Defending the Free Will Defense: A Reply to Sterba

Abstract

James Sterba has recently argued that the free will defense fails to explain the compossibility of a perfect God and the amount and degree of moral evil that we see. I think he is mistaken about this. I thus find myself in the awkward and unexpected position, as a non-theist myself, of defending the free will defense. In this paper, I will try to show that once we take care to focus on what the free will defense is trying to accomplish, and by what means it tries to do so, we will see that Sterba’s criticism of it misses the mark.

1. Introduction

James Sterba (2019, chapter 2) has recently argued that the free will defense fails to explain the compossibility of a perfect God and the amount and degree of moral evil that we see. I think he is mistaken about this. I thus find myself in the awkward and unexpected position, as a non-theist myself, of defending the free will defense. In this paper, I will try to show that once we take care to focus on what the free will defense is trying to accomplish, and by what means it tries to do so, we will see that Sterba’s criticism of it misses the mark.

I begin by outlining and explaining Sterba’s argument (Section 2). Next, I outline and explain the dialectic between the logical problem of evil and the free will defense (Section 3). I then argue that Sterba’s argument relies on a false premise (Section 4). I conclude with a brief discussion of how I think the considerations raised by Sterba help advance the problem of evil in other ways.

2. Sterba’s Argument against the Free Will Defense

Sterba’s argument depends on three central claims. (The claims I describe below are not explicitly articulated by Sterba in the same way I do.) The first claim is this:

Zero-Sum Freedom Cases: Some decisions regarding whether or not to interfere with someone’s attempt to Φ are zero-sum decisions between whether to protect one of their freedoms or a freedom of someone else’s.

On zero-sum freedom cases, there are no available courses of action that are not subsumed under “deciding to interfere” or “deciding not to interfere”, and there are no available courses of action that do not constitute a sacrifice of someone’s freedom. There are both familiar and surprising instances of zero-sum freedom cases.

Here is a familiar instance. Political states must balance which freedoms to secure and which freedoms to interfere with. Most of us would agree, for example, that a “just” political state would contain policies and procedures that protect, even if they do not guarantee, someone’s freedom from assault—namely, by creating both the relevant incentives and the relevant institutions. On certain occasions, however, this protection manifests itself as the forceful violation of someone’s freedom from interference by the state in their private affairs. This is a physical freedom that a just political state should generally care to protect as well. Unfortunately, there are many dilemmatic occasions where a just political state is in a zero-sum freedom case with respect to these two important freedoms: their agents must choose to protect either someone’s freedom from assault or someone’s freedom from interference, and they just do not have the option to protect both.

A more surprising example is the tension between the freedoms of the rich and the needs of the poor. According to Sterba (2019, p. 16), political libertarians—those whose views on public policy tend to prioritize the protection of our freedoms—misunderstand this tension by not seeing it as a zero-sum freedom case as well:

“When the conflict between the rich and the poor is viewed as a conflict of freedoms, we can either say that the rich should have the freedom not to be interfered with in using their surplus resources for luxury purposes, or we can say that the poor should have the freedom not to be interfered with in taking from the rich what they require to meet their basic needs. If we choose one freedom, we must reject the other.”

(My emphasis)

Unfortunately, there are many dilemmatic occasions where a just political state is in a zero-sum freedom case of this kind as well: their agents must choose to protect either the poor’s freedom from interference or the rich’s freedom from interference, and they just do not have the option to protect both.

The second claim behind Sterba’s argument is the following:

Morally Loaded Cases: On certain zero-sum freedom cases, it is necessarily true that one of the relevant freedoms is morally more significant than the other(s).

Not every zero-sum freedom case is clear cut. There is certainly a role for uncodifiable practical wisdom in our individual and collective pursuit of justice. But some cases, on the other hand, are such that the greater moral significance of one of the relevant freedoms is clearly engraved in our intuitions and ideals. On these cases, again, whatever you decide constitutes a choice of which freedoms to protect (so you cannot escape making a freedom-related choice). What is peculiar to them, however, is that if we focus exclusively on the moral value of the relevant freedoms, there is necessarily only one right thing to do.

