This is part nine of eleven in a series responding to the essays in Ben Arbour’s edited volume, Philosophical Essays against Open Theism (Routledge, 2019).
In this post I tackle chapter 9 by Greg Welty, “Open Theism, Risk-Taking, and the Problem of Evil” (pp. 140–158). Welty’s argument, roughly, is that open theism enjoys virtually no advantages over theistic determinism when it comes to the problem of evil. This argument, if cogent, would supply a significant undercutting defeater for open theism. That is, it would neutralize one of the main selling points for open theism (namely, the problem of evil) over against providential models like theistic determinism and Molinism in which God specifically intends (for the sake of some greater good) every single instance of evil that ever occurs.
I begin by examining Welty’s understanding of open theism (§1). I then consider his discussions of open theism and gratuitous evil (§2), meticulous providence (§3), moral motivation (§4), and divine risk-taking (§5). I argue throughout that Welty’s efforts to blunt the prima facie advantages of open theism over theistic determinism with respect to the problem of evil fail. With respect to evil, open theism remains significantly more plausible than theistic determinism.
1. Welty on open theism and risk vs. no-risk models of providence
At the very beginning of his essay, Welty characterizes open theism as the view that “God lacks foreknowledge of undetermined future events, such as the knowledge of how humans will or would use their libertarian freedom” (p. 140). One nice thing about this characterization is that it is not limited to creaturely libertarian choices but applies to undermined events generally. One downside is that it puts a negative (“God lacks”) spin on the open theistic content of God’s knowledge rather than offering a neutral description (e.g., “God knows the future as open-ended because it is open-ended”). As such, it is question-begging if understood as assuming that there are truths, such as ones about how undermined future events will or would turn out, the knowledge of which an open theistic God lacks.
In the middle of his essay, however, Welty characterizes the knowledge of God at the moment creation on open theism as one of “massive divine ignorance” (p. 148). This is a total straw man, for many open theists like myself would say that God is unqualifiedly omniscient at every moment. It’s just that the content of God’s knowledge changes as contingencies are resolved.
Welty rightly characterizes open theism as “risky” model of divine providence. It is risky in the sense that, by delegating aspects of how creation plays out to His creatures, God opens up the possibility that His creatures might specify those aspects of creation in ways that are objectively less than ideal—for example, by misusing their God-given freedom. In contrast, theistic determinism and Molinism are risk-free models of providence because, on those models, God doesn’t delegate to creatures at all but sovereignly ordains “whatsoever comes to pass,” as the Westminster Confession (3.1) famously put it. God, in short, is an omni-specifying micro-manager and is guaranteed always to get exactly what He ordained. As for non-deterministic divine timelessness and simple foreknowledge, they barely qualify as models of providence at all for in them, upon choosing to create, God simply finds Himself with a completely specified creation history. Welty summarizes:
[T]heological determinism and Molinism require risk-free providence, open theism and ‘incremental’ simple foreknowledge require risky providence, and divine timelessness and ‘complete’ simple foreknowledge are (by themselves) not systems of providence at all. (p. 140)
Welty seems inclined to discount Molinism because it arguably faces “insuperable, independent objections” (p. 140). In light of my own reflections on Molinism, I’m happy to concede the point. Setting Molinism aside allows Welty to restrict his focus to open theism, on the one hand, and theistic determinism, on the other.
2. Open theism and gratuitous evil (pp. 141–146)
In the context of the problem of evil, an “evil” may be understood as anything that, when considered without regard for the “big picture,” would seem on its face to be a net negative for reality. So understood, evils may be either gratuitous or non-gratuitous. A non-gratuitous evil is one that, when considered with regard for the “big picture,” is not a net negative because it (or something equally bad) is necessary for realizing a greater good or preventing a worse evil. In contrast, a gratuitous evil is one that, when considered with regard for the “big picture”, remains a net negative for reality because it is not necessary for realizing a greater good or preventing a worse evil. In other words, gratuitous evils are evils that, all things considered, the world would have been better off without.
