Prior Symmetry and the "Evil-God Challenge"

According to Stephen Law's "evil-god challenge," the hypothesis that there exists an omnipotent, omniscient, and maximally good creator of the universe (henceforth, theism) is no more plausible than the hypothesis that there exists an omnipotent, omniscient, and maximally evil creator of the universe (henceforth, maltheism). In defense of this contention, Law argues that theism and maltheism are a priori approximately equally plausible (call this claim prior symmetry) and that the available evidence (especially the evidence regarding the distribution of good and evil in the world) does not favor one hypothesis over the other (call this evidential symmetry). From here, Law claims that (rightly) no one takes maltheism seriously. It is surely a very implausible hypothesis. But given prior symmetry and evidential symmetry (which jointly entail what Law calls the symmetry thesis, according to which theism and maltheism are approximately equally reasonable), it follows that the same holds true of theism. Thus, theism is a very implausible hypothesis that should be rejected.

In this short post, I want to focus on the prior symmetry claim. In Bayesian terms, prior symmetry amounts to the claim that the prior probabilities of theism and maltheism are equal (or at least are very close to equal). There are a number of objections one might raise against prior symmetry, but I will restrict myself to just one in this post.

The objection is this: Maltheism is just one version of a more general hypothesis that we might call imperfect theism. According to imperfect theism, God exists but does not have all perfections. This hypothesis contrasts with perfect theism, which says that God exists and has all perfections. Perfect theism is essentially equivalent to theism in the above sense. Now, suppose that there is prior symmetry between (perfect) theism and imperfect theism. Then there will be (significant) prior asymmetry between theism and maltheism. Why? As already mentioned, maltheism is just one version of imperfect theism, one of many versions in fact. There are many ways to be imperfect, and being maximally evil is just one of those ways. In addition to maltheism, there is also what we might call impotent theism (according to which God is exceedingly weak) and ignoramus theism (according to which God is exceedingly ignorant). There are also various versions of imperfect theism according to which God is good but not maximally good, evil but not maximally evil, powerful but not omnipotent, and knowledgeable but not omniscient.

Maltheism competes with all of these other imperfect theism hypotheses and so plausibly only gets a small portion of the total probability space of imperfect theism. Consequently, the prior probability of maltheism is significantly less than the prior probability of imperfect theism. So, if there is prior symmetry between (perfect) theism and imperfect theism (meaning that the prior probability of theism is equal to the prior probability of imperfect theism), it follows that the prior probability of maltheism is significantly less than the prior probability of theism, in which case Law's prior symmetry claim is false. If Law is then right about evidential symmetry, theism will handily come out on top of maltheism on total evidence, and Law's overall symmetry thesis is false. And with the falsity of the symmetry thesis comes the demise of the evil-god challenge.

If the foregoing is right, then while there may be an "imperfect-god challenge" (which is something that David Hume raises in his posthumously published Dialogues Concerning Natural Religion—see especially Part IX), there is no evil-god challenge. Of course, Law may say that the foregoing is not the right way to set up our partition of hypotheses and that there is not prior symmetry between (perfect) theism and imperfect theism. But since he is the one raising the evil-god challenge, the burden is on him to show that that is the case. For my part, perfect/imperfect seems like a more natural and fundamental division to draw such that if there is prior symmetry, it obtains with respect to this division and not with respect to the theism/maltheism division. I would like to further justify this contention in the future; for now, I shall end the present post here.