Introduction
Stephen Law developed a challenge to theism, known as the evil-god challenge, in his 2010 Religious Studies article ‘The evil-god challenge’ (Law (Reference Law2010) passim). This challenge, which Law suggests is unlikely to be met by the theist, has been criticized on several grounds. The aim of this article is to extend Law's evil-god challenge and to defend it against these objections.
The evidential problem of evil is that the profusion of natural and moral evils may provide us with good reason to reject the claim that there exists an omnipotent, omniscient and maximally good god. Many theists contend, however, that the problem of evil can be addressed successfully, and that belief in a supremely good god is at least not unreasonable.
The evil-god challenge to theism is to explain why belief in a supremely good god is significantly more reasonable than belief in a maximally evil god. Diabolists – those who believe in a supremely evil god – must deal with the problem of good, but the answers available to the diabolist very closely parallel those available to the theist when addressing the problem of evil. Furthermore, many positive arguments for the existence of a good god can be mimicked by arguments for the existence of an evil god. If there is symmetry between the theist and diabolist positions – these positions being exclusive but not exhaustive – then the case for theism cannot be very strong.Footnote 1
Law has done an admirable job in developing the evil-god challenge, but the case can be strengthened; the extent to which the theist's ploys can be parodied by the diabolist has not been fully elucidated. Additionally, some have objected that the theist can meet the evil-god challenge and that the symmetry thesis – the thesis that there is a broad symmetry in the plausibility of the theist and diabolist positions – is false. Those objections will be addressed here.
The evil-god challenge
What follows is a brief summary of Law's evil-god challenge.
Law shows how popular theistic responses to the problem of evil can be parodied as diabolist responses to the problem of good. Theodicies purport to explain why an omnipotent, omniscient, and maximally good being would permit or cause either pain or morally blameworthy actions to occur. Consider, to give just several examples of familiar theodicies: (1) God gave us free will in order to make possible morally virtuous actions, even though this inevitably also makes possible morally vicious actions; (2) suffering is required for our moral and spiritual growth; and (3) the suffering we undergo in our lifetimes is more than compensated for by the eternal bliss of the afterlife.
These theodicies are isomorphic to the ‘reverse theodicies’Footnote 2 available to the diabolist. After all, free will is necessary for freely chosen morally vicious actions to be possible, although it inevitably results in free morally virtuous actions as well. Happiness is necessary for the destruction of our souls. While a life without happiness or joy is just a drudge, the loss of happiness or love can cause even greater despair and meanness. And whatever good we enjoy in our lifetimes, we get our comeuppance afterward by an eternity of suffering. (Law also parodies several other theodicies.)
The theist may also appeal to the vast gap between our understanding and God's of the fundamental goods and evils and how they are related. Our inability to think of a reason that justifies God in permitting suffering or immorality to occur is little reason to think that there is no such reason, given our cognitive limitations. But of course this is just what a diabolist can say about evil-god's unfathomable reasons for permitting goods to occur.
Law also argues that the evil-god hypothesis does not suffer much by comparison with the good-god hypothesis with respect to more general conceptual issues like simplicity or coherence.
Although the main emphasis of Law's article is about the symmetry of the theistic and diabolistic approaches to the problems of evil and good, respectively, he also shows how the diabolist can adapt various positive arguments for the existence of God. For instance, the ontological argument can be parodied by an evil twin, the argument for the existence of a being than whom no worse can be conceived. An evil being existing only in the understanding would be outdone in the evil department by a real evil being, so the evil god must be real to be that than which no more evil can be conceived. Law also argues that even if the argument from miracles has no evil mirror argument – involving observed violations of the laws of nature that do great harm – an evil god would have good reason at times to falsely represent itself as good, and to leave instructions for people in one part of the world that contradict those it has left elsewhere, ‘a recipe for ceaseless conflict, violence and suffering’ (Law (Reference Law2010), 363).
This has been but a sketch, not comprehensive, of Law's case for the symmetry thesis: that there is a broad symmetry between the reasonableness of the good god and evil god theses. What follows will extend the case for a symmetry between the theist and diabolist positions and defend it against a number of objections.
The evil-god challenge, extended
In this section of the article I will try to show that other compelling responses from the theist to the problem of evil also can be parodied effectively by the diabolist, and that some of Law's own parody arguments can be buttressed. Although Law argues, as part of his case for symmetry, that in general the positive arguments for the existence of a good God can be mirrored by arguments for evil-god, I will not defend that part of the article here. The sheer volume and complexity of arguments for the existence of God, and indeed just of versions of, say, the ontological argument, make such a task more suitable for a longer work, if not altogether quixotic.
The privation view of good
One theistic response to the problem of evil, associated with St Augustine, is the privation view of evil. Evil is held not to be something substantial, with positive existence in its own right, but rather it is the mere absence or corruption of substance or goodness. Although the privation view of evil is not by itself a theodicy, it is often motivated by the desire to reconcile the existences of God and evil. Since evil is the lack of something, God cannot be said to be the direct cause or creator of evil; all of God's creation is good. Nor, on this view, need we resort to Manichaean dualism to explain the existence of evil.
The following, from chapter 11 of the Enchiridion, is a representative example of Augustine's formulation, and defence, of the privation view of evil:
For what is that which we call evil but the absence of good? In the bodies of animals, disease and wounds mean nothing but the absence of health; for when a cure is effected, that does not mean that the evils which were present – namely, the diseases and wounds – go away from the body and dwell elsewhere: they altogether cease to exist; for the wound or disease is not a substance, but a defect in the fleshly substance, – the flesh itself being a substance, and therefore something good, of which those evils – that is, privations of the good which we call health – are accidents. Just in the same way, what are called vices in the soul are nothing but privations of natural good. And when they are cured, they are not transferred elsewhere: when they cease to exist in the healthy soul, they cannot exist anywhere else.Footnote 3
The diabolist may avail herself of a similar argument for a privation theory of good, in which being is evil and good the absence of being. Note that Augustine says when a cure of the body or soul is effected, the evil – disease or wound in the body, vice in the soul – does not go away and dwell elsewhere. Poor as this argument is, it can be aped as follows: when a body becomes ill, or a soul becomes vicious, the body's health or the soul's virtue is not transferred elsewhere. The body is a substance, and thus something evil, and the goods are accidents, privations of the natural evil.
More recently, Bill Anglin and Stewart Goetz have revived the privationist view. Some of their arguments assume as a premise the existence of a supremely good God, and some do not. An example of the former:
(1) Any function which something inherently [i.e. positively] evil might serve in the fulfilment of a surpassingly good divine purpose could equally well be served by a privative evil.
(2) The only way an omnipotent being's goodness would not be compromised if he created something inherently evil would be if that thing were logically necessary to the fulfilment of every surpassingly good purpose.
(3) Necessarily, God is good.
