DRAFT: final version now out at Philosophia, here.
Review of James Sterba, Is a Good God Logically Possible? Palgrave MacMillan, 2019. 209 pages. $29.99 (soft cover). ISBN 978-3-030-05468-7.
Soon after the appearance of Alvin Plantinga’s free will defense (PFWD), the consensus among analytic philosophers of religion was that Plantinga permanently and decisively put to rest the so-called logical problem of evil, according to which God and evil are logically incompatible. Since then, few have attempted to defend a deductive version of the problem of evil, and have focused instead on probabilistic formulations of the problem.
However, the tide appears to have turned. A recent wave of articles indicates that philosophers of religion have become increasingly skeptical of the success of PFWD against the logical problem of evil. To date, three main sorts of worries have been raised. According to the first, PFWD seems to be
incompatible with other central theistic doctrines.According to the second, one or more components of PFWD seems false. According to the third, even if PFWD succeeds in showing that God’s existence is compatible with some types of evil, it’s incompatible with others.
incompatible with other central theistic doctrines.According to the second, one or more components of PFWD seems false. According to the third, even if PFWD succeeds in showing that God’s existence is compatible with some types of evil, it’s incompatible with others.
James Sterba’s Is a Good God Logically Possible? is mainly concerned with the third sort of worry, and, so far as I’ve been able to determine, is the first book-length defense of its kind. Sterba deploys recent work in ethics (esp. work on the trolley problem and related topics) to argue that a certain class of evils, which I shall call significant evils—e.g., brutal physical assaults—are logically incompatible with God’s existence. The result is an important contribution to the problem of evil.
The basic structure of the book is as follows. Chapter 1 motivates the book’s main thesis and provides a brief overview of the book’s basic structure. Chapters 2 and 3 argue that both free will defenses and standard theodicies (e.g., soul-making theodicies) fail to establish the compossibility of God and significant moral evils. Chapter 4 explicates and defends the so-called Pauline Principle—according to which one must not do evil so that good may come of it—to argue that God’s existence is logically incompatible with significant moral evil. Chapter 5 argues that skeptical theism fails to deflect the problem of significant moral evil. Chapter 6 argues that the problem of significant moral evil cannot be dissolved by denying that God is a moral agent, and offers a systematic argument for the claim that significant moral evils cannot be justified by means of an appeal to a greater good. Chapter 7 argues that the biblical account of redemptive history lacks adequate resources to justify God’s permission of significant moral evil. Chapter 8 explicates and defends a deductive argument from natural evil that’s roughly parallel to his main deductive argument from significant moral evil. Finally, Chapter 9 summarizes and consolidates the main lines of reasoning developed in the book.
Space doesn’t permit me to discuss all of the ideas and arguments in this rich and insightful book, but given that primary and (arguably) most significant aim of the book is to revive the logical problem of evil, I will (for the most part) limit my discussion to explicating the latter. I will close by offering some constructive comments on how Sterba’s important case might be strengthened further.
Sterba offers several formulations of his argument. However, we can state the core of his argument simply as follows:
1. If God exists, then significant moral evils do not exist.
2. Significant moral evils exist.
3. Therefore, God does not exist.
Perhaps few will deny that brutal physical assaults (and worse) occur, in which case few will deny (2). Turn, then, to (1). In support of (1), Sterba offers two main considerations: the analogy of an ideally just and powerful political state, and the Pauline Principle.
Start with the former. According to Sterba, an ideally just and powerful state would reliably secure its citizens’ basic interests. Such interests include not only opportunities for soul-making or character-shaping that theists cherish, but also the freedom from being the victim of significant moral evils, such as brutal physical assaults or worse. The catch is that an ideally just political state cannot ensure the freedom required for soul-making without carefully protecting the latter sort of freedom, which sometimes requires the state to constrain the freedoms of potential assaulters through a policy of intervention. Such constraints are loose enough to allow for the freedom to commit insignificant moral evil (e.g., cutting in line for tickets to a concert), and even to imagine, carefully plan for, and take important initial steps toward causing significant moral evils. However, they are tight enough to prevent citizens from carrying out the final steps of those evils.
According to Sterba, the reason why such constraints on freedom are justified is that in cases of assault, “the freedom of the assaulters, a freedom no one should have, is exercised at the expense of the freedom of their victims not to be assaulted, an important freedom that everyone should have” (p. 13), resulting in “an unacceptable distribution of freedom.” (ibid.). Now if an ideally just and powerful state would constrain the freedom of potential assaulters so as to secure the weighty freedom of its citizens from significant harm, then a fortiori would an ideally just and powerful god similarly do the same as far as possible. But unlike finite persons, God has the knowledge and power to prevent everyinstance of significant moral evil, and therefore, like an ideally just and powerful state, would do so. But that is just to say that (1) is true.
