Major Gaps in Sterba’s New Atheological Argument from Evil

Abstract

In this essay, I first offer several scenarios where Sterba’s argument based on the Pauline Principle fails: specifically, one in which we all consent to living an earthly life in some prior existence (prior-consent scenario), one in which the victims would approve of the evil being done to them for some greater good (would-approve scenario), and one that combines one of these two scenarios with the stipulation that the greater goods redound to the victims (victim-beneficiary scenario). Along the way, I claim that a version of the Kantian principle that persons should not be treated as mere means, but as ends in themselves, better captures the intuitions used in support of the Pauline Principle. After this, I present two further significant problems with Sterba’s arguments. First, I argue that his claim that God should prevent the serious evil consequences of our free choices fails to consider the degree to which such a policy would make us aware of God’s monitoring of our every move. This in turn would greatly diminish our ability to make morally significant choices. Second, I point out flaws with his argument for the applicability of the Pauline Principle to God’s choices, particularly objecting to his argument that any greater-good theodicy implies that God would desire that people sin so that good may come, something he claims is morally perverse.

This essay was written in reply to James Sterba’s invitation to respond to his purported new logical problem of evil. In my reply, I consider both his summary version of his core argument as presented in his essay “Is a good god logically possible?” (Sterba 2020), along with his much more elaborated version given in his book Is a Good God Logically Possible? (Sterba 2019). I have not consulted additional elaborations and defenses that he has made of his argument since then, though I have been in touch with him about a couple of points. I want to thank him for presenting an original thought-provoking perspective on the problem of evil.

1. The Context for the Problem of Evil

I will first sketch the overall context of discussions of the problem of evil. Usually, these discussions assume a minimal requirement for calling God “good,” specifically the so-called Greater Good Principle (GGP). This principle can be roughly stated as follows:

GGP: For all evils E, an all-good, all-powerful God would only allow E if God’s allowing E is necessary for the existence of a greater good.

(For articulating the GGP, I consider the prevention of a greater evil to be equivalent to bringing about a greater good, though when we discuss Sterba’s argument it will be important to distinguish the two). Using the GGP, the atheist argument from evil can be roughly stated as follows:

1. There are many evils in the world for which we cannot find (after much thought) the kind of greater good required by the GGP.

2. The fact that in many cases we cannot find such greater goods gives us a good reason to believe there is no such greater good.

3. Hence, by the greater good principle, an all-good, all-powerful God does not exist.

In response, theists either offer theodicies, which attempt to provide plausible reasons for why God might allow the evils in question, or they offer what are called defenses. There is some controversy about how to define a defense. As I define it, a defense denies premise #2 either by providing a reason for God’s allowing evil that one argues is logically possible and not highly implausible, or by invoking the skeptical theist strategy. The latter involves claiming that since God’s mind is infinite, and ours is finite, even if there exist the required greater goods, we would not expect to find them. Thus, the fact that we cannot find them does not give us sufficient reason to think they do not exist. As an analogy, the fact that we have not yet found intelligent extraterrestrial life does not mean it does not exist since the universe is so vast that even if such life existed we would not expect to find it.

2. Sterba’s Argument

Recently, some atheists have responded to the theodicy and defense strategies based on the GGP by appealing to some non-consequentialist ethical principle that implies that for some evils in the world, it is morally wrong for God to allow them even if they are necessary for a greater good. The idea is similar to the claim that it would be wrong to execute a person known to be innocent to mollify a mob demanding justice and ready to burn down a large city, even if one were to save more innocent people by the execution.

Sterba offers one of the most developed, if not the most developed, version of this type of atheist argument, making his work of particular interest. The fundamental non-consequentialist ethical principle he appeals to is what he calls the Pauline Principle according to which one should “never do evil that good may come”(Sterba 2020, p. 204). He then argues that this principle should be expanded to include that one should never allow evil that good may come if one can easily prevent the evil. From now on, when I refer to the Pauline Principle, I implicitly refer to this expanded version of it.

In his 2020 summary article, Sterba begins by distinguishing between goods to which we have a right and those we do not. As far as I can tell, goods to which we have a right are goods others are morally obligated to give us if they can reasonably do so without violating other obligations. However, I do not think that this distinction ultimately matters much for his argument, and thus I will only look at the case involving greater goods to which no one has a right; this is the harder case for theists to defend anyhow, since if one has a right to a particular greater good, that gives God an additional reason for permitting those evils that are necessary for that greater good.

