I Flew To Tennessee To Debate Bryan Caplan About God. Here's What Happened.

1 Introduction

I had my first in-person debate recently, against against Bryan Caplan at Vanderbilt University. Jonah Franks did a wonderful job organizing it, as did the Vanderbilt philosophy club, and it was wonderful meeting all the cool people who showed up. I also was able to meet Gavin Ortlund, for he happened to be near the debate—meeting him was wonderful. His public persona is absurdly nice, and you sometimes wonder with people like that whether they are really that nice or it’s an act. In his case, he really is that nice and it was very fun talking about the existence of God and the truth of Christianity with him (I’m sold on the first now—but only about 60% sure—though not yet on the second).

The debate itself went, I think, pretty well—there were never any times when either of us were stumped, and neither of us had any crazy gotcha moments. It was also a pleasure meeting Bryan—he’s quite quick on his feet and fun to chat with. The breadth of his knowledge is particularly impressive—he knows a lot about a lot of things. While during the debate he was a bit dismissive—more on that later—outside it he was super nice and genuinely seems like a great guy.

We argued about whether God existed; I argued the affirmative and Caplan the negative. You can find my opening statement here and the whole debate is here. We covered a lot of ground and it was pretty fast-paced, so here I thought it would be helpful to give a run-down of where we disagreed and why I was right.

But first, I learned a few things from the debate. For one, there are always curveballs. Bryan’s main objection to my arguments wasn’t one I’d thought about much beforehand, and while I think it quite demonstrably fails (for reasons you’ll discover soon), it would have been helpful to have thought of responses ahead of time.

Bryan seemed a lot more confident during the debate than I did, which I think hurt me a little bit. He declared that one of my arguments—and the anthropic argument no less!—was one of the ten worst arguments he’d ever encountered. Throughout the debate, Bryan seemed to be completely, 100% sure he was right, as if we were debating about the existence of leprechauns or the tooth fairy. I think, however, that this might have ended up looking bad for him when it became clear that my arguments were a lot more philosophically sophisticated than he gave them credit for being.

Another thing I learned: don’t speak right into the mic. Doing so makes you much too loud.

I suppose the final lesson that I learned was that it’s important to come up with questions ahead of time so that you’re not asking them on the fly. I was thinking of questions to ask Bryan during cross-examination while he was giving his opening speech, and was thus able to ask some well-refined questions.

2 Prior probability of theism

My first core argument in the debate was that theism has a very high prior probability, meaning that before you look at any specific evidence, you should think it’s decently likely. I argued that this is true for three reasons: first, theism is an utterly ontologically unique view of reality, positing a wholly unlimited being; second, theism is very simple; third, theism lacks arbitrary limits, unlike naturalism. Theism posits that goodness exists without limit in the character of God, which is very simple and has no arbitrary limits, and because is such a unique view of reality, it would be overly dogmatic to declare it much less likely than alternatives, just like it would be dogmatic to think the prior of moral platonism is, say, 1 in a million.

Bryan had three “objections” to this. What was curious was that only one of them was really an objection to the core claim. His response was to raise the problems of evil, hiddenness, and the paradox of omnipotence. These are real problems (with the exception of the paradox of omnipotence) but they’re not about the prior probability of theism—they're about theism’s poor fit with the evidence (with the exception of the paradox of omnipotence).

In response to them, I gave a brief version of the theodicy I’ve defended here and argued that because the future is so vast and our knowledge is so limited, we’re not in a position to know if the world’s various evils are outweighed by greater goods. I additionally pointed out that hiddenness can’t count against God unless one thinks that the fact that so many believe in God counts for God—something that one wouldn’t expect if all they knew was that there was no God—and so these arguments, once we’ve accounted for evil, mostly cancel each other out.

Bryan accused me of using a double standard, claiming that evil doesn’t count against God because we can’t forecast infinitely far into the future but that good counts for God. But this isn’t so, as I pointed out. My arguments for God weren’t about specific features that might produce great benefits—like, say, the existence of nuclear energy which powers homes. They were about much broader features about the world, such as its ability to maintain agents, which are required to have any value in the first place. Saying that you need agents to be in a world for it to be valuable doesn’t involve making any kinds of specific predictions about which things will be good many years down the line.

Additionally, I agreed evil was evidence against God. I claimed merely that it was outweighed by the evidence for God.