Looking at these kinds of scenarios, Sterba (2019, pp. 13–14) thinks our sense of justice identifies what he calls “morally unacceptable distributions of freedoms”:

“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.”

“The practice of constraining the freedom of would-be assaulters in favor of the freedom of their would-be victims is characteristic of societies that are strongly concerned to be just.”

Assault, however, is just an example. Sterba has also argued that careful attention to our moral sense reveals that a just political state would favor the freedom of the poor over the freedom of the rich, in the kinds of dilemmatic cases described above. The poor’s freedom not to be interfered with in taking from the rich what they require to meet their basic needs is morally more significant than the rich’s freedom not to be interfered with in using their surplus resources for luxury purposes (c.f., Sterba 2014).1

The third and final pillar supporting Sterba’s argument is the following:

Lesser Freedom Cases: Many instances of evil in the history of the world are direct and indirect consequences of the morally less significant freedom prevailing in morally loaded zero-sum freedom cases: cases where someone’s morally less significant freedom to perpetrate some relevant evil prevailed over someone’s morally more significant freedom from suffering that evil.

Referring us to the morally loaded zero-sum cases discussed above, Sterba (2019, p. 18) defends this third claim by noting that “we have not yet achieved a morally acceptable distribution of significant freedom in most societies around the world” and that “this has been true throughout most of human history.” He has a controversial argument for this claim, but that matters very little for his overall point.2 The claim is independently plausible. Brief reflection on the historical ubiquity of slavery, for example, reveals a pattern of morally unacceptable distributions of freedoms that has characterized human societies throughout time. Certainly, the morally less significant freedom has prevailed in nearly all those cases. What is worse, Sterba (2019, p. 18) is no doubt right that the vast majority of humans have either perpetrated such injustices, benefitted from them, or simply not done their part to correct them.

We now have all the ingredients we need to see the force of Sterba’s argument against the free will defense. Let us assume that God exists and is all-powerful and all-knowing, and therefore that it is appropriate to say He has foreseen and has permitted all of the evils of history. If Sterba’s three claims are all true, then it seems we cannot justify God’s protecting the morally less significant freedom, in the many morally loaded freedom cases productive of the evils in our history, by appeals to the moral significance of freedom itself, as the free will defense purports to do (c.f., Sterba 2019, pp. 23–24). There may well be other ways to justify God’s permission of those evils—e.g., with non-freedom-related considerations that outweigh the reasons provided by the comparative moral significance of the relevant freedoms in each case (although the rest of Sterba’s book is devoted to challenging this claim as well)—but appealing to any such non-freedom-related considerations is tantamount to giving up on the free will defense. Once we understand the nature of far too many of the evils in our world, in other words, we see that freedom, logically speaking, cannot do the God-justifying work the free will defense needs it to do. As Sterba (2019, pp. 29–30) summarizes:

“We cannot say that God’s justification for permitting the moral evil in the world is the freedom that is in it because God could have reduced the moral evil in the world by increasing the significant freedom in the world, and that has not been done. Hence, there is no Free-Will Defense of the degree and amount of moral evil in the world.”

Deciding to interfere with someone’s morally less significant freedom to perpetrate evil on morally loaded zero-sum cases, after all, would amount to protecting someone’s morally more significant freedom from suffering that evil, thereby increasing the amount of morally significant freedom in the world, and reducing the amount of evil as well.

This is a compelling argument. I think it is mistaken, nonetheless. In the next section, I clarify what the original free will defense was trying to accomplish and distinguish it from what I call the “extended” free will defense, which is Sterba’s proper target. In the final section, I explain where I think the argument above goes wrong.

3. The Free Will Defense (and Its Expansion)

According to J.L. Mackie (1955, p. 200), classical theism is in a tight spot: it must accept, definitionally, that God is omnipotent, wholly good, and that evil exists, but it cannot accept these three claims consistently. This alleged problem for classical theism is internal. As Mackie (1962, pp. 153–54) clarified a few years after his canonical exposition:

“The question is whether God’s being what the theist calls wholly good, and omnipotent, is compatible with the existence, which he recognises, of what he calls evil.”

