The problem of evil is a moral problem. Once properly situated, most disagreements regarding the problem of evil can be identified as being distinctively moral disagreements. To disagree about whatever conclusion might be drawn from the problem of evil is to disagree about some moral claim or claims. The purpose of this article is to establish these points by pointing out two issues that arise from a recognition of the fundamental role that evaluative claims play within the problem of evil. I will begin by showing that the problem of evil contains some evaluative claims. Second, I argue that the underlying normative consistency necessary for the success of any argument from evil might be a difficult thing to achieve. Thirdly, I align J. L. Mackie's criticisms for the moral argument for the existence of God with the problem of evil, and show that Mackie's criticisms serve to act against his own presentation of the argument from evil, which occurs just two chapters later in The Miracle of Theism.
A ‘meta-formulation’ of the problem of evil
Let's assume this ‘meta-formulation’ of the problem of evil is an accurate representation of all variations of the problem of evil:Footnote 1
-
1. There exists a maximally good, maximally powerful creator of the universe.
-
2. A maximally good, maximally powerful creator of the universe would not create or permit any pointless or unconscionable evil in its creation.
-
3. Some pointless or unconscionable evil exists.
These three propositions seem to constitute an inconsistent set, such that they cannot all be true together; at least one of them must be denied. (This should not be taken to be an assumption of the ‘logical’ formulation of the problem of evil over the ‘evidential’ version. The meta-formulation is intended to capture both variations; there is still an incompatibility between God and evil expressed within the evidential problem of evil, after all, even if that incompatibility is not seen as being necessarily true. But these issues are not relevant for the purposes of this article.)
Analysis of the meta-formulation
A little analysis of the three propositions that constitute the problem of evil reveals them to be complex propositions, propositions which we can reduce down to being conjunctions or implications that are made up of simpler propositions. So, for example, the claim ‘There exists a maximally good, maximally powerful creator of the universe’ can, at the very least, be broken down into a conjunction of the various constitutive elements: ‘There is an x’ and ‘x is maximally good’ and ‘x is maximally powerful’ and ‘there is a universe’ and ‘x created that universe’, etc. These more complex propositions might take the form of conjunctions, but they might also take the form of entailments, in some way. For example, it might be the case that ‘God is maximally powerful’ is a claim that follows from ‘God created the universe’. That is, these ‘simple’ propositions might be assertions in their own right, or they might be representations of lines of reasoning.
In either case, the components of these more complex propositions are revealed to be either positive, descriptive, ‘fact-type’ claims, or else evaluative, normative, ‘value’, or broadly ‘moral’ claims; which is to say that the propositions within the problem of evil are made up of both evaluative and non-evaluative content.
I will define ‘evaluative content’ as being any claim that makes some sort of value judgement about a fact. A shortcut to identifying the most obvious of these will be to pick out any statement that contains the terms ‘… is good’ or ‘… is evil/bad’. We can, for simplicity, define ‘non-evaluative content’ in relation to the definition of ‘evaluative content’. A ‘non-evaluative claim’ is any claim that does not make a value judgement. (I have in mind here the classical distinction between ‘facts’ and ‘values’, between the ‘is’ and the ‘ought’; these are traditionally thought of as being distinct, such that you cannot deduce a claim in one from a claim in the other. This distinction will become more relevant later, when we consider J. L. Mackie's criticisms of the moral argument for God's existence.)
Typically non-evaluative claims include propositions such as: ‘God is maximally powerful’, ‘God created the universe’, etc. These are simple statements of states of affairs, or ‘facts’; they contain no obvious value judgements.Footnote 2 Recognizably evaluative claims are those such as: ‘Pointless evil exists’, ‘God is maximally good’, etc. These claims contain a value judgement of the form ‘… is good/bad’.
So, for instance, the third proposition – ‘Some pointless or unconscionable evil exists’ – is actually a conjunction of both evaluative and non-evaluative content. There is a non-evaluative claim – such as, for example, that an earthquake occurs – and an evaluative claim attached to that non-evaluative claim – i.e. the value judgement that it is bad that an earthquake occurs. Abstracted beyond any particular example of evil, the third proposition looks like this:
3*. ‘Some physical occurrence z occurs’ and ‘It is bad that z occurs’.
