Against the New Logical Argument from Evil

Abstract

Jim Sterba’s Is a Good God Logically Possible? looks to resurrect J. L. Mackie’s logical argument from evil. Sterba accepts the general framework that theists seeking to give a theodicy have favored since Leibniz invented the term: the search for some greater good provided or greater evil averted that would justify God in permitting the type and variety of evil we actually observe. However, Sterba introduces a deontic twist, drawing on the Pauline Principle (let us not do evil that good may come) to introduce three deontic side constraints on God’s choice of action. He then splits the possible goods into four categories: first- vs. second-order goods, goods to which we have a right, and goods to which we do not have a right. He argues that his deontic constraints rule out each combination, thereby showing that no God-justifying good is on offer. To defuse the argument, I draw on a pair of ideas from Marilyn McCord Adams: (i) God is outside the bounds of morality, and (ii) God can defeat evils by incorporating them into an incommensurately valuable friendship with each human. Properly appreciated, these show that the new logical argument relies on a false premise that is not easily repaired.

1. Introduction

In Is a Good God Logically Possible?, Jim Sterba aims to bring resources from moral philosophy—most notably the principle of double effect—to bear on the argument from evil for atheism. Opposing the general consensus in the philosophy of religion over the last half-century,1 Sterba presents what he contends is a successful logical argument from evil. In contrast to so-called “evidential” arguments from evil, which aim to show that evil constitutes strong (perhaps compelling) evidence against the existence of a good god, logical arguments from evil aim to show that there is a contradiction between the existence of the type and variety of evil we actually encounter and the existence of a good god. In a sense, this is less ambitious than J. L. Mackie’s (1955) logical argument from evil, which aimed to show the inconsistency of a good god with any evil whatsoever. However, the conclusion is still significant, and the stronger premise does not make the argument any less of a threat to theism.

I will develop a response to Sterba’s argument, drawing heavily on the twin ideas that God is exempt from moral norms and that God can defeat horrendous suffering by willingly joining us in it while using it to form unique relationships, making existence overall a great good to every person who exists. For both, I am heavily indebted to the work of Marilyn McCord Adams. I begin by laying out Sterba’s argument and the proposed moral principles it turns.

2. Sterba’s Argument

Sterba offers us the following summary of his argument, meant to “approximate the form Mackie should have used in his famous exchange with Alvin Plantinga.2”

1.	There is an all-good, all-powerful God	assume for reductio.
2.	If there was an all-good, all-powerful God then necessarily he would be conforming to Moral Evil Prevention Requirements I–III	divine goodness premise.
3.	If God were adhering to Moral Evil Prevention Requirements I–III then necessarily significant and especially horrendous evil consequences of immoral actions would not be obtained through what would have to be his permission	evil prevention premise
4.	Significant and especially horrendous evil consequences of immoral actions do occur all around us, which, if God exists, would have to be through his permission	actual evil premise.

Therefore,

5.

It is not the case that there is an all-good, all-powerful God

conclusion.

The argument is valid. So, the theist must deny one of the premises. In order to be in a position to assess the key divine goodness and evil prevention premises, we must state Sterba’s Moral Evil Prevention Requirements I–III, which are intended as consequences of the Pauline Principle: let us not do evil that good may result.3 While Sterba is not committed to the Pauline Principle in its full generality, he does believe that the Moral Evil Prevention Requirements are exceptionless consequences of it. We state them below.

• Moral Evil Prevention Requirement I: prevent rather than permit significant and especially horrendous evil consequences of immoral actions without violating anyone’s rights (a good to which we have a right) when that can easily be done.
• Moral Evil Prevention Requirement II: do not permit rather than prevent significant and especially horrendous evil consequences of immoral actions on their would-be victims in order to provide would-be beneficiaries with goods they would morally prefer not to have.4
• Moral Evil Prevention Requirement III: do not permit rather than prevent significant and especially horrendous evil consequences of immoral actions (which would violate someone’s rights) in order to provide such goods when there are countless morally unobjectionable ways of providing those goods.