(emphasis original)

Call the difficulty of demonstrating that classical theism is internally consistent, in this sense, as the logical problem of evil.3

Mackie’s attempt to articulate this difficulty in more detail, of course, was fraught with its own set of well-known challenges. Indeed, one of the many lessons from Alvin Plantinga’s (1977, p. 22) famous reply to Mackie is that the logical problem of evil has no bite unless something like the following ambitious claim is true:

Mackie’s Key: Necessarily, if God is omnipotent, then God can prevent any evil without bringing about an even greater evil or eliminating some even greater good.

This is not a wholly unintuitive claim. Yet Mackie’s original discussion of the issue did not show it to be the case. In this sense, the logical problem of evil is, perhaps, unmotivated or incomplete. The aim of Plantinga’s free will defense, however, is to move beyond this merely defensive maneuver. Instead of merely claiming that the theist is under no rational pressure from the logical problem of evil until presented with an argument for Mackie’s Key, Plantinga aimed to show that Mackie’s Key is demonstrably false.4

The primary goal of the free will defense, then, is demonstrating that the following is true:

Plantinga’s Lock: Possibly, God is omnipotent, and God cannot prevent every evil without bringing about an even greater evil or eliminating some even greater good.

Although there are many other evil-related problems in the vicinity that are left unaddressed (more on this below), if Plantinga’s Lock is true, then Mackie’s version of the logical problem of evil, at least, is done for good.5

Plantinga’s defense of Plantinga’s Lock—i.e., his version of the free will defense—begins with a version of libertarianism about what I will call individual freedom—the kind of freedom that is a necessary condition for genuine desert and moral responsibility:

P₁: S is individually free, at a certain time t and with respect to a certain action A, only if, at t, there are no antecedence conditions and/or causal laws that determine that S will or will not perform A.

(c.f., Plantinga’s 1977, pp. 29–30)

This is Plantinga’s attempt to capture what is required for it to be “within our power” to both A and not A, at that time, such that genuine desert and moral responsibility could be appropriate on account of our A-ing or not A-ing. It is possible, however, to have individual freedom and yet exist in a world in which we are never in a position to freely choose between doing what is right and what is wrong. This would be a world where we would be individually free—we would have the relevant powers of choice and would exercise them on various trivial occasions—but where we would not be what Plantinga (1977, p. 30) calls significantly free: “[individually] free with respect to a morally significant action.” The second element of Plantinga’s free will defense, then, is a claim about the value of there being creatures with significant freedom in this technical sense—i.e., creatures with individual freedom and the opportunity to exercise it with respect to morally significant actions:

P₂: It is possible that a world with creatures who are significantly free and some evil is more valuable than a world without any evil and without creatures who are significantly free.

This is a comparative claim. Although individual freedom is a necessary condition for desert and moral responsibility, we only get to enjoy the things of extreme moral significance that are connected to desert and moral responsibility if we are actually placed in situations that constitute genuine opportunities to deserve praise and blame for our choices—situations where it is within our power to do what is right or wrong. Indeed, it is not an egregious stretch to suggest that, possibly, eliminating evil at the cost of all significant freedom, in this sense, would produce a less valuable world than ours. That would be a world, after all, where no one ever deserves praise or blame for anything, where no one is ever morally responsible for anything, and where none of the attitudes and relations that gravitate around these properties exist as well—e.g., admiration, honor, dignity, courage, loyalty, etc. The third and final element in Plantinga’s free will defense, then, is a claim about the impact of these considerations on God’s powers:

P₃: It is possible that God could not have actualized a world with creatures who are significantly free and without some evil.6

Plantinga’s reasons for accepting P₃ are a combination of his endorsement of libertarianism about individual freedom, his endorsement of Molinism—the view according to which God’s creative activity is constrained by the truth-values of the counterfactuals of creaturely freedom7—and his endorsement of an essences-based metaphysics of necessity that is most fully developed in Plantinga (1974). In short, Plantinga’s view is that whether or not God can actualize worlds with significantly free creatures that never do wrong depends on the interconnected branching pattern of truth values of the counterfactuals of creaturely freedom—which are both contingent and independent of God’s will, if we are to be truly free—and it is at least possible that this pattern makes all possible worlds with significantly free creatures that never do wrong unactualizable by God.8

