Why I Can’t Be An Atheist

Almost a year ago I started posting about my exploration of what I see as the strongest arguments for and against the existence of God, and I started writing these posts several months earlier than that. Almost half of the posts on my blog so far have been directed at this question. (And on top of that, apparently my average word count per post has been higher this year than last!) So it has been a fairly long project.

Looking back over all of the thoughts I’ve gathered, here is where I find myself: I cannot be an atheist.

I can imagine why someone might be an atheist. I can imagine how you might weigh the considerations differently so that atheism is reasonable, even. But for me, that is not an option. For me, the arguments for God’s existence substantially outweigh those against.

What I Have to Believe to be an Atheist

Here is what I would have to believe in order to be an atheist:

• The principle of sufficient reason is false, and the fact that the universe seems to work like it is true is just one huge coincidence. Abductive and inductive reasoning is invalid, and most of our scientific reasoning is undermined.
• There is ultimately no reason why the universe (or anything at all) exists, or is the way that it is.
• There is no explanation for the finely-tuned structure of the physical universe, except perhaps for vastly improbable chance.
• The existence of consciousness is inexplicable, and in fact, because of this, the connection between our experiences and any external reality becomes dubious.
• There is no such thing as objective morality, no right or wrong. Morals are just subjective impressions foisted upon us by our evolutionary and cultural history, and we have no fundamental obligation to follow them.
• There is no such thing as objective beauty. Any sense of the transcendent we have in the sight of something beautiful; any experience of awe and wonder at the world around us; is empty and illusory.
• There is even no such thing as objective rationality, and it is dubious whether our ways of reasoning, even deductively, are capable of reliably producing truth.
• In a universe void of the transcendentals of truth, goodness, and beauty, with no teleology in our creation, living insignificant lives in the overall scheme of things, and universally destined for non-existence no matter what we do, life is ultimately without value, purpose, or meaning. (And we can pretend otherwise, but it doesn’t change the reality.)

In contrast, in exchange for the meaningfulness, explanatory coherence, and firm foundation that theism brings to one’s worldview, I only have to accept these tensions in order to be a theist:

• I cannot fully understand why God would create a world containing evil and imperfection (though I can come up with some plausible reasons why he might do so, in general).
• I cannot fully understand why God would not make more complete revelation of himself universally available, instead revealing himself in particular places and times in history and allowing many false religions and ideas about him to propagate (though again, I can come up with some plausible reasons why he might do so, in general).
• I cannot fully understand how God has acted in regards to salvation, eternal life, and those who are excluded from that destiny (but logically, I see no barrier to things turning out in a way that accords with his goodness).

I could, perhaps, be accused of playing things up a little bit here for rhetorical effect. I am not saying that these conclusions are obvious and rationally compelling for everyone. I don’t think all atheists are inherently irrational. But at the end of the day, I do think they are mistaken. Under the light of the arguments for God’s existence, I find that atheism results in absurdity.

I admit haven’t spent as much time exploring and responding to the arguments against God’s existence on this blog as I would have liked to – there is much more that could be said (and has been said, elsewhere on the internet) regarding those subjects. And ideally, in a wrap-up post like this I would be including a quantitative assessment of the cumulative strengths of the reasons for and against belief in God, accounting for the degree of dependence or independence between the different arguments.

I will at least attempt to do that quantitative assessment, briefly, here. (“Briefly.” Ha.)

Quantifying Belief

In the comments on a post a few weeks back, one gentleman (who his own blog worthy of reading) suggested a high-level categorization of the arguments for and against God’s existence, and I think it is a good way of looking at things:

• Argument from Reality: theism is the best way to explain some fundamental features of reality (encompassing the cosmological, teleological, noetic, and axiological arguments).
• Argument from Imperfection: atheism better explains the fact that reality is full of imperfection (encompassing the problem of non-god objects, the problem of evil, and dysteleological arguments).
• Argument from Revelation: theism is supported by experiences of God and the evidence of God’s actions in history (encompassing the epistemological and historical arguments).
• Argument from Indifference: atheism better explains the apparent indifference of religious revelation (encompassing the problem of divine hiddenness, the problems of religious pluralism and religious disagreement, and the problem of exclusivity).