Some object that preventing significant evils would result in a world with “toy” freedom. But toy freedom is unacceptable with respect to the aims of a good God, since it would greatly diminish our status as moral agents, and it would greatly curtail the opportunity for soul-making. In reply, Sterba argues that the objection is based on a false dichotomy—either unlimited freedom or toy freedom—since at least one other alternative is possible, viz. freedom that’s properly constrained through a policy of limited intervention. So, for example, suppose God exists and brings about a world W that conforms to the practices of the ideally just state sketched above. Then while God endows us with free will—perhaps even libertarian free will—and provides opportunities for soul-making, he intervenes when necessary to prevent brutal assaults and worse. Suppose further that human beings in W have had enough experience to become aware of God’s practice of limited intervention. Finally suppose that in W, you observe that someone has abducted a child and starts to drive away with them. Then you’d likely assume, rightly, that God will soon intervene (say, by causing the kidnapper to be pulled over by a police officer for a broken tail-light) to prevent the person from doing significant harm to the child. Still, in W, you have the free will and the opportunity for soul-making by freely choosing to intervene on behalf of the child before God does, and thereby prevent the child from the psychological trauma of being briefly abducted. The proposal of limited intervention thus does not result in toy freedom.
The lynchpin in Sterba’s second line of support for (1) is the Pauline Principle (henceforth ‘PP’), according to which one may not do evil so that good may come of it. After apparent exceptions are accounted for, we get:
(PP) One shouldn’t do evil so that good may come of it unless the evil is (i) trivial (e.g., stepping on someone’s foot on the subway so as to get off at your stop in time), (ii) easily reparable (e.g., lying to a depressed friend to keep them from committing suicide), or (iii) (more controversially) the only way to prevent far greater harm to innocent people (e.g., shooting one in twenty civilians so as to prevent, in the only way possible, the shooting of all twenty).
The principle should be acceptable to non-theists, on the grounds that many of the key moves in (e.g.) the literature on the trolley problem can be plausibly explained in terms of at least a tacit acceptance of PP. It should also be acceptable to theists – especially orthodox Christian theists – given that (i) the principle originates from the apostle Paul, in his epistle to the Romans, and (ii) the principle underlies Aquinas’s doctrine of double effect.
One might of course object that Sterba’s formulation of PP fails to take into account the morally relevant distinction between doing and allowing: PP says nothing of the moral acceptability of merely allowing evil that good may come of it, and so PP leaves open the epistemic possibility that it’s permissible for God to merely allow evils that fall crosswise of PP (and the ideally just state). Sterba anticipates this objection. In reply, he argues that permitting evil that’s significant and easily preventable is just as bad or wrong as doing it (cf. allowing a family member to be sexually assualted when this can easily be prevented).
Sterba also anticipates the objection that PP fails to do justice to the morally relevant distinction betweenforeseeing an evil and intending it, such that it’s permissible for a person to permit an evil she foresees -- even if it falls crosswise of PP -- so long as she doesn’t intend it. But perhaps this is what’s going on with God: perhaps (say) He intends us to have significant freedom when he causes us to have it, but he merely foresees that we’ll abuse it. Sterba replies that since God permits significant evils, and permitting is an intentional act, it follows that God thereby intends not to stop the significant and especially horrendous evil consequences of our actions.
From PP, Sterba teases out what he takes to be three minimal exceptionless components of it, which he labels ‘Moral Evil Prevention Requirements’ (henceforth ‘MEPR’):
(MEPR 1) 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), as needed, when that can easily be done.
(MEPR2) Do not permit significant and especially horrendous evil consequences of immoral actions simply to provide other rational beings with goods they would morally prefer not to have.
(MEPR 3) Do not permit, rather than prevent, significant and especially horrendous evil consequences of immoral actions on would-be victims (which would violate their rights) in order to provide them with goods to which they do not have a right, when there are countless morally unobjectionable ways of providing those goods.
