Category Archives: Philosophy of Religion

Resolving the Conflict Between Maximal Justice and Mercy

It is commonly thought that there is a tension between mercy and justice, and it is indeed not obvious how to make sense of God being both perfectly just and perfectly merciful. In this blog post, I characterize mercy and justice and discuss how God might be said to maximize both of these, building on recent developments in perfect being theology by Yujin Nagasawa, Mark Murphy, and Daniel Hill.

What are Mercy and Justice?

Mercy and justice are responses by an agent to the actions of an individual who is the recipient of mercy or justice. An agent responds to a recipient depending on whether the action is good or evil (positive or negative). A good action deserves reward, and an evil action deserves punishment. In short, justice is getting what you deserve (good or bad), and mercy is not getting something negative (i.e., punishment) you do deserve.[1] We can extend the characterization to add grace, which is getting something positive you do not deserve, but I will focus my discussion on mercy and justice for simplicity. There are two questions to ask to distinguish justice, mercy, and grace:

  1. Is the response deserved? (Yes: justice, No: mercy or grace)
  2. Is the response positive, negative, or non-negative? (Positive: grace or justice, Negative: justice, Non-negative: mercy)

The quadrants of answers to these questions can be summarized in Table 1 below. It is just to pay an agreed-upon wage for a job, or to give a misbehaving high school student detention. It is gracious to give someone a Christmas gift, especially one whom has never given you a gift (where there is definitely no obligation). It is merciful for a police officer that pulls you over for speeding to only give a verbal warning, or for a judge to drop or reduce a charge or sentence for someone who committed a crime.

ExamplesWages from jobJail, detentionChristmas gifts, free samplesNo sentencing, verbal warning from officer
Table 1: A summary of the attributes of justice, grace, and mercy

Thus, justice is deserved punishment or reward, mercy is withholding deserved punishment, and grace is giving undeserved reward. As it stands, it appears as though mercy and justice are incompatible in the sense that an action cannot be both merciful and just, at least at the same time in the same way with respect to the same people. In the case of justice, the result is deserved, while in the case of mercy, the result is undeserved; these seem mutually exclusive.

The Incoherence of Theism

Justice is good, and mercy is good. It is good to perform acts of justice and acts of mercy. So, just and merciful are likely attributes, or properties, that make agents good, or great. We naturally think that just leaders are better than unjust leaders, and merciful bosses (or friends) are greater than merciless bosses (or friends). Thus, we can call justice and mercy great-making properties (or good-making properties, or perfections). At the same time, God is commonly understood as the greatest possible being (or the greatest conceivable being). This is a commitment of perfect being theology. If God is the greatest possible being (GPB), then it seems as though God would possess all great-making properties (GMPs) to the highest degree possible. In other words, a GPB maximizes all GMPs.

If an action cannot be both merciful and just in an important way due to an inherent incompatibility, then there appears to be a problem with maximizing both of these properties. If there is a problem with maximizing all GMPs, then it looks like a problem for the existence of a GPB. If no GPB exists, and if God has to be a GPB, then God does not exist. This can perhaps be formalized (adapted from a previous Twitter conversation I had) as:

  1. God is (defined as) the greatest metaphysically possible being (GPB)
  2. Justice and mercy are great-making properties (GMPs)
  3. A GPB maximizes all GMPs
  4. It is not possible for both justice and mercy to be maximized in one being
  5. It is not possible for all GMPs to be maximized in one being
  6. If (5), it is not possible for a GPB to exist 
  7. It is not possible for a GPB to exist 
  8. Therefore, God does not exist

Let’s go through the premises to determine where we will focus in this post.

We will grant perfect being theology as characterized by (1) for the sake of the post, though I think it is quite reasonable to reject (1) and avoid the problem entirely, opting instead for a variant of perfect being theology or an entirely different fundamental characterization of God (i.e., metatheology). Alternative metatheologies include the creator of all else (creator theology), a worship-worthy being (worship-worthiness theology), the combination of creator and worship-worthiness theology (as Jonathan Kvanvig defends in his book, Depicting Deity, dedicated to exploring these options), or some mysterious fourth thing. I briefly motivated and do accept (2). I will briefly discuss later a reason to possibly exclude justice or mercy as GMPs due to their being unable to be maximized coherently.

