Does God intend evil for good?

Prima facie, many theists are committed to the view that God intends evil states of affairs as a means to God's ends. This is because many theists endorse theodicies on which God seems to intend that evils occur for the sake of some ‘greater good’. For example, Hick (Reference Hick2010) and Dougherty (Reference Dougherty2014) propose that God created us in a ‘vale of suffering’ as a means of producing certain virtues and manifestations of virtues among us.Footnote ³ Swinburne (Reference Swinburne1998) has God permitting natural evil as a means of imparting certain knowledge to us and giving us opportunities to make morally significant choices. And Stump (Reference Stump2010) suggests that God employs suffering to promote union between human persons and God.Footnote ⁴

Then there are theodicies in the tradition of Leibniz which propose that evil is a necessary component of the best world or best kinds of worlds, so God must bring about evil to bring about a world of that sort. Witness Hudson's (Reference Hudson2005, ch. 7) creative speculation that evil contributes to a highly aesthetically valuable multiverse; Plantinga's (Reference Plantinga and van Inwagen2004) supralapsarian view that God ordains evil for the sake of a world containing the great good of divine incarnation and atonement; and McCann's (Reference McCann and Kvanvig2009, Reference McCann2012) proposal that God ordains evil because the defeat of evil features in the best possible world (cf. Chisholm Reference Chisholm1969).

And even those who – like me – are not very keen on the greater-good approach to theodicy as a general strategy for solving the problem of evil may want to say that God intended at least some particular evils as a means to certain goods. For example, theists in the Christian tradition may want to say that God intended the unjust crucifixion of Jesus of Nazareth solely or primarily as a means of securing atonement for the sins of human persons.

What all of these views have in common is the claim that (i) there is a good state of affairs, G, that can only be attained by means of an evil state of affairs, E; (ii) it is permissible to bring about a world in which G and E both obtain, for the sake of either G or another good state of affairs which includes G (e.g. the world as a whole); and (iii) God performs this permissible action. Following other authors, I'll call theodicies of this sort ‘greater-good theodicies’.Footnote ⁵

Prima facie, greater-good theodicies commit their adherents to the view that God intends at least some evils as a means of achieving certain ends. For suppose with the greater-good theodicist that God intentionally brings about some good, G, which can only be brought about by means of some evil, E. If G is only brought about by means of E, and God brings about G, it seems to follow that God brings about G by means of E. And that seems to imply that God intends E as a means of bringing about G.Footnote ⁶

Unfortunately for friends of greater-good theodicies, there is a good reason to think that God would not intend evil – not even as a means to a good end. For it is plausible that intending evil is an imperfection, but God is an absolutely perfect being and so suffers no imperfections.Footnote ⁷ Thus, Murphy (Reference Murphy2017) argues in detail that intending an evil either as an end or as a means to an end involves choosing the evil such that it becomes one of the success conditions of one's action. And according to Murphy, any action which includes evils among its success conditions is ‘subpar, defective, in some way not all an action can be’ (ibid., 87). These actions are simply ‘unfitting’ for an absolutely perfect being, so an absolutely perfect being has decisive reasons not to intend evil (ibid., 99–100).Footnote ⁸ And a perfect being always acts on its decisive reasons.Footnote ⁹

I find Murphy's argument persuasive, but even apart from his argument it is plausible that God would not intend evil. Some formulations of the well-known Doctrine of Double Effect are witness to common intuitions that there is something particularly morally objectionable about intending evil.Footnote ¹⁰ So, one could have worries about certain steps in Murphy's detailed argument and yet still suspect on the basis of the intuitions that motivate the Doctrine of Double Effect that God would not intend evil.Footnote ¹¹

If the greater-good theodicist feels the pull of these considerations, then she will also suffer some cognitive dissonance. On the one hand, her theodicy seems to imply that God intends evil as a means of realizing greater goods. On the other hand, there are reasons to think that God would not intend evil, even for this purpose. It would be nice if the greater-good theodicist could dissolve this tension by showing that, contrary to initial appearances, greater-good theodicies do not require that God intends evil. That is what I will attempt to do in the remainder of this article.

The core idea: unintended means

Fortunately for the greater-good theodicist, there is a lacuna in the argument that greater-good theodicies imply that God intends evil. That God brings about a good which can only be brought about by means of an evil does not imply that God intends the evil in question, and we can see this with the help of a backward-looking counterpart of the distinction between intended and merely foreseen consequences of an action: namely, a distinction between intended and merely foreseen means to an end.Footnote ¹² In this section I defend this distinction and begin to develop a model of divine providence which exploits it to the greater-good theodicist's benefit.