What is the significance of these principles in particular? Sterba accepts the general framework for theodicy that most philosophers of religion work in—in order for God to be justified in permitting some evil, God must either thereby gain some greater good or prevent some greater evil—but with a twist. Sterba believes in deontic constraints on the situations in which a greater-good justification would excuse God permitting some evil. He divides the possible goods into four categories: first-order goods to which we have a right, second-order goods to which we have a right, first-order goods to which we have no right, and second-order goods to which we have no right. The Moral Evil Prevention Requirements are intended to show that providing us any good in those categories at the price of allowing the horrendous effects of evil acts would violate a deontic constraint. Each has a unique role: Principle I is meant to rule out first-order goods to which we have a right, together with the resources of omnipotence; Principle III is meant to rule out first-order goods to which we do not have a right; and Principle II is meant to rule out both kinds of second-order goods.5

The first and fourth premises of Sterba’s argument are beyond question: the first is just an assumption for reductio, and the fourth is readily discernible from today’s New York Times. So, the second and third premises are where all of the argumentative action lies.

3. The Divine Goodness Premise

The key second premise will be the main focus of our discussion. First, I will review arguments from the literature for the claim that God has no moral obligations. Prima facie, this falsifies the second premise. I will then explore options for repairing the second premise. The options we examine will fail, and their failure will let us see what kinds of constraints there are upon any successful revision of the second premise. In the process, we will also see a failure of the third premise.

3.1. Exempting God from Moral Norms

We can find three general lines of argumentation in the literature opposing the divine goodness premise. Marilyn McCord Adams denies that God has any obligations concerning creatures because of the “size gap” between them.6 Daniel Rubio makes a decision-theoretic argument concluding that holding God to even a minimal normative standard results in a paradox. Mark Murphy argues that the kinds of reasons we plausibly have to align with the standard canons of morality are not applicable to the kind of being that God is. Brian Davies provides an updated Thomistic account of God as beyond morality.7 If any of these arguments succeed, then the divine goodness premise is false.

Adams’s argument trades on the sheer difference between the God of the Abrahamic religions (the one she is interested in exploring) and human beings. This difference, she argues, places God outside of the network of rights and obligations that constitute morality as commonly understood.8 This is not to say that God is not an agent, one who sets goals and executes plans in pursuit of them. But not every agent falls within our moral community. It is common to distinguish between moral agents and moral patients, and both of these from nonmoral entities. Moral agents have both rights and obligations: things that they must (or must not) do, as well as things that moral agents must (or must not) do for (to) them. The prototypical example of a moral agent is an adult human in good health. Moral patients, on the other hand, have rights without obligations. Sentient animals and human infants are both prototypical examples of moral patients. Nonmoral entities, on the other hand, are more like rocks or hurricanes: without right or obligation.9

Just using prototypical examples, it is plain that some rational agents are not moral agents. All that is required for an agent to be rational is for it to exhibit teleological thinking: setting of goals and then executing plans aimed at achieving those goals. Its performance can then be measured by the norms of rationality. Intelligent animals such as octopi, dolphins, birds, and primates all exhibit these capacities in spades. However, it is rare (though not unknown) for philosophers to attribute moral agency to them.10 The bird who kills and eats another—or even the cat who hunts for sport—is not thereby wicked. They have violated no obligation, because they are the wrong sort of thing to be obliged.

Human infants are another interesting case. While infants are not moral agents, they are clearly moral patients, and they are clearly rational agents. There is evidence of teleological and even empathetic thinking among human infants. However, attempts to analyze the behavior of infants with a moral lens leads to self-parody, as illustrated by this passage from Book I of St. Augustine’s Confessions:

for in Thy sight none is pure from sin, not even the infant whose life is but a day upon the earth. Who remindeth me? doth not each little infant, in whom I see what of myself I remember not? What then was my sin? was it that I hung upon the breast and cried? for should I now so do for food suitable to my age, justly should I be laughed at and reproved. What I then did was worthy reproof; but since I could not understand reproof, custom and reason forbade me to be reproved. For those habits, when grown, we root out and cast away. Now no man, though he prunes, wittingly casts away what is good. Or was it then good, even for a while, to cry for what, if given, would hurt? bitterly to resent, that persons free, and its own elders, yea, the very authors of its birth, served it not? that many besides, wiser than it, obeyed not the nod of its good pleasure? to do its best to strike and hurt, because commands were not obeyed, which had been obeyed to its hurt? The weakness then of infant limbs, not its will, is its innocence. Myself have seen and known even a baby envious; it could not speak, yet it turned pale and looked bitterly on its foster-brother. Who knows not this?11

The difference between human infants and human children, who do possess moral agency, is minuscule. Yet, it is the difference between having moral obligations and lacking them. It does not take much to stand apart from morality’s grasp.