To deal with greater goods for which we do not have a right, Sterba invokes what he calls Moral evil prevention requirement II:

Do not permit rather than prevent significant and especially horrendous evil consequences of immoral actions simply to provide other rational beings with goods they would morally prefer not to have

(Sterba 2020, p. 206).

Although in the above quotation Sterba is unclear whether “other rational beings” includes the person experiencing the evil consequences, in an email he said he revised his principle to include any “would-be beneficiary.” Using the specific example of rape, he then claims that any such goods that might come from evil are goods that the beneficiary of such goods would morally prefer not to have, a claim I will call the beneficiary preference claim (BPC):1

Consider the opportunity to console a rape victim. No one is entitled to be provided with such a good and its very existence depends upon God’s permission of a rape. Given then that the would-be beneficiaries of this good would morally prefer that God had prevented the rape rather than that they receive this good, God should have acted to respect their moral preferences. Even the perpetrators of such wrongful deeds, who later have the opportunity to repent them and seek forgiveness would always morally prefer that God had prevented the external consequences of their immoral deeds

(Sterba 2020, p. 207, Italics mine).

Sterba does not clarify what it means to say that the goods are ones that they would “morally prefer not to have,” but I assume this could mean one of two things: (1) they are goods that they would prefer not to have and it is morally permissible for them to prefer not to have them; or (2) they are goods that they would prefer not to have and it is morally obligatory for them to prefer not to have them. Under interpretation (1), if the good is a greater good and there is no better alternative for achieving that good than allowing the evil in question, then any consistent consequentialist would prefer that God did not prevent their deeds, contrary to what Sterba clearly wants to say. Thus, I assume interpretation (2) is correct, not (1).

3. First Set of Problems: Three Scenarios

For now, I will grant that the Pauline Principle (which Sterba bases his other principles on) applies to God, though I will question this at the end of this essay. To get his argument to work, he would need to show that there are no counterexamples to the BPC. However, I can immediately think of three scenarios that potentially violate the BPC: namely, one involving consent before having an earthly life (prior-consent scenario), one in which the victims would approve of the evil being done to them for the greater good in question (would-approve scenario), and one that combines one of the last two scenarios with the stipulation that the greater goods redound to the victims (victim-beneficiary scenario). These are elaborated next.

3.1. Prior-Consent Scenario

Suppose, as the early Christian theologian Origin thought, we existed before our life on earth. Further, suppose that we freely chose to be born in a situation where we could potentially be subject to substantial evils, or even commit such evils, and we did so for the possibility of some set of greater goods, either for ourselves or others. Finally, suppose that we all choose under a “Rawlsian veil of ignorance” in which we do not know what situation we will be born in, including our sex and genetics. In such a scenario, I see nothing wrong in God allowing such evils to occur. Furthermore, it seems that all parties would prefer for God to allow the evils for the goods to exist.

In reply, Sterba might claim that it is morally wrong to consent to being in such a situation. He suggests such a reply in a footnote, claiming that “in standard informed consent cases, one is consenting to something that would not be wrong if one consents to it” (Sterba 2019, p. 99, n. 15). This might be true under some appropriate definition of “standard,” but even if this is true, it says nothing about whether there are cases in which such consent would be morally permissible. To argue that there are such cases, I first consider a more mundane example in which it does seem such consent is morally permissible. (I should note that this example, and the ones for the other scenarios, are a little complicated because I am constructing them so that it is clear that the evils are being allowed for the existence of greater goods, not for the prevention of greater evils, since Sterba allows for the latter sort of exception.2) In further support of the moral permissibility of this scenario and the other two below, I will also show how their moral permissibility can be justified by Kant’s categorical imperative that a person should never be treated as merely a means to an end, something I argue undergirds the Pauline Principle.

Here is my mundane example. Suppose a group of beings want to reach perfection in love, a perfection that will give them bliss for all eternity. Some beings are already perfect in love, but others are not since they still have evil tendencies in their character. For those who are not, perfection in love can only be reached in two ways: (1), by allowing their evil tendencies to be acted out in a full-fledged way, and then recognizing the evil of their actions, repenting, and performing sufficient penance; or (2), by their turning to love in the face of temptation. Furthermore, because of the kind of connection the beings have with each other along with other constraints, these greater goods cannot be accomplished by a mere simulation or any other means that does not involve the victims actually suffering the consequences of their actions. The morally perfect beings want to help the others become perfected. So, they agree to create a world in which both the imperfect beings and perfect beings enter, and in which the imperfect beings are strongly tempted to do evil.