The argument outlined above is a great opportunity to explore these ideas more in-depth, with a particular focus on premises 3-6. The first helpful item to explore is (3) regarding what sense a GPB maximizes GMPs. Are GMPs maximized individually or collectively? What are the interaction effects between the properties on their maximum values? With respect to (4), what does it mean to individually be either maximally just or maximally merciful? Is it possible to maximize them collectively? Finally, (6), if all GMPs cannot be maximized (in whatever sense we determine to be relevant), does that imply there is no GPB? We will discuss each of these in turn.

Introduction to Great-Making Properties

Before moving too far into our investigation, we need to understand exactly what is meant by great-making properties, what types of properties can be GMPs, and how to understand their maximums and their most valuable degrees. We can use some vocabulary developed by Daniel Hill in Divinity and Maximal Greatness to (hopefully) offer some clarity on this concept. Hill distinguishes between properties that have a maximum, a highest degree, and those that do not (such as set-theoretic cardinality due to the power set axiom), as well as those that have an optimum, a most valuable degree (not necessarily the highest possible degree), and those that do not.

The difference between the optimum and the maximum is that the maximum is just what is possible for a being to have more of, so a being has something maximally if it is not possible to have more of that thing (e.g., power). The optimum is where a property becomes less valuable when it is possessed to a greater or lesser degree than the optimum. In analytic terms, a being has property F maximally if and only if it is not possible that there be a being that has more F, and a being, x, has great-making property F optimally if and only if nothing could be greater than x in virtue of having more or less F.[2]

The optimum, as far as I can tell, is identical to what has previously regularly been called the intrinsic maximum of the value of that property. I find the language of optimum vs maximum much more helpful than the clunky ‘intrinsic maximum of value’ (not to mention misleading, since, if Atomism is false (see next section), then the ‘intrinsic’ maximum can be affected by external properties).

There are some properties whose maximum is its optimum, but other properties will have their optimum that is less than their maximum. Hill calls properties where the optimum = maximum maxi-optimality properties, and he calls properties where the optimum is not the maximum duality properties because they have a dual nature of a distinct optimum and maximum. In other words, a being possessing some duality property to its optimal degree is greater than a being that possesses that property to a greater degree (higher than the optimum).

Maximize Collectively or Individually?

A greatest possible being is said to maximize all great-making properties, but what if the GMPs themselves affect the maximum, or at least the maximum greatness, of one of the other GMPs? A common example is God’s omnipotence and omnibenevolence: if God can do anything metaphysically possible, can God sin? This is a standard question. A standard response is: no, God cannot sin, but that is not a problem; it would actually be a kind of weakness in God to be able to sin, not a strength. It is a liability, not a capability. Thus, there is no true capability, as opposed to liability, that God lacks. Is this response adequate for showing that God can, in fact, maximize both power and goodness in the relevant sense? We will explore this question in a general way in this section. 

Overall, we need to assess two theses regarding how GMPs are maximized (Distribution and Atomism as termed by Mark Murphy in God’s Own Ethics[3]):

  • Distribution: for each great-making property that God exhibits, God exhibits that property to the intrinsic maximum of its value
  • Atomism: for each great-making property, what constitutes the intrinsic maximum of the value of that property is independent of that great-making property’s  relation to other divine great-making properties

Our stance on Distribution will determine if we think God has to maximize each GMP individually (independently), where each GMP is possessed to its most valuable degree, or collectively, where the GPB has the highest overall value and maximizes the set of GMPs rather than each GMP individually. For example, a Distributivist may say that a GPB can have more power, but not more power in a more valuable way. See Figure 1 for a visual depiction between a Distributivist and a non-Distributivist set of GMP values and maximums.

Figure 1: Left is a Distributive set of GMPs since all GMPs are possessed to the optimum (aka the most valuable) degree. On the right, one GMP, knowledge, is not at the optimum level and thus rejects Distribution. Note that if Atomism is true, then it seems the maximum values for power and knowledge would need to be reduced below the optimum values and would need to be motivated by the metaphysical impossibility of possessing that GMP to a higher degree. If Atomism is false, then it could be that the interaction between power, knowledge, and goodness results in a decrease of the optimum values of power and knowledge to their current levels.

Yujin Nagasawa in Maximal God rejects distribution and says that a GPB has the maximal consistent set of all GMPs. Thus, the GPM maximizes greatness overall, independent of whether each individual GMP is at its intrinsic maximum of value. It is perfectly acceptable to have less power than as much as one could most valuably have if the amount of power one can most valuably have produces an inconsistency with another maximized GMP, such as omnibenevolence. Murphy challenges Nagasawa’s view and defends Distribution, but I will not assess that argument here. (I think it fails, and Distribution is plausibly rejected for a parallel reason that one can reject Atomism.)