A certain kind of rationalist might object that infants, children, and animals are not moral agents because, while they have sophisticated rational capacities, they lack something that human adults have. That extra rational capacity is what puts adults “over the line”. A divine agent would have superior rational capacities to human adults and so would already be “over the line”. I find the “line” conception of who is and is not a moral agent unconvincing. We can reflect on this with a little science fiction (though a subject that is becoming more science and less fiction every day). One of the central problems of technology ethics is that of AI alignment. It is uncontroversial that an artificial general intelligence (AGI) could possess rational capacities superior to that of a human adult. Would this automatically make an AGI a moral agent? Not clearly. The alignment problem is the problem of designing AGI that are also moral agents. As researchers working on alignment readily admit, it is far from trivial.12

In fact, upon reflection on the variety of beings with which we are familiar, only humans who have achieved at least a few years of development count uncontroversially as moral agents. The circle of rational agents exceeds the circle of moral agents. Given that even very small differences exempt rational agents from the moral norms that govern us, we should expect that vast differences such as between the Abrahamic God and us would have the same effect.

Next, Rubio’s argument takes the form of a proof.13 Rubio starts with the following assumptions: first, the decision of which world to create has no optimal strategy (no best of all possible worlds); second, any world that it would be wrong for God to create is thereby impossible; third, there is at least one world which it would be wrong for God create. Then, we can use two fairly weak principles of rational choice theory to prove that no world is possible. These principles are (i) no worst—if there are more than two options available, the worst option (if one there be) is impermissible to choose; and (ii) the independence of irrelevant alternatives—if in the choice between a and b, option b is not permissible to choose, then in the choice between options a, b, and c, option b is not permissible to choose. Rubio canvasses a number of ways out of this proof and argues that the best one is to discard the option that there is a world that it would be wrong for God to create. This effectively places God’s actions beyond the reach of both morality and rationality.

Next, Murphy’s argument surveys the ways facts about our well-being could connect to reasons for acting.14 Major theories of the (normative) foundation of morality invoke facts about human community, psychology, or nature to explain why promoting the well-being of humans provides reasons that require action. A being like God, by contrast, is not part of a human community, does not have a human psychology, and has a divine nature.15 Consequently, Murphy argues, we should not expect facts about these things to provide God with reasons requiring God to act to promote our well-being. God is not the type of thing to be bound by the moral norms you or I may be.

If even one of these arguments is sound, then God is not a moral agent. If that is true, then Sterba’s second premise fails and so too does the argument.

3.2. Modifying the Premise

In response to these kinds of arguments, Sterba suggests two moves. One: retreat to a “condemnation” of God’s failure to prevent the horrendous consequences of moral evil. Two: argue that even without obligations, we might still expect that God is good overall in God’s relationship with others, and that a failure to observe the Moral Evil Prevention Requirements is a failure to be good in God’s relationship to others. We will consider these in turn.

We will begin with the “condemnation”. In summarizing his first line of response after considering Brian Davies’s account of God as outside the bounds of moral agency, Sterba writes:

Even assuming that God were not subject to moral requirements as Davies contends, God’s failure, if he existed, to prevent the horrendous evil consequences of all the immoral actions in the world when he could easily have done so without either producing a greater evil or failing to secure a greater good is still to be condemned. It would have resulted in far more evil consequences than has been produced individually by all the greatest villains among us.16

At first glance, it is a somewhat curious stance to take that we condemn the acts of a nonmoral agent. In failing to prevent horrendous evil, God has not failed in God’s obligations to us. We do not condemn hurricanes, fires, earthquakes, and other nonmoral causes that set back our interests. Sterba’s condemnation seems here like a category error.

We do, of course, sometimes perform condemnatory behavior toward nonmoral agents for reasons other than expressing moral blame, such as to alter future behavior. The parent scolds their toddler for running into the street, not because the toddler has morally erred but in hope that the relatively mild unpleasantness of the scolding will cause the toddler to be more cautious around streets. But when it comes to behavior—even behavior that does not meet the moral standard, such as an infant disrupting their parents sleep in order to be fed a few hours earlier—that ought not be deterred, condemnation is inappropriate. Premature insistence on sleeping through the night is bad parenting.

It is also unclear how this condemnation repairs Sterba’s argument. In denying that God is subject to moral obligations, the theist denies the divine goodness premise. In order for ”condemnation” to repair the argument, it must provide a replacement premise. Here is the natural option:

(2*)

If God were all good and all powerful then necessarily God would not do anything subject to human condemnation

divine human approval premise.