Finally, the kind of skepticism that I defend means that one shouldn’t be super confident that theism would predict anything. Maybe you should only be, say, 60% confident that theism would result in conscious agents. But because the odds conditional on naturalism are around 0%, conscious agents still give evidence for naturalism. You don’t need theism to make any particular thing especially likely for things that are very unlikely on naturalism to favor theism.

Bryan additionally claimed that omnipotence is paradoxical because if a being is omnipotent they can create a stone too heavy for them to lift. But this is like objecting that the North poll isn’t far enough North! It’s logically impossible to create such a stone—it involves, quite straightforwardly, a contradiction. It’s not a limit on God’s power—for God is the most powerful possible being.

Thus, I think my first main argument was not satisfactorily addressed by theism.

3 Agents, shmagents

Conscious agents exist. I’ve seen one. I am one. But why? Theism has a nice story to tell about it—conscious agents are either the only things of value or the primary things of value, so God would create them. In contrast, naturalism has a vastly implausible story, wherein there happened to be physical stuff, laws, the laws applied to the stuff, they generated interesting things, the constants were finely-tuned, there were laws that gave rise to consciousness, they produced interesting consciousness, they produced consciousness that paired with the world, and the things (brains) that give rise to consciousness existed. Each stage is unlikely, so because naturalism has to posit that each of them happened, it is very unlikely.

Here Bryan’s responses were, totally unconvincing. He suggests that because we’ve never measured the laws of the universe change we should assume that they’re necessary. Because they’re necessary, they can’t be evidence for theism.

But this is crazy. For one, it would imply that if the laws of nature made every atom have in small font “made by God” that wouldn’t be evidence for God. I pointed this out to Bryan. He agreed that it wouldn’t be evidence for God. Now, if one’s level of confidence in atheism is so extreme that atoms saying made by God doesn’t make them think theism is even a bit more likely, it’s hard to know what to say to such a person.

Second, this is just totally confused as a matter of probability. If you observe something always happens, that might give you evidence that deviations from it are physically impossible. But physical possibility just denotes consistency with the laws of physics—pointing out that deviations from the physical laws are physically impossible tells you nothing interesting. Of course, different physical laws are inconsistent with the physical laws—but it’s mysterious that the physical laws, as well as half a dozen other things, fell in an extremely narrow and unlikely range, so as to produce conscious life.

Bryan in response to this claimed that kinds of possibility other than physical possibility don’t matter much to him. But those are how one assesses prior probabilities of things. If you observe some event, and one theory makes it likelier than another theory, then it happening favors the first theory. This is standard Bayesian inference and remains so whether or not he happens to have an affinity for that kind of inference.

Bryan declared that we should count things like the laws of physics in the background evidence, and thus not update on it. But that’s a totally arbitrary deviation from Bayesian reasoning. The way to reason about things in general is to look at whether some theory explains them better than another.

To see this, imagine that there are two theories of physics. One of them says that there are these very simple strings—I don’t know enough physics to know if this is actually the story with string theory, but I think it might be—that vibrate and when they vibrate, you can show mathematically that that would result in the various laws and constants being like they are. The other theory says that the law happened to be there for no reason, and be at the values it is at for no reason. The first theory would obviously be better. But by Bryan’s logic, you can just declare the laws necessary and then not take the first theory’s superior explanation of them to count in favor of it.

In fact, elsewhere, Bryan accepted this broad way of reasoning. He accepted, for example, that evil was strong evidence against theism. But if evil is strong evidence against theism, because theism makes evil less likely than naturalism, then improbable but good things must be evidence for theism.

Here’s another example I gave in the debate: suppose you accept necessitarianism, where you think that whatever happens is necessary (both physically and metaphysically, because determinism is true). That means that, on Bryan’s view, getting 500 royal flushes wouldn’t be evidence that you’re cheating because you getting those royal flushes is physically necessary—it can’t be otherwise consistent with the physical laws. Nonetheless, if a gambler got 100 consecutive royal flushes, not even Caplan would buy this excuse.

He next asserted, in response to the point about the finely-tuned constants, that we haven’t measured the constants to any precise degree. Even if he were right about this, the other things needed for moral agents would still be evidence for theism—like laws, interesting laws, constants, and so on. Furthermore, he’s totally wrong and presented no evidence for this claim, while I cited the Stanford Encyclopedia of Philosophy’s discussion of the consensus of physicists. I found his claim quite bizarre—we have been able to measure the cosmological constant, for instance.

