Soul-Making Theodicies and Lack-of-Character Data

Soul-making theodicies aim to defeat the problem of evil. In broad outline, they argue that moral virtues  (e.g., patience, kindness, compassion, etc.) are among the greatest possible goods, and that God must allow suffering in order to give us the opportunity to develop virtue (e.g., developing patience requires undergoing hardships; developing courage requires facing danger; developing compassion requires experiencing suffering yourself (to empathize) and seeing and responding to the suffering of others, etc.). Therefore, God is justified in permitting evil or suffering in order to allow for these goods.

The problem is that, as John Doris and others have recently argued, there is a robust set of data regarding human behavior that casts serious doubt on the hypothesis that humans have the capacity to develop virtue. And if that's right, then soul-making theodicies are thereby undercut.

It seems to me that the point can also be used as the basis for an argument against theism. For by similar reasoning, theism predicts an arena for free moral choices which in turn serve as the basis of moral development. It's therefore surprising on theism that character formation for virtue is ineffective. By contrast, such phenomena is not at all surprising on naturalism. For on that hypothesis, there is no antecedent reason to think evolution would aim at producing bodies capable of cultivating stable virtuous character traits. Therefore, lack of character data provide at least some confirming evidence for naturalism vis-a-vis theism.

Two Notions of Necessity (and the Theistic Arguments that Conflate Them)

There are two notions of necessity floating around that easily get conflated: (i) exists in all possible worlds, and (ii) can't not exist. But (ii) can't be captured by (i); (ii) is more fine-grained than (i).  Indeed, it's epistemically possible that a being is necessary in sense (i), but not in sense (ii). 

To see this, say that a world stub is some initial temporal segment of a possible world (whether beginningless or not). Now consider that it it's epistemically possible for a god (an uncreated, metaphysically independent being) G to exist in the world stub of every possible world, and yet go out of existence at some time downstream of the world stub of at least one -- but perhaps many, and perhaps even every -- possible world (say it commits suicide due to eternal boredom, or it's annihilated by some other being downstream of one or more world stubs). It's therefore epistemically possible for G to be necessary in sense (i), but not in sense (ii).

This has non-trivial implications for some theistic arguments. Some contemporary theistic arguments --  "minimal modal ontological arguments" (as van Inwagen defines them), certain Leibnizian cosmological arguments, etc. --  deploy S5 modal logic to show that an Anselmian being currently exists. In particular, they aim to show that

1. A necessary being exists in at least one possible world.

and then infer from (1) and Axiom S5 of S5 modal logic to infer that

2. A necessary being exists in every possible world.

And finally, from (2) they infer that

3. A necessary being exists.

Now of course many -- myself included -- have raised doubts about (1). But the preceding discussion raises a problem for the inference from (2) to (3). For as we've seen above, (3) doesn't follow from (2). Therefore, even if one establishes that there is a necessary being in the sense captured by sense (i) above -- viz., the necessity operator of modal logic --, one has not thereby established that such a being currently exists. And because of this, theistic arguments of the sort mentioned above that rely on an inference from (2) to (3) to establish God's existence are bound to fail.