Introduction

Perfect being theists, i.e., those who conceive of God as the greatest metaphysically possible being, claim that God exemplifies the greatest set of compossible great-making properties (Hill, 2005; Hoffman & Rosenkrantz, 2002; Leftow, 2012; Morris, 1987). Typically, great-making properties like omnipotence, omniscience, perfect goodness, and necessary existence are ascribed to God. A being with such properties—and others—is said to be maximally great or unsurpassable, or, perhaps the most fitting, perfect. Traditionally, theists have believed that there could only be one perfect being, and that being is God. Let’s say that a being is perfect iff that being is the sole possessor of the largest collection of compossible perfections, i.e., great-making properties.Footnote 1

On perfect being theism, God is that than which none greater is metaphysically possible, the majestic and supreme creator and ruler of all.Footnote 2 I venture to suppose that these theists will find the following thesis attractive:

Perfect Immunity Thesis (PIT): Necessarily, God is perfectly immune to luck. (For any possible being x, x is perfectly immune to luck iff x is not and cannot be subject to any luck.)Footnote 3

After all, it seems that given the divine perfections, God is the best candidate to be fully insulated from the workings of luck—unlike you and me, mere creatures living in a world saturated with luck; intuitively, a perfect being would be completely shielded from the possibility of luck infecting his life, actions, and plans. So, on the one hand, if God is perfect, then PIT seems at least prima facie plausible. On this view, God isn’t subject to luck as finite, frail, and contingent beings are. Perhaps just as God transcends space and time (at least on some views), God transcends subjection to luck. As Nicholas Rescher sees it:

God is exempt from the operation of luck: luck is something that has no place in the affairs of. . . an omnipotent being who controls all outcomes. . . Luck inheres in incapacity: in its absence there is no place for luck. (Rescher, 1995: 58, emphasis original)

Given the divine perfections, luck is banished from God’s life; at first pass, this is a reasonable position and I think many theists will be sympathetic to Rescher’s view. There’s something attractive about the idea that God is completely insulated from luck, perhaps because subjection to luck would diminish divine greatness.

Here, I’ll mention one motivation for PIT that arises from a consideration about divine perfections. I offer this motivation without endorsing it because I seek merely to give some reasons why PIT might be attractive. Possibly, there are other, stronger motivations in favor of PIT. So, why think that subjection to luck might decrease some aspect of God’s greatness? Because, as I’ll explain next, luck decreases control.

It’s widely accepted by philosophers—explicitly or implicitly—that control is a necessary condition on free will and moral responsibility, obligation, praiseworthiness, and blameworthiness. If possessing the ability to exercise control is a prerequisite for being an agent, acting freely, and deserving praise or blame for one’s actions, then control is something that increases the possessor’s greatness. Consequently, if control is a perfection, then control would be included in the set of divine perfections because God exemplifies the maximally greatest set of perfections. Moreover, theists generally think God is an agent who acts freely and is responsible (in some sense) for his actions.

Additionally, control is a degreed (or scaling) perfection: if x and y are both F but x has more F than y, all else being equal, x is greater than y. Control could be degreed in the following way: the greater the agent’s immunity from luck, the greater the agent’s control. Alternatively, if x is subject to luck less than y, then x has greater control than y. This is because luck at the very least diminishes control; since luck comes in degrees, so does control.

But if control is a perfection, and if it’s required that God possess each of his perfections to the highest degree possible,Footnote 4 then God must possess control to the maximum degree: maximum control rules out luck. Moreover, if praiseworthiness comes in degrees, and if the degree to which you’re praiseworthy for x correlates (at least weakly) to how much control you have over x, then a lucky God cannot be the most praiseworthy being. In other words, a lucky God could be surpassed in greatness by a being who is less lucky than God, all else being equal: this is an unacceptable conclusion. Therefore, subjection to luck would be an imperfection for God and thus stain divine greatness. On the plausible assumption that a perfect being lacks any imperfections, we reach the conclusion that God is perfectly immune to luck. Again, these are rough reasons to think PIT is true and thus if you have antecedent commitments to the falsity of PIT, or if you don’t see issues with a lucky God, I’m afraid my arguments won’t sway you.

The purpose of this paper is to cast doubt on PIT. Specifically, I argue that instead of thinking that divine perfections entail PIT—and of course, some will simply reject this entailment—surprisingly, divine perfections give us a reason to think that PIT is false. This is because, as I’ll argue, if God is perfect, then God is subject to a certain kind of luck and thus PIT is necessarily false. Whether or not the falsity of PIT raises problems for perfect being theism I leave it for others to debate. My only aim in this paper is to show that there is a least one source of luck that even God cannot escape.

