Why God Exists | The Intrinsic Probability of Theism

Theism is something like the view that there exists an omnipotent, omniscient, and omnibenevolent God. When I was in high school and college, I became interested in philosophy largely because I wanted to know whether theism was true. I had been raised as a Christian, but as a teenager, I was unsettled by arguments like those given by the “New Atheists” and their internet adherents. These arguments were mostly awful, but I didn’t see this at the time, and their proponents were so confident. At times, the arguments seemed so persuasive that I was nervous even about talking to other people about them. I thought that if I explained them to a pastor, say, and asked for guidance, they might say “Huh, I never thought of that” and convert to atheism themselves.

In principle, I guess I could have just ignored the arguments which bothered me. I didn’t. It would sound better to say that I didn’t because I bravely chose to value seeking truth over all else, even if it meant giving up comforting falsehoods and coming face to face with devastating truths about reality. But I think that, really, I’m just not temperamentally well-suited to being a theist–I am extremely skeptical, critical and pessimistic [1]–and this meant that, once the issue was raised, I couldn’t help second-guessing this belief which seemed so wonderful and on which I’d staked so much, and abandoning it if the reasons weren’t convincing. [2]

I suspect that the temperament I describe isn’t healthy, psychologically speaking, but it’s the one I have. Like many people with unhealthy temperaments, I wound up becoming a professional philosopher. My interests eventually expanded to include ethics and political philosophy, rather than just philosophy of religion. But along the way, I came to think that there was a good philosophical case to be made for theism after all. I’ll sketch part of that case in a series of three posts.


Before beginning, I will issue three caveats. The first is that this is only a sketch of part of the case. Each of the major points I make has had entire books, sometimes entire libraries, written about it. In my presenting my case, I will have to leave a lot unsaid, and to simplify, gloss over, or ignore certain details.  Further, there are some arguments which I won’t go into, such as the cosmological argument, the argument from religious experience, and the argument from consciousness. [3] (I think that the argument from consciousness, in particular, is interesting and important, but it probably requires that some form of dualism be true, and I am currently not sure how to present the arguments for dualism in an accessible, succinct, and persuasive way.) This is therefore, as much as anything, an invitation to consider certain lines of thought. Interested readers can find some additional resources on the topics I cover by clicking the hyperlinks I’ve embedded throughout.

The second is that I won’t address some important arguments against theism, most notably the problems of evil or divine hiddenness. My own research in philosophy of religion has mostly focused on these problems–either responding to them or exploring what it would take to solve them or even, in true skeptical fashion, trying to strengthen them. I think these are serious problems which provide some (not decisive, obviously) evidence against theism. I understand why some people find arguments based on them persuasive. While they need to be addressed, doing so isn’t my purpose here. (If you’re concerned about the problem of evil, I encourage you to read this. I have serious disagreements with it, but I think it is the one genuinely great book on the problem of evil which analytic philosophy–the kind of philosophy I do–has produced.)

Finally, a complete case would also involve a more thorough investigation of alternatives to theism than I can undertake here. Such alternatives involve philosophical naturalism–the view that the natural world is all there is–but also, for instance, the views of fundamental reality found in various non-theistic religions. Each of these views has its own merits and problems, and a complete case would involve identifying these and evaluating how they compare to the merits and problems of theism. But, again, this would be a massive project.

That being said, I think the case I outline is quite strong. I don’t expect it to convince everyone: evaluating philosophical arguments requires making subtle judgment calls, and there’s no getting around the fact that reasonable people of good will can wind up disagreeing. But when fleshed out and combined with some other things I would say if I had more time and space, it’s enough to convince even someone as temperamentally skeptical as myself. Perhaps it can also help convince others. In the rest of this post, I will consider the intrinsic probability of theism–the probability we should assign to God’s existence before considering the evidence–and argue that it should not be too low. In the next post, I will consider some important lines of evidence provided by science. In the final post, I will consider further evidence from our ability to obtain moral knowledge and from the events surrounding Jesus’ death and purported resurrection.

Intrinsic Probability

To see how probable theism is, we need to consider the evidence for and against it. But we also need to consider its intrinsic probability–its probability before taking any evidence into account. [4] To illustrate the idea of intrinsic probability, consider two theories, A and B. Theory A is that the fundamental physical equations which govern how matter behaves are the same, all throughout the universe. Theory B is that they are the same throughout the vast majority of the universe, but there is one tiny patch somewhere which, for no particular reason, is governed by totally different equations (maybe there is a planet where humans would float unsupported, where clapping two erasers together would cause a nuclear explosion, etc.).