In addition, Sterba suggested that we might replace talk of divine morality with talk of divine goodness in relationship to others, giving us a second replacement option:

(2**)

If God were all good and all powerful then necessarily God would be good overall in God’s relationship with others17

divine relational quality premise.

However, should the theistic denier of divine moral obligations find 2* or 2** plausible? I think not. We may start with 2*. According to 2*, God avoids doing things that humans condemn. Yet in several of the world’s religions, we find approval for protest against divine (in)action. Much of this originates from the “Book of Job”, sacred scripture to Jews and Christians, which tells the story of a man whom God allows the devil to visit with ruinous evil to demonstrate that his faith is disinterested. Job protests his mistreatment to God, and “Job [spoke of God what is right].18” Any conception of God shaped by the Hebrew Bible must reckon with stories like this. The God of these religions neither fears nor works to avoid human condemnation, even condemnation shaped by a moral code.19

This matters for a logical argument from evil. The goal of a logical argument from evil is to derive a contradiction from a conception of the divine inherent in the beliefs of the theist and our experience of evil in the world. As Mackie understood, his own project was

[to show], not that religious beliefs lack rational support, but that they are positively irrational, that the several parts of the essential theological doctrine are inconsistent with one another, so that the theologian can maintain his position as a whole only by a much more extreme rejection of reason than in the former case. He must now be prepared to believe, not merely what cannot be proved, but what can be disproved from other beliefs that he also holds.20

To hold any probative force, the logical argument must address the beliefs of actual theists. Its goal is to show a contradiction. For that purpose, an assumption that divine goodness requires acting to avoid moralized human condemnation is not fit to task. Premise 2* may be a problem for some theologies, but not for the ones we find in the world’s Abrahamic monotheisms.

This brings us to Premise 2**. This premise seems more promising. It is, plausibly, a religious commitment that God is good in God’s relationships to others, most saliently humanity. The idea that God is “for us” is an important part of most of the world’s theistic religions (although for any for which it is not, 2** would not be the right replacement premise; against, say, an Aristotelian Deist, it would not go far). So, the question then becomes: are there analogs of Moral Evil Prevention Requirements I–III for being good in a relationship?

We will start with Moral Evil Prevention Requirement I. Is it possible to be good overall in a relationship while having violated Moral Evil Prevention Requirement I? I contend that it is and argue by example.

The reformed baker: In a small village lives a baker. Because of the village’s environment, the baker’s bread is crucial for the population to consume sufficient calories and nutrients to avoid the evils of malnutrition and starvation. Being a savvy businessowner, the baker had secured several years’ worth of wheat futures guaranteed at good prices. One year, war came to the land and caused a massive demand shock in the wheat market. Like a savvy businessowner, the baker raised the price of bread as market forces dictated. Unfortunately, most of the villagers could not now afford bread and faced malnutrition and ultimately starvation. Because of the wheat futures, the baker could have maintained prices at prewar levels without compromising what was a profitable business, until the war’s end, when wheat prices would revert to normal. One night, perhaps at the urging of a trio of spirits, the baker had a change of heart, lowered prices, and saved the villagers from hunger. For the rest of the baker’s life, the baker provided the villagers with nutritious and reasonably priced bread and used the war profits to build a school for the village children.

In the reformed baker, we have a clear violation of Moral Evil Prevention Requirement I on the part of the baker. Sustenance is a basic right, and during the war, the baker could have easily provided it. Even though the baker was blameless in the circumstances that caused starvation in the village—the demand shock in food prices—he could have prevented its consequences by maintaining prices instead of war profiteering. He chose not to. But, in a change of heart, he decided to save the villagers and atone for his actions.

Was the baker good overall in his relationship to the villagers? I contend that he was. After his moral awakening, he relieved their suffering despite lacking any practical or self-regarding reason to do so. Furthermore, he took his war profits and invested them in bringing a public good to the community. The village was better off after the demand shock than before. Of course, the baker’s track record was not one of uniformly acting in the villagers’ interest. However, the standards for an overall good relationship are not so high. Despite violating Moral Evil Prevention Requirement I, the baker’s actions were still on the whole benevolent.