(4) Thus God does not create anything inherently evil.
(5) Thus nothing is inherently evil. (Anglin & Goetz (Reference Anglin and Goetz1982), 9) (The premise numbers in this argument and the next have been changed from the original text.)
This is an example of the latter, which Anglin and Goetz say can be gleaned from St Augustine's The Writings against the Manichaeans and against the Donatists:
(1) Nothing is evil unless it is destroying (or corrupting) something.
(2) Thus it is not possible that something evil not be destroying something.
(3) If there were something inherently evil it could exist apart from other things, destroying neither them nor itself.
(4) Thus there cannot be anything inherently evil.
(5) Thus evil is just privation. (ibid., 10)
Here, too, the diabolist may offer mirror arguments for a privation theory of good. The parody of the first argument: evil-god's evil would be compromised if it created a positive, or inherent, good, since a privative good could equally well serve the fulfilment of a surpassingly evil divine purpose. Thus God creates no positive evils and thus there are no positive evils. The parody of the second argument: nothing is good unless it is promoting, benefiting, or rectifying something. But something inherently good could exist apart from other things, and benefiting nothing. Thus there cannot be anything inherently good. Thus good is a just a privation.
The privation view of good is indeed difficult to defend. One might object to it that the refraining, by an indifferent person, from performing a hateful act, and the performance of a loving act are very nearly equal with respect to their lack of hatefulness and wrongness. But the loving act is a greater moral good, so moral good must be more than a privation of hatefulness and wrong action. But this objection itself is analogous to one, due to G. Stanley Kane, against the privation view of evil (Kane (Reference Kane1980), 52).Footnote 4
It might be argued that good cannot be a privation of evil because good is conceptually prior to evil, but there is no clear road from conceptual priority to ontological priority. With that caveat, I will note that Schopenhauer has argued for the conceptual priority of pain and evil, and he takes his point to have metaphysical implications:
[T]he nature of man and animal is such that we never really become conscious of what is agreeable to our will; if we are to notice something, our will has to have been thwarted, has to have experienced a shock of some kind. On the other hand, all that opposes, frustrates, and resists our will, that is to say all that is unpleasant and painful, impresses itself upon us instantly, directly, and with great clarity. Just as we are conscious not of the healthiness of our whole body but only of the little place where the shoe pinches, so we think not of the totality of our successful activities but of some insignificant trifle or other which continues to vex us. On this fact is founded what I have often before drawn attention to: the negativity of well-being and happiness, in antithesis to the positivity of pain. I therefore know of no greater absurdity than that absurdity which characterizes almost all metaphysical systems: that of explaining evil as something negative. For evil is precisely that which is positive, that which makes itself palpable; and good, on the other hand, i.e. all happiness and all gratification, is that which is negative, the mere abolition of a desire and extinction of a pain. (Schopenhauer (Reference Schopenhauer and Hollingdale1850), section 2)Footnote 5
Whatever the strengths and defects of the privation theory of good, they are paralleled by those of the privation theory of evil.
Reverse theodicies
Additional reverse theodicies are available, too.
The test of faith is a theodicy of enduring popularity, dating back at least to the Book of Job in the Old Testament. Perhaps God permits the suffering of someone like Job to test the strength of his faith. But the test of faith actually makes a better reverse theodicy. Irrationality is bad and knowledge is good, so an evil god might want our worship, but would prefer we bestowed it irrationally. The presence of good in the world makes it irrational to believe in a supremely evil god, the diabolist might say. If the world contained nothing but evil, faith in an evil god would be impossible. The evil-god allows good so as to make the evil of irrational belief in its existence possible.
The reversibility of the test of faith theodicy need not depend on such a negative characterization of faith, though. Some argue that faith should be understood not as a belief that God exists – particularly a belief that goes beyond the available evidence – but rather as a commitment of sorts. The commitment could be trust in God's goodness, or loyalty to a theistic framework of values. This commitment is not irrational as it acts as a bulwark against events that might lead one to turn away from God, or from those values.Footnote 6
Jeremy Koons argues, though, that it would be wrong for God to punishingly test us to see if we have a particular characteristic, unless our having this characteristic is morally obligatory (Koons (Reference Koons2010), 15–28). (He gives the example, which he credits to Bob Seltzer, of Evander Holyfield suddenly beating you up to test how good a boxer you are, and then giving you some money afterwards if you passed the test. It would be wrong for Holyfield to do this without your consent.) And it is difficult to make the case for trust in God being obligatory, when there is considerable evidence that the trust is misplaced. Trust or loyalty, after all, can be irrational too. It is unreasonable to inflict evil on someone to test their trust or loyalty and expect their trust not to be diminished.
So it is difficult to envision a test of faith theodicy that is not either morally dubious or too valuing of irrationality. Thus, such a theodicy is well-suited to being adapted to diabolist ends.
Other theodicies relate to the metaphysics of evil and the epistemology of evil. Some have argued that the very existence of good requires the existence of evil, just as outside is a necessary counterpart to inside. Alternatively, some contend that without evil we could not recognize, know, or appreciate the good. These too are obviously easily adaptable to reverse theodicies.
Another theodicy is that at least some of the suffering that God permits or causes is morally justified as a matter of justice: some suffering is deserved punishment. The diabolist can adapt this idea and contend that it is in pursuit of injustice that the evil god allows evil people to experience undeserved goods.
The point about injustice can also bolster Law's position on evil-god and the afterlife. Suppose, as our hypothetical diabolist might, that evil-god consigns some people permanently to heaven. Why would a maximally malevolent being do such a benevolent thing? There is no greater injustice than sending the wicked to eternal bliss after they die. But since systems of punishment and reward are made worse by arbitrariness, nobody can ensure that they avoid eternal torment in the afterlife by becoming, say, a serial killer. Although evil-god would approve of such actions, it might just send such perpetrators to hell anyhow.
In fact, a popular binary picture of the afterlife, in which (1) heaven and hell are the only final destinations in the afterlife, (2) hell is much worse than heaven, (3) heaven and hell are each populated by at least one person, and (4) God determines the criteria that sort people into heaven or hell, is more indicative of a world run by a supremely evil god, rather than by a supremely good one. As Theodore Sider argues, any suitable criterion that a good God might employ – say, generosity, kindness, or justice – will be something that can be had in degrees, on a continuum. Almost certainly, the line that is drawn between the heaven-bound and the hell-bound will separate two nearly morally identical beings, sending one to eternal bliss and the other to eternal torment. It is an injustice to treat nearly morally identical beings so drastically differently (Sider (Reference Sider2002), 58–68).Footnote 7 The arbitrariness and injustice of the binary afterlife suits evil-god very well, though, so evil-God might send at least some people to heaven to make possible this sort of comparative injustice.