Sterba argues that these components of PP are acceptable to deontologists and consequentialists alike – at least when applied to God, since his omnipotence guarantees that he can obtain analogous utility-maximizing goods without significant moral evils. Once Sterba has taken himself to have shown that these moral requirements are acceptable to theists and atheist alike, and to consequentialists and deontologists alike, Sterba presses them into service to argue that the principles entail that since God is a moral agent, he cannot permit significant moral evils unless he meets one or more of the qualifications in them (and in PP). But all such qualifications are mere excusing conditions that make provision for human limitations, and thus do not apply to an omnipotent god. But again, that is just to say that (1) is true.
Now of course Sterba is aware of the many theodicies and defenses that have been developed by theists to defeat (1)-style premises in arguments from evil (e.g., free will, soul-making, etc.), which aim to show that God is, or might be, justified in permitting such evils for the sake of some greater good -- including perhaps greater goods that are beyond our ken (cf. skeptical theism). He therefore devotes several chapters to argue that such theodicies and defenses fail. His response to such theodicies culminates near the end of chapter 6, where he consolidates the key points to construct a general argument against the very possibility of a greater good response to premise (1).
A key piece of machinery in Sterba’s argument against greater-good responses is his fourfold analysis of types of goods. Thus, Sterba argues that all goods can be divided into two basic categories: goods to which we have a right and goods to which we don’t. Each of these two categories can then be divided further into two more: those that presuppose the existence of evil (second-order goods) and those that do not (first-order goods). From these two distinctions, his fourfold categorization falls out: (i) first-order goods to which we have a right (e.g., the good of the freedom not to be brutally assaulted), (ii) second-order goods to which we have a right (e.g., the good of receiving medical treatment for wounds from a brutal assault), (iii) first-order goods to which we do not have a right (e.g., the good of a freely given blissful afterlife), and (iv) second-order goods to which we do not have a right (e.g., the good of the opportunity to provide medical treatment to someone who has been brutally assaulted).
With his fourfold categorization of goods in place, Sterba applies his three moral evil prevention requirements and the key points from his criticisms of standard theodicies and defenses to show that God cannot rightly permit significant moral evil for the sake of obtaining a greater good. We can summarize his general argument against the possibility of greater good justifications for significant moral evils as follows. If God exists, then if some significant moral evil exists, then God can rightly permit it to secure some greater good. Now such a good is either (a) a first-order good to which we’re entitled, (b) a second-order good to which we’re entitled, (c) a first-order good to which we’re not entitled, or (d) a second-order good to which we’re not entitled.
Suppose the good is of type (a). Then by MEPR 1, God cannot rightly permit a significant moral evil to secure such a good (e.g., freedom from brutal assault) unless it’s logically impossible for God to do so without violating someone else’s rights by depriving them of such a good. But it’s not logically impossible for God to secure such a good without violating someone else’s rights by depriving someone else of such a good. So, God cannot rightly permit a significant moral evil for the sake of a good of type (a).
Suppose the good is of type (b). Then by MEPR 2, God cannot rightly permit a significant moral evil unless its victim morally prefers to be such a victim in order to receive the second-order good at issue. But no victim of a significant moral evil morally prefers that they are such a victim in order to receive their second-order goods (e.g., no one morally prefers to be brutally assaulted so as to receive medical treatment for a brutal assault). So, God cannot rightly permit a significant moral evil for the sake of a good of type (b).
Next, suppose the good is of type (c). Then by MEPR 3, God cannot rightly permit a significant moral evil unless it’s logically impossible for God to secure the good for a person without permitting the evil to occur. But it’s not logically impossible for God to secure the good without permitting the evil to occur. So, God cannot rightly permit a significant moral evil for the sake of a good of type (c).
Lastly, suppose the good is of type (d). Then by MEPR 2, God cannot rightly permit a significant moral evil unless its victim morally prefers to be such a victim in order for someone receive the second-order good at issue. But no victim of a significant moral evil morally prefers that they are such a victim in order for someone to receive such a second-order good (e.g., no one morally prefers to be brutally sexually assaulted so as to provide the opportunity for another person to give them medical treatment for one’s own brutal sexual assault). So, God cannot rightly permit a significant moral evil for the sake of a good of type (d). Therefore, there is no good for the sake of which God can rightly permit significant moral evil.
Crucial to the success of Sterba’s case here is his claim that it’s logically possible to secure any relevant good he can rightly permit without permitting a significant evil. In support of this claim, he argues that its denial can be shown to generate contradictions. Sterba discusses two examples. According to the first: If God’s omnipotence doesn’t rule out cases of type (a) above, then since such cases are only causally or nomologically or technologically impossible for us, then it would follow that what is merely causally or nomologically or technologically impossible for us is logically impossible for God. But if so, then agents limited in power (humans) have more power than an omnipotent God, which is absurd.