These three claims are not, of course, obviously true.9 If they are true, however, then so is Plantinga’s Lock. Mackie’s version of the logical problem of evil, in that case, is dead. But Mackie’s version of the logical problem of evil is the boldest possible formulation of that particular problem. Consequently, it is the least difficult version to overcome (where “least difficult” does not imply “not difficult”). One can, naturally, grant Plantinga’s Lock and nonetheless insist that what is logically incoherent is rather the existence of the perfect God of classical theism and the amount and distribution of evil that we see in the actual world. This is a concessive articulation of the original problem, in a sense, but it packs a stronger punch in turn. What the free will defender needs to show now, after all, is not simply that the great value of individual freedom justifies God in permitting some evil, but rather, as Anthony Flew (1973, p. 232) put it, that all the actual exercises of free will we see in history are “items in that sum of actual alleged higher values to which the Free Will Defender appeals in hopes of offsetting, with plenty to spare, the sum of all actual evils in what is supposed to be the creation of his God.” Call this the expanded logical problem of evil. This is the version of the logical problem of evil that Sterba finds compelling.

Borrowing from our discussion just above, we can say that the burden of the proponent of this expanded version is showing that the following is true:

Mackie’s Expanded Key: Necessarily, if God is omnipotent, then God could have prevented many of the actual evils without bringing about an even greater evil or eliminating some even greater good.

Similarly, we can say that the burden of the theist who wants to move, with Plantinga, beyond a merely defensive maneuver—beyond challenging the proponent of this version of the problem of evil to show that Mackie’s Expanded Key is true—is demonstrating that the following is true instead:

Plantinga’s Expanded Lock: Possibly, God is omnipotent, and God could not have prevented any of the actual evils without bringing about an even greater evil or eliminating some even greater good.

This is certainly an ambitious claim. Providing a free will defense for it, in turn, requires modifying the second and third of Plantinga’s original claims into the following:

P_2*: It is possible that a world with the amount of significant freedom that we see and the amount of evil that we see is more valuable than worlds with much less evil but much less significant freedom too.

P_3*: It is possible that God could not have actualized a world with any less evil without also actualizing a world with much less significant freedom.

Call the combination of P₁ and these two new claims the expanded free will defense. This is the version of the free will defense Sterba thinks is hopeless.

Plantinga (1977, pp. 55–57) himself was willing to explicitly endorse P_3*. In so doing, I assume he was implicitly endorsing something like P_2* as well. His argument, moreover, is the same as the argument for the original free will defense above: whether or not God can actualize any given possible world depends on the interconnected branching pattern of truth values of the counterfactuals of creaturely freedom—which are both contingent and independent of God’s will, if we are to be truly free—and it is at least possible that this pattern makes all possible worlds with a valuable enough amount of significant freedom but less evil than ours, unactualizable by God. If one is willing to grant this argument for Plantinga’s Lock, it seems one should be just as willing to grant it for Plantinga’s Expanded Lock as well. Has Sterba’s argument from Section 2, at any rate, given us any reason to think otherwise?

4. Particular vs. Global Judgments of Comparative Value

Let us recall and restate Sterba’s argument more carefully, in the following way:

Sterba’s Argument Against the Free Will Defense.

S₁. If a perfect God exists, then many of His decisions regarding whether or not to interfere with someone’s attempt to Φ were freedom-problematic decisions, decisions such that:

(i)

They were necessarily decisions between whether (a) to protect someone’s freedom to do evil or (b) to protect someone’s freedom from suffering an evil (i.e., God was in a zero-sum freedom case);

(ii)

They were cases where it was necessarily true that the freedom from suffering an evil was morally more significant than the freedom to do evil (i.e., God was in a morally loaded case);

(iii)

They were cases where God decided to protect the morally less significant freedom to perpetrate an evil instead of protecting the morally more significant freedom from suffering that evil instead (i.e., God brought about a lesser freedom case).