Can an agent intentionally bring about a state of affairs without intending the means by which she brings about that state of affairs? Yes. In fact, it turns out that we do this all the time when we perform actions of a certain familiar and mundane kind: voluntary bodily movements.

Suppose I form the intention to raise my hand. If all goes well, this sets into action a chain of physiological cause-and-effect that runs from my brain and down my arm, eventuating in the raising of my hand. But the content of the intention that sets this causal chain in motion is just that I raise my hand, or perhaps that I raise my hand without assistance; it does not include all of the intermediate physiology. Indeed, I might even be entirely ignorant of the physiological events that mediate between my intention and my hand-raising. And even if I am not, these events typically won't so much as cross my mind when I go to raise my hand.

Evidently, I can fail to intend the means by which I perform a bodily movement. But it is one thing to show that the means to an end can be unintended; it is another thing altogether to show that they can be unintended yet foreseen. Do I fail to intend the physiological details of hand-raising when I raise my hand, even if I am fully cognizant of the physiology?

Plausibly, I do. To see this, consider the controversial but well-known distinction between intended and merely foreseen consequences of actions. Prima facie, it certainly seems that a rational agent can foresee and yet not intend the consequences of her actions. Someone with a stutter may intend to speak, foreseeing that she will stutter, yet not intend to stutter; I may intend to change lanes, foreseeing that this will provoke a rude gesture from another driver, and yet not intend to provoke the gesture.Footnote ¹³ So I am going to assume that the distinction between intended and merely foreseen consequences of actions is legitimate. But once we have seen that means to ends can be unintended (at least in the case of basic bodily movements), it is hard to understand why there could be foreseen but unintended consequences, and yet there could not be foreseen but unintended means.

Moreover, it certainly seems to me that I can raise my hand while reflecting thoughtfully on the physiology of hand-raising without intending those physiological details. But suppose I am mistaken about this. Murphy (Reference Murphy2017) argues persuasively that an action is in some sense unsuccessful if it does not produce the agent's end by the means that the agent intended. But surely, if we one day learned that scientists have been badly mistaken about the physiology of hand-raising, and that it doesn't work at all the way we think it does, it would not follow that my thoughtful hand-raising act was in some sense unsuccessful. It may have been misunderstood, but it wasn't unsuccessful. Or so it seems to me.

So, in the case of voluntary bodily movements, there is a real distinction to be drawn between intended and merely foreseen means to ends. Can this idea be applied in the context of greater-good theodicies? Can we develop a model of providence on which God intentionally brings about good by means of evil, but where the evil is a merely foreseen and not intended means of bringing about the good? I think we can. I will start with the core idea that motivates the account: an analogy between basic bodily movements and divine action. Then, in the following sections, I will develop two versions of the basic proposal in detail.

Others before me have supposed that the relationship between God's volitions and the world is similar to our basic bodily movements. Just as I can raise my hand simply by intending to, God can bring about a state of affairs simply by intending to. God doesn't have to go looking around for a means by which to bring about a state of affairs any more than I have to go looking around for a means by which to raise my hand. When I want to raise my hand, I just do it; when God wants to bring about a state of affairs, God just does it (Swinburne (Reference Swinburne1993), ch. 7).

But if my hand-raising intention is effective by means of an unintended physiological causal chain, then perhaps some of God's volitions are also effective by means of unintended intermediary causes. In particular, perhaps God can cause certain evils which are a necessary means to some good simply by intending that good or a broader, good state of affairs which includes that good, while merely foreseeing and not intending the evils themselves.

This proposal is prima facie very plausible. For since God is omnipotent, anything that God intends comes to pass. And once we've granted the distinction between intended and merely foreseen consequences of actions, it is hard to see why God could not form an intention to bring about some good state of affairs without also intending an evil state of affairs which must accompany that good, even when the evil is a necessary means to the good. But suppose God does form such an intention. Since it is metaphysically necessary that the good comes to pass if God intends it, and since it is also metaphysically necessary that the good comes about by means of certain evils, God's intention will cause not only the good, but also the evils and other causes leading up to the good, even though God doesn't intend those evils.

There is more than one way to flesh out this idea, but the approach that strikes me as the most elegant is to imagine God forming the intention to bring about a single very large state of affairs, such as a whole possible world or a large portion of a possible world. In my remaining space I will develop two different versions of this view: a ‘risk-free’ version and a ‘risk-taking’ version.