These two conjuncts combine to give the ‘simpler’ (which is to say, of course, more complex) proposition ‘Evil exists’. (Given the rather more complex form of the third proposition that I have placed in my meta-formulation, there will also be additional claims within the grand conjunction of 3 that refer to the ‘pointless’ and ‘unconscionable’ elements. But that is beside the point for the current discussion.) Further, it might be misleading to present these two conjuncts as existing entirely independently, as a brute conjunctive assertion. There might be a line of reasoning that connects them, and this line of reasoning would seem to move from the initial purely empirical (non-evaluative or ‘fact-type’) observation of some state of affairs (e.g. ‘earthquakes occur’), towards the conclusion that ‘evil exists’, which contains the evaluative claim that (e.g.) ‘it is bad that earthquakes occur’. The move from facts to values here requires the deployment of some kind of underlying normative assumption about what kinds of things are to be considered ‘good’ or ‘bad’, and this brings the problem of evil into the scope of concerns such as the ‘fact−value’ distinction, and other such moral-philosophical issues. But all I wish to point out, for now, is that the propositions within the problem of evil are made up of some value-type claims and some fact-type claims. Therefore, the problem of evil's propositions are made up of both evaluative and non-evaluative content.
A common response to my claim is to insist that the problem of evil can be constructed without any appeal to any kind of evaluative claim. That is, we can talk about nothing but physical states of affairs – ‘pain’, for example – and make the stipulation that God has a certain attribute of preventing this state of affairs – ‘God would eliminate unnecessary pain’, for example. We can then construct a version of the problem of evil that sounds intuitive and viable, but that lacks any kind of ‘evaluative content’ as I have defined it.
This response fails, because to strip the problem of evil of any kind of evaluative content is to lose the justification for the claim that God's existence is incompatible with whatever state of affairs is being identified (e.g., in this case, ‘pain’). The immediate question we would ask, of this non-evaluative version of the problem of evil, is: ‘why would God eliminate unnecessary pain?’ That is, why is it the case that God would eliminate this particular state of affairs and not others? One might be tempted to say, ‘because God is good!’, but this still gets us nowhere. We would then need to ask why it is that a good thing would eliminate unnecessary pain? The answer that must come is, ‘because it is bad that unnecessary pain occurs’. This is the only way we can non-arbitrarily establish an incompatibility between the existence of God and the existence of this particular state of affairs, by establishing some kind of connection between the specified state of affairs and the ‘goodness’ of God. If we were to insist upon removing any kind of evaluative content, then we would have no more justification for the claim ‘God would eliminate unnecessary pain’ than we would for the claim ‘God would eliminate unnecessary fluffiness’. Without evaluative content, in some form, we have no reason to take ‘the problem of pain’ any more seriously than we would take ‘the problem of fluffiness’.
So, there are two strands of content running through the problem of evil. On the one hand, we have the non-evaluative content claims: God is powerful, some z occurs in the universe, etc. And on the other, we have the evaluative content claims: z is bad, God is good, etc. These two strands are strung together somehow in order to compel some sort of contradiction between the three, and the expression of this connection comes in proposition 2. To anticipate my conclusion a little, it is the strength and validity of this string, this thread of reasoning that runs through the components of the problem of evil and seeks to use them to establish the conclusion of the argument from evil, that ultimately I want to challenge. Within the component propositions of the argument from evil, entailments are somehow drawn between evaluative and non-evaluative propositions, sometimes moving from facts to values, but sometimes also from values to facts. The former of those might be problematic, and the latter even more so.
In this meta-formulation, the second proposition only emerges as a product of the interaction between the contents of propositions 1 and 3. This is an important point; proposition 2 does not have any additional fact or value claims within it apart from identifying the interaction between the claims of 1 and the claims of 3. Proposition 2 is just a further expression of the meanings of both 1 and 3; it is, as Plantinga would have it, a merely qualifying statement, a ‘quasi-logical rule’ (Plantinga (Reference Plantinga1977), 16).Footnote 3 All the relevant evaluative claims have already been made in their entirety in propositions 1 and 3. Saying ‘If a maximally good, maximally powerful creator of the universe exists, then there would be no pointless or unconscionable evil in the universe’ carries exactly the same meaning as saying ‘There exists a maximally good, maximally powerful creator of the universe’ if one properly understands the claims being made in that complex proposition; that is, if one properly understands what is being meant by ‘maximally good’, and ‘maximally powerful’, ‘creator of the universe’, etc., in this context. Understanding the second proposition to emerge as a ‘qualifying statement’ entails that to combine the claims ‘maximally good’, ‘maximally powerful’, and ‘creator of the universe’ just means that there is no pointless or unconscionable evil in the universe; it is not an additional fact to state that the one entails the other, the entailment is contained within the meaning of the terms. That is what is required to operate within a Plantinga-style ‘qualifying statement’ demand.