Perhaps evil-god sends some to heaven to make possible the most depraved of pleasures. Saint Thomas Aquinas thought those in heaven could witness the sufferings of the damned so that their happiness may be all the more delightful to them (Aquinas (1272), supplement to part III, question 94, article one).Footnote 8 Evil-god could appropriate this as requisite for the great evil of the saved taking joy in the (probably unjust) torments of the damned. This seems to be what Friedrich Nietzsche had in mind when he said:
Dante, I think, committed a crude blunder when, with a terror-inspiring ingenuity, he placed above the gateway to hell the inscription ‘I too was created by eternal love’ – at any rate, there would be more justification for placing above the gateway to the Christian Paradise and its ‘eternal bliss’ the inscription ‘I too was created by eternal hate . . .’ (Nietzsche (Reference Nietzsche and Kaufmann1887), 1, §15)
Now, it might be objected here that no principled account has been given of what a maximally evil being's aims would be.Footnote 9 Here the diabolist suggests that it is maximal injustice or arbitrariness. There the diabolist plumps for suffering, imperfection, or moral evil. The diabolist ascribes to evil-god whatever nefarious goal suits her present purposes. This is a very legitimate objection to the evil-god hypothesis, to be sure. But this objection is mirrored by one against the theistic responses to the problem of evil, where God is depicted, alternately, as desiring for people to be good, or for them to be faithful, or for them to be treated well, or for them to be treated justly, depending on the theodicy in question.
The worst of all possible worlds?
One version of the argument from evil is that if this world were created by an omnipotent, omniscient, and supremely moral being, it would be the best of all possible worlds. But it is possible for this world to be better. Thus the world has no such perfect creator.
Leibniz, in his Theodicy, famously advanced a claim along the lines of the first premise of that argument:
(P) If a perfectly good moral agent created any world at all, it would have to be the very best world that it could create. (Leibniz (Reference Leibniz, Farrer and Huggard1710), part I, section 8)
Some theists, however, have responded to the modalized version of the argument from evil by arguing against (P), on several different grounds. Robert Merrihew Adams is one such theist. Adams thinks it is unlikely that there even is a best possible world. If good is additive, why couldn't it be that, for every possible world, there is another that is better? He also argues that if there is in fact no best possible world, it would not be better for God to create no world at all, for there being no created world at all would be a less excellent state of affairs than the existence of some of the worlds God could have created (Adams (Reference Adams1972), 317). Putting aside, though, his contention that there might not be a best possible world, Adams argues that (P), with its maximization requirement, tacitly assumes an act-utilitarian (or some similar act-consequentialist) theory of morality, which a theist of the Judaeo-Christian tradition is certainly not forced to accept (ibid., 318).
Consider this claim:
(Q) If a perfectly evil moral agent created any world at all, it would have to be the very worst world that it could create.
The diabolist can parody each of these responses with respect to (Q). She could make a Leibnizian move and argue that this really is the worst of all possible worlds, for the reasons already discussed. Or, like Adams, she could argue that there is little reason to think that there is such a thing as the worst of all possible worlds. If this is the case, then whatever world the evil god creates, there will be people who ask ‘But why isn't it worse?’ Thus it faces a choice between creating no world at all and thus having nobody suffer or be wicked and creating a world where people are wicked and suffer horribly but some wonder why they don't suffer more. It is clear what a supremely evil being must do if faced with that choice.
The modalized argument from good (or at least some versions of it) against the existence of an evil god also seems to depend tacitly on an act-consequentialist conception of morality as much as the modalized argument from evil does. The diabolist may reject the view that the morally worst action is necessarily the one that maximizes suffering (or some other bad thing), and thus avoid falling into the trap of having to find a greater evil for which each good is necessary.
As alternatives to the consequentialist moral approach, Adams considers the question from the standpoint of virtue theory and deontology. For the virtue approach, he cleverly deflects the question why God has not been more benevolent to us, and asks instead why God has not created only better beings than we are, like angels. He argues that if God intentionally creates beings who are not as excellent as God could have created, this need not mean that God has a defect of character (as Plato suggested it didFootnote 10). Rather, God's willingness to create and love morally (and otherwise) very imperfect creatures is a sign of God's grace. The virtue of graciousness is the disposition to show love that is not dependent on the merits of the being who is loved. Alternatively, Adams says, ‘the gracious person sees what is valuable in the person he loves, and does not worry about whether it is more or less valuable than what could be found in someone else he might have loved’ (ibid., 324). Adams seems to regard God's creating the greatest possible world as consistent with God's showing grace, but not required by it (ibid., 324).
The diabolist may ask why evil-God did not make only the worst beings that it could create, such as demons. What moral vice is shown by one who creates either beings morally better than it could have created, or beings with equally immoral intentions but less ability to execute them? Perhaps evil-god is jealous; it wants people to be evil but jealously reserves maximal evil for itself.
For an answer that more closely parallels Adams's, though, one might argue that in order for evil-god to have the worst of all kinds of hate, hatred of the good (or hatred of that which especially ought not to be hated), it must create beings who are at least somewhat good. Thus evil-god can have a vice (to my knowledge unnamed) that is counterpart to the virtue of grace: the disposition to show hate that is not dependent on the demerits of the one hated.
For the deontological approach to (P), Adams argues that God need not wrong anyone by creating a world that is not the best possible world. More specifically, God does not wrong anyone in creating a world with the following characteristics:
(1) None of the individual creatures in it would exist in the best of all possible worlds.
(2) None of the creatures in it has a life which is so miserable on the whole that it would be better for that creature if it had never existed.
(3) Every individual creature in the world is at least as happy on the whole as it would have been in any other possible world in which it could have existed. (ibid., 320)Footnote 11
God does not wrong those creatures from the best possible world that it chose not to create, as it is impossible to wrong or treat unkindly those who never exist. Nor does God wrong any of the created beings, argues Adams, as none of them would have been benefited by God creating any other world instead.
The diabolist can help herself to similar deontological considerations. She can argue that evil-god did not ‘right’ the tormented denizens of the worst possible world by not creating them, since there are no such beings. Nor does evil-god right the beings it did create, provided that they all have lives so miserable that they would be better off if they had never existed, and all of them are as poorly off as they could have been in any other possible world in which they would have existed.
It might be objected that evil-god fails to meet these conditions, as many people are happy enough that they would not be better off never existing, and many people would have been worse off in other possible worlds. This is surely true, but it is a point equally applicable to the theist. I take Adams's argument in this article to be just that we need not accept (P), and not that God in fact has been a perfectly good moral agent. Likewise, the diabolist wanting to mimic Adams can restrict herself to arguing against (Q).
It looks, then, like the diabolist's case against (Q) enjoys parity with the theist's case against (P). In fact, if Adams is correct that God's grace does not require God to create beings not as excellent as God could have created, but evil-God's jealousy does require that it create beings less evil than itself, there is an asymmetry that favours the diabolist.