Two claims seem highly plausible for this scenario. First, the reason for the consent is based on the existence of a greater good, not merely the prevention of a greater evil. Second, the perfect and imperfect beings do nothing morally wrong in together creating such a world and entering into it for the greater good to be realized. A primary reason for this is that all beings involved have consented to the situation. Since Sterba is claiming that the existence of evil in the world contradicts the existence of God under highly plausible assumptions, he must show that there are no plausible prior-consent scenarios in which it would be morally permissible to offer such consent. As is, he does not even consider this type of scenario.3

3.2. Would-Approve Scenario

In this scenario, the victims do not give their prior consent to God’s allowing their victimization but God knows they will give their uncoerced approval in the afterlife when they are fully informed of the greater good resulting from their victimage. I will now argue that this after-the-fact approval is sufficient to make it morally permissible for God to allow their victimage. I argue this by both giving a this-worldly example and then justifying it using the Kantian Principle that in the last subsection I claimed underlies the Pauline Principle.

Consider a case in which two sisters—Jane and Kate—are in an automobile accident. The accident damages one of Jane’s kidneys in such a way that it needs to be removed, causing her to lack sufficient stamina to compete in the Olympics and win a gold medal, which she otherwise would almost certainly win given her athletic abilities. Since Jane recognizes that most people are not able to win a gold medal, Jane has stated many times to Kate and her parents that she would not consider such a loss an evil, but only the loss of a good.

Now Kate and Jane both remain in a coma for a week after the accident. Their parents must decide the day after the accident whether to ask the doctors to remove one of Kate’s kidneys and transplant it into Jane, knowing that Kate will experience a month’s worth of suffering but fully recover afterward without any permanent damage. If they delay, the window of time for the transplant will disappear. Her parents know that given how much Kate loves Jane, along with Kate being a convinced utilitarian, she would agree to the transplant if the potential greater good of Jane’s winning the gold medal (and the consequent deepening of the love between them) would outweigh the suffering caused by the kidney removal. Further, they know that given the outweighing condition is met, Kate would very likely be angry with them for the rest of her life for not going through with the surgery.

After talking to the doctors, the parents become convinced that the good outweighs the evil, and hence give their approval for the surgery, assuming the law allows for this. In this case, it seems morally permissible, perhaps even morally required, for the parents to agree to the transplant. On the other hand, if they knew that Kate would not approve, then their approval would be on much shakier moral ground.

Of course, there are some disanalogies here to the envisioned afterlife case. For example, most victims in our world would not agree during their earthly lives to their victimage. However, this disanalogy can be dealt with by slightly modifying the above case by having Kate wake up before the surgery with temporary brain damage which makes her unable to comprehend the greater good that would be realized, and for that reason refuses the surgery. Since the parents know that once her brain is healed, she will understand and approve of what they have done, the parents allowing the surgeons to do the operation still seems morally justified. Finally, I argue below that the moral permissibility of this kind of “would approve” case can be justified by the Kantian Principle mentioned above, thus giving us a general reason for thinking that the afterlife would-approve scenario is morally permissible even if it is disanalogous in certain ways to the mundane “would approve” example just presented.4

3.3. Victim-Beneficiary Scenario

What I will call the victim-beneficiary scenario combines one of the previous two scenarios with the additional stipulation that the greater good ultimately redounds to the victims. Intuitively this makes it even more plausible that it is morally permissible for God to allow the evil in question in the above two scenarios. In fact, concerning the “would approve” version of this scenario, there are many cases of people who say that they are glad that a certain evil befell them because of something it did for their character or view of life. In such cases, it seems implausible to say that it would be wrong for God to have allowed the evil if that is the only way the greater good in question could be obtained and God knew that in the future the victims would approve of God’s allowing the evil.

As a specific everyday example, suppose that to get the best sort of education that will open up great opportunities, a parent sends their child to a school that has harsh standards and bullying that will cause their child significant suffering. However, they also know the school has excellent counseling services that largely will prevent the suffering and bullying from permanently psychologically harming the child. Further, suppose this is by far the best school available for getting that education. Even if the child does not consent to go, the parents have good reason to think that when the child gets older, he would thank them for sending him to this school so that he could have better opportunities in life. In this case, there does not seem to be anything wrong with the parent knowingly allowing their child to suffer for this greater good since it benefits the child and they have good reason to believe the child would approve of it in the future.

