The argument from suboptimality

Let's start with a few assumptions. Suppose that possible worlds can sensibly be thought to bear overall value. Suppose that each one has an overall axiological status, and that there are no ties. Suppose that all worlds are comparable and commensurable. Suppose, finally, that that there is one unique best of the lot.Footnote ¹ Against this ontological and axiological backdrop, the following argument for atheism has been proposed:Footnote ²

1. (1) If God exists, the actual world is the best possible world.Footnote ³

2. (2) The actual world is not the best possible world.

   Therefore,

3. (3) God does not exist.

Robert Adams (Reference Adams1972) offers a partial, indirect argument against (1), by criticizing two claims that might be thought to support it:Footnote ⁴

1. (Q) A creator would necessarily wrong someone (violate someone's rights), or be less kind to someone than a perfectly good moral agent must be, if he knowingly actualized a less excellent world instead of the best that he could.

2. (R) Even if no one would be wronged or treated unkindly by the actualization of an inferior world, the creator's choice of an inferior world must manifest a defect of character.

Adams argues that God could actualize a suboptimal world with the following characteristics:

1. (S) none of the individual creatures in it would exist in the best of all possible worlds;

2. (T) none of the creatures in it has a life which is so miserable on the whole that it would have been better for that creature if it had never existed;

   and

3. (U) every individual creature in the world is at least as happy on the whole as it would have been in any other possible world in which it could have existed.

Against (Q), Adams thinks it obvious that if God were to actualize a world with characteristics (S), (T), and (U), God would neither wrong anyone nor be less than perfectly kind to anyone. Against (R), Adams claims that, far from manifesting a defect of character, God's choice of an inferior world could manifest the Judaeo-Christian virtue of grace, which he defines as ‘a disposition to love which is not dependent on the merit of the person loved’ (ibid., 97–98).

Adams's attacks on (Q) and (R) have not been well received. With respect to (Q), Michael Levine argues that ‘the criterion of personal identity Adams's argument rests upon is both stipulative and counterintuitive’ (Levine (Reference Levine1996), 31). David Basinger (Reference Basinger1983) grants Adams's account of personal identity, but insists that even if God actualizes a world with characteristics (S), (T), and (U), it may still be the case that God wrongs individuals in that world.Footnote ⁵ With respect to (R), many authors have criticized Adams's appeal to grace. Most of these object either to Adams's understanding of this virtue or its application to this issue (e.g. Basinger (Reference Basinger1983); Thomas (Reference Thomas1996); Gale (Reference Gale1998); Grover (Reference Grover2003) ). In contrast, Erik Wielenberg suggests that even if God exhibits the virtue of grace in actualizing a world other than the best, God also exhibits the absence of a different virtuous trait: the disposition to pursue intrinsically valuable states proportionally (Wielenberg (Reference Wielenberg2004), 51–52). Accordingly, Wielenberg thinks, God's action in selecting a suboptimal world would manifest a defect of character, just as (R) says.

Moreover, William Rowe argues that even if Adams achieves his stated goals of undermining (Q) and (R), this is still not enough to resist (1):

Rowe holds that if God fails to actualize the best possible world, then God's action in selecting a suboptimal world is surpassable. And, he says, if God's action is surpassable, then God is surpassable (Rowe (Reference Rowe and Stump1993), (Reference Rowe2004) ). Others have agreed, sometimes by suggesting that God's choosing the suboptimal over the optimal would indicate surpassability with respect to goodness, and sometimes by suggesting that it would indicate surpassability with respect to rationality.Footnote ⁶

One way to resist the idea that God must choose the optimal over the suboptimal is to invoke satisficing. Defenders of satisficing, after all, hold precisely that it can be morally or rationally acceptable to choose the worse over the better. Indeed, Adams's argument involves an implicit appeal to divine satisficing.Footnote ⁷ This part of his argument begins with the claim that human satisficing, in relevantly similar choice situations, is morally and rationally acceptable – and then Adams suggests that, by parity of reasoning, divine satisficing is likewise acceptable. Thus, for example, Adams considers the case of a man who decides to breed goldfish instead of more excellent beings such as cats or dogs. Adams suggests that the breeder satisfices by choosing a good enough species to breed, and says that there is nothing immoral or irrational in so doing, even if the breeder could have chosen a more excellent species instead (Adams (Reference Adams1972), 329). Adams thinks that God can likewise satisfice in his selection of a world, by choosing a suboptimal one that is nevertheless good enough. Adams's (S), (T), and (U) then constitute a partial account of the sort of world that would be good enough for God to actualize.

