Follow The Converging Lines of Evidence
Reflections on Douthat's article and why you should believe in God
Ross Douthat recently wrote about me in the New York Times. That’s not a sentence I expected to write. Hello to the roughly 200 of you who subscribed from Ross’s article! I’m especially pleased to have Ross writing about me because I’ve always found him quite impressive; most journalists make an ass out of themselves when they talk about philosophy, but Ross really doesn’t. On philosophy of religion, he’s consistently extremely sharp—and usually right!
(If other New York Times journalists would like to write about my blog, let the record show: I have never spoken an ill word about the Times!)
Anyway, Ross’s article was titled My Favorite Argument for the Existence of God. Shockingly, it’s about his favorite argument for the existence of God.
After briefly mentioning a few arguments that he was fond of—fine-tuning, the argument from religious experience, arguments from miracles—he settled down on the one he liked most: the argument from the discoverability of the universe. Under theism, we have some reason to expect to be able to discover the universe; if the universe was made for us and creatures like us, it’s not unlikely it would be the sort of thing we could figure out. On naturalism, the world is just the blind unfolding of physical laws; there’s no reason to expect that we’d be able to figure out the detailed and complex interworkings of it. As Einstein famously said “The most incomprehensible thing about the world is that it is comprehensible.”
I think this is a pretty good argument. It has even more force in light of the following four considerations:
Many of the things that we can understand—morality, facts about which states of affairs are possible, facts about the relative probability of different states of affairs each of which explain our experiences—by their very nature are not the sorts of things that can move around atoms. If, therefore, you think that our beliefs all come from the unguided movement of atoms in our brain, without some being making sure they track the truth, it would be shocking if these beliefs were correct. I’ve been meaning to write in more detail about this argument—the technical details get a bit tricky—but I think this has a lot of force.
On atheism, even if the universe is in principle discoverable, there’s no reason at all to expect we’d be able to discover it. If we are bipedal apes that evolved purely by chance (to be clear, I think we are bipedal apes that evolved—just that God set things in motion so that we’d develop certain capacities), there’s no reason to expect us to be the sorts of creatures that could figure out the universe. It’s surprising that the capacities that enabled us to survive and reproduce on the savanna also allow us to carry out the highly advanced mathematics needed to discover the universe.
Even if we do have the capacities described above, it’s surprising on atheism that the universe would the kind of thing that could be deeply investigated. Even with perfect brains, there are some things we’ll never know; how many grains of rice were eaten in 1102, for instance. Most human languages that are lost will likely never be discovered, so long as they didn’t leave behind any writing. Mark Steiner has written in some detail about the surprising connections between our mathematical practice and the universe; if he’s right, the universe’s mathematical structure was discoverable by us to an almost miraculous degree. I’m not super sure about this argument, as the physics mostly went over my head, but if it’s right, it shows another surprising way the universe is discoverable.
Lastly, Robin Collins has argued that certain parameter values in physics fall in an extremely narrow range ideal for scientific discoverability. For example, the baryon to photon ratio determines the value of the cosmic microwave background radiation (CMB)—the higher the CMB, the more discoverable the universe. The baryon to photon ratio is about 1:1 billion, which happens to be within the margin of error of the ideal value for maximizing the CMB. Now, once again, I’m hesitant to place too much weight on this because I don’t understand the physics, but Robin is very smart, likely the world-leading expert on fine-tuning, and has a Ph.D in physics—if he’s right, this is very impressive.
I thus agree with Ross that the discoverability of the universe—the fact that we can figure out the deepest level goings-on in the universe—is very impressive evidence for theism. But Ross made a different point; it was something I’d been meaning to write an article about, so I thought this would be a good opportunity. Specifically, he wrote:
I think that the most compelling case for being religious — for a default view, before you get to the specifics of creeds and doctrines, that the universe was made for a reason and we’re part of that reason — is found at the convergence of multiple different lines of argument, the analysis of multiple different aspects of the existence in which we find ourselves.
This is quite an important thing to emphasize: when it comes to arguments, cumulative cases are almost always better than single arguments.
I have a friend who is a young earth creationist. He’s an interesting guy; very posh and British, but with the most batshit insane views of American evangelicals (thinks evolution didn’t happen, the earth is only 6,000 years old, the Catholic Church is the “Mother Of Harlots And Abominations Of The Earth,” discussed in the Book of Revelation, and that the King James Version is the only inspired English translation). I was recently arguing with him about evolution (~10,000 word upcoming article arguing for evolution coming out soon!)