Most of us think that accepting A would currently be much more reasonable than accepting B. But consider: our scientific evidence is about equally good for both A and B. If there is just one tiny patch governed by different equations, then it is (literally) astronomically unlikely that we would have found it, so our observations of unity are only astronomically weak evidence for A against B. If we really are justified in thinking A is much more probable than B, it must be because of something about A which A possesses prior to taking any empirical evidence into account, since the empirical evidence only negligibly favors A over B. In other words, A’s advantage is primarily because its intrinsic probability is higher.

Intrinsic Probability and Theism

To see more specifically how claims about intrinsic probability are relevant to the existence of God, consider that several popular lines of argument for atheism rely on claims about the intrinsic probability of theism. I can think of at least three types of arguments like this. One involves directly arguing that the intrinsic probability of theism is very low, so that we can be justified in believing it only if the evidence for it is very strong (if even then). Richard Dawkins’ so-called “Ultimate Boeing 747 Gambit” is an instance of this kind of reasoning. As far as I understand it, Dawkins’ idea is that God would need to be maximally complex, and theories are less intrinsically probable to the extent that they posit unexplained complexity. Accordingly, appealing to God to explain some phenomenon will always or nearly always be illegitimate, since theism will always or nearly always be less likely than other explanations, or than the phenomenon to be explained just existing without any explanation.

A more serious and sophisticated version of this type of argument comes from the philosopher Paul Draper, perhaps the keenest philosophical critic of theism today. Roughly speaking, Draper claims that theism is extremely immodest, in the sense, roughly, that it rules out very many other possibilities (it rules out naturalism, but also every possible form of polytheism, etc.). All else being equal, being compatible with fewer possibilities makes a theory less intrinsically probable. (For instance, all else being equal, it’s more likely that I draw a heart than that I draw a jack from a standard deck of cards, since there are more hearts, and therefore more ways the heart hypothesis can be true.) For theism to be rational, the evidence for it would then need to overcome both the evidence against it and its intrinsic improbability, which Draper thinks is implausible.

A second form of argument involves attempting to parody theism by arguing that some intentionally stupid alternative (the Flying Spaghetti Monster, etc.) both predicts the evidence for theism equally well and is no less intrinsically probable. Since we shouldn’t believe in the intentionally stupid alternative, and have no more reason to accept theism than to accept it, we shouldn’t believe in theism, either. If these intentionally stupid alternatives really do predict the evidence equally well, then showing the superiority of theism would require showing that it is more intrinsically probable than they are.

A third type of argument involves appealing to parsimony. The claim is something like this: theists believe in everything naturalists believe in (namely, the natural world), but they also believe in an additional thing (God). Because the naturalistic picture involves one less thing, it is simpler, and therefore intrinsically more probable. (Notice that, unlike the Ultimate 747 Gambit, this doesn’t require the claim that God would be immensely complex and therefore improbable, just the claim that the overall theistic worldview is more complex than the naturalistic one in virtue of containing an additional thing, regardless of how complex that additional thing is.) The objector claims that, because the theistic worldview is intrinsically less probable, the “burden of proof” is therefore on the theist to produce evidence for theism strong enough to overcome this improbability.

An Assessment

When I was younger, I worried a lot about arguments like these. (Well, I never worried about the flying spaghetti monster, but I worried about the others.) They made it seem like theism could be adopted only as a last resort, like every other alternative, no matter how intrinsically implausible or how bad a fit for the evidence, should be preferred if at all possible. And since, of course, it is always possible to reconcile just about anything with anything, this made it seem like any arguments for theism were doomed from the start. I had a strong a priori bias against theism, but I thought it was rationally justified. Was it?

In thinking about how to evaluate the intrinsic probability of theism, it may help to consider how we evaluate the intrinsic probability of A (that physics is uniform) and B (that there’s a weird patch out there somewhere). First, recall Draper’s point about modesty. A is far less modest than B, since there are many, many more ways that B could be true than that A could be true. Suppose we know which set of equations governs things around these parts; given that, there is only one way A could be true (that set of equations holds everywhere), whereas there are an infinite number of ways B could be true. This fact counts against A, but it clearly isn’t a decisive consideration. There must be something about A which is epistemically powerful enough to outweigh the fact that A is much less modest than B.

Unfortunately, there is philosophical disagreement about exactly which feature or features of A this might be, and I don’t plan to offer any very detailed account here. But it’s plausible that it has something to do with the fact that A is simpler than B, which I understand, roughly, as meaning that it takes much less to completely describe the fundamental elements of the theory in A’s case than in B’s. (Roughly, the fundamental elements are those taken as given by the theory, rather than explained by other elements of it.) Describing physics on A requires just one set of physical equations. Unless there is some way of subsuming the two sets of laws posited by B under some deeper set of laws, then describing physics on B requires describing the set of laws posited by A plus another set which governs some small patch plus a description of which set governs which place.