As Welty points out (p. 141), on a risk-free model of providence like theistic determinism, there can be no gratuitous evils, for God, being perfectly good, simply would not ordain any evil at all unless it (or something equally bad) were necessary either to realize a greater good or to prevent a worse evil. In contract, on a risky model of providence like open theism, there can be gratuitous evils because what’s necessary for the greater good is not that each and every evil (or something equally bad) be actual, but merely that such evils be possible. For example, an open theist might say that it was not necessary for any greater good that Adam and Eve actually sin in the Garden of Eden. All that was necessary is that it be possible for them to sin—because God’s creating humans with moral freedom is necessary for great goods that would not be obtainable by wholly determined creatures.
Welty goes too far, however, when he claims that, for open theism, “[n]o moral or natural evil must occur … for any of these great goods to be realized” (p. 142). He’s saying that, for open theism, all evils are gratuitous. This is false. Some open theists may wish to endorse such a claim, but it’s not an entailment of open theism as such. An open theist might say, for example, that while all moral evil is gratuitous—the world would have been overall better had no sin ever occurred—not all natural evil is gratuitous. Perhaps some instances of natural evil are necessary to create a “vale of soul-making” so that morally free creatures can develop virtues like courage and generosity that presuppose the existence of real dangers and real scarcity.
In any case, that open theism allows for gratuitous evil seems to give it a significant advantage over theistic determinism when it comes to the evidential problem of evil. As formulated by William Rowe, this problem takes the following form:
- Theological premise: If God existed, then there would be no gratuitous evils.
- Evidential premise: There are gratuitous evils.
- Therefore, it is not the case that God exists.
In terms of what’s already been noted, theistic determinism is committed to (1) and the falsity of (3) and so has to reject (2). But (2) certainly looks very plausible in that many evils not only do not seem to be necessary for realizing any greater goods or for avoiding worse evils, but also seem not to be necessary in those ways. The need to reject (2) thus commits the theistic determinist to a strong form of skeptical theism. Unlike moderate skeptical theism, which claims merely that we can’t reliably reason from <Evil E does not seem to be non-gratuitous> to <Evil E is gratuitous>, strong skeptical theism claims that we can’t even defeasibly reason from <Evil E seems to be gratuitous> to <Evil E is gratuitous>. In other words, even if we’ve carefully considered evil E and it really looks gratuitous to us, we should dismiss that as evidentially irrelevant. The main problem with strong skeptical theism is that it promotes moral skepticism by denying our ability to reason defeasibly from seems overall bad to is overall bad. It thus undermines our ability to ascertain whether states of affairs are compatible or incompatible with overall goodness. As such, it also undermines our ability meaningfully to say that God is good. This is a high cost to pay to defend theistic determinism. It’s also a cost that Welty completely overlooks.
Open theism, in contrast, denies (1) and affirms (2). By affirming (2), open theism rejects strong skeptical theism and so does not undermine our ability to reason defeasibly from <Evil E seems to be gratuitous> to <Evil E is gratuitous>. So, open theism does not promote moral skepticism. The main advantage open theists have with respect to the evidential problem of evil is not, as Welty says, that they can contest both premises of the argument (p. 143), but rather that they can concede the highly plausible (2) while denying the less plausible (1), whereas theistic determinists have to deny the much more plausible (2) in order to protect the less plausible (1). Theistic determinists have a much bigger bullet to bite.
[As an aside, it might be helpful to reflect on just how implausible denying (2) is, especially on theistic determinism. After all, why couldn’t an all-determining God obtain whatever “greater goods” He’s aiming at without far less evil than there actually is? Even if those goods require that there be some sins, why so many? Even if those goods requires that there be some suffering, why so much? On the face of it, it emphatically does not seem as if every evil is specifically necessary for some greater good. Moreover, for serious moral evils at least, it emphatically seems that such evils are not specifically necessary for any greater good but are in fact tragic. They are net minuses, things the world would have been overall better without.]