Bryan later said that for something to count as evidence it needs to be a correct prediction of a theory. Well at least that’s what he seemed to suggest in his opening statement, but later agreed that other things can count. For example, we can be confident in historical facts even though historians don’t make novel testable predictions. He simply said that it takes very convincing evidence. But he never explained clearly why the long series of improbable events that theism naturally explains isn’t very convincing evidence.

Bryan later said that evolution explains consciousness because many of the things that we can use our conscious minds to do are things that mere matter can’t do as well. But this is like arguing that evolution explains gravity because if there weren’t gravity, everyone would fall upwards which would be bad for their survival. The mystery isn’t why beings evolved to be conscious given that consciousness has the effects it does. The mystery is that there happen to be, as Bryan agrees, as a dualist, fundamental laws that make consciousness appear when certain brain states arise. That’s really weird. Even weirder, not only are there consciousness laws that say brains produce consciousness, the laws produce a rich and complex consciousness that can interact with the physical world harmoniously, and there also happen to be brains. That’s really weird—you could imagine much simpler or epiphenomenal consciousness, and as Crummett and Cutter have shown, the overwhelming majority of possible pairings between the mental and the physical would produce nothing of value.

The mystery is that there are laws that give rise to consciousness. Evolution can’t explain the existence of fundamental laws like that, any more than it can explain H20 being water or gravity existing.

In response to this, Bryan leaned into the microphone and said, of the argument that evolution explains gravity, “That’s not that bad of an argument actually.” It’s unclear what to say to the claim that evolution explains why gravity, which predates evolution by billions of years, exists. My friend Amos had the appropriate response.

4 Moral and inductive knowledge

To know things about some subject there has to be a relationship between one’s knowledge and the facts themselves. For instance, I can know that there’s a table because I see it. However, if I know I’m hallucinating the table, so that I’d see it even if it weren’t there, then I can’t justifiedly assume there’s a table based on seeing it. Therefore, to have moral knowledge, or knowledge that induction works, there has to be a relationship between the moral facts and our knowledge. But moral facts can’t move around stuff physically and neither can facts about the future affect our current state of knowledge, so therefore, to have knowledge that the future will resemble the past or about morality, we must have some non-physical faculty by which we grasp the moral facts and facts about induction. But theism’s a much better explanation of that—you wouldn’t expect that to exist on naturalism.

Bryan claimed that it was circular to use induction to justify induction. But I wasn’t doing that. I claim that we can know induction is right because it seems right, and then theism explains why we have brains that have accurate seeming about that. No circularity involved.

He claimed additionally that evolution makes us smart which allows us to figure out the moral facts. But this doesn’t address the argument. Even if evolution makes us smart, unless we have the ability to grasp non-natural facts, we can’t justifiedly believe in induction and morality. So clearly we can grasp non-natural facts through a non-physical faculty of intuition. But that’s best explained by theism.

Just like it wouldn’t make sense to say that evolution makes us smart, and therefore able to know what’s happening thousands of galaxies away in places we can’t see, because there’s no mechanism connecting our knowledge to the things happening in far-away galaxies, it doesn’t make sense to say that evolution making us smart explains the moral facts.

These were the only things Bryan said about that argument. So, if I’m right, nothing he said was at all convincing or responsive.

5 The anthropic argument

"I came seeking a challenge. All I found was you."

—Zurgo, khan of the Mardu (flavor text on the Magic: The Gathering card Utter End).

The anthropic argument claims that theism best explains why you in particular exist. There are uncountably infinite possible people, so if there is no God, you’d only expect a few of them to exist—maybe infinite people if we’re lucky. But given that the number of possible people is a very large infinity—at least Beth 2—if there is no God, only 0% of possible people would exist, and so the odds you’d be one of the lucky few people to exist are 0%. In contrast, if there is a God, he’d either create all possible people, or some obscene number, so it’s much likelier that you’d exist. If 5x as many people exist, your existence is 5x as likely, if a view of anthropics called the self-indication assumption is right (which it is). Therefore, because creating a person is good (as Bryan’s written a book arguing), if God exists, a huge number of people would be created, so it would be much likelier that you’d exist. Therefore, your existence is very good evidence for God—in fact, your existence gives you infinitely strong evidence that uncountably infinite people exist, which is much better explained by the existence of God than alternatives.