The paper proceeds as follows. In the ‘An Analysis of Luck’ section, I provide a brief analysis of luck because it’s a crucial concept in my argument. Then, I explain the concept of constitutive luck in the ‘Constitutive Luck’ section and consider some reasons to think that constitutive luck is incoherent; this conclusion, if true, would bring my argument to a swift end. However, in the ‘Constitutive Luck: Uncommon Property Condition’ section, I examine a different approach to make sense of constitutive luck but unfortunately, the new approach faces a serious obstacle, which I consider in the ‘An Objection to the Uncommon Property Condition’ section; there, I attempt to remove the obstacle. But not all is lost because in the ‘Direct and Indirect Constitutive Luck’ section, I distinguish between two kinds of constitutive luck and argue that even if God isn’t subject to one kind, evading the second kind appears unsuccessful. I offer two ways that a perfect being is constitutively lucky so that even if one way fails, the other could succeed. I make concluding remarks in the ‘Conclusion’ section.

An Analysis of Luck

Philosophers of luck have made tremendous progress (even if major disagreements persist) in clarifying the nature of luck, and I must take time to explain what I’ll mean by the term ‘luck.’ All extant accounts of luck deal explicitly or implicitly with non-divine beings, so to examine whether a maximally perfect being could be subject to luck, I’ll take a widely endorsed account of luck and modify it as needed to analyze luck vis-à-vis God. Moreover, since I’ll be concerned only with constitutive luck, my account of divine constitutive luck will tell us under what conditions God would be subject to constitutive luck, regardless of whether or not this account is adequate to handle other species of luck.

To say that an agent is lucky is to attribute a certain feature or quality to that agent; it is to say that the agent exemplifies the property of luck (or technically, being subject to luck). But agents cannot be inherently lucky tout court; agents are lucky only in relation to, say, events or states of affairs. Accordingly, locutions like ‘God is lucky’ should be interpreted as ‘God stands in a direct relation to (or experiences) a lucky state of affairs.’

There are four main approaches to explicating the nature of luck. A good way to grasp the basic difference between these accounts of luck is to think about how they (roughly) explain the paradigmatically lucky event of winning the lottery. According to the probabilistic account, winning the lottery is a matter of luck because the chance of winning was very unlikely and something which you could not reasonably predict. According to the modal account, winning the lottery is a matter of luck because in a nearby possible world, you lose because, say, you’re one number away from the winning ticket. According to the lack of control account, winning the lottery is a matter of luck because it was beyond your control. The fourth account, the so-called hybrid or mixed account, combines the lacking control account with the modal account.

I begin with a condition that most accounts of luck include: the significance condition. Imagine you’re flipping a coin and it’s beyond your control whether the coin lands heads or tails. If you’re flipping the coin just to alleviate boredom, the coin’s landing whichever way is insignificant for you, and thus this event won’t be considered lucky for you. But if an assassin tells you ‘tails you live, heads you die,’ this toss is of great significance and thus the landing of the coin will be considered a matter of luck precisely because now it matters which way it lands. The conclusion we may draw from this vignette is that luck always involves agents:

All agree that for an event to be significant for someone, she must have interests . . . The consensus just is, at bottom, that if an event is lucky for an individual, then it’s somehow good for or bad for her. (Ballantyne, 2012: 320)

Since I’m interested only in divine luck, my version of this condition needs slight modification. I don’t need to claim that something is significant for God only if it’s either beneficial or harmful to God. All that’s needed is that God cares or has a preference or interest in the state of affairs. For example, surely God cares or has a preference or interest in the state of affairs (a) all humans accepting him as Lord obtaining rather than (b) all humans rejecting him as Lord obtaining; to think that God could be indifferent as to whether (a) or (b) obtains seems highly dubious at the very least. My significance condition tries to incorporate this very broad caring or preferring or having an interest in a state of affairs. Hence, if God simply doesn’t care about, say, how many times you blink today, then that state of affairs can’t be considered a matter of luck for God; if E is not significant for God, then E is not lucky for God. Here’s the condition:

Significance Condition: a state of affairs E is significant for God iff God cares about E.Footnote 5

Next, we move to the second condition. Suppose you shoot an arrow at a target 100 yards away. Bullseye! Was that a lucky hit? I wonder if you can do it again. Suppose you miss on the next ten tries: I form the reasonable belief that you were just lucky. The widespread intuition is that if an event occurs as a matter of luck, then it’s not up to anyone’s control whether or not it occurs. In other words,

when something happens that you have no control over, its happening is a matter of pure luck with respect to you. The converse seems even more obvious: when something happens because you intentionally made it so, then its happening is no accident (Riggs, 2019: 125).

However, if you’re able to hit the target, again and again, my suspicion that you’re just lucky will evaporate. This is because I’ll come to realize that you have some control over your success. So there seems to be a connection between luck and control, and more precisely, between luck and lacking control.