The risk-free version

Theists often talk of God selecting a possible world to ‘create’ or ‘actualize’. This notion of God actualizing an entire world seems to presuppose what Hasker (Reference Hasker and Perszyk2011) calls a ‘risk-free’ model of providence, according to which ‘God's decision between two or more alternative courses of action [is] informed and guided by knowledge of the specific consequences that will follow from each course that might be chosen’ (ibid., 283). Thus, from God's point of view, there is no uncertainty in deliberation, no risk. Whatever God does, God knows exactly what the outcome will be. Let's say that God ordains an event, e, iff God performs an action knowing that e will occur as a consequence of that action. Then a risk-free model of providence is one on which God ordains all of the details of history. As I hope to show, risk-free models provide a nice context for the view of divine intentions that I am developing.

Risk-free models of providence include such venerable views as Molinism and theological determinism, as well as newer proposals like Kvanvig's (Reference Kvanvig2011) variation on Molinism featuring epistemic rather than counterfactual conditionals, Byerly's (Reference Byerly2014) time-ordering model, and perhaps also Peels's (Reference Peels and Kvanvig2016) creative hybrid view. There are many subtle differences between these views (e.g. whether God strongly or weakly actualizes a worldFootnote ¹⁴), but what is most important for our purposes is what they all have in common: the thesis that God specifically ordains each event that comes to pass.

Let's suppose with the proponent of risk-free providence that God sovereignly brings about (ordains) one particular possible world among all others. While it is controversial whether there is a best possible world, and whether it is necessary that God would create it if there were one, there is wide agreement that God would only create a very good world, i.e. a world that may contain evils, but which is very good on the whole.Footnote ¹⁵ I take it that greater-good theodicists will want to say that at least some of the very good worlds which are candidates for divine creation are worlds which contain the goods featured in one or more greater-good theodicies, as well as the evils which those goods require.

For example, the soul-making theodicist will want to say that, among the good worlds that God might choose to make are worlds with the good of soul-making, as well as the evils of suffering and hardship that soul-making needs. The supralapsarian theodicist will want to say that, among the good worlds that God might choose to make are worlds containing divine incarnation and atonement, as well as the evils this presupposes.Footnote ¹⁶ And so on. If God were to create one of these worlds for the sake of the goodness it contains, this would be sufficient for the truth of a greater-good theodicy.

But can God create one of these worlds without intending any evil states of affairs, either as ends or as a means to an end? Yes. Let W be a world of the sort we have been describing: a world that is good on the whole, and in which some greater-good theodicy is true. Though W may have evil states of affairs among its proper parts, the property of being evil does not transfer from these parts to the whole (Pike (Reference Pike and Rowe2001) ). This is significant because, as I shall now argue, God can intentionally bring about W without intending W's proper parts.Footnote ¹⁷

Return to the hand-raising analogy for a moment. Why is my intention to raise my hand automatically fulfilled despite the fact that I do not intend the physiological events by means of which it is fulfilled? The answer seems to lie in certain nomic necessities.Footnote ¹⁸ The laws of nature ensure both that the intention to raise my hand will cause certain physiological events, and that those physiological events will cause my hand to rise. The upshot is that my intention to raise my hand automatically triggers the physiological cascade that culminates in and explains my hand-raising. I don't even have to think about it.

These nomic necessities are paralleled in the divine case by metaphysical necessities. First, just as it is nomically necessary that my intention to raise my hand initiates appropriate physiological events, so it is metaphysically necessary that, if God forms the intention that W obtains, then God's intention brings about the sequence of events that constitutes the history of the world in W. Why? Because, necessarily, if God intends that W obtains, then W obtains, and, necessarily, W obtains iff the sequence of events in W occurs. And what happens in the course of history must conform to God's intentions; not the other way around. So, it seems that God can bring about the sequence of events in W simply by forming the intention that W obtains (as God is surely free to do).

Second, just as it is nomically necessary that, if the physiological events triggered by my intention to raise my hand occur, this causes my hand to rise, so it is metaphysically necessary that, if the sequence of events which constitutes the history of the world in W occurs, this brings it about that W obtains. For, in general, if a possible world obtains, it does so because reality is as the world represents it. Thus, each of the nomic necessities crucial to the hand-raising story is paralleled by metaphysical necessities in the God story.

Pulling all of the foregoing considerations together, the result is this. Just as my intention to raise my hand automatically triggers the physiological causal sequence that culminates in and explains my hand rising, so God's intention that W obtains automatically triggers the causal sequence that culminates in and explains the fact that W obtains. For simplicity I will assume that God's intention causes the sequence of events which constitutes W's history, though some prefer to think that God's intentions ground their effects (e.g. Pearce (Reference Pearce and Kvanvig2017) ).