3.4. Pauline or Kantian Principle?

Before looking at potential responses Sterba could give to these three scenarios, it should be noted that each of them satisfies the Kantian principle that persons should never be treated merely as a means to an end, but as ends in themselves. Roughly, for Kant, to treat a person as an end in themselves is to treat them in a way that recognizes their inherent worth, particularly as an autonomous moral agent that can choose in a non-coerced way based on their own precepts. In contrast, for one to treat a person as a mere means is to treat them in a way that uses them as an instrument for some other purpose and does so by treating them as something less than an end in themselves. (See O’Neill 2016 for this understanding of Kant’s ethics.)

This principle is satisfied in the prior-consent scenario since God treats the persons as free agents capable of deciding for themselves, which would be treating the persons as ends in themselves. This is also true of the would-approve scenario since God could still be thought of as treating them as autonomous moral agents with their own precepts, just not at or before the evil occurs. Put differently, God is allowing the evils based on the knowledge that their future selves would choose that God allows them to have been victimized in the past given that, per impossible, God gave them such a choice regarding these past facts. In fact, only in the afterlife would they be fully aware of the context and potential goods and evils that might have occurred because of their victimage, and so arguably only then would they be most fully moral agents. Finally, for those potential greater goods that redound to the victims of the evil, God’s allowing the evil not only values them as autonomous moral agents, but further values them by doing it for the victim’s sake, not merely for the good of others. This in turn should help remove any qualms one might have regarding the prior-consent and would-approve scenarios based on the victims still being treated merely as a means to someone else’s good, even if they consented to, or would approve of, such treatment.

I bring up the Kantian principle since it gives us a principled reason for why the above scenarios are morally acceptable. Further, I believe it is more fundamental than the Pauline Principle. A fundamental ethical principle should be stated in such a way that the principle tells you whether an action violates the principle or not, given that one properly understands the meaning of the words used in the principle. Arguably, once we understand what it means to treat someone “merely as a means,” and the idea of treating someone as an end, the Kantian principle tells us whether or not it morally prohibits an act.

In contrast, as stated—“never do evil that good may come”—the Pauline Principle does not itself tell us what acts are prohibited. (This is also true when modifying it by replacing “do evil” with “allow evil that one could easily prevent,” a modification that Sterba argues for as noted above.) Instead, we must consult our moral intuitions to determine the exceptions to the principle, not merely to understand the principle (as in the Kantian case). For example, suppose I understand evil as doing something intrinsically evil. Then, as stated, the Pauline principle implies, for instance, that a plastic surgeon causing a patient significant suffering (an intrinsic evil) to make them look better is wrong, even if the patient consents to the surgery and the happiness it brings the patient outweighs the suffering. Further, counterintuitively, it would say the type of non-consent case in which parents allow suffering to occur to their child because it is necessary for their greater good is wrong, as in the school example above. To save the Pauline principle, one must then consider these exceptions.

On the other hand, the Kantian Principle shows why these and other exceptions hold.5 Of course, one could always adopt the Pauline Principle and then add exceptions based on our moral intuitions regarding specific examples. Two things should be noted about this approach, however. First, the above scenarios appear to fall into the pattern of exceptions that Sterba allows, and thus Sterba needs to consider them in his argument. Further, if one takes this “adding exceptions” approach, then one does not know whether there are other exceptions that no one has thought of. In that case, one is open to an analog of the standard skeptical theist response to the problem of evil: just as the skeptical theist says that as far as we know there might be greater goods we have not discovered that justify God’s allowing the evils in the world, one could argue that there are further exceptions to the Pauline Principle that we have not discovered that justify God’s allowing such evils. Finally, to clarify, I am not advocating the Kantian Principle but merely arguing that it better captures the intuitions behind the Pauline Principle.

4. Sterba’s Potential Responses

In this section, I consider various responses Sterba could make to the above scenarios and arguments. I begin with the prior-consent and would-approve scenarios and then address the victim-beneficiary scenario.