The identification of luck with lacking control is common:

[I]n the realm of action, to say that an outcome or event is ‘lucky’ for someone is to say that it was, to some important degree, out of his or her control—it is not something that the agent brings about. (Riggs, 2007: 334)

[S]omething which occurs as a matter of luck with respect to someone P is something which occurs beyond P’s control. (Zimmerman, 1993: 231)Footnote 6

Many philosophers writing on luck argue that lacking control is at least a necessary condition (but not sufficient) for luck.Footnote 7 But what kind of control? Suppose God exercises his powers by performing a basic action and efficaciously materializes a velociraptor ex nihilo into existence in Times Square; this is an exercise of direct control. God exercises direct control if he e.g., makes the hurricane dissipate, gives sight to the blind, or annihilates one of Neptune’s moons. God’s creative acts (like the creation of the universe) and miraculous interventions are plausibly interpreted as exercises of direct control. Define direct control as follows:

Direct Control: S exercises direct control over E if S performs a basic action and thereby efficaciously brings about E’s obtaining.

But there’s a distinct sense of control that God possesses which I’ll refer to as indirect control. An example of indirect control is when God allows certain events to unfold without his interference if things are going according to plan, but he’s capable of intervening and exercising direct control as needed. God could permit me to do A knowing that it’ll lead to B, and B is exactly what God desires to happen. Define indirect control as follows:

Indirect Control: S exercises indirect control over E if S brings about E’s obtaining or prevents E’s obtaining but not by virtue of exercising direct control.

With these two senses of control in mind, here’s the other necessary condition in my analysis:

Lack of Control Condition: A state of affairs E is beyond God’s control iff (i) God lacks direct control over E and (ii) God lacks indirect control over E.

Thus far, the two conditions make up one of the main accounts of luck, known as the lack of control account of luck (LCAL). But philosophers who find LCAL problematic think that another necessary condition is required to make sense of luck. As Lee John Whittington observes, ‘LCAL still misses something about what is generally regarded to be the case about lucky occurrences—that they are in some way chancy, improbable, unforeseeable, and/or unlikely to happen’ (Whittington, 2014: 657). To fix this problem, philosophers often add a third condition on luck, called the modal condition, which appeals to modality and specifically, to chanciness.Footnote 8 Chance is explicated in terms of possible worlds: E is chancy if it fails to obtain in some possible worlds, and E is lucky for you if E obtains in the actual worlds but not in nearby worlds. In other words, E is modally fragile—it could easily have failed to obtain.

Notice that there’s one problem with adding the modal condition to an account of luck: it eliminates the existence of constitutive luck. If God is lucky with respect to E, fulfilling the modal condition requires that E is chancy. But then the modal condition fails to apply to God’s perfections—which presumably, he has of necessity—and so it seems that constitutive divine luck is impossible. However, as I will discuss in detail later, some philosophers argue that an analysis of constitutive luck doesn’t require the modal condition but rather what may be called the uncommon property condition. (Again, I will discuss this condition at length later.) Instead of resolving the debate about whether the uncommon property condition or the modal condition must be included in an account of constitutive luck (a discussion that would take us far afield), I will show that with certain qualifications, the different conditions establish two routes to the conclusion that God is constitutively lucky. For now, my analysis looks like this:

Analysis of Divine Constitutive Luck: God is subject to constitutive luck with respect to E (where E is a property, trait, or characteristic of God) iff God possesses E and (i) E is significant for God, (ii) E is beyond God’s direct or indirect control, and either (iiia) E is uncommon or (iiib) E is chancy.

If God is to be subject to constitutive luck, all three conditions must be satisfied. In the ‘Constitutive Luck: Uncommon Property Condition’ and ‘An Objection to the Uncommon Property Condition’ sections, I argue that God is lucky because he satisfies conditions (i), (ii), and (iiia); in the ‘Direct and Indirect Constitutive Luck’ section, I argue that he satisfies conditions (i), (ii), and (iiib). But before we can answer the question of whether God is lucky to be perfect, I need to explain the idea of constitutive luck and why some think that this kind of luck is absurd. To this, I now turn.

Constitutive Luck

According to Thomas Nagel, constitutive luck is found in one’s constitution, the ‘inclinations, capacities and temperament’ one has (1993: 60). If you reflect on the kind of person that you are, you’ll see that many things about you are both significant for you and beyond your control. Constitutive luck, good or bad, could be found in what may be called physical traits (e.g., being beautiful, athletic, nearsighted), mental traits (e.g., being intelligent, witty, indecisive), and character traits (e.g., being compassionate, generous, pessimistic). Additionally, many factors beyond your control at least partially contributed to your moral, cognitive, psychological, spiritual, etc. formation. You didn’t control whether you were born into a wealthy or poor family, during an economic expansion or recession, or during a time of peace or war. At first pass, the existence of this kind of luck might strike you as plausible and even intuitive.

However, some find the idea of constitutive luck puzzling, to say the least. Suppose one of your essential characteristics is being a member of the species Homo sapiens: in every world in which you exist, you’re a human. In other words, this characteristic is not chancy, and thus there are no possible worlds in which you fail to exemplify it. But if that’s the case, then it seems that you couldn’t have existed and been something different than a human. It’s not as though you first came into existence without any characteristics and then by luck became a human being instead of, say, a gnat; no, you came into existence already as a human (since humanness is essential to you), and thus it seems that luck couldn’t have been involved whatsoever. Based on these kinds of considerations, some argue that constitutive luck is incoherent.