This is not to say that any of these claims are true, of course, merely that this is what they are meant to mean. This is really just a way of restating Plantinga's requirement that any additions or qualifying statements be necessarily true. Whether they are actually true or not is open for debate; whether the combination of God's goodness and power in actuality entails that there would be no evil of a certain kind in the world is another discussion. My point is only that we can work with Plantinga's requirement that these ‘qualifying statements’ must be necessarily true, and an easy way of doing this is to say that the non-existence of a certain type of evil is entailed by the combination of claims made about God, in exactly the same way that ‘bachelor’ is entailed by the combination of ‘unmarried’ and ‘man’. Framing this as an ‘if-then’ statement, as proposition 2 does, makes it sound like it is saying more than it is. This entailment does not imply any kind of causal relationship here (though that would be the most natural way of reading it: ‘If God exists, then He would cause there to be no evil of a certain kind’); instead, this entailment only reflects the intended meaning of the terms given in proposition 1. So just as the intended meaning of the terms ‘unmarried’ and ‘man’ combine to entail ‘bachelor’, so too the intended meanings of the terms ‘good’, ‘creator’, ‘powerful’, etc., combine to entail that there would be no evil of a certain kind in the world. To disagree with the entailment is to disagree with the meanings assigned to the terms involved.
So this is how the problem of evil works, generally speaking: Proposition 1 says something about the existence of God, proposition 3 says something about the existence of evil, propositions 1 and 3 are understood to conflict with each other, and this contradiction is captured and expressed in proposition 2.
The argument from evil has a non-evaluative conclusion
The problem of evil, which is made up of both evaluative and non-evaluative content, when used as an argument for God's non-existence, works its way towards a non-evaluative conclusion:
4. ‘Therefore, God does not exist.’Footnote 4
Although this seems like an obvious and intuitive conclusion to draw from the problem of evil, appearing to be a simple negation of the first proposition, I will argue that it is not so simple as it may seem. Recognizing that the problem of evil contains some evaluative claims offers up two significant problems for the argument from evil.
The First Problem
The evaluative claims within 1 and 3 need to actually connect up in the way that 2 expresses. This might not be the case. If different kinds of underlying evaluative claims or normative assumptions are being deployed in 1 and 3, then the interaction between 1 and 3 expressed in 2 will not hold. If, for example, one is being a divine command theorist in 1, yet a consequentialist in 3, then these claims might not connect up properly, since the underlying normative assumption operating in 1 does nothing to justify the evaluative claim made in 3; 1 and 3, although both talking about ‘good’ and ‘evil’, are in effect talking about different kinds of ‘good’ and ‘evil’. Some sort of underlying normative consistency must be achieved both within the propositions of the problem of evil, and between those who are discussing it; further, one must remain consistent in the application of these underlying normative assumptions. This is particularly problematic when we consider that there is likely to be a great deal of disagreement about which underlying normative assumptions ought to be deployed in discussion of the problem of evil. Identifying the First Problem serves to address two issues. It seems to me both that (a) people often fail to realize that they are even deploying normative assumptions in this case, and (b) if they are aware that normative assumptions are being deployed, people often begin by showing philosophical quasi-sympathy with their interlocutor, framing their version of the problem of evil as a strict reductio, working with their opponent's normative assumptions, only to shift (unwittingly) to radically different normative territory half way along the line of argument.
The Second Problem
More attention needs to be paid to the fact that the problem of evil casually and repeatedly skips across the is−ought gap or fact−value distinction over the course of its propositions. Even if it is not fatally problematic to skip across the is−ought gap, there might still be a major problem for the argument from evil in drawing non-evaluative conclusions from evaluative premises. Skipping across the is−ought gap the wrong way – i.e. arguing for non-evaluative claims from evaluative premises – is, at least according to J. L. Mackie, decidedly not on. It is ironic that this claim comes from Mackie, who argues for the truth of this point during his criticisms of the moral argument for God's existence in The Miracle of Theism, merely two chapters before he discusses the problem of evil. If I am correct in my analysis, then this was a serious oversight on his part. He cannot have both his criticisms for the moral argument for God's existence succeed, and retain the success of the argument from evil.
