I've heard this argument before (it has sometimes been attributed to J. L. Mackie, but Mackie himself never articulated the argument in its full form or analyzed any of the objections to it), along with many others that rely on so-called "possible worlds" models, and I'm still baffled as to why anybody would take it seriously. It tries to draw a distinction along the following lines. Let [W1] and [W2] stand for two possible worlds as follows:
[W1] The actual world as we know it.The obvious (and simplistic) claim is that [W2] is a manifestly better world than [W1], in which some human choices are for evil rather than for good. The less obvious (because merely implicit), but more important claim, is that [W2] is a possible world.
[W2] A world that is identical to this world, only all human choices are freely for the good.
Clearly, this sort of proposal addresses only my category (c) of possible evils, so as a general response to the Catholic critique of PE it is worthless, but it certainly raises an interesting question: regardless of what the logical difference is between [W1] and [W2], what possible ontological difference could there be? In short, even if we grant that [W2] is logically possible, do we have any reason to grant that it is possible simpliciter, i.e., could such a logically possible world ever be an actual world?
Obviously, from an ontological point of view, [W1] is identical to [W2]. There is no reason at all why all human beings in [W1] could not always and freely choose only the good. So the only difference between [W1] and [W2] is that in [W1] some human beings have, as a matter of historical fact, sometimes chosen other than the good, while in [W2]...well, [W2] doesn't actually exist, does it? So in [W2] nobody has ever chosen anything, either good or evil. So one difference between [W1] and [W2], perhaps, is this: in [W1] it is clear that the agents really are free, whereas in [W2] we have no reason to believe that the agents are really free, other than the fact that it is so stipulated by the objection. The objection is, of course, a purely logical one, namely, that there could be a world in which some people freely choose only the good, but to take the libertarian notion of freedom seriously we need to move beyond the realm of purely logical possibility and ask what the connection to reality is supposed to be in imagining a world of free agents who do not, as a matter of fact, exhibit any evidence of being free. According to our mythology, even the angels, far more perfect beings than we, are manifestly free since many of them fell with Satan. I have yet to hear a convincing story about what talk of "possible worlds" could possibly amount to in the "real world", but it is a favorite of philosophers so we'll just have to make do. It is supposed to be a convenient way to talk about possibilities without committing oneself to anything, but as a method of discussing how things really are and why--that is, as an avenue towards explaining the actual--it is basically useless. There is, after all, a possible world in which gravity has the opposite effect to the one we're used to, but thinking about such a world tells us nothing at all about this world, or why gravity works the way it does rather than the opposite way. And yet some people, even some philosophers, continue to think that talk of "possible worlds" can be a useful explanatory hermeneutic for this world.
In one sense, the world that God created already is [W2] rather than [W1]. When God created the heavens and the earth, he saw that it was good. Adam and Eve (so the mythology goes) lived for a while without choosing evil before the time at which they did, as a matter of history, choose evil. So for some finite length of time from creation until the fall it would have been true of this world that human agents were free and chose only the good, and the distinction between [W1] and [W2] would have seemed meaningless. It is really a matter of historical accident that we happen to be living at a time when the question can be raised at all, and yet, of course, it could have been the case that the fall hadn't happened yet, in which case this version of PE would have seemed absurd. So let's posit two more worlds, [W3] and [W4]:
[W3] A world identical to this world except for the fact that, so far, no human agent has ever chosen evil.Empirically, there is no difference--right now!--between [W2] and either [W3] or [W4], so if it weren't for the first sin we would have no way of knowing which of these (and infinitely many more) possible worlds we actually inhabit. But of course, we don't inhabit any of those worlds, we happen to inhabit [W1], and it is this fact that supposedly gives the atheist his purchase against the theist in PE. But it is precisely the logical character of this whole scenario that does the atheist in in the end.
[W4] A world identical to [W3] except for the fact that on 1 January 2525, some human somewhere will freely choose evil rather than good.
