden bright discharges coming from seemingly inert clouds - seems to violate natural law. If you don’t understand nuclear fusion, then the sun - a great glowing fire in a void without fuel or oxygen - seems to violate natural law. As people learned new laws, they dropped “powerful hidden entity intending this specific thing, operating on exotic substrate” explanations, one after another.

Now we have some remaining set of anomalies. The fact that the universe exists at all. The nature of consciousness. Fine-tuning (or is that one solved too?) Anthropics, probably, I can never keep track. Various supposed miracles.

So when someone sees these and says “I bet I know the answer to this one! It’s a powerful entity who intended for those specific things to happen! No, you can’t see it, it hid all evidence of its own existence. Yeah, the outcome doesn’t quite match any sane entity’s plausible goals, but I can sort of tell a story for why they kind 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,” then before I’ve even started listening to your arguments, I’m rolling my eyes so hard that I’m giving myself oculogyric crisis.

II.

There’s an embarrassing element of respectability and defensibility in the argument above. It talks of “applying penalties” and rejecting things that seem too “optimized” or too conspiracy-theory like. The whole thing feels a bit midwit, as if I’m saying you’re not allowed to consider this because it sort of looks like other things that give me the creeps. Shouldn’t I be talking about the epistemics, using terms like likelihood and simplicity and “Occam’s Razor”?

This would be virtuous, but people’s intuitions are famously bad here. You shouldn’t multiply entities beyond necessary. But the aliens-caused-Napoleon theory only needs one alien, and the usual theory requires literally thousands of Frenchmen! The right way to do things is with Kolmogorov complexity, but 1) most people don’t know what that is, and 2) the few people who do know what it is don’t know how to apply it. I will, however, at least briefly try to respond to one especially bad argument.

Some religious people say that powerful hidden entity explanations, especially in the form of conspiracy theories, generally deserve a penalty only because the conspiracy is a complex object with moving parts. But God is simple. In fact, He is the simplest possible being. Therefore, He incurs no penalty. For example, here’s Adelstein in his recent debate (my emphasis):

The world is . . . filled with three kinds of things: minds, physical stuff, and abstract objects. I think of theism—of the sort I advocate—as the hypothesis that simply one of these things (mind) exists totally without limit. God is a mind without limit. Being unlimited, nothing restricts his power or knowledge. Because he has no rational constraints, following Parfit, he’d be motivated to always do the right thing.

Now, it’s true that a limitless mind is just one of infinite conceivable designers. But crucially it’s a much simpler and more modest hypothesis than the other designers with random built in desires, powers, and bits of knowledge. It posits one of the fundamental things exists without limit. Just as the theory that space has uniform laws all throughout is much better than any of the infinite theories on which the laws of physics break down in a random patch, I think the theory that there’s a totally unlimited mind has a higher prior than the limited god hypotheses. Then, when we observe that it fits the data way better than alternatives, it’s worth believing.

If I understand correctly: God is the simplest possible entity: mind without limits, or even existence without limits. If, as they say, “the only numbers that need no justification are zero, one, and infinity”, then God is infinity (for example, infinitely powerful, infinitely intelligent, etc) and therefore needs no (or very little) justification.

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. There should be giant factories where they import aircraft-without-limits in and, after thousands of man-hours of work installing more and more constraints, extrude a few biplanes per year. My two-year-old children would deserve recognition as far greater artists than Michelangelo, because Michelangelo just tacked nearly-unlimited beauty onto the canvas and called it a day, but my children are such perfectionists that they refuse to draw anything without limiting its beauty in countless different ways!

Using this “subtracting limits” trick, you can trivially prove that any complex thing is simpler than any simple thing1 (since it will have fewer limits). This allows you to posit the most complex entity, call it the simplest, and invert Occam’s razor entirely, making your “powerful hidden entity intended this exact outcome” theory look like the least overfit hypothesis!

III.

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. From the same source:

Now, regarding why I think God is good—I’ve sketched this out in a bit more detail here, but let me give the brief outline. I’m a dualist (I know you are too Richard, but a lot of the audience won’t be—so I’ll link Richard’s powerful arguments for dualism). I think that mind and matter are, therefore, distinct and don’t reduce to each other.

But most good arguments for dualism rely 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:

1. If the brain is damaged, eg by strokes or a railroad spike, computation deteriorates or changes character, generally in a way predictable from the specific part of the brain that was damaged (eg language from Broca’s Area)
2. If the brain is altered by drugs (eg LSD), computation deteriorates or changes character. In some cases (eg antipsychotics treating psychosis), drugs can even improve the quality of computation, in ways that are predictable based on brain chemistry and anatomy.
3. People with various genes (eg for high IQ, or high impulsivity) do computation differently than people with the opposite genes, and although it makes complete sense for genes to affect the brain, they shouldn’t be able to affect an exotic substrate directly.
4. If the brain turns off (eg sleep, anaesthesia, comas) and then turns on again, people report that no computation happened in the interim. If they do report computation in the interim, it’s of diminished character, to the exact degree that their brain activity is diminished (eg dreams during light sleep but not most comas).
5. There are a few simple brain circuits where neuroscientists genuinely understand exactly what’s going on (eg certain types of visual edge detection), and they’re implementing types of computation which match the types of computation that would generate the capabilities which we in fact have2.
6. AIs are structurally similar to the brain (eg neural nets sculpted by experience), and they can do computation very similar to the sorts of computation that the brain does. We build them, and know that there is no exotic substrate involved.

Some dualists try to avoid this conclusion by arguing that the brain could be nothing more than a complex receiving device for information sent by the soul. I think this weakly fails 1-5 on a “god of the gaps” argument. That is, before we understood the visual cortex, we might have thought that the brain receives completed visual information from the soul. But now we know that certain cells detect edges in images. So (under the receiver theory) the brain must receive some kind of preliminary visual information from the eyes, do some preliminary processing on neural edge compute, send a packet to the soul for the really hard processing, receive another packet back to the soul, do some additional post-processing to the soul’s results, then output it as nerve impulses that form a reasonable reaction to what it saw. But as we learn more and more, we find that more and more of the computation must be part of the “preliminary processing” and the “post-processing” stages, such that it’s very hard to figure out what the soul is adding, and no longer obvious why this should be preferred to more-obviously-false “receiver” theories like that the liver doesn’t truly metabolize toxins, but only teleports them to some ethereal liver behind the veil that does the true metabolism. In any case, Point 6 is direct and fatal.

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.

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!)

1

I’m eliding the distinction between “complex” and “capable” here, on the grounds that for anything with an artificial or evolutionary purpose (like airplanes or animals), complex ones should usually be more capable than simple ones, since we assume the designer is trying to achieve the purpose efficiently, and more capabilities therefore take more complexity than fewer capabilities. A theist might counter that God is not designed, and so we have no reason to expect this to apply. This gets into some kind of meta-meta-meta-logic where we can’t possibly have any intuitions, but I would still suggest that it would be a weird coincidence for the most capable being to be the simplest, unless you’re using a limits argument like the one discussed above.

2

The best treatment of this I’ve ever read is Stephen Grossberg’s Conscious Mind, Resonant Brain. I never even got to the juicy stuff attempting to explain consciousness because there were so many circuit diagrams for trivial visual features first and I eventually gave up. Still, read one hundred of these diagrams and you will have good intuition for how low-level features in the brain work and how they might potentially get built up into higher ones.