S₂. If many of God’s decisions were freedom-problematic decisions, then it is not possible that, on those occasions, God was justified in permitting the relevant evil on account of considerations pertaining to the moral significance of freedom.

S₃. If it is not possible that, on those occasions, God was justified in permitting the relevant evil on account of considerations pertaining to the moral significance of freedom, then the free will defense fails to justify God’s permission of many instances of evil.

C. So if a perfect God exists, then the free will defense fails to justify God’s permission of many instances of evil.

Notice how Sterba’s S₂ seems to imply that if S₁ is true, then P_3* is false:

P_3*: It is possible that God could not have actualized a world with any less evil without also actualizing a world with much less significant freedom.

As we noted at the end of Section 2, this is because deciding to interfere with someone’s morally less significant freedom to perpetrate evil on morally loaded zero-sum cases would necessarily amount to protecting someone’s morally more significant freedom from suffering that evil, thereby increasing the amount of significant freedom in the world and reducing the amount of evil as well.

This is a tempting but, as far as I can tell, mistaken interpretation of the dialectical import of Sterba’s points. To see this, we must first clear up some obfuscating terminology.

When Sterba talks about “significant freedom” in his S₁ he is talking about something very different from the kind of “significant freedom” Plantinga is talking about in his P_3*. If we are going to suggest that the truth of S₁ is to have any implications for the truth of P_3*, therefore, we first need to clarify how these two different notions—annoyingly picked out by the same expression—are supposed to be related. Sterba (2019, p. 12) is well aware of this, of course, and he is careful enough to tell us both what he means by “significant freedom” and how he understands the relation between his use and Plantinga’s:

“For me, significant freedoms are those freedoms a just political state would want to protect since that would fairly secure each person’s fundamental interests… significant freedoms for me are like the freedom from assault, whereas Plantinga’s significant freedoms include those freedoms and also include freedoms like the freedom of not having someone cut in front of us in the line for the movies. Clearly it is God’s failure to secure significant freedoms in my sense and not God’s failure to secure the additional freedoms captured by Plantinga’s more expansive notion of significant freedom that gives rise to the problem of evil.”

According to Sterba, in other words, the kind of “significant freedom” he cares about—i.e., the kind “a just political state would want to protect”—is a subset of the kind of “significant freedom” Plantinga cares about. Plantinga is thinking of things such as “freedom from assault” and things such as “the freedom of not having someone cut in front of us in the line for the movies,” whereas Sterba is only thinking of things such as the former.

This, however, is a mistake. As we have seen above, Plantinga (1977, pp. 29–30) takes “significant freedom” to consist in the combined existence of individual freedom—the kind of fundamental freedom required for desert and moral responsibility—and the existence of a genuine opportunity to exercise it with respect to morally right and wrong actions. Neither “freedom from assault” nor “the freedom of not having someone cut in front of us in the line for the movies” are examples of the kind of thing Plantinga is talking about. The state of being free from some harm H, in general, is not the same kind of thing as having a certain power and opportunity. Despite what Sterba tells us in this passage, therefore, set-and-subset is not the proper understanding of the relationship between their shared term. It is rather unclear, in fact, what Sterba ultimately means by his S₂, such that it could indicate some tension with P_3*.

Taking our cue from a later passage, however, one where Sterba (2019, p. 27) gives us the heart of his concerns, it seems to me more accurate to think of their different uses of the same term as different selections of non-overlapping subsets of the broader set of morally valuable freedoms:

“Plantinga fails to take into account that there are two ways that God can promote freedom in the world. He recognizes that God can promote freedom by not interfering with our free actions. However, he fails to recognize that God can also promote freedom, in fact, promote far greater significant freedom, by actually interfering with the freedom of some of our free actions at certain times.”