For example, Elizabeth Hurley argues that the notion of constitutive luck commits us to believe that each person has ‘an equal chance from some relevant perspective, of having any particular constitution, hence of being any particular person’ (Hurley, 1993: 197) and thus constitutive luck ‘requires us to make sense of the nonsensical idea of a constitutionless self’ (Ibid.: 198).

Previously, I mentioned that most accounts of luck include the significance condition, which states that the target state of affairs or event must be significant (or of some value) for the agent. But if Hurley is right, then the idea of constitutive luck comes into conflict with the significance condition or, more specifically, the significance condition casts doubt on the possibility of constitutive luck. If some agent S is lucky with respect to E, then it must be the case that E is significant for S. S is lucky to walk away from a high-speed crash unscathed because it affects her, and likewise unlucky if she becomes paralyzed due to the crash; in both scenarios, luck (either good or bad) affects the same person. Hurley argues that luck is ‘identity-dependent,’ such that

there must be someone whose identity is constant between various possibilities that would count as good or bad luck, in order for them to count as good or bad luck for that someone (as opposed to: just impersonally good or bad, or good or bad for mankind as a whole). (Hurley, 2002: 87)

The identity-dependent conception of luck could be illustrated as follows: two buttons are before you, one will give you a pleasurable sensation and the other a painful shock. If you press the pleasure button, that’s good luck for you, and bad luck for you if you press the shock button. In these scenarios, it’s coherent to say that you were either lucky or unlucky, and that’s in part because both scenarios involved the same person, you. Suppose that when you pressed the shock button, the person experiencing pain wasn’t you but your duplicate: then it wouldn’t make sense to say that you were unlucky. Hurley concludes that your identity must be constant across alternative possibilities if it’s really you who enjoys good luck in one alternative and bad luck in the other.

But when we think about constitutive luck, the identity-dependent condition is violated. If being F is essential to you, then you couldn’t have existed without this trait. The person who is in all regards like you but lacks F isn’t you, and thus it’s incoherent to say that you’re lucky to be F. Daniel Statman agrees:

luck necessarily presupposes the existence of some subject who is affected by it. Because luck in the very constitution of an agent cannot be luck for anyone, the idea of one being lucky in the kind of person one is sounds incoherent. (Statman, 1993: 12).

Similar criticisms are raised by Rescher. For example, he maintains that

[A person’s] dispositions and talents are part of what makes her the individual she is; it is not something that chance happens to bring along and superadd to a preexisting identity. . . there is no antecedent, identity-bereft individual who draws the lot at issue with a particular endowment. One has to be there to be lucky. (Rescher, 1995: 30-1)

If you’re born with the disposition to be generous, this character trait wasn’t chosen by you because, well, you weren’t there to choose it! So, it seems that choosing your character—at least your essential traits—is impossible. I think that all these considerations favor the view that constitutive luck, at least in this form, is incoherent. Although aimed to show the incoherence of creaturely constitutive luck, I think the objections raised above mutatis mutandis apply to divine constitutive luck as well. On this understanding of constitutive luck, it’s difficult to see how God could be lucky to be perfect given that if a being lacked any of the divine perfections, that being would not be God. Divine perfections are not a matter of contingency and chance, and thus talk of luck seems confused. Here’s Rescher again:

It makes no sense to envision a prior featureless precursor who then has the good (or bad) luck to be fitted out with one particular group of character traits rather than another. With persons, as with objects of any sort, there is no appropriate place for ‘bare particulars’ devoid of any and all descriptive properties. Identities are not allocated by a lottery of some sort to otherwise nondescript individuals. (Rescher, 1995: 157).

The lesson we learn from Hurley, Statman, and Rescher is two-fold: first, luck is identity-dependent, and second, more importantly, identity precedes luck because there cannot be ‘bare individuals,’ lacking a constitution. A formal argument against the coherence of constitutive luck—call it Against Constitutive Luck—would look something like this:

  1. (1)

    If S is lucky with respect to her constitution, then S exists prior to possessing a constitution.

  2. (2)

    But S cannot exist prior to possessing a constitution.

  3. (3)

    So, S cannot be lucky with respect to her constitution

  4. (4)

    So, constitutive luck is incoherent

(3) and (4) are conclusions, and (2) is extremely plausible for its denial would mean that the existence of ‘constitutionless selves’ would be possible; but as I’ve argued, it’s difficult to make sense of the idea of choosing one’s constitution (or character). This leaves (1) as the premise that advocates of constitutive luck need to reject. If Against Constitutive Luck is sound, then my argument cannot possibly succeed. In the next section, I consider some reasons in favor of rejecting (1) and affirming an updated analysis of constitutive luck.

Constitutive Luck: Uncommon Property Condition

Now, notice that Against Constitutive Luck assumes essentialism about identity; specifically, origins essentialism, the thesis that the origins of material objects (including human beings) are essential to them (Kripke, 1980). If origins essentialism is true, then, for example, you couldn’t have had different parents or even been born a year later to the same parents. This is because there are properties essential to you—particularly, your origin—that you couldn’t fail to possess.