I will discuss The First Problem first and conclude with The Second Problem. If my analysis is correct, then I will have established an inconsistency within Mackie's work. Further, if Mackie's criticisms of the moral argument for the existence of God are correct, then the argument from evil will be left fatally undermined. The problem of evil cannot then be used as a basis from which to argue for the conclusion that ‘God does not exist’.
The First Problem: underlying normative inconsistency
In order for the problem of evil to work as a problem – i.e. give rise to a prima facie contradiction – the evaluative contents of 1 and 3 need to connect up properly via 2 such that (when combined with the non-evaluative content) if one (of either 1 or 3) is true then the other will be false. This might be more difficult than is intuitively thought. For if different types of underlying normative assumptions are operating in 1 and 3, then they will not connect up properly.
Consistency within the problem
The problem of evil will collapse internally if the normative assumptions underlying the evaluative claims of 1 and 3 do not match up. If one evaluative claim, such as ‘God is maximally good’, depends upon a certain normative assumption, such as ‘good is defined as whatever God does/commands’, whereas the other evaluative claim (i.e. ‘evil exists’) depends upon a different and incompatible normative assumption (e.g. ‘good and evil are to be defined primarily in terms of pleasure and suffering’), then the two will not connect up. This is because the normative assumption underlying the evaluative claim of 1 does nothing to provide any justification for the evaluative claim contained within 3, and as such the claims of 1 and 3 do not contradict each other since they are not referring to the same version of ‘good’ or ‘evil’.
This would be a fatal problem for the problem of evil, since to maintain its validity in light of the underlying normative inconsistency would be to commit the fallacy of equivocation, but it is probably quite unlikely ever to happen; at least not within someone who is proposing the problem of evil. Most people are going to be sensible enough to recognize that underlying normative assumptions affect the nature of any evaluative claims that are derived from these assumptions, and so though this problem is a potentially serious problem for those posing the problem of evil, it is easily solved, simply by remaining internally consistent in one's normative assumptions and by applying them consistently.
Having said that, I identify this possibility to highlight that there remains the danger of this mistake occurring as a failure to recognize the proper domain of a discussion concerning the problem of evil. Say an atheologian wishes to present the problem of evil as an internal critique via reductio. She therefore wishes to begin with the theist's normative assumptions, which she does. But then, half way through the discussion, the atheologian asserts: ‘surely a maximally good God would prevent pointless suffering!’ If we assume that the atheologian is engaging with a theist who is particularly bold regarding his normative assumptions as they relate to God – the absolute priority of His goodness and justice, etc. – then this is a ‘surely’ too far. The atheologian has supplanted the normative assumptions originally adopted for the sake of reductio with her own, and has lost the feature of an ‘internal’ critique. The proper domain of discussion is no longer whether (or not) pointless evil shows that God does not exist, but whether the theist's normative assumptions are acceptable. In this discussion, the internal critique via reductio has gone out of the window; the problem of evil has gone out of the window. We are now in a strictly moral-philosophical discussion, and the problem of evil has failed to compel any kind of conclusion regarding God's existence.
Additionally, there certainly remains an argument to say that those responding to the problem of evil (i.e. theodicists) are guilty of a kind of internal normative inconsistency within their solutions to the problem of evil. Given that many of these theodicies (for example, the ‘Soul-Making Theodicy’) are recognizably consequentialist in their normative nature, yet come from the mouths of theists who do not commonly deploy consequentialism in their other ethical judgements, nor think that God is a consequentialist, it would seem that some ‘internal inconsistency’ is at work here. On the one hand, it is reasonable to construct a refutation of the problem of evil on consequentialist grounds; on the other, there is a general resistance from traditional theism against the ‘doing of evil so that good may come’ (Romans 3.8). But this is an argument for elsewhere since it would, at best, establish an internal inconsistency in the theodical response to the problem of evil, rather than in the problem of evil itself.
But internal consistency is only one factor here; there is a more relevant version of this problem for the problem of evil, and it emerges when we consider the dialectical expression of the problem of evil.