The evil god challenge, defended
Several articles have raised objections to the evil-god challenge, and they are addressed here.
The objection from simplicity
One might object that a being with only positive attributes (omniscience, omnipotence, and complete moral goodness) is simpler than one with a mix of positive attributes (omniscience and omnipotence) and a negative one (complete moral evil), and thus that an explanation that invokes the former is, at least in that way, simpler, more economical, and likelier to be correct than one that invokes the latter.
Law indeed considers this objection, and addresses it by making two points. First, he reminds us that he claimed only that the moves available to the diabolist were approximately isomorphic to those available to the theist, and not that there was a perfect symmetry. Furthermore, he argued that an asymmetry with respect to simplicity need not confer a significant advantage on one of two competing theses. If two theses are both highly implausible, the simpler thesis is no better off than the less parsimonious one. (Law compares the claim that Swindon is populated with a thousand elves with the ontologically less economical claim that Swindon is populated with a thousand elves, each with a fairy on its head (Law (Reference Law2010), 372).
A theist might argue that it is question-begging here for Law to help himself to the claim that theism is so implausible. A better reply is available, though. One could follow Kant in arguing that the only unconditional good is a good will, and thus that omniscience and omnipotence are positive attributes only on the condition that they are conjoined with moral goodness, and that they are negative attributes when present in evil-god.Footnote 12 As with the coolness of Kant's villain, evil-god's omniscience and omnipotence make it all the more abominable.
The simplicity objection was revived by Calum Miller in a recent Religious Studies article (Miller (Reference Miller2016), passim). Miller makes a strong case for the epistemic weight of simplicity, and he argues, following Swinburne (Swinburne (Reference Swinburne2004), passim), that infinite degrees of properties are simpler than extremely large but finite degrees of properties (called mega-properties). He contends (although this is but a very small part of his article) that this supports the view that theism is simpler than parodical alternatives, like Law's evil-god. However, if Miller is correct that omni-properties are simpler than mere mega-properties, this would show at best that theism is simpler than so-called ‘mega-theism’, which Miller defines as the view that God is extremely-but-not-all-powerful and very knowledgeable but not omniscient, and that it is simpler than ‘mega-diabolism’. But the evil-god under consideration here, and in Law's article, has omni-properties. Miller's argument may strike a blow against a Humean morally indifferent God – though it is not clear that it does – or Cleanthes’ ‘finitely perfect’ deity, but it is irrelevant to the case against the symmetry of theism and diabolism, absent some argument that maximal evil is not an omni-property.
So how might one otherwise argue that theism is simpler than diabolism? One way would be to hold that on the theist hypothesis the way the world appears is a more reliable guide to how it is, compared to how it would be on the diabolist thesis. Provided that that is true, and some sort of phenomenal conservatism – that if it seems to one that p, then, absent defeaters, one is justified in believing that p – as well, then theism enjoys greater epistemic justification. This might explain why few people believe in a Cartesian evil demon or evil-god, while many believe in a good God.Footnote 13
The difficulty in this line of argument, I think, is in making the case that the appearances are any worse guide to reality if diabolism is true. The problem of evil itself, after all, is motivated by a pretty huge disparity between what appears to be gratuitous evil and the putative reality that all evils are justified. Keep in mind, as well, that we are not considering Descartes's demon, bent single-mindedly on deception, but rather an evil-god bent on evil in its various forms. Granted, deception is, typically, one kind of moral evil. But it is not at all clear that evil-god's promotion of evil in the world requires either that perception be an especially poor guide to reality, or that evil-god lie to us. Just as many theists take God to be hidden or silent, and justified in so being, it is open to diabolists to accept and rationalize diabolical silence. Furthermore, appearance being a fairly good guide to reality seems required for people to effectively pursue various evils, just as it is required for them to effectively pursue goods. (A brain-in-a-vat can neither help nor harm anyone.)
The objection from the intrinsic goodness of free will
If freedom of the will is an intrinsic good, then the good-god and evil-god free will theodicies do not precisely parallel each other. Indeed, the good-god theodicy would be decidedly stronger.
Law concedes both the asymmetry and that the asymmetry favours the theist, but he argues that this does not significantly harm the symmetry thesis. First (again), Law never contended that there was a perfect symmetry between the cases for theism and diabolism. Second, he reminded us that the advantage here for the theist might be counterbalanced or even outweighed by asymmetries that favour the diabolist. (Law mentions the diabolist justification for miracles. I might add the test of faith reverse theodicy.)
The reply to this objection can be strengthened, however, by attacking directly the idea that free will is an intrinsic good. Again, one may follow Kant in arguing that it is only a good will that is unconditionally good, and that a bad will is the only unconditional evil. Free will itself is neither an intrinsic good nor an intrinsic evil, so the free will theodicy and reverse theodicy are perfectly parallel.
Most, perhaps, are inclined to see free will as intrinsically good, however. But the idea that a world in which people pursue evil of their own free will is worse than one in which people are ‘programmed’ to be evil sounds (to my ears, at least) just as plausible as the corollary that a world in which people pursue good of their own free will is better than one in which God simply causes them to do so. And just as freely chosen moral good is better when it is a choice between good and evil, rather than only between two goods, when it is chosen despite the presence of temptation to do evil (as argued in Swinburne (Reference Swinburne and Rowe2001), 49), freely chosen evil is worse when it is chosen over good, rather than over a different evil, when it is chosen despite temptation to do good. It makes just as much sense for evil-god to give people free will as it does for a good God to do so.
One could even press the argument that there is an asymmetry but that it favours the diabolist, because free will is an intrinsic evil. Consider Nietzsche's contention that free will is a ‘trick’ of the theologian:
Today we have no sympathy anymore for the concept of ‘free will’: we know only too well what it is – the most disreputable of all the theologians’ tricks, designed to make humanity ‘responsible’ in the theologians’ sense, that is to make it dependent on them . . . Here I am simply offering the psychology of all making-responsible. – Wherever responsibilities are sought, what tends to be doing the seeking is the instinct of wanting to punish and rule. One has stripped becoming of its innocence when some state of being such and such is traced back to will, to intentions, to acts: the doctrine of the will was essentially invented for the purposes of punishment, that is, for purposes of wanting to find people guilty. All the old psychology, the psychology of will, is predicated on the fact that its originators, the priests in the elites of ancient communities, wanted to create a right for themselves to inflict punishments – or wanted to create a right for God to do so . . . Human beings were thought to be ‘free’ so that they could be ruled, so that they could be punished – so that they could become guilty. (Nietzsche (Reference Nietzsche and Hollingdale1889), ch. 5: The Four Great Errors)
Although Nietzsche thinks that belief in free will is a pernicious error, his point could be adapted to support the claim that free will is given to us by an evil god. A God who really makes people responsible for what they do might do so in order that they could become guilty and be punished. If the impulse to find someone responsible for what they do is predominately in order to place blame, rather than to give credit, then perhaps this can be extended to making people responsible for their actions.