4.1. Prior-Consent and Would-Approve Scenarios

Sterba recognizes that cases of consent might be potential exceptions to the Pauline Principle, stating that “one possible way that the infliction of harmful consequences on us might be justified is if we were to give our informed consent to them” (Sterba 2019, p. 98). Sterba rightly notes that in this world informed consent is rarely given for the evils victims suffer. However, he fails to consider the prior-consent scenario I elaborated above in which such consent is given before this life on earth, and he also fails to directly consider the would-approve scenario. Regarding the former, many Christians might find this scenario unappealing because it assumes the preexistence of souls. However, as noted above, this view was held by one of the greatest early Christian theologians, and regardless, finding it unappealing is not an argument against it. Further, Sterba purports to be offering a logical problem of evil, which claims there is no way of reconciling the existence of the types of evils we find in the world with the existence of the God of traditional theism. So, for his argument to work, he would have to eliminate the prior-consent scenario. Yet, since we have no independent evidence one way or another about such a preexistence, the only way he could eliminate it is by claiming either that there are no such greater goods or that it would be wrong for us to ever give such consent. Regarding the former, that would be subject to the skeptical theism objection according to which just because no one can find such greater goods, that does not mean they do not exist. Regarding the latter, Sterba hints at this response by claiming that the Pauline Principle prohibition of God’s permitting the consequences of significantly evil actions “is even grounds for thinking that victims may never be able to give their informed consent to or find reasonably acceptable the infliction of such consequences on themselves” (Sterba 2019, pp. 99–100). Contrary to Sterba’s claim here, it seems to me perfectly moral for a person to offer such informed consent, particularly in the prior-consent scenario. Further, as far as I can tell, Sterba never offers an argument that it is not moral, whereas as explained above, my intuition for this case can be supported via the Kantian Principle, which I argued undergirds the Pauline Principle.6

4.2. Victim-Beneficiary Scenario

As I defined it, the victim-beneficiary scenario consists of either the prior-consent or would-approve scenarios with the additional stipulation that the greater good redounds to the victims of the evil. As noted above, this additional stipulation significantly increases the case for thinking that it is morally permissible for God to allow evil for a greater good in those two scenarios. Since Sterba does not directly address the first two scenarios, it follows that he does not address the victim-beneficiary scenario since it incorporates at least one of those two scenarios. However, he does consider those potential goods elaborated in a theodicy by the late Marilyn Adams in which the goods benefit the victim, goods that he appears to understand as a type of soul-making. (For the latter, see Sterba 2019, chp. 3, “An Attempt at a Theodicy.”) The main way in which Sterba responds to the goods that Adams cites is that they are not “organically related” to the horrible evils that give rise to the goods. This response, however, is not sufficient for rejecting my victim-beneficiary scenario since he would have to show that there are no plausible greater goods that would be so related. One cannot do this by merely looking at some of the major goods that people have cited that could redound to the victims.

In fact, I have developed a theodicy in which the greater goods redound to the victims and are organically related to the evils in the sense that the occurrence of the evils are not only necessary for the goods but the memory of their occurrence forms an integral part of the greater good. Thus they avoid the objection Sterba raises to Adam’s theodicy. I call this theodicy the connection-building theodicy (CBT), though some have suggested calling it the “love theodicy” (See Collins 2014). In brief, the CBT claims that our virtuous responses to evils result in connections of appreciation, contribution, and intimacy (ACI) that occur between the victims of evil and those who aid them, and that these connections last forever due to our having ongoing memories in the heavenly state of what others have done for us.7 It postulates that some of the connections—what I call evil transformative connections—require the existence of corresponding types of evil. For example, person A suffering to benefit another B can result in a connection of appreciation of B for A, along with a connection of contribution resulting from A’s contributing to B. Such connections could not exist if no one suffered. I then argue that we experience these sorts of connections as intrinsically good, and postulate that the greater the sacrifice of the benefactor—such as deeply sharing in the victim’s suffering—the deeper the connection of ACI.

I further postulate that each type of connection forms its own good. Thus, even though there could be a connection of ACI without any evil—e.g., in the heavenly state, Jill could bake a cake with a trillion candles for Jane’s birthday resulting in a corresponding connection of ACI—the connection formed would not be the same good as that formed by, for instance, enduring suffering on another’s behalf. This means that if God prevented a certain type of evil, then the corresponding type of connection would never occur, and hence a certain type of good would never exist. Moreover, given these connections are intrinsically good, and they last forever, I argue that it is plausible to think they outweigh the finite evils necessary to produce them. I also attempt to show how this theodicy could deal with a variety of objections, including that of horrendous evils and that of victims that no one appears to help in this life, in which case their victimage does not appear to lead to any positive connection. Finally, because the connections exist between the beneficiary and the benefactor, both gain from being part of the connection; accordingly, it is a case of the victim gaining a potentially greater good that would not exist if God prevented the evil, with the good being organically related to the evil.