Now just as the most immediate effect of my intention to raise my hand is not my hand rising, but rather some other physiological event causally upstream of my hand rising, so the most immediate effect of God's intending that W obtains is not W obtaining, but rather the first event that occurs in W – i.e. W's initial, instantaneous state.Footnote ¹⁹ In the hand-raising case, the laws of nature ensure that the rest of the causal sequence following the initial physiological event will play out in the right way so that my hand rises. In God's case, the rest of the history of the world following W's initial state will play out in the right way, too, so that W obtains. But the story about why this is so differs slightly depending on whether the history of the world according to W is deterministic or indeterministic.

Suppose first that W is a deterministic world. Then the story is easy: the initial state of the world will causally determine that the rest of the sequence goes as W represents it (and so as God intended it). Suppose instead that W is an indeterministic world. Here one might worry that, after God's intention to create W generates the initial state of the world as W represents it, history could veer off in the ‘wrong direction’, so that instead of W being actualized, some other world, W*, is actualized instead.

But since we are assuming a risk-free model of providence, we must also assume that God is equipped with resources like middle knowledge or something similar that enables God to infallibly predict indeterministic processes.Footnote ²⁰ Thus equipped, God would know whether an attempt to actualize W would fail because history would veer off course, actualizing W* instead. (In Molinist terms, God would know that W is not a feasible world – Flint (Reference Flint1998), 51.) And being perfectly rational, God would not attempt to actualize such a world. Instead, God would either form the intention to create W*, or God would choose a world that doesn't share its initial state in common with W or W*.Footnote ²¹

So God's intention to create W automatically triggers the causal sequence that constitutes the history of the world in W. Perhaps it begins with a bang. A physical universe expands out of a singularity, stars and galaxies and planets emerge, and then life appears and evolves into rational animals, who form communities, build empires, fight wars. . ., and so on. But which, if any, of these details does God intend?

Just as I do not intend the physiological causal chain that is automatically initiated by my intention to raise my hand, and which results in my hand rising, we are free to suppose that God does not intend the causal chain that is automatically initiated by God's intention to create W. Instead, we may suppose that, when God forms the intention that W obtains, God merely foresees, but does not intend, that W's proper parts obtain too. Since we stipulated that W is a good world, and that it has evil states of affairs among its proper parts, it follows that God can bring about all of the evils that occur in W without intending either those evils or any other evil. More generally, God can bring about worlds that contain such greater goods as soul-making and atonement without intending the evils these goods presuppose.

Let me now address some objections to this proposal. The first objection I will consider distinguishes between bringing about an evil, E, as a means of bringing about a greater good, G, on the one hand, and appropriating or ‘making use of’ E to bring about G, on the other hand. In the former case, E obtains only because God aims to bring about G, while, in the latter case, E occurs independently of the fact that the God aims to bring about G, and God merely takes the existence of E as an opportunity to bring about G. Now it's clear that God does not intend evils that God merely appropriates for good purposes. But, on the present, risk-free view, all evils are ordained by God. And since G is supposed to bear the weight of theodicy, we should not say that God has other adequate reasons, independent of G, for ordaining E. So, it looks like God does not merely appropriate E. Instead, God brings about E precisely to bring about G. And one might object that, if this is so, then surely God intends E.Footnote ²²

However, this latter inference is the sort of inference that the hand-raising case exposes as invalid. When I intentionally raise my hand, I am not merely appropriating (for hand-raising purposes) the physiological events involved in hand-raising. After all, those events did not occur for any reasons independent of hand-raising. They were not triggered by any causes independent of my intention to raise my hand. Nor did I have any other reasons for bringing about those physiological events aside from whatever reasons I had to raise my hand. Apart from my intending to raise my hand, they would not have occurred at all. So, it seems that, instead of merely appropriating the physiological events, I bring them about.

Yet, as I argued in the previous section, I can intend to raise my hand by means of those physiological events without intending them. They are an unintended means to an intended end. Similarly, though God does not merely appropriate the evils in the world to bring about G, but rather brings them about as a means of bringing about G, nevertheless, God does this without intending the evils in question. They are an unintended means to an intended end.

Another objection begins with the claim that, to intend that W is actual, God must form an intention with a long conjunction as its content, where the conjuncts of the conjunction are all of the propositions that are true in W. But surely God cannot intend that conjunction without intending each of the conjuncts, so God cannot intend W without intending the evils W contains.Footnote ²³

In light of this objection, consider the following case. Suppose God gives me a choice between several lives that I could live while on earth. All contain some suffering, but they are good on the whole, and one – call it L – is clearly better than the rest.Footnote ²⁴ In L I have one serious car accident that leaves me injured, but I eventually recover, and otherwise I do not suffer seriously. There is more suffering in the other lives. God tells me that, whichever life I choose, God will bring it about that I live that life. Naturally, I choose L. And presumably, when I choose L, I form the intention that I will live life L (what else could it mean to choose L?). But my intuition is that, in choosing L, I merely foresee and do not also intend that I will have a car accident. It would be more appropriate to say that I merely accept or tolerate this feature of L in making the choice to live life L than to say that I intend it.