Welty argues, however, that open theists can’t coherently accept (2) and deny (1):
Although open theists reject the view that any particular evil is necessary in God’s providential scheme …, it seems they must accept a parallel view: with respect to any particular actual evil, God’s following his general policy of nonintervention in this case was necessary to maximize opportunities for great goods that could only be obtained by God’s following such a non-interventionist policy. … (Why else wouldn’t he intervene when he is fully able to do so, except for a judgment of this sort on God’s part?) It follows that for any evil in God’s universe, God had to permit it upon pain of violating a policy grounded in his moral goodness. In this sense, no evil is gratuitous. (p. 143)
To make matters more concrete, Welty considers an actual example of a five-year-old girl Sue who was “beaten, raped, and strangled by her mother’s boyfriend” (p. 144). Why didn’t God intervene to prevent this? Welty sees only four possibilities (I’ve adapted them slightly):
- God didn’t intervene because Sue’s suffering was specifically necessary for a greater good.
- God didn’t intervene because He lacked the power, goodness, or knowledge to do so.
- God didn’t intervene because intervening would have undermined a general policy that is necessary for a greater good.
- God didn’t intervene even though He could have done so without undermining any general policy that is necessary for a greater good. (p. 144)
Welty attributes option (a) to theistic determinism, option (b) to “sub-theistic” views like process theism, and option (c) to open theism. He anticipates that open theists will reject (d) because “on that option permitting the gratuitous evil remains wholly unjustified” (p. 144). But, if the open theist goes with option (c), then “the demands of morality required [God] to permit the evil” (p. 144). But then, “how is this so different” from the theistic determinist’s position? On both views “no evil is gratuitous” because every evil is “necessary to obtain outweighing goods,” whether specific goods or general goods (pp. 144–145).
Welty’s reasoning is faulty, though. The problem is that (d) is ambiguous. It can be read as saying either (d–) or (d+):
d–. God didn’t intervene despite having no good reason whatsoever for not intervening.
d+. God didn’t intervene because His general policies require that He permit evils of that type.
When read as implying (d–), Welty is correct that open theists would reject (d). God, being perfectly good, cannot allow evil without good reason. But what Welty misses is that (d) is also compatible with (d+), and the latter is something that open theists can accept. What (d+) says is that God’s reason for permitting Sue’s suffering need not be token-specific. Open theists, recall, are committed to the idea that, for at least some token evils (particularly moral ones), it is only their possibility that is necessary for a greater good, not their actuality. In Sue’s case, therefore, we can suppose that her actual suffering was neither token-necessary for some particular greater good (a) nor token-necessary for upholding some general policy aimed at a greater good (c). Rather, we can suppose that the possibility of suffering like Sue’s was type-necessary for a greater good. After all, if God has good reasons for giving humans significant moral freedom, including the freedom to physically harm one another, then He can’t dictate precisely how, when, where, and how often they exercise that freedom. God may decide to prevent some instances of Sue-like suffering, but His general freedom-giving policy requires that He allow some such evils to occur subject only to whatever frequency and distribution constraints His perfect goodness requires. Those instances that God decides not to prevent remain, by definition, gratuitous because they are not token-necessary for a greater good, but only type-necessary with respect to their possibility, not their actuality.
So, contrary to Welty, open theists can consistently accept (2) and deny (1). Open theism and theistic determinism are thus not in the same moral boat when it comes to gratuitous evil.
3. Open theism and meticulous providence (pp. 146–149)
Welty next argues that the difference between theistic determinism’s meticulous providence and open theism’s general providence is much smaller than might be supposed. Open theists, he contends, are committed to “meticulish” providence (p. 147). Let’s look at his argument.
Welty first tries to correct a “misleading” impression about the meticulous providence / general providence contrast. Meticulous providence says that, at the moment of creation, God unilaterally ordains whatsoever comes to pass. General providence says that, at the moment of creation, God unilaterally ordains only the initial conditions and a “world-type” (p. 146), leaving many of the details of that type as yet-to-be-filled-in by free creaturely agents and God’s responses to those agents. So far so good, but Welty thinks we need to move the comparison point away from the moment of creation and that, when we do so, the contrast between meticulous and general providence starts to blur:
[S]ince on open theism God is temporal, he is continually making decisions as history unfolds and as his awareness of present human tendencies changes. And that implies that … [w]hether or not good decisions actually occur, or bad decisions actually occur, depends quite a bit on God’s ongoing decisions about what to actually permit. … Yes, “the particular way things will go” cannot be known by God at the moment of creation. But it will be known by God (or believed with a high likelihood of truth) just prior to the choices themselves, at precisely a time when God is in a position to prevent those choices. (p. 146, underlined emphasis is mine)
What Welty is arguing is that, for open theism, God specifically permits each and every evil that occurs and does so under conditions where, just prior to each evil, He has virtually certain knowledge of what’s going to happen.