In the context of the argument from suboptimality, it is helpful to consider the general question of whether divine satisficing is acceptable. Doing so allows us to prescind from the description of ‘good enough’ worlds that Adams defends with his (S), (T), and (U). After all, perhaps divine satisficing is acceptable, even if Adam's account of ‘good enough’ worlds is not. In fact, if divine satisficing were defensible, this would provide strong grounds for resisting (1) – even without any account of just what sort of suboptimal world(s) God could or would choose.Footnote ⁸

In a recent article, however, I offered several reasons for thinking that the coherence and propriety of divine satisficing have not been established (Kraay (Reference Kraay2013)). I won't rehearse these here, but if my arguments succeed, it is not clear that Adams – or anyone else, for that matter – can rely on divine satisficing to resist (1) after all. Chris Tucker (Reference Tucker2016) agrees with me that, given the present state of the literature, the appeal to divine satisficing is at best underdeveloped.Footnote ⁹ But Tucker also claims that Adams shouldn't be construed as relying on satisficing in the first place. According to Tucker, Adams actually appeals to a different idea: motivated submaximization. Moreover, Tucker thinks that certain responses to another important argument for atheism can also best be construed as relying on motivated submaximization: this is the ‘problem of no best world’.Footnote ¹⁰ In the next section of this article, I set out this argument briefly. Then, in the following section, I introduce Tucker's account of motivated submaximization. Officially, Tucker does not aim to show that the appeal to motivated submaximization can defeat either argument for atheism; his goal is primarily exegetical (Tucker (Reference Tucker2016), 139). But, of course, the philosophical question that ultimately matters is whether motivated submaximization can indeed defeat these arguments. In the penultimate section of this article, I examine whether motivated submaximization can be deployed successfully against the problem of no best world, and in the final section, I examine whether it can be deployed successfully against the argument from suboptimality. I conclude that in both cases, this is highly doubtful.

The problem of no best world

So far, we have assumed that there is one unique best of all possible worlds, and we have looked at one argument for atheism that proceeds on that assumption. But many contemporary philosophers, following Aquinas, have suggested that there is no best possible world, but rather an infinite hierarchy of increasingly better worlds.Footnote ¹¹ Against this alternative ontological and axiological backdrop, some philosophers have mounted a different argument for atheism.Footnote ¹² This argument can be expressed with reference to the following inconsistent set of propositions:

1. (NBW) For every world w, there is a better world, x.

2. (P1) If it is possible for the product of a world-actualizing action performed by some being to have been better, then, ceteris paribus, it is possible for that being's action to have been (morally or rationally) better.

3. (P2) If it is possible for the world-actualizing action performed by some being to have been (morally or rationally) better, then, ceteris paribus, it is possible for that being to have been better.

4. (G) There possibly exists a being who is essentially unsurpassable in power, knowledge, goodness, and rationality.

Critics of theism have urged that since this set of propositions is inconsistent, and that since (P1) and (P2) are plausible, those who endorse (NBW) ought to reject (G). This amounts to an a priori argument for the impossibility of an essentially unsurpassable God on (NBW), and it has come to be called ‘the problem of no best world’.