However, whether this Kripke-style essentialism is true of humans or not is beside the point: the divine constitution is essential to God; it is what makes him God. That is, given that the divine perfections are essential to God, there’s no possibility that God could exist without them; no place for contingency within divine perfections means no place for constitutive luck.Footnote 9 In other words, if Against Constitutive Luck is sound, then God isn’t lucky to be perfect; this is a problem for my argument.

On the supposition that God is essentially who he is, then his perfections remain stable across all possible worlds. And if God couldn’t exist without his perfections, this means that God isn’t lucky to be the way he is (Coffman, 2015: 42.) By contrast, an accidental property—traditional candidates are properties like creator and redeemer—is one which God could have lacked and still exist. So, on the assumption that God’s nature is immutable and therefore not chancy, it seems that divine constitutive luck is impossible; it makes no sense to claim that God is lucky to be perfect since it’s a necessary fact about him that he’s perfect and thus couldn’t fail to be perfect. But as I’ll show next, this conclusion is too hasty.

Andrew Latus (2003) argues that contrary to Hurley, Statman, and Rescher, we can make sense of constitutive luck. On Latus’ view, you’re not subject to constitutive luck because you possess a trait that you might have lacked; instead, you’re constitutively lucky because the trait in question isn’t common in the constitution of persons. According to Latus, we should stop thinking about the possibility or chance of you being constituted differently and instead in terms of the possibility or chance of a person being constituted that way (Latus, 2003: 472). Latus believes that Rescher (and others) focus on the wrong comparison class. Suppose you’re a genius (by this I’ll mean possessing an IQ of over 160), the newest member of Mensa. Well, instead of comparing this uncommon trait to your counterparts in other possible worlds, you just need to broaden the comparison class and see whether high intelligence is uncommon for other persons; if it is, you’re lucky to be a genius. Levy offers a similar understanding of constitutive luck: you’re lucky with respect to those traits which vary significantly in human experience or across one’s society (Levy, 2011: 496), or across the relevant reference group, ‘with the relevant reference group being fixed by the context’ (Levy, 2011: 33–34). Interpreted this way, constitutive luck is not obviously incoherent.

To recap: critics of constitutive luck argue that constitutive luck is impossible because you couldn’t have possessed different essential traits; otherwise, it would no longer be you. In response, Latus and Levy argue that the concept of constitutive luck should be expanded to include also your non-essential traits which vary across the relevant comparison class. For this reason, they would see the Against Constitutive Luck argument as unsound because the first premise is false: there is more than one way to explicate the nature of constitutive luck, and their preferred way doesn’t require you to exist prior to possessing a constitution because on their view, you could be lucky to possess a certain train if that trait is uncommon.

Let’s see if Latus and Levy’s updated conception of constitutive luck can help my argument move forward. On their approach, God wouldn’t be subject to constitutive luck because he happened to be perfect and there was a chance that he could’ve been imperfect. Rather, God would be constitutively lucky in this sense: being perfect is an uncommon property that varies significantly across the relevant comparison class. Here’s the condition:

Uncommon Property Condition: a property is uncommon if it varies significantly across the relevant comparison class.

Conjoining this condition to the significance and lacking control conditions, we get a kind of luck that Levy (2011) calls non-chancy luck because it doesn’t require the property in question to be chancy, i.e., manifesting in one possible world but not another. On this understanding of constitutive luck, the modal condition plays no role and is replaced by the uncommon property condition. The other two conditions remain the same: the property must be (i) significant for the agent and (ii) beyond her direct and indirect control.

Let’s define the property of being perfect as follows:

PERFECT: solely possessing the greatest set of all compossible perfections.

Is God lucky with respect to PERFECT? Well, since we can assume that PERFECT is significant for God, beyond his direct and indirect control, and, although not chancy, PERFECT is uncommon (and even rare), all conditions on luck are satisfied and we get the conclusion that God is subject to constitutive luck.

An Objection to the Uncommon Property Condition

Unfortunately, this revised understanding of constitutive luck faces a difficulty that merits closer inspection. To see the problem, let’s go back to what Latus and Levy say. Latus thinks that the relevant comparison class is that of persons, and once we see that possessing some property is unlikely in that class, we have everything we need to attribute constitutive luck to the possessor of that property. Levy is a bit more specific: the degree to which you’re constitutively lucky will vary depending on the proportion of worlds in which that property occurs in the human population: if the property is uncommon then you’re lucky to have it, and the more common it is, the less lucky you are. But which worlds are relevant to determine whether you’re lucky? According to Levy,

Rather than relativizing worlds to the individual, to understand attributions of nonchancy constitutive luck we should understand the relevant worlds as those in which human beings like us exist: I am (non-chancy) constitutively lucky in those traits and dispositions that vary significantly in human experience. (Levy, 2011: 33)