Based on this conclusion, Welty contends that open theism arrives at a “meticulish” model of providence:
[On open theism] although God cannot guarantee the actualization of any specific possible world through any creative decision he makes at the outset of world history …, God can virtually guarantee the actualization of a specific possible world—and not just a ‘world type’—by way of his providential decisions as world history unfolds. (p. 148)
On these grounds Welty concludes that open theism is committed to “literal micromanagement” (p. 149) and therefore, just like theistic determinism, open theism cannot consistently accept (2) and deny (1) (p. 149).
In response to Welty I have two main points to make.
First, when Welty describes the open theist God as “continually making decisions,” this is incorrect. As I argue in my Open Theism book (Cambridge, 2024), the best version of open theism will affirm that God does exhaustive contingency planning explanatorily prior to actually creating. On this version of open theism, all of God’s decisions have already been made under the same epistemic conditions that God is in prior to creation. Welty is wrong, therefore, that the particular way future contingencies will go cannot be known by God at the moment of creation but somehow can be known by God (or believed with high likelihood) in such a way as to inform God’s decisions just prior to the resolution of those contingencies.
Second, Welty is simply wrong that probabilities invariably rise (or lower) to near certainties (i.e., either 1 or 0) “just prior to the choices themselves.” Probabilities can certainly change as contingencies resolve. This is particularly true where there is a chain of future contingencies. For example, if an indeterministically fair coin is tossed five times, the probability that it lands heads all five times is 1/32 at the outset but becomes 1/2 after four heads have already been tossed and zero if it lands tails at any point during the first four tosses. In contrast to such contingency chains, individual indeterministic events have a base probability defined by the type of event in question. In our coin toss example, the base probability of head versus tails is 1/2. It doesn’t matter how close we get to the time of the coin toss, that probability is going to stay at 1/2. Likewise, the probability of a individual libertarian free choice cannot deviate significantly from one over the number of parallel live options. So, if there are exactly two live options, and the free agent is in a torn condition between them, then the agent could just as easily choose either way up until the very moment of the choice itself. The only reasonable probability estimate under these conditions is 1/2. In general, as long as we’re still dealing with future contingencies, probabilities do not converge to either zero or one. They remain at the base probability for the relevant type of event.
If we put these two points together, we can see that Welty’s argument that open theism is a “meticulish” model of providence morally on par with theistic determinism’s meticulous model of providence abjectly fails. Unlike meticulous providence, God does not specifically ordain whatsoever comes to pass. Unlike meticulish providence, God does not specifically permit whatsoever comes to pass with near-certain confidence that it is going to come to pass. What open theism affirms, rather, is that God permits a wide range of future contingencies, including various evils, without any advance certainty, or near-certainty, as to how those contingencies are going to play out. He engages in contingency planning to pre-decide how He will respond to each contingency, but even if He does exhaustive contingency planning, this is not “literal micromanagement.” Responses, after all, come too late for that.
4. Open theism and moral motivation (pp. 150–153)
Welty next considers an objection against theistic determinism by noted open theist philosopher William Hasker. Hasker argues that those who reject the possibility of gratuitous evils are committed to an offsetting good principle:
Offsetting good principle (OG): Any harm resulting from a morally wrong action will be offset by a ‘greater good’ that God could not have obtained without [ordaining] the evil in question. (p. 150)
My insertion of “ordaining” here replaces Hasker’s “permitting.” The point is that, for theistic determinism, nothing is merely “permitted” by God. In any case, Hasker argues that OGP is at odds with another principle that is much more fundamental to the Christian worldview:
Divine moral intention principle (DMI): It is an extremely important part of God’s intention for human persons that they should place a high priority on fulfilling moral obligations and should assume major responsibility for the welfare of their fellow human beings” (p. 150)
According to Hasker, OG and DMI are in conflict, for “if we really believed that the … [OG principle] obtained in God’s universe, then that knowledge would undermine our moral decision-making, and that would be at odds with the … [DMI principle]” (p. 150). The thinking seems to go like this: If we thought that every bad thing was specifically necessary for a greater good, in most cases a greater good that—per strong skeptical theism—is utterly beyond our ken, then we would have no reason for trying to change the status quo by helping people to improve their lot (e.g., by campaigning against human slavery or sex trafficking). For all we know, working to change things would be working contrary to the offsetting goods that God is aiming at. Hence, best to accept the status quo and passively wait for God to change things. In such a way, it may be argued, OG undermines DMI.