Several important recent responses to this argument implicitly or explicitly invoke divine satisficing.Footnote ¹³ Bruce Langtry, for example, asserts that given (NBW), ‘there are infinitely many pairs of worlds V and W such that both W is better than C and an omnipotent and omniscient being can [actualize] W, but his [actualizing] W need not be a morally better action, or a better all-things-considered or more rational, than his [actualizing] V’ (Langtry (Reference Langtry2008), 75, emphasis added). Langtry denies (P1) on the grounds that, given (NBW), God must and can satisfice by selecting a world that is surpassable, but nevertheless good enough.Footnote ¹⁴ As noted above, I have recently argued against the coherence and propriety of divine satisficing. In response, Chris Tucker (Reference Tucker2016) has urged that responses like Langtry's should not be understood to invoke satisficing, but rather motivated submaximization. If Tucker is right, then my criticisms miss the mark. I now set out his account of this idea.

Tucker on motivated submaximization

Chris Tucker (Reference Tucker2016) agrees with me that appealing to divine satisficing is not a promising way to respond to either the argument from suboptimality or the problem of no best world. But he also argues that I am mistaken to think that the responses described in the first two sections of this article actually invoke satisficing. According to Tucker, they invoke motivated submaximization instead. In this section, I set out Tucker's view, and in the next two sections, I turn to the question that matters most in these debates: can motivated submaximization defeat either argument for atheism?

Here are Tucker's official definitions of satisficing and motivated submaximization:

1. Satisficing (with respect to good G1): an agent A satisfices with respect to G1 in a transparent situation iff:Footnote ¹⁵

1. (a) A aims, purely for its own sake, at promoting G1 to degree D but not as much as A can, and

2. (b) A chooses a suboptimal option with respect to G1 that has a value greater than or equal to D because A knows it satisfies the aim in (a). (Tucker (Reference Tucker2016), 133).

1. Motivated Submaximization: an agent A submaximizes with motivation in a transparent situation iff:

1. (i) A aims at getting as much of good G as A can, but

2. (ii) A chooses a suboptimal option with respect to G because of some countervailing consideration (ibid., 132).

Suppose that Tucker is right that authors like Adams and Langtry are best understood as deploying motivated submaximization instead of satisficing.Footnote ¹⁶ Tucker says that ‘the claim motivated submaximization can be appropriate is widely endorsed and well defended. The claim satisficing can be appropriate is rarely endorsed and poorly defended’ (134). I agree with both parts of this judgement, at least with respect to ordinary human choice situations. But the key question is whether motivated submaximization is plausible with respect to the divine world-choice situations at issue in these arguments for atheism. This is what I will explore in the final two sections of this article.

I must highlight two points about countervailing considerations that will be important for what I say below. First, Tucker identifies a key structural feature of these considerations: they ‘operate independently of and against what they countervail’ (Tucker (Reference Tucker2017), 1377, and see ibid., 1372, n. 10). An example from Slote illustrates this point (Slote (Reference Slote1989), 22; 32–36). One might initially aim to increase one's well-being as much as one can, but subsequently come to realize that one just doesn't need quite so much. A moderate agent might, accordingly, submaximize. In such a case, non-need countervails well-being, because its reason-giving force operates against and independently of the reason-giving force of well-being.

Second, ‘countervailing’ seems to be a success term, such that a consideration is genuinely countervailing only if it actually justifies the agent in selecting a suboptimal alternative. In the divine case, this justification must be robust enough to preserve the intuition that God is a world-chooser who is unsurpassable with respect to goodness and rationality even while he selects an inferior world in lieu of a superior one. In this context, any proposed countervailing consideration must be held against this (very, very high) standard.

Motivated submaximization and the problem of no best world

So, can motivated submaximization be harnessed in an effective response to the problem of no best world? Tucker suggests that the countervailing consideration might simply be this: there being no optimum. Here is how this is supposed to work. God, so to speak, finds himself in the choice situation depicted by (NBW), and has to pick a world to actualize. The good at issue is the overall axiological status of the selected world. As per Tucker's (i), God aims to get as much of this good as he can. As per Tucker's (ii), God chooses a suboptimal option with respect to this good, given the countervailing consideration at issue, which is this: there being no optimum level of world-goodness (Tucker (Reference Tucker2016), 136).