Consistency between problems
It is one thing to be consistent within one's normative assumptions, it is another to have that consistency extend to others around you. Consider, for example, a not too far-fetched situation such as this: an atheist is presenting the problem of evil to a theist, and wishes to set the theist ‘the task of clarifying and if possible reconciling the several beliefs which he holds’ (Mackie (Reference Mackie1982), 150). The atheist is a brutally utilitarian consequentialist; she thinks that ‘there is one ultimate moral aim: that outcomes be as good as possible’ (Parfit (Reference Parfit1986), 24), and that outcomes are to be judged on the basis of the maximization of pleasure and the minimization of pain or suffering. On the basis of this ethical outlook, and these underlying normative assumptions, the atheist frames the problem of evil in such a way that the overwhelming presence of seemingly pointless and unjustified suffering in the world is deeply contradictory with the notion of a maximally powerful, maximally good creator God. God ought to maximize pleasure, and minimize suffering, and He has the power to do so, and yet (it seems) clearly does not; this is a valid rendition of the problem of evil, and it is quite sound from a utilitarian point of view.
But the theist, on the contrary, is a rather simple and normatively bold divine command theorist. Not only does he think that what can be considered morally ‘good’ depends entirely upon God's commands – he has opted for the spiky horn of the Euthyphro dilemma: ‘it is good because God commands it’ – but he also believes that whatever God creates He deems good, and therefore everything that is, is good (with the possible exception of freely chosen rejection of/falling away from God). All suffering is deserved; it is all just; whatever is, is right. Is the theist vulnerable to the atheist's version of the problem of evil? Of course not. The suffering that the atheist appeals to as support for her assertion of proposition 3 is in no way contradictory with the type of ‘maximal goodness’ that the theist holds to in proposition 1. Internally, the atheist's version of the problem of evil retains its contradiction, since there is an expectation (under that normative view) that God behave a certain way (i.e. minimize suffering), but it falls on deaf ears to anyone who does not share the atheist's normative assumptions. The type of evil presented as evil by a utilitarian atheist is not contradictory with the notion of goodness possessed by the Euthyphro-bold theist. For the theist, such evil simply does not exist according to his value-system; it could not, for suffering would always assuredly be just desserts – even the term ‘pointless’ has no purchase as a concept in this system, since good is not being defined in terms of outcomes – and so the problem is solved. There simply is not the link between ‘suffering’ and ‘bad’ that the utilitarian atheist would assume to be self-evident. Raimond Gaita puts the point succinctly: ‘what a person counts as harm depends on his understanding of value’ (Gaita (Reference Gaita2004), 196)
One could also run the story in precisely the opposite manner. Imagine an atheist who thinks that certain instances of suffering are so horrendous as to be considered ‘unconscionable’, not even in principle capable of being justified by appeal to greater goods. In contrast, we have a brutally consequentialist theodicist who thinks that anything can be justified if subsequently compensated by greater goods. Assuming that one can reasonably believe, as no doubt many people do, that all the sufferings of humanity are subsequently compensated, is the theodicist troubled by the atheist's challenge? Of course not. The evaluative claims of the atheist (re: the existence of unconscionable evils) are not shared by the theodicist, and so her challenge has no bite. But equally, the consequentialist theodicist's response holds no weight for the morally minded atheist. Again, though the various versions of the problem of evil might retain an internal contradiction, the contradiction is not transferred across the atheist/theodicist divide.
In each of these cases, the evaluative contents of 1 and 3 connect up within the various versions of the problem of evil, but they do not connect up between the various versions. So this is the first point to extract from all this: In order for the argument from evil to be persuasive, it seems that either interlocutor must first convince the other to share her evaluative assumptions. Any resolution on the problem of evil would amount to an evaluative resolution of this kind.
Perhaps this seems like a trivial conclusion of the ‘First Problem’ for the problem of evil,Footnote 5 but I do not intend merely to point out that ‘in order for an argument to be persuasive, an interlocutor must persuade the other to accept his assumptions’. I want to highlight that people do not even seem to be aware that they are deploying assumptions in this context. These normative assumptions are often framed not as logical assumptions – for the sake of reductio, say – but are simply propositions that are assumed. My point is that, in the context of the problem of evil, we cannot assume the truth or applicability of such normative claims. This is partly because religious and non-religious ethical world-views are likely to be quite different, but also because the problem of evil has a peculiarly reflexive normative character. The existence of evil can be seen to tell us as much about God's goodness as it can tell against His existence, and the existence of God is invariably seen as a fundamental and foundational variable in formulating any kind of value judgement.