My point here is not to argue for the adequacy of the CBT, but only to point out that there are other potential theodicies than Adams’s in which the greater good redounds to the victims and that potentially avoid the objections Sterba raises to Adams’s theodicy. Thus, to make his case, he would either have to consider all of these such theodicies or offer some general argument against all greater-good theodicies. He attempts to do the latter with the perverse-incentive argument that I critique below.

5. Second Set of Problems

In this section, I first argue that Sterba’s proposal that God should prevent sufficiently serious evil consequences of our free choices fails to consider the degree to which such a policy would make us aware of God’s monitoring of our every action. This in turn would greatly diminish our ability to make morally significant free choices. Second, I point out flaws with his argument for the applicability of the Pauline Principle to God’s choices, particularly what I call his perverse-incentive argument, according to which God would be perversely incentivized to want people to do evil if God allowed evil actions for a greater good.

5.1. God’s Intervening to Stop Consequences of Evil Free Choices

To preserve the type of morally significant free will necessary for soul-making while at the same time eliminating horrendous evils, Sterba proposes that God should prevent horrendous evil consequences of bad acts but allow some sufficiently minor evil consequences (See Sterba 2019, chp. 4). He does not tell us where the dividing line is between minor and major evil consequences, but for the sake of argument, I will assume there is some natural line that can be drawn. He then goes on to claim that the virtuous would not object to the consequences of their acts being restricted in this way; only the “supervicious” would. He compares his proposal to what a superhero or a perfectly just, benevolent state would do. Further, he argues that the regularities of nature required for morally significant free choice could still be maintained.

A critical issue Sterba fails to address, however, is our enormously increased awareness of God under such a proposal. Suppose every time one decided to do some evil action that had horrible consequences, such as one person stabbing another in a fit of anger, something would prevent one from doing it. In such a world, we would be hearing all the time about the latest miracle on the news and through reports of friends, and we would know that they were the result of supernatural agency because they would be counter to the natural regularities in the world. An awareness of God’s watchful eye would be always there, not just for the acts that could result in substantial harm, but even for minor acts. This would largely prevent one from doing what is right for its own sake, something necessary for morally significant free will.

As an analogy, this would be like continually seeing a police car in your rear-view mirror while driving. This would largely prevent you from driving the speed limit because it is the morally right thing to do, instead of doing so to avoid getting a speeding ticket. In contrast, even though in our world one might strongly believe God is watching everything we do, it rarely reaches this level of awareness. Or, as another analogy, Sterba’s proposed world would be like a future “helicopter” mother who installs “under-the-skin” security cameras on her son’s body along with a device that allows her to control his body when necessary. She tells her son that she is only trying to keep him out of serious trouble and assures him that she will not interfere with his more minor missteps. It seems clear to me that this would seriously undercut her son’s free choices even about minor things since he would be acutely aware that his mother is watching him.

Finally, consider Sterba’s benevolent state analogy, according to which the kind of benevolent state we should desire is one that only allows evil consequences of an action if the consequences are less serious than a certain cutoff. To make this state truly analogous to what he claims God should do, it would have to be a state that had security cameras and microphones everywhere: in every room of one’s house, in one’s car, on one’s body, and so forth; otherwise, it would not be able to intervene to stop many horrible acts. (Or, at least people would have to think it is likely they had such monitoring and control devices.) Although this benevolent state would only intervene for acts that met a certain threshold of bad consequences, such interventions would constantly be reported in the news and by one’s friends and family. This would make people almost constantly aware of the monitoring of the state, and the possibility of significant consequences for many morally problematic choices that were under the threshold of prevention. Once again, it seems clear to me that people’s morally significant free choice would be greatly hampered even for minor bad choices.

Indeed, we can imagine deciding whether to be born into this benevolent-state world instead of our world while behind a “Rawlsian veil of ignorance” in which we do not know who we will be born as. It is not obvious to me that being born into such a benevolent-state world is preferable. I think I would rather take my chances, risking the possibility of enduring greater suffering for similar reasons to why many would be willing to risk the sufferings inherent in our world over a “dream-machine” world in which they only experience pleasure.