By design, this case is closely analogous to God's situation as I have envisaged it. The differences are just that the agent making the choice is me rather than God, the state of affairs chosen is a life rather than a world, and the efficacy of my intention is due to God rather than certain metaphysical necessities. Because of their similarity, thinking about the life-choosing case can help us to think about the God case. In particular, I want to suggest that there are possible explanations for how I manage to intend L without intending the car accident, and those explanations carry over to the God case (with suitable modifications).

First, it might be possible in some cases to intend a conjunction without intending all of its conjuncts. Radical though it may seem, some tempting objections to this suggestion turn out to be spurious. For example, an objector might claim that the conjunction just is all of its conjuncts, so, necessarily, if the former is the content of an intention, then the latter are also the content of that same intention.Footnote ²⁵ But this requires something like a ‘composition as identity’ thesis for propositions, and so it requires that one thing is identical to many. I find this result repugnant, and so I deny that a conjunction is identical to all of its conjuncts.

Alternatively, an objector might claim that intention is closed under logical entailment, or under known logical entailment, so an agent must intend the known conjuncts of any conjunction she intends. But this closure principle is false. Suppose I throw a dart, aiming to hit a bullseye, and instead hit my friend who is standing nearby. Though I know that hitting the bullseye entails hitting something, I do not think that I have intentionally hit something.Footnote ²⁶ The objector might reply that all she needs is the weaker closure principle that intention is closed over conjunction. But I think it is easier to doubt this weaker principle once we've seen that the similar, stronger principle fails.

So maybe what is going on in the life-choosing case is that I intend the conjunction of what happens in L without intending the car accident, even though the latter is a conjunct of the former. And if I can do that, then it's plausible that, in God's closely analogous situation, God can intend the conjunction of every proposition true at some world without intending each of its conjuncts.

But maybe something else is going on in the life-choosing case. A different account of why it seems that I can intend L without intending the car accident is that I can intend L without intending the conjunction of everything that happens in L. Perhaps I can somehow pick out L by ostension and form the intention that I will live that life. Or perhaps L satisfies a short definite description like ‘the life on offer that includes a car accident’ or ‘the best life on offer’, so I can form an intention the content of which is just that I live the best life on offer. If this is what is going on in the life-choosing case, then there need be no worries about intending a conjunction without intending its conjuncts.

Once again, this strategy carries over to God. Instead of intending the conjunction of all the propositions true at a certain world, God could proceed in another way. For example, perhaps God is directly acquainted with possible worlds. (E.g. Pruss (Reference Pruss2011) suggests that worlds are divine thoughts, which in turn suggests that God could be acquainted with them by introspection.) Then God could pick out a particular world by something like ostension and form the intention that that world obtains.Footnote ²⁷ Or perhaps the world of God's choice satisfies a short definite description, such as ‘the best possible world’, or ‘the one and only world that contains exactly n units of value’ or ‘God's favourite world’ or ‘the world God is thinking of now’,Footnote ²⁸ etc. Then God could form an intention with the content that, e.g., the best possible world is actualized, without thereby intending the particular evils in that world.Footnote ²⁹ I won't take a stand on exactly what to say here, but I conclude that there are promising ways to address the objection about the content of God's intention.

Here's a final objection. Some might aver that God intends at least some of the specific events that occur in the world, and not only the world as a whole. For example, many theists may want to say that God specifically intends such things as answers to prayer, divine revelations and visions, public miracles, and the atonement. Typically, these are regarded as specific acts that God performs in history, and not merely as unintended but foreseen consequences of acts that God performs. Yet one might fear that my model of providence makes them more like the latter than the former.Footnote ³⁰

But actually, the model I have been developing can accommodate the view that God specifically intends these things. What I have argued so far is that God can intentionally create W without intentionally bringing about its proper parts. This does not entail that God cannot intentionally bring about W's proper parts. For all I have said, God may, if God wishes, intend some of the states of affairs which are proper parts of W, in addition to intending W itself. And in the same way that God can intend W without intending, but merely foreseeing, the means by which W is actualized, God may intend proper parts of W without intending, but merely foreseeing, the means by which those parts of W are actualized. In short, God can selectively intend or not intend individual states of affairs in W as God wishes.