Examples of evaluative claims in the problem of evil
It is not difficult to find some examples of these evaluative claims that are fundamental to any version of the problem of evil, or a response to it. Consider this from John Hick:
Virtues which have been formed within the agent as a hard-won deposit of her own right decisions … are intrinsically more valuable than virtues created within her ready made. (Hick (Reference Hick and Davis2001), 43, emphasis added)
It is noteworthy that of all the claims and complex arguments made within the proposition and rebuttal of the problem of evil, these evaluative claims are often asserted rather than defended. As Hick says, these evaluative claims express ‘a basic value judgement that cannot be established by argument but which one can only present, in the hope that it will be as morally plausible, and indeed compelling, to others as to oneself’ (ibid.). This shows a rather pessimistic take on the prospects of moral philosophy in this context.
Likewise, consider this from Alvin Plantinga:
A world containing creatures who are significantly free (and freely perform more good than evil actions) is more valuable, all else being equal, than a world containing no free creatures at all. (Plantinga (Reference Plantinga1977), 30, emphasis added)
Again, this claim is not defended, but only asserted, and yet it is absolutely fundamental to the success of Plantinga's Free-Will Defence.
From Marilyn Adams:
Christians never believed God was a pleasure-maximiser anyway. (Adams (Reference Adams and Peterson2007), 366)
Here Adams rejects the implicit assumption of hedonism present in almost any version of the problem of evil. The ‘evils’ identified invariably align with pain and suffering, but pains and pleasures are not the whole of the story of good and evil from a Christian point of view. In a similar yet slightly more extreme vein, we have this from Simone Weil:
He [God] did not reserve the cross for Christ… (Weil (Reference Weil1987), 101)
For Weil, not only do Christians not believe God to be a pleasure-maximizer, some might even consider it a virtue, obscurely understood, to suffer terribly, if that terrible suffering allowed them to be more Christ-like and thus brought them closer to God.Footnote 6
Each of these statements clearly represents a kind of evaluative claim, derived from an underlying normative assumption, and one that is unlikely to be shared by all participants in the debate. To say with Simone Weil that ‘God did not reserve the cross for Christ’, and imply that suffering, even horrendous suffering, can actually be seen as an overwhelmingly good thing, since it allows one to be all the more Christ-like, is clearly not an evaluative claim that many non-Christians or atheists are likely to agree with. But if this counts as a response to the problem of evil, it can successfully rebuff any challenge, since it resolves the apparent inconsistency within the problem. God would create or permit horrendous suffering, because ‘He did not reserve the cross for Christ’ alone. The normative assumption that the atheologian was relying upon – the one that generated the contradiction between the evaluative claim of 1 (God is maximally good) and that of 3 (evil exists) – is not shared by Weil, so there is no impact to the atheologian's challenge. Either the atheologian has failed to construct a successful reductio, or else she has chosen to work with her own normative assumptions over and above her opponent's. Either way, her formulation is toothless.
Given the fundamental role that these evaluative claims play, the problem of evil only seems to have a hope of being viable if the various sides of the debate can come to some agreement over the underlying normative assumptions that are to grant these evaluative claims legitimacy, or at least agree upon which normative assumptions are to be deployed in the debate. Such agreement is never going to be an easy thing to achieve. As a result, the problem of evil lacks the universal consistency between versions of it that would be required for it to be successful.
Many philosophers seem to be aware of this, displaying the sort of pessimistic view of moral philosophy that Hick demonstrated earlier, and yet carry on regardless. Marilyn Adams says: ‘Once theorizing begins … the hope of universal agreement in value theory is shattered’ (Adams (Reference Adams1999), 11) Similarly, Brian Davies says: ‘It is very hard to see how we are to settle the question, for what is now at stake is a fundamental moral option’ (Davies (Reference Davies1993), 38). Given this pessimism regarding the prospects of moral philosophy in this context, it is always going to be an open option for a theist to respond to any version of the problem of evil with contrasting evaluative claims. This is ‘The First Problem’ for the problem of (and argument from) evil that emerges from my analysis: unless some sort of underlying normative consistency can be agreed upon, the argument from evil cannot hope to be successful.
This is not the final word on problems for the argument from evil that stem from my analysis. Suppose we adopt a more optimistic attitude to the prospects of moral philosophy, and assume we can all come to some (at least minimal) normative agreement, and further assume that the argument from evil does manage to connect up the evaluative claims of 1 and 3 in such a way as to compel a conclusion. This conclusion will be a non-evaluative conclusion: ‘God does not exist.’ It will, therefore, appear to be a non-evaluative conclusion that was deduced from evaluative premises, and this is something that J. L. Mackie considers to be essentially problematic.