If Distribution were false, there is a fairly easy way to avoid consistency issues by making God have the maximal consistent set of justice and mercy. If maximizing GMPs just means take the maximum of two things combined in some way that requires consistency, then coherence is baked into the definition.

  1. God is (defined as) the greatest metaphysically possible being (GPB)
  2. Justice, mercy, & grace are great-making properties (GMPs)
  3. The GPB has the maximal consistent set of all GMPs
  4. God has the maximal consistent set of justice, mercy, & grace

This sounds like a nice (though cheap) save, but, as Jeff Speaks has pointed out in The Greatest Possible Being, something fishy must be going on here. There is still more to explore about how to maximize the consistent set, whatever that is, of the two properties. It remains unanswered if this GPB meets some absolute standard of greatness (in case the GMPs are in tension to the extent that the “maximum consistent set” is not any greater overall than a human[4] or even a rock). I am not going to assume that Distribution is false in the rest of this, so I consider alternative ways to respond to the challenge of the consistency of the GMPs.

Even assuming Distribution is true, to say that each GMP is at their intrinsic maximum is not to say that there are no interaction effects between the GMPs. Namely, the GMPs might interact in such a way that reduces the intrinsic maximum of value of another GMP. Thus, our stance on Atomism determines if we think that the value maximum of a GMP can be affected by another GMP.

Without Atomism, we have no problem saying God can have more power with the ability to sin, but not power in a more valuable way, which means that God still maximizes all GMPs in the relevant way. An atomist would say that the maximum value of power does not depend on anything to do with the goodness of an action or as a property of that agent, which appears to imply that God would be greater if he could sin. Murphy does not think this (and Atomism generally) is implausible, but that seems to me to be pretty clearly incorrect. We can, in fact, say that one who can sin is not greater than one who cannot sin. Plausibly, one is greater if one cannot sin due to their impeccability (not due to a lack of power or ability) than if one can sin. Thus, power cannot be realized in a more valuable way by adding the ability to sin. 

The key takeaway from this section is that GMPs can interact with each other in such a way that affects what degree of that property is most valuable to possess, which is to say that Atomism is false. I am neutral on Distribution, but I find no issue with rejecting it. As Murphy contends, we need some absolute minimum overall standard to be met without Distribution; however, (as Murphy also defends) we might should include that absolute minimum either way.

This section applies to mercy and justice because (1) without Distribution, mercy and justice do not need to reach their individual maximum values for the GPB to maximize them collectively, and (2) without Atomism, mercy and justice can interact in such a way that their value maxima are changed compared to considering them individually.

How to Maximize Justice or Mercy?

When maximizing a great-making property, we need to understand what it means for that attribute to be exhibited to a greater degree (assuming it is, in fact, a degreed property), and what it could look like to exhibit that attribute at the greatest degree. In the case of justice and mercy, maximization is likely over numerous dimensions, including degree or magnitude, quality or type, recipients, times, and/or possible worlds.

Justice and mercy are relational properties, and they require multiple parties to be involved. Agents are the only objects that can be just or merciful, and agents are also the only objects that can be recipients of justice and mercy. Every act of justice and mercy has a recipient, someone who receives mercy or justice from an agent.

Let us consider a natural starting point for maximizing over all the dimensions listed earlier:

  • Maximal justice1 = just in the highest degree with respect to all people at all times in all possible worlds in the retributive, compensatory, and restorative aspects of justice.

Now, we will go through this piece by piece.

The highest “degree” aspect is probably redundant because one can only be just or unjust. If one gives less than the reward that is earned, it is weird to call that “partially just.” Punishment or reward is either fitting or it is not. We might instead understand that justice is exhibited to the highest degree if it is just in the way described by the rest of the definition. Consequently,  justice in the highest degree just is justice with respect to all people at all times in all possible worlds in the retributive, compensatory, and restorative aspects of justice.

Regarding the different qualities, or aspects, of justice, this would depend on our independent account of justice, but I think it is reasonable to say there are multiple aspects of justice, including retributive and restorative aspects.

The maximizing over all possible worlds makes sense, which is to say that God is necessarily just. We would need to clarify that, since justice is relational, it would be necessary relative to worlds in which God created agents. (See the next section for an objection along these lines).