Now compare this benevolent-state world with our world. In our world, God might be preventing some of the worse evils—such as a global nuclear war—but these preventions are hidden enough that we are not aware of them. Further, the universe is structured so that there is a limit to the evils that can be inflicted—we can only endure so much pain, and only for a limited time because we have a finite lifespan. So, our world is structured with limits to evil while at the same time these limits do not make us aware of any divine monitoring. Of course, this opens up the possibility that God could create a world with a different law structure that allowed for less severe evils. However, I do not think any of us are in a position to conceive of such a world in sufficient detail to evaluate whether such a world would be better than ours. That said, there is one possibility for decreasing the amount of horrendous evil in our world that does not run into the above difficulties, or any other obvious ones: namely, the possibility of God’s implanting in our nature some deep-seated aversion to committing horrendous evil acts, an aversion that makes us much less susceptible to perpetrating such acts. If we did have such an aversion, it would not immediately make us aware of God’s watching over us—for instance, it is not obvious that it could not be explained as a product of our evolutionary development. In any case, this possibility deserves further exploration.

5.2. God’s Obligations and the Perverse-Incentive Argument

Another weakness in Sterba’s argument is his assumption that the moral obligations that apply to us also apply to God. Says Sterba,

So for human agents, given that such intrinsically wrongful actions would significantly conflict with the basic interests of their victims, there are no exceptions to the Pauline Principle for cases of this sort where the significant evil that is to be done is just a means to securing a good to which the beneficiary is not entitled. Moreover, if there are no exceptions to the Pauline Principle for humans in such cases, then the same should also hold true for God. If it is always wrong for us to do actions of a certain sort, then it should always be wrong for God to do them as well. So for contexts where the issue is whether to permit a significant evil to achieve some additional good, God, like us, would never be justified in permitting evil in such cases

(Sterba 2019, p. 78).

In reply, I note that certain moral obligations might hold for us, but not for God since God is not only our creator but our sustainer. The relation of God to us is like that of parents to their children but much more so. Simply assuming that the moral requirements that apply to us also apply to God would be like assuming that since we do not have the right to make other adults do something for their own good (except perhaps in special circumstances), we do not have the right to do this for our children. Indeed, ordinary people implicitly recognize the difference between our moral obligations and God’s when they say a doctor does not have the right to play God in allowing someone to die when they could have prevented it, which implies that God does have a right to do so—that is, God has a right to play God!

In the context of the quotation from Sterba, he does not provide an argument for his claim that moral obligations given by the Pauline Principle apply to God as well as us. He does say in footnote 10 to the above quotation (Sterba 2019, p. 78) that he will provide such reasons in subsequent chapters but does not specify exactly where. As far as I can tell, the underlying reason he gives for his claim is that if the Pauline Principle did not apply to God, God would be incentivized to want us to do evil. He states this in the following passage:

First, it would be morally inappropriate for our receiving a Godly opportunity for soul-making to be conditional on God’s permitting significant and especially horrendous evil consequences of immoral actions. This is because it would give us the incentive to commit, and want others to commit, significant and even horrendous evil actions, virtually without limit, so that God would permit their consequences and thereby make possible our receiving a Godly opportunity for soul-making. It would also support perverse incentives for God as well. Assuming that God wanted to provide us with a Godly opportunity for soul-making, God would also have to perversely want us to commit significant and even horrendous morally evil actions, virtually without limit, so that God could then permit their consequences and thereby make possible our receiving a Godly opportunity for soul-making”

(Sterba 2019, p. 116, italics mine).

Although in the above passage Sterba presents what I call the perverse-incentive argument in the context of soul-making, later he applies it to other greater goods such as those offered by what he calls the felix culpa theodicy (Sterba 2019, p. 112, n. 48).8 I have three major criticisms of his argument. First, the argument assumes that it would always be wrong for God to desire that someone intend evil so that good may come from it. However, this implies that a non-consequentialist ethic is correct; for, under consequentialism, God should desire someone do evil if the ultimate result is better than it otherwise would be. Since God follows correct ethical principles, this entails that God is not a consequentialist. On the other hand, a greater good resulting from someone’s evil act would in itself only incentivize God to want the person to do evil if God were a consequentialist. So, for the argument to work, one must assume two contradictory things: that God both is and is not a consequentialist.9

Second, it seems that one could consistently hold that in some cases it is morally permissible to allow a person to commit evil so that a greater good can result, but it is not morally permissible to want to person to commit the evil. That is, for the cases in question, God could be what could be called a mere-permissibility consequentialist (i.e., a consequentialist about what is morally permissible to allow) without being either a desire-obligatory consequentialist or a desire-permissibility consequentialist (i.e., a consequentialist about what it is obligatory or permissible to desire, respectively). In any case, Sterba needs to offer an argument that for the relevant cases, mere-permissibility consequentialism entails desire-permissibility consequentialism.