I begin with two relatively minor worries. First, it is not obvious that in the choice situation expressed by NBW, God can sensibly be understood to have the aim described in Tucker's (i). After all, why would God aim to maximize the axiological status of the actual world, when it is perfectly clear to him, given his omniscience, that this simply cannot be done? This way of construing the situation imputes an irrational aim to God – and this is a problem, given the goal of defending God's rationality. Second, it is not obvious that Tucker's proposed countervailing consideration (there is no optimum) meets the independence requirement he sets out. After all, it refers to the very same good (the overall axiological status of the chosen world) that (i) picks out.

But suppose we waive these concerns. The deeper problem is that to deploy motivated submaximization in this way would be to beg the question. I take it that the goal here is to criticize (P1). Now, it would clearly be illegitimate, in this dialectical context, to grant (NBW) and (P2) for the sake of argument, and then to reject (P1) merely by presuming that (G) is true. After all, the denial of (G) is precisely what this argument for atheism means to establish, so it would be question-begging to respond simply by presuming that (G) is true – that God is possible. But this is what the appeal to motivated submaximization subtly does. To see how, recall that ‘countervailing’ is, for Tucker, a success term. Given this, to claim that there being no optimum is a countervailing consideration is, in effect, to stipulate that God can coherently choose a suboptimal world. And to stipulate this amounts to presuming that God is possible – which, of course, is the very claim at issue. The critic of this argument for atheism needs to do more than assert that there being no optimum is a countervailing consideration. The critic needs to show that it is more plausible to think that it is really a countervailing consideration than it is to think that (P1) is true, rather than illicitly presuming that God is possible.Footnote ¹⁷

But perhaps this seems a little unfair. After all, Tucker does offer some reason for thinking that there being no optimum can be a countervailing consideration. Specifically, he considers what ordinary human agents are rationally permitted to do in so-called EverBetter cases (Tucker (Reference Tucker2016), 130, 136 and Tucker (Reference Tucker2017), 1367).Footnote ¹⁸ Suppose, for example, that a genie offers to ensure that your life enjoys any degree of welfare you choose. And suppose, further, that there is no maximum degree of welfare that you can enjoy: for every degree of welfare you can have, you could have some better degree. Tucker holds that in such a scenario, your aim to maximize your welfare is countervailed by the fact that there is no optimum. Accordingly, you are rationally permitted to submaximize. And, on Tucker's interpretation, Langtry (Reference Langtry2008, 74–78) should be read as suggesting that since it is rationally and morally permissible for ordinary human agents to submaximize in EverBetter cases, it is likewise rationally and morally permissible for God to do likewise on NBW: God's goal of maximizing world-value is countervailed by the fact that there is no optimum.

This, of course, is an argument from analogy. It succeeds only if two conditions are met. First, it must indeed be rationally and morally permissible for human agents to engage in motivated submaximization in EverBetter cases. Second, the human case must be relevantly and sufficiently similar to the divine case.Footnote ¹⁹

With respect to the first condition, it is worth noting that some philosophers have argued that no choice is rationally or morally permissible for human beings in EverBetter cases.Footnote ²⁰ If their arguments are sound, then the first condition is not met. Those who wish to defend the propriety of divine motivated submaximization by analogy with human cases must engage with and defeat these arguments. As for the second condition: even if human agents are rationally or morally permitted to submaximize in EverBetter cases, it is far from clear that God is likewise permitted. This is because the standards for what counts as rationally or morally permissible for ordinary human agents may well be significantly lower than the standards for God. God, after all, is held to be essentially unsurpassable in rationality and goodness. Simply to presume that human standards for permissibility are identical to God's is, in my view, to fail to take these divine attributes sufficiently seriously – and it can even amount to begging the question in this context. An argument is needed to show that the standards for permissibility in the human EverBetter cases are not lower than the standards for permissibility in the divine case.