Both Latus and Levy include the uncommon property condition or something like it in their analysis of luck and rely on the condition to explicate the nature of constitutive luck. But there’s a potential problem with this understanding of constitutive luck. The uncommon property condition seems to assume a frequency interpretation of probability: if some property P does not occur frequently in the comparison class, P is uncommon and thus the probability of someone possessing P is low; accordingly, the possessor of P is lucky with respect to P. But frequentism faces the well-known reference class problem. Here’s a clear statement of the problem:

If we are asked to find the probability holding for an individual future event, we must first incorporate the case in a suitable reference class. An individual thing or event may be incorporated in many reference classes, from which different probabilities will result. (Reichenbach, 1949: 374)

In other words, some property P’s frequency must be relativized to a reference class, but frequentism cannot (non-arbitrarily) pick out the single relevant reference class in which P will be uncommon (Hájek, 2019). Suppose P is classified as a member of set S1. But P could be classified also as a member of set S2, S3, and so on. As a member of S1, P’s probability is x; as a member of S2 and S3, P’s probability is y and z, respectively (where x ≠ y ≠ z). So by changing the reference class of P, we can change P’s frequency and thus P’s probability, and if there are different and equally good claimants for the probability of P and the probabilities differ significantly from each other, the reference class problem presents a formidable—and to some, decisive—challenge to frequentism and by extension, to any conception of constitutive luck which relies on the uncommon property condition.

The crucial question for Latus and Levy is this: why focus on the reference class consisting of persons, or humans, or even beings similar to us? It seems arbitrary to privilege one class over the other. For example, if we restrict the class to human beings, P indeed might be uncommon in comparison to other humans, and thus possessors of P are lucky. But what if we examine the frequency of P in the broader class consisting of persons? Candidates for membership in this class might include non-human persons, such as extraterrestrial life forms or angels or spirits or whatever. In this class, P might not be uncommon, and thus on the whole, possessors of P aren’t lucky. The basic problem is that P belongs to many classes, and a non-arbitrary relevant reason must be offered for why we should focus on one class rather than the other.

In the case of human constitutive luck, we can’t privilege a specific class of beings because whether or not some trait is lucky for a human being will be determined by the class. For example, is Sam lucky to be a genius? If we place Sam in the class consisting of all actual human beings, he is lucky indeed. But if we expand the class to all possible human beings, Sam won’t be lucky. Why? Because there’s an infinite number of possible human beings with an IQ over 160, 161, 162, and so forth. If we expand the class further to include non-human highly intelligent life forms, then again, Sam won’t be lucky to be a genius because there are an infinite number of possible non-human geniuses. At this point, it might not even make sense to ask whether some trait is lucky given that these classes consist of an infinite number of members.

Does the reference class problem present a challenge to understanding divine constitutive luck that uses the uncommon property condition in its analysis? Here, things get a bit tricky. Return to God’s perfections. We want to know whether God is lucky to be perfect, that is, whether PERFECT is uncommon. So how uncommon is PERFECT? To find the answer, we first need to find the relevant reference class. Would this class consist of all possible or merely actual beings? Only divine beings? Or does it include non-divine beings as well? Only persons and agents? It seems that we could specify requirements for membership in the class so that all other beings are part of it and PERFECT is uncommon because it’s a property only had by God, in which case God is lucky. But perhaps we could just as easily alter the requirements so that no other beings are part of the class (e.g., the class of all greatest conceivable beings or metaphysically necessary being) in which case God is not lucky because PERFECT is not uncommon; every member of the class possesses PERFECT because there is only one member: God!Footnote 10

Upon further reflection, I think the reference class problem doesn’t present a serious challenge to the possibility of divine constitutive luck. First, because God is a being (as opposed to e.g., an object or event), we can limit the classes under consideration only to those classes containing beings. This means that we have a good reason to privilege a class consisting of only beings. Still, there are perhaps an infinite number of different kinds of beings: human beings, angelic beings, divine beings, sentient beings, etc. So, which class is the non-arbitrarily relevant one? This brings me to the second consideration. The property we’re analyzing is PERFECT. By its very nature, PERFECT is uncommon; plausibly, there could only be one being who possesses the greatest set of all compossible perfections. Now, it seems to me that in order to determine which being has the greatest set of perfections, that being needs to be compared to other beings who possess perfections (be it one perfection, or ten, or a thousand). Here then we have two non-arbitrary criteria for inclusion in the relevant membership class: all members must be (i) beings, and these beings must be (ii) possessors of at least one perfection. As a result, God is subject to constitutive luck precisely because only he possesses PERFECT in this relevant reference class.

I conclude that although the attempt to analyze constitutive luck in the light of the uncommon property condition fails with respect to humans, an analysis of divine constitutive luck which includes the uncommon property condition seems sensible. So, the result of the discussion thus far is this: if we employ the uncommon property condition rather than the modal condition in our analysis of divine constitutive luck, it’s reasonable to conclude that he’s subject to luck precisely because PERFECT is significant for God, beyond his direct and indirect control, and uncommon. This is my first attempt to show that God is constitutively lucky. However, suppose that my response to the reference class problem fails. That is, let’s suppose that I’ve failed to show that a relevant non-arbitrary comparison class exists and hence the uncommon property condition doesn’t justify ascribing constitutive luck to God. Still, the overall conclusion of this paper—namely, that God is constitutively lucky—could be reached via another route. As I’ll argue in the next section, there are two varieties of constitutive luck and thus even if one doesn’t apply to God, the other one does.