The idea here would be that while what Plantinga calls “significant freedom” is certainly a morally valuable kind of freedom, given its role as a necessary condition for desert and moral responsibility, what Sterba calls “significant freedom” is a morally valuable kind of freedom as well, given its role as a necessary condition for, or reliable means to, securing each person’s fundamental interests. In what follows, I will assess Sterba’s argument, at any rate, with this interpretation in mind. But clear thinking calls for the use of different terms when talking about different things, especially in cases where their difference matters to the argument. And since Plantinga’s usage is older and more established, in what follows I will use “significant freedom” to refer exclusively to the property he has in mind, reserving “political freedom” (a term used by Sterba himself) for the different properties Sterba has in mind. I will refer to both as “morally valuable”, indicating that they both matter morally and that God has prima facie reasons to promote both.

Now that we are clear on the differences between significant freedom and political freedom, we need to clarify the content of S₂—the bridge premise connecting Sterba’s S₁ to Plantinga’s free will defense. Indeed, built into S₁, notice, is an evaluative claim suggesting that, at least on some noteworthy occasions—though not necessarily always—protecting political freedoms is more morally valuable than protecting significant freedoms. Those are what I have labeled morally loaded cases. With this in mind, we can re-phrase S₂ as the following claim:

S_2*. If many of God’s decisions were freedom-problematic decisions, then it is not possible that, on those occasions, God was justified in permitting the relevant evil on account of considerations pertaining to the moral value of significant freedom.

In other words, S_2* is telling us that we cannot appeal to the moral value of significant freedom to justify God’s failure to protect a political freedom which, on this occasion, had even more moral value. Interestingly, however, we can now see that, properly disambiguated, Sterba’s S_2* is not the claim that if S₁ is true, then P_3* is false. P_3*, after all, is just a claim about the possible relationship between the amount of evil in our world and the amount of one kind of morally valuable freedom, namely significant freedom. S_2* is not the claim that this relationship does not hold; it is instead the claim that this relationship does not have the normative power to justify God’s actions, since other more morally valuable freedoms—political freedoms—must be placed in the balance as well. On this interpretation, that is, S_2* is implying that if S₁ is true, then P_2* is false:

P_2*: It is possible that a world with the amount of significant freedom that we see and the amount of evil that we see is more valuable than worlds with much less evil but much less significant freedom too.

When we take S₁ as a claim about occasions when protecting political freedoms would have been more morally valuable than protecting significant freedoms, we see that worlds with much less evil but much less significant freedom can nonetheless be worlds with a lot more morally valuable freedom than worlds with the amount of significant freedom that we see and the amount of evil that we see. Even though Sterba talks as if he is challenging Plantinga’s claims about how significant freedom constrains God’s powers to actualize various valuable worlds, disambiguating the relevant terms leads us to understand his argument as rather the claim that those constraints, possible as they may be, cannot do the normative work Plantinga needs them to do in the end.10

We are now finally in a position to see where Sterba’s argument goes wrong: S_2* is false. The mistake it makes, in fact, is a common misunderstanding of the mechanics behind the free will defense.

First, notice how Sterba has simply re-labeled something that P_2* was already taking into account: namely, the relation between significant freedom and the kind of evil that consists in not being free, on some occasion, from some kind of harm (e.g., assault). Freedom from harm at any time, after all, is a good, and the lack of that kind of freedom at any time, of course, is an evil. So, when Plantinga makes a positive comparative assessment of our world, given worlds with less evils and less significant freedom, he is already including in that comparison worlds with less significant freedom but more freedom from assault. This is not to say that Plantinga’s assessment here is right, but rather to say that Sterba is not identifying anything new in the balance of considerations. What he has done, instead, is simply increase our intuitive sense of what is more valuable than what, as if to ask: how did the free will defender trick us into thinking that my significant freedom to assault you was more morally valuable than the evil that is your lacking the freedom from assault? And the implied answer to this question seems to be: maybe by referring indiscriminately to the lack of the latter freedom as just another non-descript generic “evil”.