It is obviously logically possible for there to be a world composed entirely of free beings who always choose good rather than evil. However, it does not follow from the fact that this is a logically possible world that God could actually cause such a world to exist. In order to be susceptible to the objection at issue here, any possible world under discussion will also have the property "created by God", since the objector is trying to establish that there is some sort of contradiction involved in a possible world that has both the property of having been created by God as well as the property of containing evil, or containing human agents who have, as a matter of historical fact, chosen evil rather than good. Now, it is essential to see that in spite of his omnipotence, there are certain logical constraints on what God can and cannot create. For example, although there is no logical contradiction involved in the concept of a being that is ontologically dependent on no other being, clearly God cannot create such a being, since the very fact of being a created being is a sufficient condition for not being a being that is not ontologically dependent on any other being. Indeed, God is the only being that is not ontologically dependent on any other being, and God did not create himself, and no other being created him, else he would not be such a being.
So the fact that a certain state of affairs is logically possible does not entail that an omnipotent being could bring about such a state of affairs. So let's return to the state of affairs described by [W2]. Ex hypothesi, it is an essential property of [W2] that it is "created by God". As in our own mythology, we may imagine this world starting off with human agents who freely choose only good. What then? Does God wait around, hoping for the best? Does he keep his fingers crossed, hoping against hope that Eve does not eat that fruit? This would, of course, be logically consistent with what actually did happen: in [W1] things began well and then went south as free beings exercised their freedom to choose evil rather than good.
In order for God to create [W2]--a world in which all the agents are free but in which it is necessarily the case that no free agent will ever choose evil--God would have to insure that the free beings in [W2] never use their freedom to choose evil rather than good. Since God's will cannot be resisted (this is entailed by omnipotence), if God were to insure that these free beings never choose evil it would follow by logical necessity that these beings are not, in fact, free to choose evil and, hence, they are not really free and, hence, their choices for the good are not free. It is here that the incoherence of the objection becomes apparent: the atheist asks us to imagine a world in which the agents are free in a libertarian sense and yet unable to do evil. But it is part of the libertarian conception of freedom that agents may choose to do evil as well as good. Hence, it is logically contradictory to suppose that God could have created free beings who inevitably did only good and never evil. So although a world in which all choices are freely for the good is a logical possibility in the sense that it does not contain any internal logical contradictions, it is impossible for such a world to be brought into existence in the order of time in such a way that all beings would perdure in goodness until the end of time, for being actualized in the way necessary for the objection to go through would be logically incompatible with the libertarian thesis that is part of the description of the world as required by the objection.
But this is only half the story, the half answering to the property of omnipotence. The other half of the objection relies on God's omniscience. Just as God is believed to be able to do any act that is able to be done, so too we may define God's omniscience this way:
[K] For any and every object of knowledge k, God knows k.We define omniscience this way in order to avoid sophistic questions about whether God can know the unknowable, just as we defined omnipotence in such as way as to avoid sophistic questions about whether God can do the impossible.
So now imagine a story along the following lines (I am adapting an example from the book of my colleague, James Petrik, that I mentioned in an earlier post). You are planning a party at your house, and you are going to invite your colleague Joe, and you know that Joe has a very wicked temper. In fact, it is extremely likely that Joe will insult your other guests and make them extremely uncomfortable, because he just can't keep his tongue to himself, and he says things that are insulting and hurtful to other people almost as a matter of course. But you also know this: when Mary is around, Joe is a completely different person. He admires Mary very much, and when she is around he behaves himself perfectly. So you invite Mary to the party as well, knowing that her presence will be like balm to Joe, and you see to it that she is seated next to Joe at dinner. Now as far as epistemic agency goes, you are very limited, and yet we can imagine a scenario like this working quite well: even with your limited knowledge you can bring off a fairly successful dinner party just by inviting the right people, and in doing so you can bring about a better state of affairs than otherwise would have obtained, and you have done so without interfering with either Joe's or Mary's moral freedom. If you, in your limited knowledge, can bring about a better state of affairs in this way, why cannot God, with unlimited knowledge, bring about a better state of affairs for all human agents, by arranging the denouement of our worldly affairs in such a way that we are influenced always to choose the good, just as Joe was subtly influenced to act better just because Mary was there?
This half of the objection turns on what God could have known, and when he could have known it. For you, planning a dinner party, it is possible that your success in bringing about a better state of affairs than would otherwise have obtained depends upon your knowing things that can only be known empirically, that is, you know Joe fairly well because some of his behaviors are habitual and you have observed them over time or heard reports from others about them. The fact that some of Joe's behaviors re habitual need not be construed as compromising his moral freedom: I think, in fact, that there are plenty of versions of virtue ethics out there that construe virtue as a matter of habit and that are also fully compatible with libertarianism. God, of course, is said to be omniscient in the sense defined above and so we must wonder: could he, at least in principle, know how people will behave prior to their actually behaving in a certain way, if we assume that the behaviors follow from the free choice of the will?