I regard this flexibility as an important virtue of the model. Thanks to it, theists can say that God specifically intends miracles, prophecies, answers to prayer, etc. But more than that, the idea that God can selectively intend (or not) the parts of W opens the door to all sorts of interesting hypotheses. For example, perhaps being intended by God is a distinguishing feature of so-called special divine action. Alternatively, the theist could propose that God specifically intends every good state of affairs which is a proper part of W, but not the evil ones. This would be a nice way of spelling out the traditional idea that God relates to the good things that happen in the world in a different way than God relates to the bad things, even though God is in some sense provident over all of them.Footnote ³¹

All of this is compatible with the view I am developing. My proposal is simply that God does not need to intend any of the states of affairs which are proper parts of W in order to bring either them or W about, and moreover, God does not intend any of the evil states of affairs which are proper parts of W. To bring about these states of affairs, it is enough for God to intend W as a whole.

The risk-taking version

I have just sketched a risk-free model of providence on which God selects one particular possible world among all others and creates it. Hasker (Reference Hasker and Perszyk2011) contrasts risk-free models of providence with ‘risk-taking’ models of providence, on which it is not the case that ‘God's decision between two or more alternative courses of action [is] informed or guided by knowledge of the specific consequences that will follow from each course that might be chosen’ (ibid., 283). Thus, on a risk-taking view, God does not specifically ordain all of the details of history. Rather, God's actions serve to narrow the range of possible worlds which might be actual, and the narrowing succeeds in selecting just one world among all others only in conjunction with events that God does not ordain, such as free human actions.

Risk-taking models of providence include open theism and ‘simple foreknowledge’, each of which boast many adherents, as well as the more obscure view that Todd (Reference Todd and Kvanvig2011) has dubbed ‘Geachianism’. As with risk-free models, the differences between various risk-taking models (such as whether God foreknows free human choices) are not particularly important for present purposes. What matters is what they all have in common: that God does not ordain every event that occurs. Can the view that I am developing in this article make provision for risk-taking models of providence? With suitable modifications, I think it can.

Imagine a set of possible worlds, W1, W2, . . . Wn, all of which share certain features that God wants the actual world to have, but which also differ in many details.Footnote ³² Peter van Inwagen (Reference van Inwagen and Morris1988, 227ff.) speculates that God can issue disjunctive decreesFootnote ³³ which underdetermine their consequences – decrees of the form ‘Let either X, or Y, or Z . . . be the case.’ If God issues this decree, then one of the disjuncts must obtain, but which disjunct enjoys this privilege is not causally determined by the decree, or by anything else. Applying van Inwagen's idea to the present case, we might imagine God issuing the following decree: ‘Let either W1, W2, . . . or Wn be’. Then one of the worlds in the disjunction would obtain, but God's decree would not determine which one. In the same way, if God were to decree the state of affairs which consists of all and only what is common to every disjunct of this disjunction, one of the worlds in the disjunction would obtain, but God's decree would not determine which one.

Below I will argue that this kind of decree enables God to bring about goods by means of evils without intending the latter, just as on the risk-free model above. But first it is important to show that a disjunctive decree of this sort also gives the advocate of risk-taking models of providence what she wants.

To that end, note that the advocate of a risk-taking model of providence will almost certainly want to say (with van Inwagen (Reference van Inwagen and Morris1988) and against Plantinga (Reference Plantinga, Bergmann and Brower2016) ) that God does not know which world would in fact result from the decree that either W1, or W2, . . . or Wn obtains. For God to know this would be for God to know a subjunctive conditional whose antecedent underdetermines its consequent. Knowledge of such a conditional would be analogous to middle knowledge,Footnote ³⁴ and therefore not something the risk-taker will be keen on. And if God does not know which world would result from God's disjunctive decree, then God does not ordain – even by merely weakly actualizing – any of the details which vary from one world to another in the disjunction.Footnote ³⁵

It is not difficult to formulate our disjunction such that the events which God does not ordain when God decrees that disjunction are precisely the events which advocates of risk-taking views normally claim that God does not ordain (e.g. free human actions). We need only suppose that the details which vary from one world to another across the worlds W1, W2, . . . and Wn in our disjunction include the free actions performed by human persons, the decay of radioactive isotopes, and other events that one might think are indeterministic.Footnote ³⁶

Since these events vary from world to world in the disjunction, God's decree will not causally determine them. And since we are supposing that God does not know what world would obtain if God were to issue the decree in question, God does not know of any event that varies from one world to another in the disjunction that it would occur if God issued the decree. So, by issuing the decree, God does not ordain any of these events.