Maybe. But once we realize that Sterba has merely re-labeled the terms in a paradigmatic comparison that was part of the dialectic all along, we see that there is nothing new in his suggestion, or in the proper reply to it. Indeed, recall David Lewis’ (1993, pp. 154–55) description of what God’s answer to a prayer from the Gulag would look like if Plantinga’s claims were right:

“No, I will not deliver you. For I resolved not to; and I was right so to resolve, for otherwise your fate would not have been in Stalin’s hands; and then Stalin’s freedom to choose between good and evil would have been less significant.”

The implication here seems to be that Plantinga’s free will defense depends on, or entails, the claim that the justification for every evil that we see is the comparative moral value of the particular exercise of significant freedom, like Stalin’s, that brought it about. If this were not the case, after all, then God could have easily actualized a world with a little less significant freedom and a lot less evil by simply staying Stalin’s hand. Lewis’ suggestion, in other words, is that Plantinga’s (a) global assessment of the value of the different worlds that would ensue from interference or non-interference in a certain case depends on a misguided (b) comparative assessment of the moral values of a particular instance of significant freedom that brought about a particular instance of evil. Sterba is making the same move. He is suggesting that Plantinga’s free will defense depends on, or entails, the claim that every lesser freedom case is a case where the justification for allowing someone to not be free from assault was the comparative moral value of someone’s significant freedom to assault. If this were not the case, after all, then God could have easily actualized a world with a little less significant freedom and a lot less evil by simply preventing the attack. Like Lewis, Sterba is suggesting that Plantinga’s (a) global assessment of the value of the different worlds that would ensue from interference or non-interference in lesser freedom cases depends on a misguided (b) comparative assessment of the moral values of a particular instance of significant freedom and a particular instance of evil (lacking a certain kind of freedom).

But Plantinga’s extended free will defense does not depend on, and does not entail, the claim that the justification for every evil that we see is the comparative moral value of the particular exercise of significant freedom that brought it about. To join Lewis and Sterba in thinking otherwise is to lose sight of the interconnected and structural nature of the puzzle God is possibly confronted with at the point of creation. What the free will defender disputes—and rightly so—is the claim that, necessarily, interfering with someone’s significant freedom at time t, and thereby eliminating certain instances of evil, would lead to a better world overall than not interfering. Possibly, this is not so. This is not, however, because of a comparative assessment of the moral values of a particular instance of significant freedom and a particular instance of evil. Rather, this is because such interference would have a wide-ranging impact on which other significant freedoms and evils exist—by an ever-expanding ripple effect on the situations other individuals would then be in—and because it is possible that this impact, given the contingent and independent truth values of the counterfactuals of creaturely freedom, would produce a world that is much worse overall. Sterba (2019, p. 26) is twice mistaken when he tells us that “Plantinga by appealing simply to the freedom of the wrongdoer alone as a justification, or possible justification, has not achieved this [i.e., provided us with a possible justification for God’s non-interference]. Something else would be needed that Plantinga does not provide.” As I have detailed in Section 3, Plantinga makes no such narrow appeal in his free will defense, and he does indeed provide us with the something else that is much needed.

Of course, many wince at attempts to spell out some of these ripple effects with the intension of identifying consequences that justify horrendous evils. Responding to an attempt by Swinburne (1998, p. 245), Laura Ekstrom (2021, p. 50) says:

“I find such comments—that it was “a great good for those who themselves suffered as slaves that their lives were not useless” since “their vulnerability to suffering made possible many free choices”—to be morally repugnant.”

I agree. But what is morally repugnant about them is the moral hubris involved in pretending to be able to even gesture at a collection of consequences that, together, could justify God’s permission of something like slavery, coupled with the moral myopia required to believe that the half-baked suggestions one has produced could truly justify such horrendous evil. By contrast, it is not morally repugnant to simply make the point that it is logically possible that this world, with all its horrors, is the least bad world compatible with a valuable amount of significant freedom, given the pattern of truth values for the counterfactuals of creaturely freedom God possibly had to work with at the point of creation. To be clear, this is not a claim about what we have most reason to expect would have happened if God had interfered to prevent the many exercises of significant freedom that were responsible for the unspeakable evils constitutive of, and ensuing from, slavery. What the free will defender wants us to see is simply that there is no logical guarantee that the possible world that would ensue from some such interference is overall better than the one ensuing from non-interference. In general, God’s choice of interfering or not interfering with anyone’s evil-producing exercise of significant freedom on any occasion, possibly, was a choice between our world (with all its horrors) and a much worse world instead.