Along similar lines, Primo Levi, in The Drowned and the Saved, has said that the creation at Auschwitz of Sonderkommandos – special squads of prisoners given temporary reprieves as long as they carry out monstrous orders – was the Nazis’ ‘most demonic crime’, since it shifted some of the burden of guilt to the victims (Levi (Reference Levi1989), 53). Although these prisoners chose to cooperate only under extreme duress, by doing so they incurred some moral culpability.
In fact, merely having the freedom to choose, however, or whether, it is exercised, can bring guilt. Consider Sophie's Choice, in which an Auschwitz doctor tells Sophie to choose which of her children is to be murdered. Although her choice – that her daughter should be murdered – is morally defensible, perhaps even morally required, making that choice implicates her, in some measure, in the atrocity. She is wracked with guilt and eventually takes her own life. Her plight would be no better had she refused to choose, and let the SS doctor pick the child to be killed, as that is a choice in itself. By giving her freedom to choose, in a world where her choice is between only greater and lesser atrocious evils, the Nazi gives her some responsibility and guilt for those evils.
Scrutton's objection from pragmatic encroachment
In a recent Religious Studies article, Anastasia Philippa Scrutton appeals to the pragmatic encroachment account of epistemic justification in support of her argument that the good-god thesis is epistemically more justified than is the evil-god thesis (Scrutton (Reference Scrutton2016), passim). In short, she contends that because we are better off believing, and acting as if, the good god hypothesis were true, but worse off believing and acting as if the evil god hypothesis were true, we are more epistemically justified in believing in the good god, even if there is evidential parity for the two hypotheses. Hence, the symmetry thesis is false.
In the work mostly closely associated with pragmatic encroachment, Knowledge in an Uncertain World, Jeremy Fantl and Matthew McGrath (Reference Fantl and McGrath2012) advance the following principles (cited by Scrutton) tying epistemic justification with practical reasoning:
KA: S knows that p only if S is rational to act as if p.
JA: S is justified in believing that p only if S is rational to act as if p.
They cite familiar thought experiments involving cases where only the stakes differ. For instance, Amy has evidence that makes it quite probable but not certain that the post office is open on Saturday, and nothing much turns on whether she is right. Beth has the same evidence that the post office is open on Saturday, but it would be catastrophic for her if she is wrong. Many have the intuition that Amy is justified in believing that the post office is open on Saturday (as she is justified in acting as if it is, and employing that belief in her practical reasoning) and knows that it is open (provided that is in fact the case), but Beth is not, since she cannot afford to be wrong. (She'd better double check, or stop by on Friday.) Basically, they purport to show that being probably right that a proposition is true isn't necessarily sufficient for being justified in believing it, as being probably right is not good enough for practical reasoning in high stakes cases.
Scrutton writes:
Notably it is JA rather than KA that is most relevant for this article, since what is at stake is the rationality of holding different religious beliefs rather than whether they are actually true or not. Most of the pragmatic encroachment literature focuses more on knowledge than on justification, but if knowledge (broadly speaking) entails justified belief, then if KA holds then JA holds too. (Scrutton (Reference Scrutton2016), 348)
This is fallacious. KA does not entail JA, even provided that knowledge entails justified belief. Knowledge entails both belief simpliciter, after all, and truth, but it would be absurd to hold that KA entails that S believes that p only if S is rational to act as if p, or that p is true only if S is rational to act as if p. KA and JA give only necessary, not sufficient, conditions for knowledge and justified belief, respectively.
That said, Scrutton is probably correct that JA enjoys just about as much plausibility as KA does. In both, it is knowledge-level-justification, or the degree of justification sufficient for knowledge provided that the belief in question is true, that is doing all the work in the various thought experiments supporting pragmatic encroachment. So let's suppose that JA is true. This is supposed to favour theism over diabolism because various studies show that people who believe in a good God or benevolent spirits are better off in a number of ways – both morally and in terms of well-being – than are people who believe in malevolent spirits. Though Scrutton allows that these benefits might be culture-relative – the Ju ‘hoansi, a nomadic Bushman people of the Kalahari, believe in evil spirits without evident distress or moral detriment – and notes that studies on this issue are limited by having focused largely on religions in the Christian tradition in the USA and western Europe, let us assume for the sake of argument that she is right about this.
But are the mental health benefits of theism relevant to an appeal to pragmatic encroachment? What Fantl and McGrath argue is that the higher the cost of one's using a believed proposition in one's practical reasoning, were the proposition to turn out false, the higher one's threshold for knowledge-level justification for that proposition So what Scrutton needs are Fantl and McGrath-style high stakes/low stakes scenarios for belief in a good God, and for belief in evil-god, respectively. Curiously, Scrutton does not provide any. Probably this is because it is very difficult to do so, since it is difficult to envision a low stakes scenario at all. There is nobody for whom it little matters whether there is a good god, an evil god, or no god at all. The prospect of one's decisions helping to bring about one's eternal bliss, or one's eternal punishment, or the lack thereof, makes the proposition – and one's belief of the proposition – one of invariantly high stakes. One lesson from the pragmatic encroachment account, then, seems to be that here the standards for justified belief in either theism or diabolism must be extremely high in all contexts.
Note also that JA provides a necessary condition, not a sufficient one, for justified belief, so even if one is better off living as though there is a good God, this does not ensure that the belief is epistemically justified. The athlete who has little chance of winning a competition might be prudentially justified in believing that, and acting as if, she will win, as this will improve (modestly) her chances of winning, but she is not epistemically justified in doing so, given the paucity of evidence that she will win. JA concerns knowledge-level justification, which requires a strong alethic component. The following biconditional is a pragmatic encroachment definition of justified belief suggested by Jonathan Jenkins Ichikawa and Matthias Steup (2001):
S is justified in believing that p if and only if no epistemic weakness with respect to p prevents S from properly using p as a reason for action.
Scrutton assumes for the sake of argument that there is evidential or alethic parity between the good and evil God hypotheses. These hypotheses are also exclusive and not exhaustive, as there is some probability that there is no God, or a God who is neither supremely good nor evil, or that there are many gods. Consequently – dialetheism aside! – the good God hypothesis is significantly likelier to be false than to be true, and that is a paradigmatic epistemic weakness. Hence, no matter the benefits to the believer of theistic conviction, JA does not apply here.
One could argue, though, even if Fantl and McGrath, and Scrutton, apparently, do not, that the insight of the pragmatic encroachment account can be extended to weaker-than-knowledge levels of epistemic justification.Footnote 14 We can adapt the high stakes/low stakes cases to be examples where the subjects have modest evidence in support of a claim. The modest evidence in the low stakes scenario will justify someone to some modest extent in acting as if the claim were true, but will justify her less so (or not at all) in the high stakes scenario. So extended, even if the evidence in favour of diabolism and that in favour of theism are both fairly weak, the practical benefits of theism might also tip the epistemic justificatory scale.