The Second Problem: crossing the is−ought gap is problematic, especially the wrong way; Mackie's criticisms of the moral argument for the existence of God
J. L. Mackie addresses these issues during his criticism of the moral argument for God's existence. This argument – put in various forms, but Mackie engages principally with Kant's – states that God's existence must be rendered more probable if we wish to maintain the rationality of a commitment to objective moral values. In order for it to be rational to pursue the satisfaction of the highest good, which is the only true aim of morality, it must be possible for this highest good to be achieved. But the only way that can happen is if God exists, as a guarantor of universal justice. Therefore, if we wish to believe that it is rational to behave morally, we ought also to commit to the truth of God's existence.
Kant retreats from claiming this argument deductively proves the existence of God, and instead claims that maintaining a belief in God's existence is necessary from a ‘practical point of view’ (Mackie (Reference Mackie1982), 108). But in any case, one can extract similar arguments from various theistic apologetics; a personal favourite of mine can be found in the works of Dostoevsky, but the argument is present in William Lane Craig's (Craig (Reference Craig2007) ), as well as many others (Evans (Reference Evans2014) ), and can be paraphrased (simply) like so:
M1. If God does not exist, then everything is permitted.
M2. It is not the case that everything is permitted.
M3. Therefore, God exists.
It is a simple argument, straightforwardly deductively valid, and with apparently true premises (at least for some). Why, then, is it not terribly persuasive for some people?
Well, partly, no doubt, because many are happy to reject the truth of the premises. Moral sceptics are quite happy to reject M2, and say that everything is indeed permitted, whilst many morally minded atheists balk at the notion of M1, that God is required for the legitimacy of even one moral obligation. But there is another problem with arguments such as these, and it is this problem that Mackie considers to be ‘the basic weakness of almost every form of moral argument for the existence of a god’ (Mackie (Reference Mackie1982), 122).
Whilst one can reject moral arguments for the existence of God such as these by rejecting the premises, it is also clear that there is something rather odd going on with the structure of the argument overall: the moral argument for God's existence moves from normative premises to a non-normative or ‘factual’ conclusion, and this is not really a rational way to go about arguing for something. As Mackie says:
A set of beliefs, even if they are called ‘intuitions’, about how one ought to act cannot be a good reason for settling a factual issue, a way of determining what is the case, or even for deciding what to ‘believe for practical purposes’. (ibid., 112)
He presents this by way of example:
[W]hat should we say about a general who accepted these three premises:
1. If the enemy are advancing in overwhelming strength, then, if we do not withdraw, our army will be wiped out;
2. We must not allow our army to be wiped out;
3. We must not withdraw, because that would mean letting down our allies; and concluded, on these grounds alone, that the enemy were not advancing in overwhelming strength? (ibid., 113)
Clearly, we would think this general were being incredibly irrational in seeming to think that he/she could deduce a claim about what factually is the case from his/her judgement of what ought to be the case. Mackie goes on:
In all such cases, what it is rational to do depends upon what the facts are; but we cannot take what we are inclined to think that it is rational to do as evidence about those facts. To use a conjunction of practical judgements to try to establish what the facts are would be to put the cart before the horse. We must rely on speculative reasoning first to determine what is the case, and then frame our practical and moral beliefs and attitudes in light of these facts. There is a direction of supervenience: since what is morally and practically rational supervenes on what is the case, what it is rational to believe with a view to practice, or to choose to do, must similarly supervene upon what it is rational to believe about what is the case. (ibid.)
Now, here Mackie is directing his criticism against the kind of moral argument for God's existence put forward by Kant, one that is supposed to be persuasive from a ‘practical point of view’. As such, Mackie's language slightly misses the mark for what I wish to say. But we can use what he says, and particularly his example of the unfortunate general, to draw out some relevant points.
First, one should note that the general's argument is deductively valid if we remove all normative content. Consider a version in which this is the case; it is easier, grammatically, to put the argument in the past tense, thereby removing any normative ‘oughts’ yet preserving a sensible set of sentences. So instead of ‘We ought not to let our army get wiped out’, which contains a sort of normative claim, we get the straightforward non-normative claim of ‘our army was not wiped out’. And so, mutatis mutandis, and though it will read a little clumsily, we get something like this:
G1. If the enemy had advanced, then if we did not withdraw, our army would have been wiped out.
G2. Our army was not wiped out.
G3. We did not withdraw.