Perhaps the atheist can make the following rejoinder: granting that the reformed baker shows 2**’s insufficiency, we might still improve it. Instead of requiring that God be good overall in God’s relationship to others, what if we require that God is perfect in God’s relationship to others? The baker is not perfect in his relationship to the villagers, and so this newly created 2*** does not face the counterexample. This, I think, is a dead end, because we lack a sufficiently developed conception of what it takes to be perfect in relationship to others. If we define perfection in relationship as including observation of all three Moral Evil Prevention Requirements, the atheist risks victory by definition—no victory at all. We might attempt other definitions, perhaps in terms of how much weight one places on the interests of the other, but these sorts of efforts risk outstripping the commitments of the theist. The world’s monotheistic religions promise a God who is for us. They do not make specific promises about how that God weighs our rights and interests against other divine cosmic purposes.

So let us stick with 2** and proceed to the second Moral Evil Prevention Requirement. According to it, one constraint on greater-good justifications for allowing horrendous evil is that the goods not be ones the recipients ”would morally prefer not to have”. The use of the term ”morally prefer” is somewhat puzzling. The idea of a preference is straightforward enough; what does the qualifier ”morally” add? Unfortunately, Sterba does not elaborate on the idea of a moral preference. Some candidate meanings are presented below.

• mp-i: an agent morally prefers option x to option y just in case she prefers x to y and in so doing does not violate any canon of morality.
• mp-ii: an agent morally prefers option x to option y just in case her ideal self prefers x to y.
• mp-iii: an agent morally prefers option x to option y just in case attention to only her moral reasons would lead her to prefer x to y.21

Before confronting a revised Moral Evil Prevention Requirement II in the context of Premise 2**, one should recall the great good that, in the world’s great religions, God provides to defeat evil.

3.3. The Good of Divine Friendship

The idea of defeating evil (with good) originated with Roderick Chisholm and was developed in the specifically Christian context by Marilyn McCord Adams. The central idea is that given some evil, there are two ways of providing some good in response to it. One way is to balance it off. Suppose, for example, in a mediocre world, a Chosen One breaks their arm. In recompense, that world’s deity raises the quality of the world from mediocre to excellent (perhaps by increasing the well-being of every individual from a mediocre level to flourishing). The world would be better still without the arm breaking, but the deity’s actions add far more good to the world than the arm breaking added evil. The evil of the arm breaking is balanced off by the deity’s miracle.

The other way is to defeat it. In contrast with mere balancing off, a defeated evil contributes positively to the good of the whole. It achieves this by entering into organic unity with the goods provided to defeat it. By destroying this organic unity, removing the defeated evil from the world would thereby make it worse, wounding the world by an act of vandalism. For example, a splotch of ugly color in an impressionist’s painting may enhance the value of the whole despite being ugly when considered alone. Or, closer to the relevant case, the ill fortunes of a life may nevertheless enhance its overall value by providing a good narrative structure. The mark of a defeated evil in the career of a rational being is a later endorsement; if down the line, considering only my own well-being, I prefer the suffering to be included in my life rather than out of it, it has been defeated.22

For Adams, defeat rather than balancing off is the standard if God is to be good to us. To provide that defeat, she proposes God Godself. Relying on the Christian doctrine of incarnation, she argues that God defeats horrendous suffering by voluntarily undergoing horrendous suffering, and thereby creating solidarity and relationship with the sufferers. Culminating in a beatific vision that is unique to each individual and a friendship with God that is unique to each individual, horrendous suffering is enfolded within the larger human–divine relationship. Friendship with God formed through solidarity in horrendous suffering is a second-order good, logically dependent upon the horrendous suffering of both human and divine participant. I will now argue that it is not a second-order good that its recipients would prefer not to have. Consequently, even if God is conforming to Moral Evil Prevention Requirements II, God would still allow victims to suffer the consequences of horrendous evil in order to forge a uniquely valuable friendship with them as co-sufferer. This leaves Moral Evil Prevention Requirement II intact, but at the price of the evil prevention premise.

3.4. Moral Preference

We begin with mp-i. According to mp-i, a moral preference is simply a preference that does not violate any canon of morality. Is there some moral objection to preferring a life that contains horrendous suffering that is defeated through a relationship that provides incommensurate good? Not obviously. Something like the Pauline Principle would frown upon using the suffering of others as a way to obtain great goods, but that is not what we are talking about. Morality does not stop us from preferring a life with suffering to one without when it is our own suffering and partially constitutes a great good.

Next, mp-ii. This interpretation of moral preference, somewhat inspired by virtue theory, interprets one’s moral preferences as the preferences of an ideal self. It too is a nonstarter. In traditional religions, the ideal self is just the beatified self. Thus, the mp-ii moral preferences are just the preferences of the beatified self.