We get a bigger issue when considering maximizing over all times and objects. The intrinsic maximum of the value of justice is likely not when all times and all objects are included. The reason is because it is likely morally better, or at least more valuable, for God to not act in this way at all times for all people. Namely, it is better if God were at least sometimes merciful, or sometimes gracious, at least with respect to some people. In the same way, it is better if God were not always merciful, but God at least sometimes gave people punishment they deserved rather than always letting people go scot-free.

Consider the videos that go viral on Facebook/YouTube of judges giving people mercy when they deserve a harsher sentence. We consider this praiseworthy but not obligatory, usually, and thus a supererogatory action on the judge. We would say that this judge is better than one who is a stickler and never lets people off the hook(never merciful).

Plausibly, it is worse to be always merciful, or always just, with respect to any given person, than sometimes merciful and sometimes just. So, in terms that we used earlier, there is a point at which justice could not be realized in a more valuable way, and the same with mercy.

What about being always just, but instead of with respect to any given individual, with respect to any persons? In other words, God may be just at all times, but not at all times for all people. This is how I used to reconcile the concept of maximal justice and mercy: God is just with respect to someone (anyone) at any given time, and God is merciful with respect to someone at any given time; alternatively, we could simply say: God, at all times, exhibits mercy and justice. Therefore, God is maximally merciful and just.[5]

However, there is a counterexample. Consider a possible world with only one person. God being always just would require God to be just with respect to this person at all times, but as we already considered, it is better if, for a given individual, there was a mixture of justice and mercy, at different times. Therefore, this cannot be the most valuable amount of justice. Obviously, we need a different analysis of the maximum value of justice.

I think the conclusion is that the intrinsic maximum value of justice maximizes over possible worlds, types of justice, and objects, but it does not maximize over all times. I think the analysis of mercy will be similar. If God never punishes the wicked, but always gives them mercy, that is worse than sometimes giving justice.

Crystallizing the Compatibility

Does justice or mercy have a true maximum? I explored in the previous section what maximizing justice or mercy might look like. One objection to this is that there is no maximum because God could always create more people to receive mercy, and that would make God more merciful to be merciful to more people.[6] A quite plausible response is analogous to what’s called “person-affecting” views in population ethics, which is where something is only important if it applies to an actual person. This means that it would not make God more merciful to create more people and be merciful towards them because if God did not create them, then they would not have done anything that warranted mercy in the first place (or existed at all). In this scenario, God is as merciful as possible per person, and that activity is all that is needed for maximal mercy.

We may say that it is greater to have mercy on 100 persons than have mercy on 10 persons, but that does not mean is greater in virtue of being more merciful by having mercy on 100 instead of 10 persons, particularly if we are comparing different possible worlds. It may be greater in virtue of other aspects of the ability to have mercy, such as the necessity of overcoming limitations in power, space, time, and knowledge to create and have mercy on 100 persons compared to 10 persons. (The same applies, even more clearly, to the parallel worry Murphy has about God being maximally loving). God can have the same level of being merciful whether he creates more people that he actualizes this ability or not, similar to how omnipotence does not mean God creates as many objects as possible because God does not need to actively use all his power to be maximally powerful. Thus, God can be maximally merciful without creating more people to be merciful toward, but by merely being maximally merciful towards the people that God does create.

A similar worry can arise when considering the necessity of God’s mercy. If mercy requires creatures, and God only contingently creates, then there might be a problem for God being necessarily just. For now, while I am engaging with Murphy on this, I adopt a parallel stance as Murphy on necessary love[7]: “given the existence of created persons, the Anselmian being necessarily [is just and merciful towards] them.”[8] One might press that even if God creates, that doesn’t imply that necessarily anyone has sinned and thus be a candidate for receiving mercy. One can appeal to a kind of transworld depravity, that any possible world with an agent will sin at some time and thus be a candidate for mercy. If all of this is unsatisfactory, I am perfectly fine with saying God is just and merciful in a contingent way, namely those worlds in which God creates moral agents, as there is no way for God to be merciful or just in worlds without agents and thus does not make God any less merciful or just.

Let us go back and reconsider properties that have an optimum that is distinct from their maximum, what Hill calls duality properties. In fact, Hill actually gives lenience (aka mercy), which he contrasts with justice, as an example of one! Hill says, “I contend that it is possible to be too lenient, i.e., that a maximally lenient being would not be optimally lenient… It is great, I think, to be lenient, and a more lenient being is greater than a less lenient being – up to a point, the point of being optimally lenient.”[9] I completely agree with him here. 