I think, though, that the pragmatic encroachment view (so extended) actually supports the opposite conclusion. If anything, it is the diabolist who can better afford to be wrong. Suppose you bank on the evilness of god, based on various sorts of evidence, and act accordingly. You worship an evil god and do what you think will be pleasing to that evil god. But you worry. ‘What if I am wrong and God is good?’ Well, a good God may well be kind and forgiving. It could well be okay for you, even if you are wrong (though of course it might not).
But the theist can hardly make the same claim. Suppose you believe in the goodness of God, and act accordingly, employing this belief in your practical reasoning by acting the way you think would be pleasing to a morally good God. But you worry that you might be wrong about God's goodness. What if God is evil? The theist can feel little reassurance that it might not be too bad if she turns out to be wrong! (Though, of course, it is possible any particular morally good theist would be better off with evil-god, it is unlikely.) If the foregoing is correct, there is indeed an epistemic asymmetry here, but it favours the evil-god hypothesis, since the theist has more to fear from being wrong than the diabolist does.
Pragmatic encroachment ties practical considerations to epistemic justification by holding that the higher the cost of one's being wrong in believing and acting on a given proposition, the higher one's threshold for epistemic justification for that proposition. It is difficult to say what will happen to a theist if it turns out that there is an evil god, just as it is difficult to say what will happen to a diabolist if it turns out God is good. But the stakes are likely higher for the theist, in that if the theist is wrong about whether God is good or evil, she is at odds with an unjust deity who values neither mercy nor forgiveness. If the diabolist is wrong about the same issue, she is at odds with a just being who probably values mercy and forgiveness. Consequently, if diabolism and theism enjoy evidential parity, the threshold for epistemic justification is, if anything, higher for theism than for diabolism.
Scrutton has a second argument that the practical benefits of believing a proposition (or of acting as if that proposition were true) can confer added epistemic justification to that belief. Scrutton considers the objection (which she credits to Roger White) that the practical factor which has genuine epistemic weight is not conduciveness to well-being but rather the self-fulfilling nature of a belief. Self-fulfilling (or probability-enhancing) beliefs, such as perhaps I will recover from this illness, are epistemically justifiable since belief of the proposition can help bring about the truth of the proposition. But beliefs that are simply conducive to well-being are epistemically dubious, as well-being considerations could be used to justify wishful thinking. The belief in a good God belongs to the latter, epistemically dubious group, as this belief does nothing to bring about its own truth (ibid., 354).
While Scrutton agrees that a belief can be epistemically justified because it is self-fulfilling, she contends that this supports her own position that well-being contributes to epistemic justification. She points out that in a situation where belief of a proposition and belief of its negation would each be self-fulfilling, it is more rational to believe the proposition the truth of which better promotes one's well-being. If I will recover from this illness and I will not recover from this illness are both self-fulfilling, and one indeed wants to recover from the illness, then it is more rational to believe that, and act as if, one will recover. So the rationality of believing even self-fulfilling beliefs is contingent upon the belief being good for one's well-being. Furthermore, she argues that regarding well-being as a justification for belief would provide support for wishful thinking only if well-being were taken to have priority over, or exclude, evidentiary concerns. That is not how she is regarding it, though, as well-being is adduced in support of one theory over another with parity of evidential support.
It is unclear to me whether this second argument is a further application of the pragmatic encroachment view, or is independent of it. Let us consider both possibilities, starting with the supposition that we are still working with the pragmatic encroachment view. Again, Fantl and McGrath's (JA) is the claim that if one is epistemically justified in believing that p, then one is rational in acting as if p were the case (or rational in using p in one's practical reasoning). Though this could be understood in a number of ways, Fantl and McGrath take it to mean that the more that's at stake, the higher the threshold for epistemic justification. So in a case where the belief that one will recover and the belief that one will not recover are both fallibly self-fulfilling, the crucial issue would be the stakes for using these beliefs in one's practical reasoning. Acting as if one will recover might involve, say, not writing a will, booking a vacation a year hence, and so forth. Acting as if one will not recover might involve writing that will, saying farewells to loved ones, checking off final items from one's bucket list, and so on. Which one is safer to use in one's practical reasoning, assuming evidential symmetry, probably depends on the situation. Consequently, I am unconvinced that Fantl and McGrath would concede that the belief that one will recover from the illness is more epistemically justified that the belief that one will not recover, because I am dubious that they would regard the former as generally more suitable for practical reasoning (though one could contrive an example in which one or the other would be). Besides the facts, already discussed, that Fantl and McGrath are talking about knowledge-level justification, and that (JA) claims only that epistemic justification is sufficient for practical rationality, not that it is necessary for it, their conception of what it is for one to be rational to act as if a proposition were true is quite different from Scrutton's well-being-conduciveness conception.
However, one might find the claim that, given that one wants to recover from the illness, it is clearly more practically rational to hold the fallibly self-fulfilling belief that one will recover, and less practically rational to hold the belief that one will not recover. Regardless of the particular way in which pragmatic encroachment theorists Fantl and McGrath understand ‘S is rational to act as if p’, there is a broader understanding of that concept that licenses favouring some self-fulfilling beliefs over others, due to well-being (including moral well-being) considerations, perhaps showing that it is reasonable to regard well-being as one of the practical considerations capable of affecting practical rationality and also epistemic justification. The implications for philosophy of religion are that practical reasons for religious beliefs, rather than being relevant only to pastoral theology and religious therapy, should be front and centre, alongside alethic considerations, with respect to the epistemic justification of religious beliefs.
But if this is not a further application of the already developed pragmatic encroachment theory, then we need an argument for this sort of practical virtue of a belief – conduciveness to well-being and positive moral consequences – being part of what constitutes epistemic justification, much as Fantl and McGrath have provided an argument for one's epistemic justification for a belief being sufficient for that belief's suitability for one's practical reasoning.
Note that I am not disputing that there may well be some sorts of justification for believing in a good God which belief in evil-god will lack. Belief in a good God may be conducive to well-being when belief in evil-god is detrimental to well-being. There may be a moral justification for believing in a good God which is not had by belief in evil-god. But the evil-god challenge relates to epistemic justification, and the pragmatic encroachment account of epistemic justification that Scrutton appeals to does not make conduciveness to well-being and good moral consequences constituents of that epistemic justification.