From this, if we hold that the premises are true, we could quite happily conclude that the enemy army did not advance, and this would be a perfectly valid conclusion. So what is the difference between the obviously rational line of reasoning evident in this non-normative general example, and the obviously irrational reasoning in the normative general example? Clearly, it is the presence of the normative content, and the attempt to draw non-normative conclusions from that content. This is what Mackie means by ‘direction of supervenience’; when we are dealing with truth-apt propositions, values can supervene upon facts, and facts can supervene upon other facts, but facts cannot supervene on values.
Mackie presents another analogous case to support this claim, in the form of a couple of syllogisms: First, ‘Eat no animal fats; butter is an animal fat; so don't eat butter’ (ibid., 112), which is a syllogism of a kind that ‘most of those who have discussed imperative logic have assumed … are valid’ (ibid.). Second, ‘Eat no animal fats; you may eat butter; so butter is not an animal fat’ (ibid.). According to Mackie, if the first of these is valid, then so too must the second be. And yet, in the case of the second, ‘such a pair of imperative premises … could not objectively establish the truth of the factual conclusion’ (ibid.).Footnote 7 According to Mackie, you cannot derive facts from values.
It is this that Mackie thinks causes the obvious irrationality in the normative general case. The general is attempting to have some fact claim about the enemy advancing supervene on his or her normative claims about what he or she ought to do, but this does not work. Similarly, Mackie argues, Kant et al. are trying to have the fact claim regarding God's existence supervene upon a normative claim about how the highest good ought to be achievable, or that there are moral obligations. But clearly the direction of entailment cannot be drawn from what ought to be the case to what actually is the case.Footnote 8 Once again, Mackie's criticisms of the moral argument for God's existence distil into one claim: There is a direction of supervenience, such that values supervene upon facts, but facts cannot supervene upon values.
If Mackie's criticisms of the moral argument for the existence of God are to be considered successful, then all that remains for me to point out is that the very same criticisms ought to apply to the moral argument against God's existence too; that is, the argument from evil. The argument from evil moves just as much from evaluative claims to non-evaluative claims as the moral argument does, and therefore ought to be considered just as vulnerable to Mackie's criticisms. In his criticisms of the moral argument contained within The Miracle of Theism, Mackie has therefore sawn off the branch that he sits on, only two chapters later, when he presents the argument from evil.
Conclusion: the argument from evil is vulnerable to Mackie's challenge
I have argued that the problem of evil is made up of both evaluative and non-evaluative content, and that (as the argument from evil) it seeks to establish a non-evaluative conclusion. I have pointed out that there might be some difficulties concerning consistency within the underlying normative assumptions of the problem of evil, but assumed that these are not insurmountable difficulties. I have also discussed Mackie's criticisms of moral arguments for God's existence, and distilled them into one simple point: it is not legitimate to argue for a non-evaluative conclusion from normative (or evaluative) claims. There is a direction of supervenience between facts and values, which such an argument would get back-to-front.
Given that the argument from evil to the non-existence of God depends upon some evaluative claims, then, if the conclusion of that argument seeks to be a non-evaluative claim, wishing to state something about God's existence, it will become susceptible to Mackie's criticisms of moral arguments for the existence of God. Mackie's problem with moral arguments for the existence of God is that they attempt to derive facts from values. But, if I am correct, this is also precisely what the argument from evil attempts to do. The argument from evil moves from some non-evaluative claims (facts), via some evaluative claims (values), and back again to a non-evaluative conclusion (fact). If deriving facts from values is not on, then neither is deriving facts from values and facts, if those values play some fundamental role – which, in the problem of evil, they do. So even if we achieve some underlying normative agreement concerning the evaluative claims within the problem of evil, the fact that the argument from evil skips across the is−ought gap into fundamentally evaluative propositions, and then back again to a non-evaluative conclusion, is something that must be considered deeply problematic for the argument from evil. To do this is to commit precisely the same error that moral arguments for the existence of God are guilty of; it would be hopelessly inconsistent to ignore the issue when dealing with moral arguments against the existence of God.
So either Mackie's criticism of the moral argument is fatally wrong in some way, or the combination of it and my analysis entails that the argument from evil is left fatally undermined. Since it is based upon evaluative propositions, the argument from evil cannot conclude with a non-evaluative claim. And since ‘God does not exist’ is taken to be a non-evaluative claim, the argument from evil cannot be used to argue for the non-existence of God.Footnote 9