According to the second: Suppose for reductio that it is logically impossible for God to prevent some lesser freely chosen evil consequence A without it leading to some greater freely chosen evil consequence B. Suppose further that God permits A to occur. Then by hypothesis, B is logically necessitated, in which case B is not a freely caused evil. But this contradicts the hypothesis that both A and B are freely caused. So, we must reject the original assumption: God can always prevent both significant moral evils. But if he can, then by PP, he should prevent both (by PP).
I found Sterba’s main argument from significant moral evil largely persuasive. I will therefore close by briefly mentioning two main ways in which his argument might be strengthened further. The first concerns the value Sterba places on free will. The second concerns the value he places on the opportunity for soul-making.
Start with the value of freedom. Sterba’s analogy of the ideally just political state assumes a view of persons that have a preference structure to do at least some evil—and for at least some, to do even significant evil—and a will to carry it out, coupled with a political practice of intervening when the evil consequences of carrying it out are significant. But mightn’t one worry that there is a better way to make persons, and a correspondingly better political state?
On the proposal I have in mind, citizens come ready-made with a preference structure that finds doing evil repulsive—or at least uninteresting. If one insists, such persons need not lack the power to do evil; they need merely have a preference structure that strongly inclines them against such actions. As such, we can allow that there are possible worlds in which they do evil, but none that are “nearby” (in the Lewis/Stalnaker sense).
Indeed, a case can be made that classical theists, no less than non-theists, have reason to think that the justification for the present proposal is stronger than the one proposed by Sterba. Why is this so? For the classical theist, God is seen as having a nature and preference structure as I’ve suggested; that is, God’s nature and preference structure are such that he finds the prospect of doing evil repulsive. Thus, Wes Morriston (2000) has argued that it’s not at all clear that theists can support the value of freedom of the proposed by Sterba. For a tension arises between the value theists place on free will, on the one hand, and the doctrine of divine moral perfection on the other. For on classical theism, God is morally perfect essentially, and thus lacks moral freedom. But if God lacks moral freedom, then it’s not clear why moral freedom is such a great good for humans to have such freedom. On the other hand, if moral freedom is a great good, then since God lacks moral freedom while humans have it, then there’s pressure to say that humans enjoy a kind of freedom that’s superior to God’s. Better, then, to avoid such worries and deny that libertarian free will that includes the ability (or at least the “nearby” ability) to do evil is a great good.
The proposal I’m suggesting is prima facie compatible with God endowing us—and not just other possible persons—with such a preference structure. If this isn’t possible, then a new problem arises for theists, viz., a problem of personal identity for the redeemed in heaven. For on standard Orthodox views of the heavenly afterlife, humans transition to a state of having a heavenly preference structure, and yet remain numerically identical to their prior selves who existed with a worldly preference structure.
So we have two competing proposals about constraints on freedom. The main difference between the two proposals is that on Sterba’s, the constraint is external (in terms of intervention to prevent evil consequences of our actions), while on the one I’m proposing, the constraint is internal (in terms of one’s preference structure). The grounds for accepting one over the other would therefore seem to come down to what reasons there are for thinking one as better than the other. Now we’ve seen that one sort of reason in favor of the internal constraint model is one suggested by theism, viz, that God himself embodies the internal constraint model, and yet this doesn’t diminish his moral status or the quality of his life. But if not, then it seems that at least theists have strong reasons to favor such a model over Sterba’s. However, Sterba suggests that the external model is preferable in virtue of (i) the intrinsic value of freedom, and (ii) the value of engaging in soul in soul-making to make oneself virtuous, which in turn requires the external model.
This leads me to my second concern which, recall, concerns the value of soul-making. It appears that a similar tension arises for the value of soul-making that arose for the value of free will. Given God’s essential moral perfection, God was morally perfect from the get-go, and thus didn’t undergo a period of soul-making. Despite this, theists do not think God’s agency, praiseworthiness, or the value of his life are therefore diminished. But if not, then it’s not clear why it’s a great good for humans to undergo a period of soul-making. On the other hand, if the opportunity for soul-making is a great good that we have and god does not, then there’s pressure to say that virtuous humans have a kind of character that’s superior to God’s. Given this dilemma, it’s not at all clear why it’s important that humans have the opportunity for soul-making, rather than being endowed with a nature like God’s, who comes equipped with a virtuous character from the get-go.