Direct and Indirect Constitutive Luck

In this section, I examine whether God is lucky on an analysis of luck that includes the modal condition instead of the uncommon property condition. I will argue that once the important distinction between direct and indirect constitutive luck is made, then even if God isn’t subject to direct constitutive luck, he is subject to indirect constitutive luck. But first, I’ll clarify the distinction, made by Enoch and Marmor (2007) in their examination of constitutive moral luck, but I’ll use their distinction to discuss constitutive non-moral luck.

We all seem to make judgments about the moral character of others, either praising or condemning them for what and who they are. But of course, there are relevant character traits that are beyond your control because they are non-voluntarily acquired. If these non-voluntarily acquired traits affect your praiseworthiness or blameworthiness, then you’re morally constitutively lucky and specifically, you’re subject to direct constitutive moral luck. In the same way, if at least some parts of your constitution are non-voluntarily acquired (and they are significant for you), you’re subject to direct constitutive non-moral luck which says something about your control over your constitution and not the extent to which your moral responsibility is affected by your constitution. Let’s define direct constitutive luck as follows:

Direct Constitutive Luck: S is subject to direct constitutive luck only if S possesses some property trait or property P, P is non-voluntarily acquired or possessed, and P is significant for S.

Accordingly, if P is good for you or valuable in some way or gives you some kind of advantage in life, then you’re directly constitutively lucky to possess P; if P is bad for you in some way or gives you a disadvantage, then you’re directly constitutively unlucky. For example, if your IQ—assuming that it’s a somewhat reliable indicator of one’s cognitive capacities—is well above the average, then you’re lucky; if it’s well below the average, you’re unlucky.

But what can be said about indirect constitutive luck? Imagine that you’re predisposed to anger. In certain situations, given your overall makeup (which includes both essential and non-essential traits as well as the environment in which you were born and raised), you’re inclined to become angry—and let’s just consider cases in which your anger wouldn’t be justified. Let’s say, somewhat arbitrarily, that out of ten scenarios in which conditions are ripe for you to manifest anger, you’ll become angry in eight of them. Now, even though the probabilities were not in your favor to remain calm and collected in a certain rage-inducing situation, we normally see you as free and responsible for either choice you make. Still, this doesn’t contradict the fact that a predisposition to anger is a trait of yours that’s non-voluntarily acquired, and this trait influences—by shaping the probabilities—your actions. But the crucial point is this: even if it’s incoherent to think that you’re not directly constitutively (un)lucky to have such a predisposition, you could nevertheless be subject to luck in this way: your non-voluntarily acquired predisposition to anger influences which actions you perform. Or, in other words, the disposition narrows which actions are likely for you, and in some scenarios, perhaps even eliminates some actions from consideration. Now we can define this sort of luck more precisely:

Indirect Constitutive Luck: S is subject to indirect constitutive luck only if some non-voluntarily acquired or possessed trait or property P of S is significant for S and partially influences how S acts.

The coherence of indirect constitutive luck is based on the plausible view that our characters play a causal (but not always determining) role in the production of our actions: we act a certain way in part because we’re constituted a certain way. And we’re constituted a certain way partly because we were born with some non-voluntarily acquired or possessed traits and partly because some traits were (and are) shaped by factors external to us, e.g., the kinds of people that exerted influence on our lives, especially during formative years. So, if your parents ever told you things like ‘A man is known by the company he keeps’ or, in my case, ‘C кeм пoвeдeшьcя, oт тoгo и нaбepeшьcя,’ they were right.Footnote 11

Here’s an important difference between the two kinds of constitutive luck. Direct constitutive luck has to do with merely possessing the property in question while indirect constitutive luck has to do with how that property influences your actions. Hence, you can be directly constitutively lucky in virtue of possessing P even if you never ‘act on’ P; if you were born with the ability to fly or become invisible, you’d be lucky even if you never had the opportunity to exercise this ability. If you’re indirectly constitutively lucky with respect to P, then P influences how you act in certain situations.Footnote 12

This isn’t to deny that we can shape our characters, at least indirectly, to some degree. Still, it seems obvious (at least to me) that there’s a strong causal connection between one’s character and one’s actions, which is why we’re often surprised when our loved ones or friends, people we tend to know very well, act ‘out of character.’ As Dana Nelkin points out, ‘since how we act is partly a function of who we are, the existence of constitutive luck entails that what actions we perform depends on luck.’ (Nelkin, 2019).

We can now see that premise (1) of the Against Constitutive Luck argument—If S is lucky with respect to her constitution, then S exists prior to possessing a constitution—is false because it assumes that there’s only one way to be constitutively lucky, and that’s by possessing one’s constitution essentially. Hence, (1) is best seen as an argument against direct constitutive luck but it fails to address the possibility of indirect constitutive luck. For the sake of the argument, let’s just assume that direct constitutive luck is incoherent. What I propose next is that there’s still a second way available to reach my overall conclusion.