In one sense it is clearly possible to know contingent matter of fact before they obtain, provided one employs the proper method of hypothetical reasoning about a deterministic system. For example, every highly skilled pool player has reliable knowledge of the future that can be schematized in the following way:
If I strike the cue ball in such-and-such a way, then the striped balls will necessarily come to rest here, here, and here.The reliability of the player's knowledge, of course, requires that the system be deterministic. If we introduce genuine randomness into the picture, then the reliability of the knowledge disappears. Suppose, for example, that we are playing with a cue ball that is of such a character as to be liable to explode at a random time. I strike the cue ball thinking to sink the 15 ball in the corner pocket; I am a highly skilled player, so necessarily the 15 ball will go into that pocket--provided the cue ball actually reaches the 15 ball. But prior to striking the 15 ball the cue ball explodes and the 15 ball remains in its place. There is one sense in which I could have been said to "know" that the 15 ball would wind up in the corner pocket--it would have come to rest there if the cue ball had struck it with the proper velocity, at the correct angle, with the correct English, etc. But the cue ball ceased to exist in a way that could not be foreknown since it was a random, i.e., uncaused, event, and so there is another sense in which I did not, in fact, "know" that the 15 ball would wind up in the corner pocket, since it is impossible to know what is not in fact the case.
Human agency, according to the libertarian, is completely free, and in order for this to have any meaning at all it must mean that human agency is uncaused. It is not enough to stipulate that human agency be random in some sense, because we can imagine random human behavior that is still caused by something external to the agent and, hence, not really free. For example, imagine a device that can be hooked up to a human brain, and this device determines our choices for us, but it does so by means of generating random numbers. If the random number is even, we take one course of action, if the number is odd we take a different course of action. This means that our behavior will be erratic, even indeterministic, but it does not mean that we are free, since our choices are still due to something external to us. So in order for moral freedom to mean anything at all it must mean that our free choices are (at least in principle) uncaused. If they are uncaused, then they cannot be objects of knowledge prior to their actual occurrence in time, hence not even an omniscient being could know them prior to their actual occurrence in time, hence an omniscient being could not "set things up" in such a way as to anticipate and subtly influence such free choices.
Returning to our dinner party for a moment, suppose you say to yourself, "I'm not all that sure Joe will even come to this party, but I'll invite Mary anyway, just in case he does come." In short, we can hedge our bets: we can admit that we don't really know for sure what's going to happen, but we can set things up in such a way that if things went in a direction that might go wrong, we've got the system set up in such a way as to minimize the net evil effect. So why can't God do this? Maybe he can't know what a genuinely uncaused free choice will be, but he can still set things up in a certain way "just in case". But as soon as it is admitted that God is doing what Einstein said he doesn't do--playing dice--the objection fails, since to admit this much is to admit that God cannot set things up in such a way that there will necessarily be no evil, any more than the skilled pool player can leave himself a fine succeeding shot as a matter of absolute logical necessity when there is an unknowable but greater than zero chance that the cue ball will explode during a shot. Hence, if any possible world should manifest evil in my category (c), it cannot be the case that God could have foreseen and prevented such evil if the agent who made the choice for evil was genuinely free in a libertarian sense.
This is the death knell for the logical version of the Problem of Evil, but it is worth noting that most atheists have already abandoned this sort of approach and rely instead on what can be called the Probabilistic Problem of Evil [PPE] which differs from [PE] insofar as it does not claim that the Christian conception of God is logically incompatible with the existence of evil, but rather PPE claims only that the existence of evil makes it less likely that God exists. Radical defenders of PPE hold that the evils of this world are such as to make the existence of God highly unlikely. It is worth noting that the defenders of PPE do not recognize evil of category (c) as being anywhere near as troubling as the Christian thinks it to be. Instead they are principally worried about the problem of physical suffering, which I have already indicated is simply an instance of the naturalist fallacy. It is tempting to write specifically against PPE, but since I just cannot motivate myself to take PPE very seriously given the abysmally naive picture of evil it employs, I will rest my case with this post.