Welty, for his part, will have none of this.
His first line of response to Hasker (pp. 150–151) is to appeal to the arguments we’ve just considered relating to gratuitous evil (§2 above) and meticulous providence (§3). But, as we’ve seen, those arguments don’t work.
Welty second line of response (pp. 151–153) is to deny that OG and DMI are in conflict:
Hasker’s key mistake is to only draw our attention to the following principle, implicit in … [theistic determinism]: [3] If I were to harm P, then the world would be better off off than if I hadn’t harmed P. But he seems to overlook the complementary principle … : [4] If I were to refrain from harming P, then the world would be better off than if I had harmed P. … Assuming that both [3] and [4] are true, the key observation to be made is that, in my moral deliberations, I don’t know which will come to pass … But then the convictions of a theological determinist … wouldn’t lead her in either direction with respect to harming P. Since these cancel each other out … any previous (independent) reasons for thinking that harming P is wrong or blameworthy come into play, and those motivations are not undermined by my convictions about divine providence. (pp. 152–153, numeration of [3] and [4] changed to avoid conflation with (1) and (2) above).
Welty’s idea here is that, even though OG by itself leaves us agnostic how to behave, we have independent reasons to pursue the welfare of our fellow human beings. These reasons are not undermined by OG. Thus, OG is compatible with DMI.
I don’t think this response works as well as Welty thinks it does. The central issue here is whether or not OG undermines our supposedly independent reasons for pursuing human welfare. If it does, then OG does undermine DMI after all. Do we, then, have any reason to think that OG would undermine whatever other reasons we may have for pursuing human welfare? Yes, we do. Recall from above that theistic determinism requires a strong version of skeptical theism in order to resist the inference from <There seem to be gratuitous evils> to <Probably, there are gratuitous evils>. We can replace “gratuitous evils” here with “all-things-considered evils” or “objective evils” as opposed to merely subjective or perspectival evils, i.e., things that seem evil when viewed from a finite, human perspective but that are not evil when viewed from God’s perspective. Indeed, given OG, all seeming evils are really objective instrumental goods because they are specifically necessary for some “greater good.” When understood in this way we can see that strong skeptical theism tells us to distrust our moral seemings for it implies that whatever in reality may seem objectively bad to us is actually good. But if we can’t trust our moral seemings, then we can’t trust whatever independent reasons we might think we have for pursuing human welfare, for those reasons must ultimately bottom out in moral seemings if they are to motivate us.
In sum, then, I don’t think Welty succeeds in deflecting Hasker’s criticism.
5. Open theism and divine risk-taking (pp. 153–155)
Throughout most of Welty’s essay his main goal is to argue that open theism enjoys no significant advantage over theistic determinism with respect to the problem of evil and moral motivation. In the final section of his paper he attempts to show that theistic determinism does enjoy at least one significant advantage over open theism, particularly with respect to the ethics of divine risk-taking.
Welty first notes that, given a risky model of providence, there is an asymmetry between God’s exposure to risk versus that of His creatures: “God’s risk-taking exposes humans to risks that are much severer [sic] than the risks that God undertakes for himself” (p. 154). Thus, God risks being disappointed, but is otherwise invulnerable to any real harm. Humans, in contrast, face tremendous risks, such as being “subject to involuntary, torturous pain for years on end” (p. 154).
