The Probabilistic Problem of Evil: Doesn't Evil at Least Make God Less Likely?

Maybe the existence of bad things in the world doesn't completely disprove God's existence. Fine. But surely, at the very least, it makes God's existence less probable, right? Let's consider it.

Lowering the Bar

Up to now, we've examined the Logical Problem of Evil, which tries to show God's existence is flatly impossible given evil — that there's no possible reason God could have for allowing it. We've spent several episodes on candidate reasons (free will, soul-making, love, the megaphone, skeptical theism), and it's fairly widely accepted, by theists and atheists alike, that at least some of these show God's existence isn't logically impossible.

So what if the atheist lowers her sights, and tries only to show God's existence is improbable? This traces back to philosopher William Rowe, whose landmark 1979 paper "The Problem of Evil and Some Varieties of Atheism" reframed the debate as the evidential (rather than logical) Problem of Evil. Rowe's famous illustration was a fawn trapped in a forest fire, dying slowly over days with no one around to witness or benefit from its suffering. Rowe argued this kind of pointless animal suffering makes it rationally more likely than not that God doesn't exist, even without strictly proving it — probabilistic, evidential reasoning, the kind scientists use.

The Thought Experiment

Imagine standing outside the universe as a neutral observer at the moment of its birth. You think an intelligent, benevolent being is causing this; I think it's just indifferent natural forces. Who's right? Both hypotheses posit invisible entities, so we can't check directly — but we can make predictions and see whose come true. I, holding the Hypothesis of Indifference (HoI), predict a mixed bag: famine, earthquakes, disease, extreme suffering. You, holding the Hypothesis of Theism (HoT), predict something closer to paradise. We watch it unfold, and evil shows up — lots of it. My prediction succeeded.

1. Hypothesis I predicts state of affairs E (evil).
2. Hypothesis T predicts ~E.
3. I and T have equal prior probability.
4. If E obtains, I is more probable than T.
5. If ~E obtains, T is more probable than I.
6. E obtains.
7. So, I is more probable than T.

(Two footnotes: technically HoI only denies God created the universe, and HoT claims more than mere theism — but they're close enough to treat as atheism vs. theism here. And remember, unlike the Logical PoE, which tried to show theism impossible, this only tries to show atheism more probable — a far more modest goal.)

Objection: Skeptical Theism Changes the Prediction

Here's where Skeptical Theism comes roaring back. If God is infinite, He may have reasons for permitting evil we can't anticipate. If so, theism wouldn't predict a paradise with no evil — it would simply abstain from predicting either way. That doesn't sink the project, just retools it:

1. Hypothesis I predicts E.
2. Hypothesis S abstains from predicting E or ~E.
3. I and S have equal prior probability.
4. If E obtains, I is more probable than before.
5. If ~E obtains, I is less probable than before.
6. Either way, S's probability is unaffected.
7. E obtains.
8. So, I is more probable than S.

Even without a specific prediction, evil showing up should still tip things toward indifferent forces, since S can never "win" from it — only stay neutral.

Qualification: Not in Isolation

One caveat first: premise (3) — equal prior probability — only holds if we pretend there are no other arguments for or against God floating around, which isn't true. Arguments for God's existence split into two kinds: those aiming at logical necessity (proving something strictly impossible or guaranteed) and those aiming only to shift credence up or down, to be weighed against everything else. The teleological (fine-tuning) argument is squarely the second kind. By some counts there are over thirty arguments historically offered for God's existence and only two commonly cited against Him (the Problem of Evil and the argument from science). If we take this argument seriously as one piece of evidence among many, the conclusion has to soften to something like: if the boost this gives atheism outweighs everything pointing toward theism, then atheism comes out ahead.

Objection: Mutual Exclusivity

Here's a sharper problem. Indifferent Forces (I) and Skeptical Theism (S) are being treated as mutually exclusive explanations of the same data — so saying evil makes I more probable automatically means it makes S less probable. But that contradicts what Skeptical Theism claims: that God permitting evil is a live, expected option. If evil is already within S's anticipated range, it can't rationally count against it.

Picture two bakers, Indigo Forces and Skip Theism, sharing one oven. A box of pastries comes out; we don't see who baked it. Indigo loves eclairs; Skip bakes eclairs and everything else indiscriminately. I predict eclairs; you say it could be eclairs, or honestly almost anything. It's eclairs. I say, "Aha, must've been Indigo!" Should you buy that? No — Skip bakes eclairs plenty too. Matching my narrow prediction doesn't make the outcome improbable on your broader hypothesis. Likewise: evil obtaining can't rationally make Skeptical Theism less probable, since Skeptical Theism already expected it as a live option. The consistent conclusion is that evil doesn't shift the probability toward either hypothesis.

Objection: The Problem of Order

Here's an even more powerful flip. Go back to the birth of the universe. Would a neutral observer, seriously holding the Hypothesis of Indifference, actually predict complex, ordered life — planets, ecosystems, intelligent beings building technology — with just some bad stuff mixed in? Or would they predict a formless, chaotic mess going nowhere? Honestly, if I were purely committed to unguided indifferent forces, I wouldn't predict stars and planets coalescing, and I'd be especially surprised by intelligent life. These hypotheses have to account not just for the bad in the universe, but for the staggering amount of order and good in it — and indifferent forces don't obviously predict that.

1. Hypothesis I predicts ~O (disorder).
2. Hypothesis S abstains from predicting O or ~O.
3. I and S have equal prior probability.
4. If ~O obtains, I is more probable than before.
5. If O obtains, I is less probable than before.
6. Either way, S's probability is unaffected.
7. O (order) obtains.
8. So, S is more probable than I.

Same structure as before, flipped entirely — because our universe turns out to be stunningly, improbably orderly.

Reply: isn't this special pleading? You might say, if evil obtaining didn't count against Skeptical Theism, order obtaining shouldn't count against Indifferent Forces either. But the cases aren't parallel: Indifferent Forces specifically predicted evil, and evil obtaining is at worst a tie for both hypotheses. But Indifferent Forces specifically predicts disorder, while Skeptical Theism has order well within its expected range — so order obtaining is a real failure for I and a real success for S. Back to the bakery: if the box held an elaborate Opera Cake — something Skip would plausibly make but Indigo, who only loves eclairs, wouldn't — that genuinely would favor Skip.

Reply: "self-organizing" forces like natural selection can produce order without a designer. But the whole thought experiment depends on pretending we don't yet know how history played out, so we can ask what we'd predict in advance. The moment you specify "there's a strong force, weak force, electromagnetism, gravity of exactly this strength, a singularity of exactly this size" — you're no longer predicting anything; you're reverse-engineering a description that fits what already happened, like a kid filling out an "All About Me" poster and marveling that it's accurate. "I predict Boston Terriers will exist... wow, they do!" isn't a prediction if you built the answer into the hypothesis. And Skeptical Theism could play the exact same trick ("God would create Boston Terriers — look, they exist!"). If the move is illegitimate for one side, it's illegitimate for both. The only fair version of the thought experiment is the genuinely blind one.

Where This Leaves Us

The Probabilistic Problem of Evil, on its own, runs into the mutual-exclusivity problem — it can't count evil against Skeptical Theism without contradicting what Skeptical Theism already predicts. And once we widen the lens to the order in the universe, the same predictive logic swings hard the other way. This doesn't settle the debate by itself — it's one piece of evidence among many — but I don't think the Probabilistic Problem of Evil, taken alone, does the work its proponents hope it does.