That brings us to mp-iii. Most similar to notions of moral preference we find in economics, mp-iii identifies moral preferences as the way preferences would be were they informed purely by moral reasons. The extent to which morality demands we eliminate self-suffering is debatable. While the usual view holds that morality places no self-obligations upon us, there are dissenters.23 Granting that a moral preference in this sense tells against accepting horrendous suffering even if it is defeated by coming to partially constitute an incommensurate good, we move now to the question of whether violating our twice-modified Moral Evil Prevention Requirement II renders one no longer good overall in a relationship. For clarity, we state the new principle as below.

Moral Evil Prevention Requirement II*: do not permit rather than prevent significant and especially horrendous evil consequences of immoral actions simply to provide would-be beneficiaries with goods they would prefer not to have were their preferences shaped only by moral reasons.

Here, it seems clear that violating the modified principle does not rule out being good overall in a relationship. In Adams’s view, God still provides every single person a life that is a great good to them and in which all suffering is defeated. Perhaps such a God violates the Pauline Principle, but it is hard to fathom what reasonable complaint the beatified may level.

This brings us to the third Moral Evil Prevention Requirement. According to Moral Evil Prevention Requirement III, it is impermissible to allow great evils in order to secure a good that could have been secured in a morally unobjectionable way. Does violating this norm rule out being good overall in a relationship?

We can modify the reformed baker example to argue for the negative. Suppose that instead of being a profit-seeking business owner, the baker is a single-minded education fanatic. He raises prices when the war causes a demand shock not because it is what the market will bear, but as a means to raise funds for a school. In the modified story, the school is not now a sacrifice on the part of a reformed profiteer, but the work of a myopic figure whose desire to see a school built blinds him to the suffering inflicted by high bread prices. The story still includes moral epiphany and apology, the eventual relief of suffering, and provision of a communal good. We may also add that the baker had an alternative, morally unobjectionable means of constructing a school—by organizing a community effort, for example. The modified story then speaks of a violation of Moral Evil Prevention Requirement III while still revealing a relationship that is overall good between the baker and the villagers.

3.5. Taking Stock

We have now exhausted the second line of response Sterba offers to the theist who denies that God is bound by moral norms. Ultimately, it is a failure. A requirement of goodness overall in one’s relationships is no substitute for morality. Having now explored all three Moral Evil Prevention Requirements, we are in a position to draw a general lesson. Deontic status attaches to acts and is all or nothing. Once an impermissible act has been performed, there is nothing more to say. It cannot be reversed. Acts are static, confined to the location of their occurrence. Relationships, on the other hand, are dynamic. They can change. Consequently, errors in relationship can be atoned. Poor relationships can become good ones with proper devotion of resources and effort. Even when divine behavior towards us seems decidedly unfriendly, there are things God can do to rebuild the relationship. This is not so with moral norms. Absolute norms when violated cannot be unviolated. To defend against eschatological answers to the challenge of horrendous evil such as Adams’s that rely on God doing something about them after the fact, the atheist requires the stern rigidity of moral law.

The question is often raised (e.g., in Murphy 2017) as to whether a God who is not a member of our moral community is one to whom we can reasonably give worship and allegiance. I do not have space here to fully discuss this question, although I endorse Murphy’s discussion as both nuanced and insightful. I will say only that a God who does not follow the rules, but nevertheless guarantees an incommensurably good life to all of God’s children, seems worthy of both.

4. Conclusions

Sterba’s new logical argument from evil aims to show that within a standard framework for theodicy, one where divine permission of evil is justified by either providing a greater good or preventing a greater evil, consideration of a few deontic constraints on permitting evil shows that no possible God-justifying good can be found. These flow from the Pauline Principle, let us not do evil that good may result. Sterba divides goods into four kinds: first-order goods to which we have a right, second-order goods to which we have a right, first-order goods to which we have no right, and second-order goods to which we have no right. Encoded as Moral Evil Prevention Requirements I–III, Sterba argues that his constraints rule out all four kinds of good as candidate God-justifiers.

My response began with arguments that show that God is not subject to the requirements of morality. I then considered the question of whether Sterba’s argument could be modified, with the Moral Evil Prevention Requirements functioning as necessary conditions on being good overall in one’s relationships. I argue that they are not. The dynamic nature of relationship allows for missteps to be corrected and compensated for. In particular, I argue that the framework for thinking about horrendous evils in the context of divine–human friendship developed by Marilyn McCord Adams shows a way for God to violate Sterba’s principles while still being good in God’s relationship to us. I conclude that we do not have a successful new logical argument from evil.24