Bryan boldly declared this one of the worst arguments he’d ever seen—in the bottom 10. His responses, however, were super lame—some of, dare I say, the worst responses I’ve ever seen; perhaps in the bottom ten. Every philosopher I’ve made the argument to has found it at least somewhat interesting and thought that it might work. This is just to say that if you think the argument is just obviously dumb, you’ve probably misunderstood it. Which Bryan did—egregiously.

He first claims that my view involves mangling probability. Whoever it is that exists can reason that it’s unlikely that they would exist if few people exist. Therefore, claims Bryan, my view violates the conservation of evidence—it results in whatever happens people who exist having a higher credence in more people existing.

Now Bryan is correct that probability was mangled badly in the debate, but incorrect about who mangled it. It’s true that if people follow my reasoning, some people will be misled. Whether or not more people exist or not, the odds that some people would exist are unchanged. However, the odds that any particular person would exist are higher if more people exist than if fewer people exist.

Here’s a helpful analogy: suppose that everyone in the world goes to sleep. Then a coin is flipped. If it comes up heads, everyone in the world is woken up. If it comes up tails, only five people are woken up. Suppose that you wake up—do you get evidence that the coin came up heads? Answer: yes, obviously. Yet by Bryan’s logic, you wouldn’t get evidence because some people would wake up either way. While this is right, the odds that any particular person gets woken up are higher if the coin comes up heads. Similarly, the odds you in particular would exist are higher if more people exist. Bryan’s response thus fails by taking an insufficiently specific look at the evidence—when looking at the odds of some evidence given two hypotheses, by the law of total evidence, you have to look at the most specific version of the evidence.

Furthermore, if Bryan’s logic was correct, it would produce the kind of absurd results that I describe here, which I also gave in the debate, in my opening statement, wherein a person can be confident that they won’t get pregnant because if they did get pregnant, they’d have a bunch of kids. This is absurd. So we therefore have multiple devastating arguments against Bryan’s view.

Bryan next asserts that the argument gives you reason to think that infinity people exist. But because they don’t, it’s wrong. But we have no evidence that infinity people don’t exist—we can’t see other worlds in the multiverse. So therefore Bryan’s view is just begging the question—it just assumes the argument fails without giving a reason why it does.

He finally claimed that the improbable existence of you can’t be evidence for theism because every event is improbable. This would be like saying that theism gets support from the fact that the chair is in the location that it is—while that’s very unlikely if there is no God, it’s not evidence for God, because something had to happen.

But this is confused! Obviously the mere fact that an improbable event happens doesn’t favor theism. It only favors theism if the event is more likely on theism than naturalism. Bryan’s response is like a gambler who gets 100 royal flushes in a row claiming that, because some sequence of cards had to be flipped over, the 100 royal flushes is no evidence that he’s cheating. Of course it is—getting that many royal flushes is much likelier if you are cheating than if you’re not. Similarly, your existence is likelier if there is a God than if there isn’t, so it massively favors the existence of God.

The reason it doesn’t make sense to believe gerrymandered theories that explain random improbable events is that the theories are as improbable as the events. While you could posit, after getting some particular sequence of cards, that a leprechaun rigs things to get that sequence of cards, because a leprechaun could rig the cards in any way, the odds they’d rig it that way are just as low as the odds one would get that sequence of cards by chance. However, if the cards turn out a way that’s likely if the cards are rigged—say, the way that happens to facilitate a particular magic trick—but unlikely if they’re not, then of course that’s evidence for rigging.

So despite Bryan’s claim that this argument was terrible, it turned out to be his responses that were terrible. This has been a trend I’ve noticed—people who are more philosophically-informed tend to take the anthropic argument more seriously (a bit like psychophysical harmony).

6 Takeaways

So I guess the main takeaway is that God’s not dead, he’s surely alive, he’s living on the inside roaring like a lion, etc. I’m pretty happy about how the debate went, though I wish I’d made my case better, but I think it went pretty well overall. I think I was pretty well-prepared, which made the debate fun and didn’t make me that nervous. All in all, I think my arguments leave the encounter relatively unscathed.