Third, even if God were a desire-obligatory consequentialist, Sterba’s claim that God would desire the person to do evil in the imagined situation does not follow. I will show this under the assumption that goods have positive utility, evils have negative utility, and the total utility for a given choice is the sum of all the utilities combined. Specifically, let G_p ≡ the utility of the good of God’s permitting person P to make a morally significant free choice of whether or not to do a certain evil action E; let G_gfc ≡ the utility of the good/virtuous free choice of resisting the temptation of evil along with any consequences that occur as a result; let E_fc ≡ the utility of the evil free choice itself, regardless of whether God prevented all the intended consequences of the choice; E_c ≡ the utility of the evil consequences of the evil choice, apart from the evil of the choice itself; and G_R ≡ the utility of the greater good that results from God’s allowing the evil consequences. Now suppose P chooses E and God does not prevent the consequences. Then,

1. Total utility of P choosing E and God’s allowing the consequences = G_p + E_fc + E_c + G_R.10

Next, suppose P does not choose E. P would still have the morally significant free choice to do so, and hence the good, G_p, of God’s permitting the free choice would still exist. Further, the intrinsic good, G_gfc, of the virtuous free action of not choosing E would exist. However, the freely chosen evil, its consequences, and the resulting greater good—that is, E_fc, E_c, and G_R—would not exist and hence have zero utility. Thus,

2. Total utility of P not choosing E = G_p + G_gfc.

If God is a desire-obligatory consequentialist, God would not want P to choose E if the total utility of P not choosing E (given by #2 above) is greater than the total utility of P choosing E (given by #1 above). That is, if:

3. G_p + G_gfc > G_p + E_fc + E_c + G_R ⟹ God does not want P to choose E

Subtracting G_p from both sides, it follows from #3 that God would not want P to choose E if

4. G_gfc > E_fc + E_c + G_R ⟹ G_R < |E_fc| + |E_C | + G_gfc,

where | | represents the absolute value of a quantity between the vertical bars. Finally, by assumption, the utility of the good resulting from God’s allowing the evil consequences outweighs the utility of the evil consequences—which is why the good is called a greater good. Hence,

5. G_R > |E_C|.

It follows from #4 & #5 that God would not want P to choose E if

6. |E_C| < G_R < |E_C | + |E_fc| + G_gfc.

Condition #6 will be met if G_R is greater than |E_C| but not so large that it is greater than |E_C | + |E_fc| + G_gfc. This in turn implies that as long as G_R is not too large, a desire-consequentialist God would not want P to choose E even if the evil consequences of P choosing E are necessary for a greater good. Consequently, Sterba’s inference fails.

What is the mistake that underlies the reasoning from the implicit assumption that God is a desire-obligatory consequentialist to the conclusion that God would want P to choose evil E if E is necessary for some greater good? My guess is that the mistake results from neglecting all the other utilities except the negative utility, E_C, of the evil consequences of P choosing E and the positive utility, G_R, of the resulting greater good. If we neglect the other utilities, then the utility of P choosing E simply becomes E_C + G_R, which by definition is greater than zero. On the other hand, the utility of P not choosing E becomes zero. Hence, if we mistakenly neglect these other utilities, it does follow that a desire-obligatory-consequentialist God would want P to choose E since that would maximize total utility.

6. Conclusions

In my critique above, I first offered three possible scenarios that are exceptions to the Pauline Principle when applied to God’s allowing horrible evils for a greater good: namely, one in which we all consented in a prior existence to live an earthly life; one in which in the afterlife we would (or will) approve of God’s allowing the evils that were done to us for the greater goods that resulted; and one in which one or both of the last two scenarios are combined with the stipulation that the greater goods redound to the victims of the evils. Along the way, I argued that the Kantian principle that one should not treat others as mere means, but always as ends in themselves, better accounts for the intuitions undergirding the Pauline Principle than the Pauline Principle itself does. Further, I argued that the moral permissibility of these scenarios is not only supported by our moral intuitions, but also by the Kantian principle. After this, I presented two further significant problems with Sterba’s arguments. First, I argued that his claim that God should prevent the serious evil consequences of our free choices fails to consider the degree to which such a policy would make us aware of God’s monitoring of our every move. This in turn would greatly diminish our ability to make morally significant choices. Second, I pointed out flaws with his argument for the applicability of the Pauline Principle to God’s choices. I particularly objected to his argument that greater-good theodicies would imply that God would desire that people sin so that good may come, something he claims is morally perverse.

I thus conclude that unless Sterba can adequately address the above scenarios, and answer the objections I raise to his other arguments, his atheological argument fails.