Hill further comments on the relationship between mercy and justice by saying, “[I]t seems that too much lenience implies not enough justice, and that being optimally lenient is compatible with being maximally (or optimally) just.”[10] Again, I completely agree. If one is always merciful, always forgiving and letting wrongdoers off the hook, then that is not enough justice. There is some optimum level, the exact level need not be identified, where value is maximized, and that optimum is not God acting always just towards every person at all times in all possible worlds, etc. Therefore, a visual depiction of the interaction effects between justice and mercy affecting the optimum values can be depicted by Figure 2.

Shows two graphs described by the figure caption. It shows how interaction between justice and mercy can cause their optimums to decrease below their original when considered individually.
Figure 2: The left depicts the GMPs of justice and mercy considered individually, while the right considers them collectively, including their interaction effects. In this case, there are interaction effects that result in a depression of the optimum degree of both justice and mercy to below the maximum degree, which constitutes a rejection of Atomism. This graphic suggests something like the optimum percentage of actions is 75% just and 25% merciful, which sounds reasonable, but I am not married to that ratio.

These considerations culminate in a new plausible account of maximal justice that is compatible with maximal mercy:

  • Maximal justice2 = justice with respect to all people in all possible worlds (that have agents) in all aspects of justice.

In this way, it appears as though maximal justice and maximal mercy, in the sense of maximizing them as great-making properties (i.e., achieving their value maximum), are perfectly compatible simultaneously. In this analysis, we made use of the interactions between the GMPs of justice and mercy, which suggest that the value maximum of each is affected by the other, and thus Atomism is false.


In this post, I proposed brief accounts of mercy and justice and analyzed what problems that might arise for God’s existence and perfect being theology. I analyzed what it might mean to be maximally just or maximally merciful, concluding that God would be just in all possible worlds (in which there are creatures that do morally relevant actions) in all senses towards all agents, but God does not need to be just or merciful toward all agents at all times. Mercy and justice have their intrinsic maximum of value affected by one another, which means that Atomism is false and each great-making property is not maximized purely individually. The optimum amount of justice and mercy is neither 100% just nor 100% merciful, but somewhere in between. Something that remains to be explored is how the Gospel so beautifully and perfectly combines justice, mercy, and grace in a way that punishes all evil and rewards all good while offering mercy and grace to all.

[1] I am neglecting a lot here and am giving a fairly naïve account, as I have not done much sustained reading on this front. This account of justice may be limited to retributive justice, which I am perfectly fine with. It may also be incompatible with compensation (e.g., for undeserved suffering), which again I am perfectly fine with for several reasons I leave for elsewhere. I also acknowledge this video for the featured image.

[2] Hill, Daniel. Divinity and Maximal Greatness. Routledge, 2004, pp. 10-11.

[3] Murphy, Mark C. God’s Own Ethics: Norms of divine agency and the argument from evil. Oxford University Press, 2017, p. 12. Definitions adapted for internal and external parallelism.

[4] This is Jeff Speaks’s Michael Jordan objection, that PBT may end in the conclusion that Michael Jordan is the greatest possible being, and thus we have clearly gone astray somewhere.

[5] There is a more general question I have here, and I’m not sure where to look for the answer (action theory? metaphysics?). How does one make inferences in either direction regarding attributes and actions? Given one’s set of actions over all time, how exactly do you make conclusions about the attributes of that individual? Or, more useful for doing perfect being theology, how do we make conclusions about one’s actions based on one’s attributes? Specifically, can one have the property of justice if one does not act just at all points in time? Does the property disappear if one does not “act out of justice”? Does the phrase “acting out of justice” even make sense? This is language that I have seen used, and use myself, when describing God’s actions. God is always loving, always just, always merciful, but that does not mean that all his actions are primarily out of mercy. Some actions are more related to his justice than his mercy. Is there a more robust way to make sense of this language? It may be that perfect being theologians have talked a lot about this in the context of maximal properties that I don’t know. There’s a lot of work here that I haven’t read yet. There is especially a lot of work on maximal love by Talbott and others that I have not yet adequately sifted through. I could add to this post based on what I have read thus far, but I’m trying to keep this short. 

[6] Thanks to Johnny Waldrop for raising this objection.

[7] The problem does seem worse than the problem of contingent love, since inter-Trinitarian relations can easily have love, but it is hard to make sense of justice or mercy as being contained in inter-Trinitarian relations.

[8] Murphy, p. 32.

[9] Hill, Daniel. Divinity and Maximal Greatness. Routledge, 2004, p. 11.

[10] Ibid.