But let's suppose that both these conditions are met, and indeed that this analogical argument establishes the rational or moral permissibility of divine motivated submaximization. Would this be enough to defeat the problem of no best world? Perhaps not. In my view, further argument is needed to show that permissibility is itself good enough for God. There might be an important distinction between an action's being rationally or morally permissible, on the one hand, and that action's being perfectly rational or moral, on the other. And if such a distinction can be upheld, it would be natural to expect God's actions not merely to be rationally and morally permissible, but to be perfectly rational and moral. (Again, God is supposed to be essentially unsurpassable in rationality and goodness.) If this is right, then mere permissibility is not good enough for God.Footnote ²¹ Pulling all these worries together, in my view it is far from clear that motivated submaximization can be deployed to defeat the problem of no best world.

Motivated submaximization and the argument from suboptimality

Finally, then, let's examine whether motivated submaximization can be harnessed in an effective response to the argument from suboptimality. Tucker points out that in ordinary human cases, submaximization can be motivated by a special connection or relationship with specific individuals (Tucker (Reference Tucker2016), 130; Idem (Reference Tucker2017), 1367). He begins with an ordinary case:

Tucker then applies this idea to the divine case:

Let's consider this proposal.Footnote ²²

I am happy to grant, for the sake of argument, that special connections or relationships with individuals can – in ordinary human affairs – constitute countervailing considerations that justify selecting a suboptimal option in lieu of maximizing. But it is not clear that this can work in the relevant divine choice situation. For one thing, in the human case, the special connection or relationship is (at least typically) with actual people, but in the relevant divine case, it is not. This is because in the envisaged scenario, God is deliberating, in the moment logically prior to his creating anything, about which world to actualize.Footnote ²³ So there are not yet any actual people available with whom God can have a special connection or relationship. But the special connection or relationship must nevertheless exist in the moment logically prior to God's choice of a world, since it is proffered as a reason for that very choice. The idea must be, then, that God has a special connection or relationship with certain possible people.

But can this really be? It is worth noting that Adams himself is emphatic that God cannot actually be in a certain sort of relationship (that of having an obligation) with merely possible people:

I see no reason to think that Adams's point here fails to apply to the sort of connection or relationship that Tucker envisions. The problem, fundamentally, is that one of the relata is non-actual, and it is difficult to believe that an actual agent can be in an actual relationship at all, let alone a special one, with somebody who is not actual.Footnote ²⁴

Suppose, however, that we waive this concern, and grant that God can indeed have a relationship with possible people. The next issue to consider is whether it makes sense to suppose that God can have a special connection to, or be in a special relationship with, some of them in a way that would generate a plausible countervailing consideration. Consider an ordinary case in which you bequeath money to your children, though you recognize that you could have done far more good in the world by using the funds differently. (For example, you could have used the money to save many lives in the developing world, by supporting charities that combat malaria or parasitic worm infections.) Tucker's idea is that common-sense morality permits you to bequeath the money to your children because of the special relationship that you have with them.Footnote ²⁵

What grounds the specialness of this relationship or connection? Perhaps it is causality: you are part of the causal chain that leads to their existence in a way that you are not for the other people you might have helped instead, and that's what makes your connection or relationship special. But this account simply cannot transfer to the divine case. That's because no matter which possible people God chooses to actualize, God will be an equally important part of the causal story that leads to (and indeed sustains) their existence. Given God's unique role as creator and sustainer, there's just no way for him to avoid it!