Now, he had ways of explaining away each of the many lines of evidence favoring evolution. Each of his explanations added an epicycle, none was plausible, but he had some response to each of the kinds of evidence for evolution. What makes his view so implausible isn’t so much that any single argument decisively killed it—though some individual arguments massively lower its probability—but that it is opposed by a broad convergence of considerations each pointing in the same direction. He had to explain away the evidence from:
Nested hierarchies.
Atavisms.
Transitional fossils.
Molecular evidence.
Embryological evidence.
Radiometric dating.
Ice cores.
Distant starlight.
And much, much more. When a view about how the world came to be is false, usually there are lots of bits of evidence that point against it. If evolution happened and explains the development of every life-form, we should expect there to be very clear remnants of that—we should expect there to be perhaps a hundred different lines of evidence for evolution. We should expect, in the words of Dobzhansky, “Nothing in Biology [to] Make Sense Except in the Light of Evolution.”
If a view purports to explain a lot about what the world is like, and it is false, we should expect many things about the world to testify to its falsity. If there is no God, therefore, we should expect there to be many converging lines of evidence against God’s existence. If God does exist, however, we should expect many converging lines of evidence to point in the direction of a God. There are many things about Harry Potter that make it quite clear it had an author.
But we have to be very careful in looking at converging lines of evidence. Often people act like there are many converging lines of evidence for a conclusion, when really they’re just quintuple counting the same basic kind of evidence. One arguing for a global flood could claim that there are many converging lines of evidence from all the ancient flood myths, but really this is just one kind of evidence being counted multiple times—whatever explains why the first four civilizations had flood myths will likely explain why the other ones did.
I use the term “evidence for a view” to mean stuff about the world that is likelier if the view is true than false—if your blood is found on the murder weapon, that’s evidence you committed the crime, because it’s likelier your blood would be on the murder weapon if you did commit the crime than if you didn’t. So I’m not counting, for instance, arguments about the complexity or incoherence of God. While I think these have some force—it would be strange if there’s a limitless agent at the core of reality—it would also be quite strange if ultimate reality is just a set of lawful, physical equations. Whether atheism or theism is true, fundamental reality is weird as heck; I don’t think you should think each one starts out more than an order of magnitude or two likelier than the other before you look at the evidence.
Taking this into account, in my view, there’s really only one strong piece of evidence for atheism: evil. More specifically, the fact that the world sucks in numerous ways. Tsunamis sweep innocent grandmothers into the ocean; earthquakes leave young children trapped under rubble; the evolutionary process has resulted in quadrillions of animals suffering horrendously for almost a billion years! There are other bits of evidence people give for atheism—widespread religious confusion, divine hiddenness, and so on—but in my view, these are broadly just variants on the wider theme of “things suck in lots of ways.”
I think this argument has a lot of force. Under theism, if you weren’t told what the world was like, you’d expect it to look like this.
Instead, it looks like this:
Obviously I’m oversimplifying a bit. But the general point is important and quite evidentially significant: much about the world is very bad, and this is surprising if there’s a perfect God.
Still, it’s not like there’s nothing theists can say about why God would make a world like this one. I’ve given, in various places, explanations of why God permits a world with so much suffering. And while my proposals are all pretty speculative, I think they’re enough to make it so that evil, while a major piece of evidence against God’s existence, doesn’t single-handedly make theism unreasonable. This is especially true given that there might be all sorts of unknown reasons for God to permit evil. Just as Magnus Carlsen makes chess moves that I can’t see the reason for, God, who has an infinitely long-lasting future plan in mind and knows infinite things I don’t, may have a good reason for allowing the evils of our world.
Crucially, I don’t think there’s much of a cumulative case against theism from all the world’s evils. Whatever it is that explains why the first nine things about the world suck will probably also explain why the next 500 things suck—whatever explains why God permits a world with bone cancer in children will likely also explain why he permits a world with floods, hurricanes, and people being ignorant of his existence.
Thus, I think there is one very good argument for atheism. But there is not any kind of broad convergence of lots of different considerations that all point in the same direction. The evidence for atheism is not like the evidence for evolution; it comes primarily from a single argument.