Another (possibly related) [5] point is that B seems to contain features which are arbitrary in a way which A doesn’t. If someone suggested that there was a patch of the universe governed by different equations, the natural question would be something like, “What’s so special about that area?” If the answer was “nothing, the laws are just different for no reason,” this seems like a good reason to reject that view. A deviation of the sort posited by B seems to cry out for explanation in a way that the uniformity posited by A does not. [6]

Analyzing Theism

What happens when we apply these criteria to theism? Draper is correct that theism is immodest, and that this lowers its intrinsic probability. But remember that simplicity and non-arbitrariness were more than enough to outweigh this feature in A. And I claim that theism possesses simplicity and non-arbitrariness in a way similar to A. Contra Dawkins, theism is extremely simple. It begins by positing a single being (God) with just one property (being absolutely perfect) which entails all of the being’s other essential properties (omnipotence, omniscience, etc.). It then suggests that all of contingent reality can be explained in terms of decisions God makes in light of these essential properties.

Perhaps Dawkins understands simplicity somewhat differently than I do. He elsewhere says that an object is complex insofar as its “constituent parts are arranged in a way that is unlikely to have arisen by chance alone,” and where that way has significance which is “specifiable in advance.” [7] Elsewhere he phrases the issue, not terms of the arrangement of parts, but in terms of having “some quality, specifiable in advance, that is highly unlikely to have been acquired by random chance alone.” The complexity of a theory would then presumably have something to do with the complexity of the objects it posits.

Employing this definition in the Ultimate Boeing 747 Gambit raises some worries. One is circularity: God is unlikely because complex, and complex because unlikely? Another is that it’s not clear how this could explain why A is simpler than B: what are the “parts” posited by B which are “unlikely to have arisen by chance alone?” What “quality” were the laws posited by B unlikely to “acquire?” But the funniest problem is that God isn’t actually complex according to this definition, either, since God isn’t supposed to have parts or to have “acquired” the divine properties.

Similarly, like A, theism posits uniformity rather than arbitrariness. Suppose someone put forward the following theory: there exists a being who is omnipotent, omnibenevolent, and knows about everything except water chestnuts, about which this being is mostly in the dark. There are a lot of things one might say to this, but an obvious one is, “What’s so special about water chestnuts?” If the answer is “nothing,” and the limitation is just being posited as an unexplained fact, this would be a major reason to favor theism over this theory. But just as there is nothing special about water chestnuts, neither is there anything special about positing a limitation in God’s knowledge concerning water chestnuts: positing any limits on divine power, knowledge, or goodness (apart from those which can be explained, such as God’s inability to make contradictions actual) raises the question of why God has precisely that amount of power, knowledge, or goodness in a way that traditional theism does not (both links are worth reading).

Serious Theistic Advantages

Considerations like these suggest that theism has a substantial advantage over intentionally stupid competitors, and that the popular parody arguments therefore fail. The Flying Spaghetti Monster, for instance, would be extraordinarily complicated and arbitrary. [8] It’s not surprising that these competitors possess theoretical vices, since they are intentionally stupid. But the fact that theism doesn’t possess these vices means they aren’t effective parodies of it after all.

In addition, the points made above similarly suggest that theism has an advantage over some serious hypotheses. For instance, they favor theism over forms of polytheism which begin by positing multiple gods with very arbitrary characteristics. Plausibly, they also favor theism over forms of naturalism which, for instance, trace all of contingent reality back to an initial singularity, but treat that singularity as a brute fact. While that singularity was much simpler and less arbitrary than the universe as it currently stands, it nonetheless must have had a number of diverse and quite arbitrary properties.

But wait–what about the parsimony argument, which claimed that the theistic worldview must be less parsimonious, since theists believe in the same physical world which naturalists do, but also believe in at least one additional thing? Notice that, if this argument worked, it would also be an extremely powerful argument for solipsism, the view that only I (or you, or whoever’s considering solipsism) exist: were I a solipsist, I would believe in nothing but my mind and my conscious experiences, whereas I presently believe in those plus all sorts of other stuff. We shouldn’t be tempted by arguments like this. I think the problem is that the person who claims that the theistic picture must be more complicated on these grounds misunderstands the kind of simplicity which is relevant to determining intrinsic probability. As I’ve said, this kind of simplicity has at least primarily to do with the simplicity of the fundamental elements of the theory.

It’s worth noting that Dawkins recognizes this. He writes that the goal of the biologist, in attempting to explain the existence of complex lifeforms, is to show how they could arise “from primordial objects sufficiently simple to have come into being by chance”–in other words, the biologist’s goal is to come up with an empirically adequate theory whose fundamental elements are quite simple, and can explain the complex ones. One of the virtues of evolutionary theory is that it can do this. I claim that something similar might hold true for theism.