If this is right, then S_2* is false:

S_2*. If many of God’s decisions were freedom-problematic decisions, then it is not possible that, on those occasions, God was justified in permitting the relevant evil on account of considerations pertaining to the moral value of significant freedom.

To think that this is true is to mistake the fabulously complex global comparison of which of two inevitably ensuing worlds would have more value for the narrow comparison of which of two particular freedoms have more moral value. Sterba at times seems confident in his powers to make the former comparison (see his discussion of the various possible consequences of God preventing the horrendous murder of Matthew Shephard, on pp. 21–24). But apart from making suggestive comparisons between the moral values of particular instances of significant freedom and the particular instances of evil that were brought about by them, Sterba gives us no reason to believe that it is logically impossible that interfering as he suggests would not, eventually, given the interconnected branching pattern of truth values of the counterfactuals of creaturely freedom, deliver us to a much worse world. Sterba’s attempt to provide support for Mackie’s Expanded Key, therefore, falls short.

Perhaps the most important thing to highlight from the considerations above—and I have indeed been trying to highlight it throughout—is modest nature of the epistemic demands made by Plantinga’s claims. As a reply to the expanded logical problem of evil, the expanded free will defense does not require that we have good reasons to believe that “a world with the amount of significant freedom that we see and the amount of evil that we see is more valuable than worlds with much less evil but much less significant freedom too,” or that “God could not have actualized a world with any less evil without also actualizing a world with much less significant freedom.” It does not even require that we have good reasons to believe the theoretical scaffolding behind Plantinga’s claims: libertarianism about individual freedom, Molinism, transworld depravity, etc. In fact, I myself believe that Plantinga’s three central claims, as well as their scaffolding, are all actually false. I also believe, in fact, that the amount of evil in our world does indeed give us good reason to doubt the existence of a perfect God. But the expanded free will defense is only insisting that these claims are possibly true. And this much I think we must concede. I certainly do not think there is anything new in Sterba that challenges this modest, but of course powerful, defense.

5. Conclusions

The free will defense is one of the hallmark achievements of the 20th century analytic philosophy of religion. I am not a theist of any kind, but I believe it nonetheless succeeds in hitting its narrow target: showing that the existence of the perfect God of classical theism is logically compatible with the existence of evil, including all the evil that we see. The fact that some of these evils are unacceptable distributions of political freedoms—or ensuing consequences thereof—raises no extra burdens for the success of this traditional argument. But this is not to say that Sterba’s book-long discussion of the difficulties of adequately justifying God’s permission of unacceptable distributions of freedoms fails to advance our understanding of the force of the problem of evil. I want to conclude by briefly explaining how this is so.

The problem of evil is strongest, to my mind, when presented in non-cumulative evidential form (c.f., Oliveira 2020). Since we are epistemically and morally fallible, we can expect to not see the moral justification for any particular instance of evil. This is why particular instances of evil, if skeptical theists are correct, do not give us any evidence whatsoever against the existence of God, and why the collective evidential force of the amount of evil that we see cannot be a function of the accumulated evidential power of each instance. We are not, however, hopelessly fallible in our beliefs and morals—at least we cannot consistently think so and still count on the strength of our arguments and moral sense, and cannot consistently think so and still justifiably believe in the goodness of God. Consequently, we can indeed expect to see the moral justification for a very large and varied collection of evils. Although our ability to correctly identify unjustified evil is fallible, our “chance of getting something right increases… with the repeated opportunities for that unlikely success” (c.f., Oliveira 2020, p. 327). In this way, the collective evidential force of the amount of evil that we see is non-cumulative: it is not a function of the accumulated evidential power of each instance, but rather a function of what they become as a collection instead. Importantly, that force increases whenever we increase our awareness and understanding of the difficulties of adequately justifying the ever more varied evils in this world. To this project, I think Sterba (2019) has indeed provided a great contribution.