The argument from imperfection is paired against the argument from reality, and the argument from indifference is paired against the argument from revelation. (One further category, which I’m not going to consider in this analysis due to the difficulty of casting them as abductive arguments, would have the ontological argument on the side of theism, and contradictions in the concept of God on the side of atheism.)

Now what I want to do is run a Bayesian analysis of the odds for theism over atheism, considering each of these arguments in turn: reality (R1), imperfection (I1), revelation (R2), and indifference (I2). Using the odds-ratio form of Bayes’ theorem, and iterative use of the fact that P(A & B) = P(A|B)*P(B):

To make estimating these probabilities a more reasonable task, I am considering them all to be conditional on some level of background knowledge which frees us from having to consider the probabilities of many irrelevant specific contingent data about the world. In this way the focus can be on whether each worldview in consideration (theism or atheism) has the resources to explain the high-level facts we are interested in.

Even with that, it is hard to pin down some of these numbers, so I will actually model them as a distribution of probabilities. More on that below.

P(G)/P(~G):

First, I think it is fair to consider the prior odds of theism (and atheism) to be one to one, or approximately so. I mentioned Paul Draper’s “low priors” argument a few posts ago, but as I said then, I don’t think it is successful. The intrinsically symmetric alternatives that he considers, “source physicalism” and “source idealism,” both need to be further specified to account for the full range of data. And I see no reason to think that theism must be further specified from source idealism than a viable form of atheism must be from source physicalism.

So P(G)/P(~G) = 1.

P(R1|G):

Now, does the existence of God provide a solid ground for the fundamental features of reality (e.g. existence, appearance of design, consciousness, objective value) that need to be explained? Without considering any of the imperfections in reality – that is the next argument, not this one – I would say the answer is a solid yes. So I will set P(R1|G) = 1.

P(R1|~G):

The argument from reality is atheism’s weak point, in my mind. In order to explain these fundamental features of reality, it faces all the difficulties that I raised above. (And I’m trying to stick to the rational difficulties, but there are serious existential difficulties as well.) My initial impulse is to rate this probability no higher than 0.01, with 0.001 being closer to what I would put it at on most days. Just the failure of the PSR alone, I think, is worth putting it below 0.01.

That is probably a bit extreme, so I will ease off by a factor of 2 and say P(R1|~G) = 0.002 to 0.02. (Median value 0.006.)

P(I1|G,R1):

Of the considerations that could go under the argument from imperfection, I have only specifically addressed the problem of evil on this blog. But the others, the problem of non-god objects (why would God, a perfect being, create anything at all?) and dysteleological arguments (why would God create things that appear poorly designed?) do not add much weight to it, honestly. And given what I think are good responses to the problem of evil, I don’t think this probability needs to be much lower than 0.5.

But recognizing that the argument from imperfection does cause some tension for theism, I will estimate P(I1|G) = 0.1 to 0.4. (Median value 0.21.)

P(I1|~G,R1):

Assuming atheism could overcome the argument from reality, can it explain all the imperfections and evils that we see? Yes, completely. P(I1|~G) = 1.

P(R2|G,R1,I1):

The question here is whether it makes sense for God to provide some revelation of himself and whether theism can account for the kind of examples of revelation that we see. I think the answer is yes, in general, and that this probability is pretty close to 1. To make things simpler, any doubt about this can be transferred to a boost to the next probability. So P(R2|G) = 1.

P(R2|~G,R1,I1):

For me, this probability would be pretty close to 1 (making the argument from revelation quite weak) if it were not for the events surrounding Jesus’ death and the origin of Christianity. I do think atheism has difficulty explaining the evidence that we have for what happened there.