What about love? Perhaps, in the ordinary case, the special relationship that justifies your bequeathing money to your children (in lieu of doing far more good in the world) is grounded in love: you just love your children more than all the other candidates for your largesse, and that makes it acceptable to favour them. Likewise, perhaps God loves some possible people more than others – and perhaps this can ground the special connection or relationship that Tucker posits. Jeff Jordan (Reference Jordan2012) writes of the ‘topography’ of divine love: he describes the idea that God's love for persons must be both wide (maximally extended) and flat (equally intense). This, as he rightly notes, is a familiar enough theme in theistic traditions, and indeed is often held to be a non-negotiable feature of perfect being theology.Footnote ²⁶ If possible people can be loved, as we are now granting, it is difficult to see why this topographical requirement would apply only to actual people: surely the width and flatness of God's love – the love of an unsurpassable being – should extend equally throughout modal space. At the very least, it seems that the defender of partiality in divine love bears the burden of showing otherwise. And if God loves all people in modal space equally, then it is difficult to see how God's love for certain possible people in a suboptimal world could justify God's choice of that world over the best one, all else equal.Footnote ²⁷

I have considered two suggestions for grounding the special connection or relationship that Tucker suggests could serve as a countervailing consideration (causality and love), and have found both wanting. A final proposal is more radical: perhaps the specialness of the connection or relationship is grounded in, or somehow constituted by, God's ungrounded or brute preference for certain possible people. On this proposal, God's chosen people (so to speak) are chosen simply because God brutely prefers them.

Let's be very clear about what it would take for this to work: it would have to be the case that God's ungrounded or brute preference can ground or constitute the postulated specialness of the postulated actual connection or relationship that God has with merely possible people in the logical moment prior to creation, in such a way as to generate a countervailing consideration that justifies God's choice of a suboptimal world in lieu of the best world.Footnote ²⁸ Now, I can accept that a brute preference can, in principle, ground or constitute the specialness of relationships, at least in ordinary human cases. (I'll check to see what my wife thinks!) And I can also accept that it can be entirely rational, at least in certain cases, for ordinary human agents to act on their brute preferences. But it's not obvious that God – who is held to be unsurpassably rational – can have any non-rational, brute preferences in the first place. Moreover, even if God were to have such preferences, it's not obvious that he could be rational in privileging them over other considerations. If either of these things are impossible for God, then this proposal is a non-starter.

Suppose, however, that God can sensibly be thought to have non-rational or brute preferences. And suppose, further, that God can sensibly be thought to act on them rationally. This is not yet enough to secure what is desired: a countervailing consideration that justifies God's choice of a suboptimal world. This is because any putative countervailing consideration in this domain must meet a very, very high standard: it must be consistent with the idea that God is a world-chooser who is unsurpassable with respect to goodness and rationality, even while he selects an inferior world in lieu of the best. Perhaps ordinary human agents can rationally act on their brute preferences in this way. For example, I might just brutely prefer to read that trashy novel or watch this silly action movie instead of (say) reading Flaubert or watching Fellini – and I might be rational in so doing. But – and this is crucial – nobody thinks I am an essentially unsurpassable chooser of novels and films. (And I mean nobody!) So the analogy fails. I just don't see how an ungrounded or brute preference for the denizens of an inferior world – a preference that is, by definition, non-rational – can be weighty enough to ground or constitute a connection or relationship that is so special that it countervails the force of reasons that a perfectly good and rational agent already has to choose the best. At the very least, it seems that the defender of brute preferences in this domain bears the burden of showing otherwise.

The foregoing generalizes into a cautionary point for any attempt to appeal to special connections or relationships in an attempt to justify God's choice of a suboptimal world. For this to work, the connection or relationship must be (a) special enough so as to generate plausibly a countervailing consideration that in fact justifies an unsurpassable being in selecting the suboptimal over the best. But it's important to see that there is another constraint operating in the opposite direction. If the connection or relationship is too special, then the proposal can court paradox by turning the suboptimal into the optimal. Let me explain. God's special connection or relationship with these possible people can't be so valuable that the world they inhabit turns out to be the best one after all, since that would undermine the entire project of showing that God can choose a suboptimal world. So the connection or relationship must also be (b) not be too special. Pulling these thoughts together, we can see that any attempt to flesh out how God's choice of a suboptimal world can be grounded in a special connection to or relationship with possible creatures must take care to ensure that this connection or relationship is (a) sufficiently special, but (b) not too special. Like Goldilocks's choices, they must neither involve too little nor too much – they must be just right. I don't see how this can be done.