The connection between one’s character and the choices ‘flowing out’ of one’s character holds true not only for humans but for God as well. I think Timothy O’Connor is correct on this point:

God’s choices reflect His character—and His character alone. He was not given a nature, nor does He act in an environment that influences the development of individualizing traits. If His character precludes His entertaining various options that are within the scope of his power, this fact cannot be attributed in the final analysis to something else (some combination of nature and nurture). Rather, their impossibility is solely and finally attributable to Him. (O’Connor, 2005: 213)

Given the nature that God has, some options are unavailable for him, e.g., the option to annihilate himself or sin. His nature delimits the possibilities, but not always. In some cases, God’s nature might incline an option without necessitating it. Suppose that God has preferences.Footnote 13 In certain circumstances, given his preferences, God is more likely to do or choose A rather than B. For example, suppose God had a preference that creatures like us exist, creatures to whom he would reveal himself and seek a relationship with. In deciding which creatures to create, God acted on his preference but could have created creatures very dissimilar to us: the preference didn’t necessitate God to act a certain way. Since this preference (we can assume) was non-voluntarily possessed, i.e., the preference wasn’t ‘formed’ within God as a result of his will, God’s acting on this preference is subject to indirect constitutive luck; God’s preference played a role in—or contributed to—God’s acting the way he did. Although there was a possibility that God wouldn’t create creatures like us, his decision was perhaps heavily influenced by his non-voluntarily possessed constitution. His constitution shaped the probabilities of each live option, and it was beyond his control how the probabilities were shaped. Thus, it’s a matter of constitutive luck that God created creatures like us.

More generally, for any non-voluntarily possessed property or trait of God that influences some action of his, indirect constitutive luck will be present. That divine perfections are non-voluntarily possessed by God is the standard claim of perfect being theism. On this model, God essentially and necessarily possesses his great-making properties; divine perfections don’t vary from world to world, and there’s nothing that God does or wills to obtain those perfections. Unless one adopts some kind of theistic activism (or ‘absolute creationism’) on which God has control over his essential properties (Morris & Menzel, 1986), this standard claim enjoys high plausibility. Therefore, if we acknowledge that in at least some cases God acts the way he does in part due to who he is, and if we acknowledge that at least some features of God’s constitution are beyond his control, then we must concede that God is subject to indirect constitutive luck. When we consider PERFECT, it especially seems plausible that God is lucky. The upshot of indirect constitutive luck is this: because PERFECT only influences but doesn’t necessitate divine actions, the actions that follow are chancy and thus satisfy the modal condition.

Now, there are significant factors beyond our control over which we didn’t and couldn’t have control, like being born on a life-permitting planet or being born healthy. Someone might think that this makes the existence of constitutive luck trivial and therefore nothing to worry about. So even if God were constitutively lucky, it would be trivially true and thus raise no problems for theists. But defenders of PIT can’t accept even the trivial existence of divine constitutive luck because admitting that God is lucky entails that there are factors beyond God’s control that make him into who he is and determine the probability of certain actions in various situations, and advocates of PIT wouldn’t want to say this. Even if the existence of constitutive luck is trivial and unproblematic for humans (or non-divine beings), those interested in fully insulating God from luck must continue to deny that God is subject to luck of any kind.

In summary, I’ve argued that even if God is not subject to direct constitutive luck, he is subject to indirect constitutive luck. This is my second route to conclude that God is constitutively lucky.

Conclusion

In this paper, I’ve considered two pathways to the conclusion that God is subject to constitutive luck. I spent some time examining this variety of luck because there were good reasons to think that constitutive luck is incoherent. As I’ve argued, some understandings of constitutive luck really are nonsensical, but once the uncommon property condition was introduced, worries about the incoherence of constitutive luck dissipated. Still, the uncommon property condition faced a serious objection (the reference class problem), but I think I’ve shown that identifying the relevant comparison class was possible, and thus the objection was neutralized. I also offered a second way to show that God is constitutively lucky, which hinged on the key distinction between two kinds of constitutive luck. In the end, the conclusion that God is subject at least to indirect constitutive luck appears promising.

Moreover, given that luck comes in degrees, e.g., there’s more luck involved in winning the lottery than winning tic-tac-toe, the more uncommon some non-voluntarily possessed (or acquired) property is, to that degree one is luckier to possess that property. Consequently, if PERFECT is a property of immense value and there could be only one possessor of this property, then God appears to be the luckiest of all possible beings. By my lights, this conclusion strikes me as problematic, although I’ll leave it for others to make the case one way or the other. Admittedly, there’s a lot of wiggle room here for theists to reject my arguments, assumptions, and intuitions. Nevertheless, tentatively—and somewhat reluctantly—I conclude that God is subject to constitutive luck and invite others to examine whether this conclusion is compatible with perfect being theism.