Furthermore, while Scrutton implies that belief in a good God has better moral consequences than belief in evil-god does, she does not really defend the claim, aside from a passage about the moral benefits – kindness, patience, humility – of believing that all people are of equal value, irrespective of differences in ability or beauty. It is not difficult to imagine how such a defence might go. One who believes in a good God and puts that belief into practice is presumably likelier to strive to obey a good God by trying to help other people, while one who believes in evil-god might be more likely to torment others. However, there are self-interested grounds for one who believes in a good God to help others – to please the omnipotent being with the power to reward or punish one eternally – that are absent for one who believes in evil-god. If helping others out of self-interest is antithetic to genuine morality, then belief in a good God may make authentic moral goodness much more difficult to achieve. There are many ways of totalling the moral consequences, and not all of them clearly count in favour of theism.
The objection from the impossibility of diabolism
Two recent articles, one by Peter Forrest and the other by Christopher Gregory Weaver, take on the evil-god challenge by arguing that an omniscient individual being motivated by evil is less likely than such an individual being motivated by good (Forrest (Reference Forrest2012), passim), or that an unsurpassably evil being – one necessarily motivated by evil – is impossible (Weaver (Reference Weaver2015), passim). Either way, theism would be more plausible than diabolism. I think these represent the strongest objections to the asymmetry thesis.
Forrest's case against the symmetry thesis is complicated. While Forrest assumes both a consequentialist morality and that (consequentialist) goodness can be predicated only analogically, not literally, of God, it is not clear to me that either of these claims really is crucial to his case, so I will concede those assumptions and discuss them no further.
Philosophers disagree about whether moral judgements or beliefs motivate by themselves or only by means of some pre-existing conative state, such as the desire for good. Roughly, moral (or reasons) internalists believe that moral judgements motivate by themselves and externalists believe they do not.
While Forrest grants that there is no more evidence or reason to ascribe to God a divine desire for the good, or a benevolent character, than there is to ascribe a divine desire for evil, or a malevolent character (39) – good news for the symmetry thesis, if moral externalism happens to be true – he contends that considerations from simplicity favour moral internalism. The internalist picture, since it posits no desires for the good, only moral judgements, is simpler both in information-theoretic terms and in ontological terms. Explanations positing the internalist's divine being are simpler and thus, by Ockham's Razor, better explanations. Furthermore, this sort of divine simplicity is in keeping with the classical theism of Augustine and Aquinas. Thus a God without desires, and indeed without virtues or character, is at least as likely as a God with those things (Forrest (Reference Forrest2012), 39–40).
Suppose then that moral internalism is true with respect to God. Suppose also that moral judgements are objective in the sense that they are true or false independently of any mind, even God's. Forrest endorses a view that combines this moral internalism with moral cognitivism, which he calls axiarchism. Axiarchism, a thesis about agents both human and divine, is the conjunction of the following three claims:
(1) Comparative evaluation and hence the judgement of which act has the best consequences is objective in the sense of being true or false independently of any mind, even God's. (Presumably it is therefore non-contingent.)
(2) That the consequences of act X appear to be better than the consequences of any other act being considered directly motivates the agent to perform X unless the agent suffers from akrasia. (The motivation would be indirect if it was due to a desire to do whatever appears the best.)
(3) Akrasia does not afflict human beings all the time, and when it occurs it can be explained by details of the human condition that should not be extrapolated to the divine condition. (ibid., 40–41)
Of course, there is another moral internalist position available, which Forrest calls antaxiarchism, much like axiarchism except that ‘best’ and ‘better’ are replaced by ‘worst’ and ‘worse’. All that Forrest says in favour of axiarchism is that our experiences of our own acts, and feelings of temptation, support axiarchism over antaxiarchism. Normal human beings are not directly motivated to perform an action by the judgement that the action is wrong. When we knowingly do wrong, our moral judgements and temptations to do wrong are pitted against each other, as they would not be, were antaxiarchism true.
So I see Forrest's argument roughly as follows:
(1) Either moral internalism or moral externalism is true.
(2) If moral internalism is true, then God is simpler than if moral externalism is true.
(3) Thus, it is at least as likely that moral internalism is true as it is that moral externalism is true.
(4) If moral internalism is true, then either axiarchism or antaxiarchism is true.
(5) Axiarchism is inductively supported over antaxiarchism by our experiences of our own actions and temptations.
(6) So if moral internalism is true, then God is motivated directly by good, not evil.
(7) If moral externalism is true, then it is equally likely that God is motivated indirectly by evil as it is that God is motivated indirectly by good.
(8) Thus, it is more likely that God is motivated by good, whether directly or indirectly, than it is that God is motivated, directly or indirectly, by evil.
(9) Thus, the symmetry thesis is false.
This ignores some niceties of Forrest's argument, but I think it captures the core idea.
One might quibble about a number of points here, but I will focus on premises (4) and (5).
The problem with premise (5), and Forrest's support for it, is that our experiences of our own actions and temptations are at best inductive evidence for a more restricted axiarchism that applies only to ourselves, or to typical people, but not for abnormal or divine beings. Both axiarchism and antaxiarchism are sweeping claims about all agents, human and divine. The diabolist need not defend anything as extravagant as antaxiarchism;Footnote 15 rather, divine motivation might function quite differently from that of typical people. Indeed, by Forrest's own lights, the motivational structure for God differs fundamentally from that of humans, since we have desires, virtues, vices, and characters, but God does not. And even though it is true that people do not always act according to their moral character, this too makes for a disanalogy with God; God never acts out of character. So Forrest's own claims about God undermine arguments that extrapolate from our being directly motivated by the good – provided that we really are – to God's being the same way.
Rather than relying on an inductive argument, perhaps Forrest can boil it down to simplicity: a theory according to which people are directly motivated by good but evil-god is directly motivated by evil is not as simple (and thus not as likely) as one according to which both people and God are directly motivated by good. Again, though, Forrest's own thesis – that's God's motivational structure differs fundamentally from that of people – makes this line of argument problematic for him. Why count it against diabolism that it holds there to be important differences between the responses of God and humans to their moral judgements, if theism – Forrest's theism, at any rate – does the same? If we drop that part of Forrest's view – that unlike people, God has no desires or character – the simplicity argument is easier to make. The diabolist view (assuming internalism) is not ontologically baggier, but it would require that evil-god responds negatively, rather than positively, to moral judgements. Perhaps the most reasonable conclusion is that considerations of simplicity do modestly favour the theist here, for reasons of the sort Forrest gives.
Weaver argues against even the possibility of an unsurpassably evil being, relying only on a claim about what it is to be unsurpassably evil, a principle of moral rationalism, and a weaker moral internalist principle than the one Forrest uses.
First, for the evil-god to be analogous to the good God of the theist, Weaver argues, it must be maximally evil, the most evil being possible. Just as God's goodness must be a necessary property of God's if God is to be maximally good, so too must evil-god's evil be necessary to it (Weaver (Reference Weaver2015), 6). A being who actually is motivated by evil but who could have been motivated by good – a being who, as we might say, is motivated by good in another possible world – is not as evil as a being who is necessarily or essentially evil, or evil in all possible worlds.