What about the evidence for theism? In my view, theism has a much broader set of cumulative considerations in its favor. Specifically, theism explains:
The fact that you exist specifically out of the infinite array of possible beings.
The discoverability of the universe (see above).
And more! I haven’t even listed the most popular arguments for God—mostly because I’m not too sold on most of them. In my view, therefore, the evidence for theism looks a lot more like the evidence for evolution—like the evidence we’d expect for the view that explains fundamental reality.
I won’t discuss the above list of arguments in any detail, but instead I’ll just mention two of them: fine-tuning and consciousness. The fine-tuning argument proceeds from the observation that the constants of physics fall in an infinitesimal range needed for the formation of complex structures; if the cosmological constant was a bit weaker or stronger, no life or even interacting atoms could ever arise. If there’s a God who sets the values of the constants, this makes perfect sense—a God who cares about life would be expected to set the values of the constants to whatever is needed for life. On atheism, however, there’s no reason at all to expect the constants to take on the values they did—values that were needed for anything valuable to form. And while atheists can explain away the relevant data by invoking a multiverse, the more you have to add epicycles to your theory to explain surprising data, the worse your theory becomes (a multiverse has other problems).
The argument from consciousness claims that consciousness is much more expected under theism than atheism. In order to get consciousness, it first must be that—in shocking and almost miraculous fashion—certain arrangements of atoms somehow give rise to subjective experience. It’s very bizarre and surprising that collecting atoms into an arrangement like that of a brain gives rise to experience; a private, qualitative inner life that isn’t even in principle describable in the language of physics.
But even if there are certain arrangements of matter that produce consciousness, on atheism it’s a coincidence that those arrangements ever arise—even if brains give rise to consciousness, the atheistic worldview gives no reason at all to expect there ever to be brains. Now, again, the atheist can explain this—perhaps there are certain brute laws that make physical states give rise to consciousness—but it will be at the cost of extra epicycles. When I was an atheist, I felt as though I was repeatedly explaining away data, rather than accommodating it. While atheism is obviously much more tenable than young earth creationism, I felt like a young earth creationism, in that I had to sweep numerous independent lines of argument under the rug.
While the above list of arguments for God’s existence aren’t totally independent, they are mostly independent. Whatever explains why there’s consciousness won’t obviously explain why there’s fine-tuning or moral knowledge. Analogously, the lines of evidence for evolution aren’t completely independent, but they are mostly independent—they’re independent enough to make the case for evolution unbelievably decisive.
The evidence for theism looks the way we’d expect the evidence to look for the true theory of ultimate reality. If reality has an author, we’d expect traces of the author to show up throughout much of reality. In my view, this is precisely what we observe. Thus, while I take the arguments for atheism seriously, and regard the problem of evil as quite forceful, in my view it cannot match the overwhelming force of the powerful cumulative case for theism. God likely exists.
Seems to me that most/all of these arguments are actually extensions of fine-tuning.
You've narrowly defined fine-tuning to be about certain specific physical laws of the universe, but more or less everything you list is some variation on "isn't it remarkable that we and reality are configured in such a way to be doing this investigation".
A sufficiently powerful selection-effect/anthropic argument thus explains all of them. That is, if we are in a type of multiverse where not just the cosmological constant but also certain facts about the nature of consciousness, the structure of our brains, etc are variables, then all of this is explained at once.
I think the strongest case you could make for one of your arguments being outside this description is "the universe could simply have been too complex for us to understand". But this has its own selection effect -- of course the explanations we have so far are ones we can understand, because... we came up with them. We have found the keys under the street lamp. If there are rules beyond our ken, out in the dark, how would we know?
Anyway, imo you can object to the idea that a multiverse is more plausible than a God, but both can explain all of this convergently.
I don't know, it seems like most (maybe not all) of the converging lines of evidence are based on a set of extraordinarily contentious metaphysical assumptions: Humeanism is false, moral and axiological realism are true, SIA isn't broken, axiological stalking horses can't work, God can really be counted as simple and that divine explanations aren't merely shifting all the improbability from the likelihood to the prior, etc. I feel a lot more comfortable rejecting the disjunction of all these than the disjunction of all molecular dating methods that geologists use, where there's essentially 100% consensus and it really would take a miracle for every single one of them to be inaccurate.