Because incompatibilist free will is important to defenders of risk-taking models, it is also worth emphasizing that human choices which vary from one world to another in the disjunction can be free in the incompatibilist's sense. To see this, it will help to consider a specific example. Imagine that, in at least some of the worlds in the disjunction, Suzy faces the choice between eating a certain cookie, or resisting the temptation to do so. In some of the worlds in which she faces this choice, she eats the cookie; in others, she doesn't. And in none of these worlds is her choice causally determined by prior events distinct from God's decree.

Now suppose that, when God issues God's decree, the world which obtains as a result is one in which Suzy is faced with the choice about whether or not to eat the cookie, and it is one in which she successfully resists temptation. Given that there are worlds in our disjunction where Suzy does not resist temptation, her choice is not causally determined by God's decree. And Given that every world in the disjunction where Suzy faces this choice is one where the choice is not determined by prior events other than God's decree, nothing causally determines Suzy's action. So, the action satisfies the distinctive incompatibilist condition on freedom.

Moreover, this account fits nicely with (though it does not require) agent-causal versions of incompatibilism in particular. For if neither God's decree nor other events determine which way Suzy chooses, then it is plausible that Suzy herself, by choosing to either eat the cookie or not, determines whether the world that results from God's decree is a world where Suzy gives in to temptation, or successfully resists. Even if one is inclined to say that God's decree indeterministically causes Suzy's action, one can take a page from Markosian's (Reference Markosian1999, Reference Markosian2012) book and suppose that it does this by causing a causal relation to hold between Suzy herself (an agent) and her action (an event).

So, a disjunctive decree of the sort we have been imagining can deliver much of what defenders of risk-taking models of providence want. This is good news, because the same kind of decree will enable us to construct a risk-taking model of providence on which God brings about good by means of evil without intending evil. For I take it that what it means for God to issue such a decree is that God forms an intention with either the relevant disjunction of worlds, or whatever is common to all the disjuncts of that disjunction, as its content. And for the same reasons as before, it is metaphysically necessary that, if God forms this intention, then the intention is automatically fulfilled.

But unlike before, there is more than one means by which God's intention might be fulfilled. Any world in the disjunction will do. This means that, if God chooses to issue the decree we have imagined, the decree will trigger a sequence of events which constitutes the history of some world in the disjunction, but many of the details of this sequence will not be ordained by God. The most immediate effect of the decree will be the initial, instantaneous state of one of the worlds in the disjunction. If every world in the disjunction has the same initial state, then God's decree will causally determine that initial state. If not every world in the disjunction has the same initial state, then God's decree will only ensure that the state which results from the decree corresponds to the initial state in one of the worlds in the disjunction, but it will not determine which one.Footnote ³⁷

The initial state which results from God's decree will be the first in an indeterministic sequence of events. Within the limits set by God's decree, many of the events that occur in the sequence will not be ordained by God and will be left to chance or to the choices of other agents. Depending on how the sequence proceeds, any one of the worlds in our disjunction might be the one that is ultimately actualized.

It should be clear that, when God forms the intention that will cause one or another of the worlds in a certain disjunction to obtain, God does not thereby intend any state of affairs which features in only some, but not all, of the disjuncts. But the greater-good theodicist may want to say that every world God includes in God's decree/intention is a world containing the good which features in a greater-good theodicy (e.g. soul-making, atonement, etc.), and therefore the evils which that good requires. Even supposing that all particular evils vary from one disjunct to another, if every disjunct is a world containing some evil, then the disjunctive decree/intention determines that there will be some evil.Footnote ³⁸ Does this imply that God intends evil?

No. I maintained above that God can intend a good world as a whole without intending each of the states of affairs which are its proper parts. I won't repeat that case here. But the point is a general one, and so it applies in the present case as well. God can intend a state of affairs such as the one which consists of whatever is common to every disjunct in a disjunction of worlds without intending any of the states of affairs which are its proper parts. Instead, God merely foresees that they will result from God's intention. As before, God can intend some of these ‘smaller’ states of affairs in addition to intending the whole of which they are parts; but God does not need to intend them to bring them about, and God does not intend any of them which are evils.

So far, so good. However, there remain some complications that are specific to certain greater-good theodicies. For example, on some greater-good theodicies that have been endorsed by advocates of risk-taking models of providence, the greater good that God is aiming to realize depends on the outcome of free human choices. But if God's decree ensures that the greater good which depends on those choices obtains, then it must also ensure that the choices on which that good depends are made. But, as we've seen, God does not ordain our choices on risk-taking models of providence.