Second, Weaver asserts moral rationalism, which he characterizes thusly: ‘Roughly put, moral rationalism is the thesis that duties and/or moral obligations either strictly imply practical reasons for action, or are identical to such reasons’ (ibid., 10) Weaver alternatively characterizes this principle thusly: a being that knows its moral obligations has good reasons to fulfil them (ibid., 14).
Lastly, Weaver also endorses this moderate principle of reasons internalism:
For any moral agent a, necessarily, if a has a good reason to ϕ in circumstance C*, then possibly a will be motivated to ϕ in C*. (ibid., 12)
The ‘possibly’ is added to the embedded consequent so as to accommodate the possibility of the amoralist. So long as the person is not essentially amoral, the existence of an amoral sociopath is consistent with this principle.
Put all of these together and you can deduce that Law's evil-god – a being who knows all the moral facts but who is necessarily such that it feels no motivation or desire to do what it knows is morally obligatory – is impossible.
It is a commonplace that God, provided it exists, is a necessary being. That necessity is often extended to its various attributes, but now we venture into muddier waters. Some divine attributes seem like they must be contingent, such as God's knowing a particular contingent truth, having answered a particular contingently existing prayer, or being the creator of a particular contingently existing world.
For God to be maximally good, must God be necessarily good, or good in all possible worlds, as Weaver contends? The received theist answer to this question is ‘yes’, but I would like to push back a little on that. One motivation for thinking that an unsurpassably moral being must be a necessarily moral being (and mutatis mutandis for a maximally evil being) is rooted in the tendency to see possible worlds as akin to distant, concrete worlds. Weaver, for example, says: ‘an entity that is evil in one world, but indifferent, or good at another, is not unsurpassably evil’ (ibid., 6). Although so-called Modal Realists like David Lewis believe precisely that, the more prevalent view is that possible worlds are just ways things could have been. (I will assume the correctness of the latter.) So if it makes any sense to say that a being is maximally good, when it could have been – but in no way is – bad or at least worse, then maximal goodness need not entail necessary goodness.
And this has some intuitive appeal, I think. In the same way that the free will theodicy concedes that God could have made everyone kind but that moral goodness has more value, or is possible only, when it is freely chosen, we might say that God is morally better if it freely chooses the good over evil, rather than being good of necessity. The nature of divine freedom is notoriously difficult to pin down, of course, including whether God's freedom is one of the libertarian sort. But if God can be maximally good even though it could have been bad (or at least worse), and evil-god can be maximally evil even though it could have been good (or at least better), then Law's evil-God would not be ruled out by the conjunction of Weaver's modest reasons internalism principle and his moral rationalism.
A. A. Howsepian has in fact argued along these lines. He argues (Howsepian (Reference Howsepian1991), 477–479) that if two beings are wholly good, but the set of good action-types of which the first is capable is a subset of the set of good action-types of which the second is capable (as with, say, an angel and God, respectively), then the first being is good to a lesser degree than is the second. Refraining from evil is a good action-type (provided good motives, perhaps), but Howsepian notes that not doing evil is not sufficient for refraining from evil. I am not squaring a circle yet I cannot be said to be refraining from squaring a circle. An oak tree does not eat a sandwich but it does not and cannot refrain from doing so. To refrain from an action, one must be able to perform the action. So if God is necessarily good, in the sense that it is metaphysically impossible for God to do what is not good, then God cannot refrain from doing what is not good. A contingently good God can so refrain, so a contingently good God is good to a greater degree than a necessarily good God is. An unsurpassably or maximally good God then must be good contingently.
If Howsepian's argument enjoys some plausibility, as I think it does,Footnote 16 then Weaver cannot help himself to the assumption that God is essentially or necessarily good, nor that the diabolist position, to be a genuine parody of the theistic one, must hold that evil-god is essentially evil.Footnote 17
But as with Forrest's argument, a simplified version of Weaver's case might be stronger. Suppose Weaver assumes only moral rationalism (and no internalist thesis). If that view – basically the second conjunct of Forrest's axiarchism – is correct, then any agent who acts wrongly suffers either from practical irrationality or from some sort of delusion about what is morally right. (On this view, sinister reasons cannot serve as proper practical reasons (or ‘decisive’ reasons) for actions (Weaver (Reference Weaver2015), 22).) The diabolist cannot accept either of those with respect to evil-god, so the plausibility of moral rationalism is a threat to the symmetry thesis.
Weaver's argument for moral rationalism, in short, is that it is consonant with Kantianism, a leading normative ethical theory. While this is well short of a proof of moral rationalism, it does confer some plausibility on the view. What can a diabolist say in reply? Direct rebuttal of moral rationalism would be a philosophical project so involved as to be beyond the scope of this article, and to require a rebuttal of moral rationalism unduly shifts the burden here onto the diabolist.
Instead, consider a diabolist parody of this Weaverian ploy. Suppose the diabolist assumes immoral rationalism, according to which the recognition that an action is contrary to one's moral obligations just is a practical reason to do that action. When the diabolist is confronted with the claim that diabolist responses to the problem of good are paralleled by theists’ responses to the problem of evil, she can take a page from Weaver, invoke immoral rationalism, and deny even the possibility of the unsurpassably good being who has practical reason to do what is right. Moral reasons, she might say, simply are not the kinds of reasons which can serve as proper practical reasons for actions.
The reply is question-begging, but is it any more question-begging than Weaver's? Both kinds of rationalism hold moral judgements to be intrinsically normative and categorical, though they differ on whether the norm requires a positive response or a negative one. The two kinds of rationalism seem equally vulnerable to the arguments of Philippa Foot (Foot (Reference Foot1972), 312–313), such as that even if moral judgements were hypothetical imperatives, our moral lives would continue on the same way, as long as we (contingently) care (one way or the other) about things like justice and suffering. After all, moral rationalism, unlike internalism, seems not to be supportable (over immoral rationalism) by any evidence from the first-person perspective about the nature of human motivation.
So the evil God challenge can be bolstered against Weaver's objection in two ways. One way is to argue that a maximally evil being need not be a necessarily evil being, as a being who could have been good but is evil (i.e. one who refrains from the good) is even worse than a being who had to be evil. The other way is to argue that a diabolist parody of Weaver's anti-diabolist strategy is available and no less effective.
The objections from Forrest and Weaver are the strongest against the symmetry thesis, and their force against it is only partially deflected. Though they may show some asymmetry in favour of theism, I argue that it is a modest asymmetry, offset by other modest asymmetries that favour the diabolist.
Conclusion
While Law's case for the symmetry thesis is strong, it is somewhat understated. I have explained more symmetries between the theist response to the problem of evil and the diabolist response to the problem of good, and bolstered the case for some of the symmetries discussed by Law. Several recent articles have argued against the symmetry thesis, but I argue here that all of them fail to establish a significant asymmetry. The evil-god challenge remains unmet.Footnote 18