Nevertheless, assessing the argument for the resurrection is a complicated matter, and I am fairly uncertain about just how much force it has. My initial thought is to put this probability anywhere from 0.01 to 0.5. To offset the value of 1 given to the corresponding probability on theism, I will set P(R2|~G) = 0.05 to 0.6. (Median value 0.22.)

(*Note: in an earlier version of this post, I instead set P(R2|~G) at 0.15 to 0.75, wanting to be generous to the other side. But upon later reflection, I felt this really underrepresented my assessment of the strength of the evidence, even taking my uncertainty about that into account. With the original numbers, the median value for my overall probability for God’s existence was about 91.5% instead of 95.3%.)

P(I2|G,R1,I1,R2):

The question that theism has to face now is why God’s revelation of himself is apparently so ineffective and localized. Why would God not make his existence more obvious? Why is there so much religious confusion? How could God leave people in this state of uncertainty, and then condemn them to hell?

The kinds of considerations that I have raised in response to these things in that past few posts go a ways to reducing the tension that the argument from indifference brings to theism – but I find it more difficult to overcome than the argument from imperfection. Let us say that P(I2|G) = 0.05 to 0.3. (Median value 0.13.)

P(I2|~G,R1,I1,R2):

Finally, can atheism explain the apparent indifference of religious revelation? Again, it can do so perfectly. P(I2|~G) = 1.

For each of the four probabilities with a range of values, I have modelled them using a logit-normal distribution with the upper and lower values listed above set to the 10th and 90th percentiles. These are what the distributions look like:

 

And once these random values are put into the Bayesian equation, and the result converted into a distribution for the posterior probability that God exists (using 10,000 sets of randomly generated numbers to estimate the distribution), this is what I end up with:

This ranges from 82% at the 10th percentile to 98.8% at the 90th percentile, with a median probability of 95.3%. (An equal probability for theism and atheism, at 50%, is down below the 1st percentile in this distribution.)

So, there you have it. Given how I’ve weighted the above arguments, I should on average be a bit more than 95% confident that God exists.

Faith and Reason

This value (or distribution of values) represents an assessment of the strength of the rational justification for belief in God. Obviously, it is a product of several subjective judgements, and different people could weight the arguments differently and get a different number out as a result. On a different day, I myself might feel that different numbers are more appropriate than the ones I have given here. So this number shouldn’t really be thought of as any kind of precise determination of my level of belief in God.

But even more than that, while rationality is extremely important, at the end of the day I have to admit that it isn’t the only consideration that goes into forming one’s beliefs. There are existential considerations as well: considerations about finding meaning and purpose in our lives, and about how our beliefs are going to impact the way we live. So while I definitely think that there are good reasons to believe in God – 95% isn’t insignificant, after all – there is more than that. And the meaning and purpose of life that I find in the theistic worldview draws me, essentially, to commit to what I see as the truth.

The existence of God is too important, and the implications too far-reaching, for it to be practical to be indefinitely weighting the reasons for and against theism, holding that left tail of the distribution in mind and wondering if it will change. This gets back to something I wrote about at the very beginning of my blog: we need critical thinking, but we also need epistemic faith. When we see good reasons to accept a belief – as I see good reasons to believe in God – we should trust that, and not linger in unnecessary skepticism.

Basically, what I am saying is that I am willing to take that last 5% on faith, and I think it is proper to do so. I will trust what I have good reason to believe.

This does not mean that I think I should seal up the issue of belief in God and never subject it to scrutiny again. If something comes up that makes me think about how I’ve weighed the evidence, or if I come across some new consideration that could affect the balance, I am willing to take a look at it. (Granted, given the scope of my exploration of this issue so far, I admit that I find it unlikely that something will ever impact my belief so much as to make me change my mind. But I want those who disagree with me to be open to changing their mind, so I should probably exhibit the same attitude!)

So, that is why I am not an atheist. In my next post, I will write a little bit more on why I believe that Christianity, specifically, is true.