Relatedly, John Doris and others have raised serious doubts about whether humans have the capacity for soul-making, due to extensive empirical data against the existence of robust moral character traits. This further calls into doubt the plausibility of soul-making theodicies, and similarly for Sterba’s appeal to soul-making to justify the basic freedoms of the ideally just state.
In light of these points, it seems to me that Sterba can strengthen his case by dropping the external constraints model of freedom for the internal constraints model. He could then argue that the ideally just political state would contain virtuous-from-the-get-go citizens who lack free will – or, if they must be free, then they only prefer what is right and good, and choose among right actions only (at least in the actual world and all “nearby” counterfactual worlds).
Given this model, the corresponding ideally just political state requires no intervention. Sterba can then argue that such an account of the ideally just state is more closely analogous to how a perfectly just god would and should relate to its creatures. But if so, then theism is even worse off than Sterba has made out. For then a case can be made that anymoral evil—whether significant or not—is incompatible with God’s existence.
Almeida, Michael. 2012. “The Logical Problem of Evil Regained.” Midwest Studies in Philosophy 36 (1): 173–176.
Anders, Paul C., Joshua C. Thurow, and Kenneth Hochstetter. 2014. “On Counterfactuals of Libertarian Freedom: Is There Anything I Would Have Done If I Could Have Done Otherwise?” American Philosophical Quarterly 51 (1): 85–94.
Doris, John. 2002. Lack of Character. Cambridge University Press.
Howard-Snyder, Daniel. 2013. “The Logical Problem of Evil: Mackie and Plantinga.” In The Blackwell Companion to the Problem of Evil, edited by Justin McBrayer and Daniel Howard-Snyder. Hoboken: Wiley-Blackwell.
———, and John O’Leary-Hawthorne. 1998. “Transworld Sanctity and the Free Will Defense.” International Journal for Philosophy of Religion 44 (1): 1–21.
Leon, Felipe. 2019. “A Perfectly Good Personal Foundation: Some Reasons for Doubt”, in Joshua Rasmussen and Felipe Leon. Is God the Best Explanation of Things? London: Palgrave MacMillan.
Mackie, J. L. 1955. “Evil and Omnipotence.” Mind 64 (254): 200–212.
Maring, Luke. 2012. “A New Problem of Evil: Authority and the Duty of Interference.” Religious Studies 48 (4): 497–514.
Mizrahi, Moti. 2014. “The Problem of Natural Inequality: A New Problem of Evil.” Philosophia 42: 127–136.
Morriston, Wes. 2000. “What’s So Good About Moral Freedom?” The Philosophical Quarterly 50: 344–358.
Otte, Richard. 2009. “Transworld Depravity and Unobtainable Worlds.” Philosophy and Phenomenological Research 78 (1): 165–177.
Plantinga, Alvin. 1967. God and Other Minds. Ithaca: Cornell University Press.
———. 1974. The Nature of Necessity. Oxford: Clarendon Press.
———. 1978. God, Freedom, and Evil. Grand Rapids, MI: Eerdmans.
———. 2009. “Transworld Depravity, Transworld Sanctity, and Uncooperative Essences.” Philosophy and Phenomenological Research 78: 1, 178–191.
Pruss, Alexander. 2012. “A Counterexample to Plantinga’s Free Will Defense.” Faith and Philosophy 29 (4): 400–415.
Rasmussen, Joshua. 2004. “On Creating Worlds Without Evil—Given Divine Counterfactual
Knowledge.” Religious Studies 40: 457–470.
———. 2007. The Wisdom to Doubt: The Justification of Skeptical Religion. Ithaca, NY: Cornell University Press. Kindle Edition.
Sterba, James P. 2019. Is a Good God Logically Possible? London: Palgrave MacMillan.
 Plantinga (1967, 1974, 1978).
 Versions of the logical problem of evil go back to at least Epicurus. The version Plantinga was responding to is the one given in Mackie (1955).
 Morriston (2000); Leon (2019).
 Howard-Snyder and O’Leary-Hawthorne (1998); Rasmussen (2004); Otte (2009); Plantinga (2009); Almeida (2012); Howard-Snyder (2013); Anders et al. (2014); Leon (2019).
 Schellenberg (2007); Maring (2012); Mizrahi (2014).
 For the purposes of this review, ‘significant moral evils’ should be understood as short for Sterba’s preferred locution, ‘significant and especially horrendous evil consequences of immoral actions’.
 Doris (2002).