We can make the problem more vivid with an example. On a soul-making theodicy, God places human beings in an environment that is designed to foster character formation, but whether we form our characters in the intended way depends at least in part on our own choices. If we suppose that every world in the disjunction W1, W2, . . . Wn features successful soul-making, then God's decree will causally determine that at least some human beings perform choices that are conducive to soul-making. And this is contrary to the typical risk-taking view on which God does not ordain free human choices.

One might think that this worry is spurious, because, even if there is successful soul-making in every world in the disjunction, it need not be the case that the very same individuals complete the soul-making process in each world. God could simply ensure that, for each person who appears in some of the worlds in the disjunction, there are some worlds where that person freely completes the soul-making process, and some worlds where she does not. Nevertheless, in every world in the disjunction, a substantial number, n, of people freely complete the soul-making process. Then God's decree will causally determine that n people complete the soul-making process, but it will not determine of any particular individual that she will complete the soul-making process.

I am not convinced that this will work. For suppose that God chooses to create only a finite number of people. Moreover, suppose that, at a certain time t after God has initiated the history of the world, so many of these people have not only failed to complete the soul-making process, but have also permanently hardened their characters against soul-making,Footnote ³⁹ that the only way n people could complete the soul-making process is if everyone born from time t onward completes the process. Then it seems that God's decree together with choices that have already been made by others causally determines that everyone born from time t forward will complete the soul-making process. And this is the kind of result we were trying to avoid. Granted, this example is very contrived, but the fact that the suggested fix will have this result in some cases makes me wary of it.

Fortunately, there is a better way to show that the advocate of the risk-taking view that I have been developing can endorse soul-making theodicies and their ilk without conceding that God causally determines that soul-making occurs. I think that most soul-making theodicists who also endorse a risk-taking model of providence either say, or would say, that God does something good when God creates a world that fosters soul-making, even if there is no guarantee ahead of time that humans will make the right choices and so form their characters properly.Footnote ⁴⁰ On this view it is a sufficiently high probability of soul-making that justifies placing humans in the ‘vale of suffering’ that is the soul-making environment.

If that is so, then it is very plausible that the following is a good state of affairs: a state of affairs consisting of all and only what is common to every world in a disjunction of worlds such that each world features rational animals in a vale of suffering, but not all of which are worlds where the soul-making project is generally successful. The soul-making theodicist might argue that this is a good state of affairs because it is one which will probably result in a great deal of soul-making. And if that is a good state of affairs, then it is one that God can decree without intending an evil. Moreover, since decreeing that state of affairs will not necessitate that soul-making occurs (since it does not occur in every world of the disjunction), there should be no worry that God ordains any free human choices.

Another concern that arises for some, but not all, greater-good theodicies is whether, on the risk-free version of my model, God can bring about suffering in response to human choices. For example, drawing on themes from Stump (Reference Stump2010), let's suppose that, if Suzy makes a certain series of poor choices that lead her away from God and fragment her psyche, and nothing short of suffering will provoke her to correct her self-destructive course in life, then God will allow her to suffer. But, on a risk-taking model, how can we make sense of this apart from supposing that God specifically intends Suzy's suffering in response to certain choices she made that God did not ordain?

There are at least two ways we can handle this problem. The first is simply a matter of being careful about which worlds we put in our disjunction of worlds W1, W2, . . . Wn. If Suzy is free to make the series of poor choices in question, then some of these worlds will be worlds in which she makes those choices. But suppose all of the worlds in the disjunction where Suzy makes these poor choices are also worlds where, upon making those choices, she suffers (in a way that is aimed at her ultimate good). Any world where Suzy makes those choices and does not suffer (in a way that is aimed at her ultimate good) is excluded from the disjunction. Then God's decree as we have imagined it need not determine that Suzy makes the series of poor choices, but it will ensure that, if Suzy makes those choices, then she will experience suffering as a result.Footnote ⁴¹

A second way to address this concern is to suppose that God issues additional decrees in response to our choices as history progresses. Once Suzy makes that series of poor choices, God issues another decree which results in Suzy suffering. But instead of simply decreeing that Suzy suffers, which would require God to intend evil, God decrees a broader and overall good state of affairs that includes Suzy's suffering. It could even be another disjunction of worlds, provided that all of the disjuncts are consistent with how history has unfolded up to the point when God issues the decree, and provided that all of them feature Suzy's suffering.

As far as I can see, then, there is no serious obstacle facing greater-good theodicists who want to adopt the risk-taking version of my proposal that God can bring about evils for the sake of certain goods without intending those evils.