Last Thoughts

So: theism is both simple and non-arbitrary, and that insofar as this is true, it has some of the theoretical virtues we should be looking for in attempting to account for evidence. Its intrinsic probability is not too low, and is probably higher than that of many of its competitors. There are therefore some reasons to place bets on theism right out of the starting gate. [9] But, of course, we do also need to consider the evidence. I will consider some of it in my future posts.


[1] I don’t mean to suggest that atheists tend to be those things. But being those things makes it harder to be a theist.

[2] In fact, people often assume that I am a militant atheist. When I told a member of my department who I’ve been friends with for many years that I was working on a series of blog posts about arguments for the existence of God, he was shocked to hear that I was defending them–despite the fact that I’ve told him before that I go to church! I think people get this impression for a number of reasons  (such as my politics, which are very left-wing), but that among these reasons is the temperament I describe.

[3] Another interesting statement of the argument from consciousness is Robert Adams’ “Flavors, Colors, and God,” found in this book.

[4] Two notes. First, we also need to consider how it fits with what’s called our “background knowledge;” it will be simpler to just paper over this and, for slightly technical reasons, it won’t make an important difference. Second, if it is the case, as many people (including me) think, that if God exists, then God exists necessarily, then the objective probability of God’s existence is either 1 or 0: either God must exist or God cannot exist. But what I am concerned about is the epistemic probability we should assign to theism prior to taking evidence into account. It is clear that sometimes we should assign an epistemic probability between 0 and 1 to a proposition’s being true, even when, in fact, it must be either necessarily true or necessarily false. For instance, suppose I think I vaguely remember hearing that a certain mathematical conjecture had been proven. Since the conjecture will presumably be either necessarily true or necessarily false, its objective probability will be either 0 or 1. However, my credence in the conjecture should be somewhere above, but not too far above, .5. If I have no evidence about the truth of the conjecture either way, it’s harder to say what my credence should be–maybe .5?–but it clearly shouldn’t be 0 or 1.

[5] It may be that the second criterion is ultimately just a part of the first–that what I’m calling arbitrariness makes a hypothesis less simple, and this is the reason it makes the hypothesis less likely. It doesn’t matter for our purposes whether this is true.

[6] If someone rejects the claim that something like these criteria are fundamental determinants of intrinsic probability, they face the question of how we can justify employing criteria like these in science, and of explaining why investigating the world using these criteria works so well. Since God would have some reason to make an orderly and elegant world, theism probably provides a better explanation of this than many other views, so that it might then constitute evidence for theism. Cf. this.

[7] The thought is that, if a tornado tears through a junkyard and creates a 747, this may not be less likely than its scattering the parts in any other, equally specific way. But the creation of the 747 has significance which is “specifiable in advance,” in a way that its scattering the parts in some equally specific but uninteresting way does not. So the 747 is complex in a way that another, equally specific and equally unlikely arrangement of parts is not.

[8] Of course, the Flying Spaghetti Monster can be given a fairly short description–i.e., he’s a flying spaghetti monster. But this isn’t a complete description. It doesn’t tell us–and what a world we live in, that I should have to write this–what sort of spaghetti he’s made of, how much spaghetti there is, the precise configuration of the spaghetti, etc., nor anything about his personality or intentions or history, and so on.

[9] For more on the intrinsic probability of theism, see ch. 6 of Joshua Rasmussen’s forthcoming book How Reason Leads to God.

Part 1   |   Part  2   |   Part 3

If you enjoyed reading this blog post, consider supporting us on Patreon!



Get updates on new posts, upcoming live discussions, and more.

Leave a Reply

2 Comment threads
1 Thread replies
Most reacted comment
Hottest comment thread
3 Comment authors
PeterDustin CrummettBrenda von Ahsen Recent comment authors
newest oldest most voted
Notify of
Brenda von Ahsen
Brenda von Ahsen

I see nothing on how you determined the intrinsic probability of theism. You will of course need first to define what kind of theism you mean and define what it’s central concepts are in such a way that they are flasifiable, coherent and contain no logical contradictions. You have not done so and until you do I am free to reject your claims.


To me the notion of divine simplicity fails because while it’s possible to wrap God’s supposed attributes in simple sounding words like omniscience, omnipotence, moral perfection and so on, when we drill down to what concepts these words really refer to, we inevitably end up in a huge amount of complexity. For example, we now know that what we have always called knowledge is actually an internal physical representation of some set of the external world stored in our brains. “Knowing things”, as we always understood it, turned out to be in itself a hugely complicated process that we’re only… Read more »