Welty then appeals to certain moral intuitions about risk-taking:
- Equity: All other things equal, “the risky environment needs to be equitable. If A subjects B to risk R, then B can rightly subject A to B.” (p. 154)
- Consent: Inequitable risk exposure can only be justified by consent. (p. 155)
Welty doesn’t make the underlined qualifications explicit, but they are required for the logic of his argument to work. Thus, he starts with the equity principle but then, in the very next paragraph, notes some obvious exceptions: “a military officer may command a soldier to take care of the next hill under heavy gunfire, while the officer stays behind in (relative) safety. Or a doctor may perform surgery on a patient who may clearly die from the procedure, but the doctor himself bears little risk” (p. 155). In the light of such clear counterexamples, the equity principle can only be sustained with a ceteris paribus or “all other things equal” qualifier. In these exceptions to the equity principle there is clearly a functional hierarchy at work. In the first case inequity in risk exposure is a consequence of a supervisor–supervisee relationship: the military officer’s job is to manage the battle whereas the soldier’s duty is to carry out the officer’s orders. Likewise, in the second case, inequity in risk exposure is a consequence of an agent–patient relationship: the doctor’s job is to actively mitigate or eliminate internal physical threats to the well-being of his patient whereas the patient’s role is more passive.
But if the equity principle doesn’t apply in cases where there is a functional hierarchy at work, as there obviously is in the Creator–creature relationship, then Welty needs a way to sideline these counterexamples in order for his argument against open theism to work. That’s where the consent principle comes into play. Welty says that both the military officer and doctor cases “involve voluntary risk-taking on the part of the one subjected to the risk” (p. 155). Of course, this isn’t always true—has Welty never heard of military conscription or of doctors operating on patients who, due to age or incapacity, are incapable of giving consent? In any case, for the consent rationale to carry over to the Creator–creature relation and to serve as the basis for an objection against open theism, Welty has to assume that consent is the only possible justification for inequitable risk exposure. Otherwise, the open theist can simply note that there are other relevant differences between God and creatures that justify differences in risk exposure. For example, God is metaphysically fundamental, ontologically necessary, and nonphysical whereas human creatures are not. So God is essentially immune to certain kinds of risks that humans, as such, are susceptible to. That source of inequity has absolutely nothing to due with consent unless we suppose, incoherently, that it is morally necessary for God to secure each human being’s consent before creating them! Seriously, do human parents need to secure the “consent” of their kids before they pursue pregnancy? If this consent principle were true, then it would be morally wrong in principle for God to create humans or for humans to have children. Should all be anti-natalists? Should God be an anti-creationist? The questions need only be posed to see how absurd they are.
Welty complains that open theists must “flout these moral intuitions” (p. 155), but that’s a laughable charge. The equity principle is plainly false without the ceteris paribus qualifier. And the consent principle (with “only” added) is false as well. In short, Welty hasn’t even begun to make a cogent case that open theism’s risky model of providence is too risky. In my Open Theism book (Cambridge, 2024) I deal with this objection in much greater detail and argue that there isn’t any serious threat to open theism on this basis. Open theism does not “flout” any evident intuitions about the ethics of risk-taking. Or, at the very least, Welty hasn’t succeeded in identifying any such intuitions.
Conclusion
Welty’s essay is a vigorous attempt by a theistic determinist to push back against the charge that theistic determinism falls short in comparison with open theism when it comes to the problem of evil. In my judgment, his arguments fall well short of their intended goal. Open theism is in fact significantly better than theistic determinism with respect to the problem of evil, mainly because it denies that God specifically ordains (and ultimately causes) every instance of evil—something that really does flout our moral intuitions. Moreover, open theism denies that God even specifically permits every instance of evil. Rather, for many or even most evils, God merely ordains their possibility. This means that many, if not most, evils are gratuitous in the sense that their actuality is not specifically necessary for any greater good, whether a particular good or a general one. The open theist can thus say, in alignment with our moral intuitions, that many evils are genuinely tragic. The world would have been overall better had those evils not occurred. Finally, open theism does not require any appeal to strong skeptical theism. It does not ask us to discount our moral seemings—which leads to moral skepticism—in order to undercut the defeasible inference from <E seems to be a gratuitous evil> to <E is a gratuitous evil>. Open theism doesn’t need to go to that extreme precisely because open theism, unlike theistic determinism, doesn’t need to eliminate gratuitous evils. Indeed, the possibility of such evils is baked into the model.