Introduction
Stephen Law has recently posed the ‘evil-god challenge’. As he explains, ‘the challenge is to explain why the hypothesis that there exists an omnipotent, omniscient and all-good god should be considered significantly more reasonable than the hypothesis that there exists an omnipotent, omniscient and all-evil god’ (Law (Reference Law2010), 353) (henceforth, ‘theism’ and ‘maltheism’, respectively). Law opens the forum for debate by pre-empting one of the more obvious responses: that the good we observe in the world is too difficult to explain if evil-god exists. He argues that a variety of the most compelling theodicies can be paralleled by maltheists. The implication, he argues, is that the problem of good for maltheism and the problem of evil for theism are equally strong. Either the problems are compelling,Footnote 1 in which case both are improbable, or the problems are soluble, but in which case theists still need to explain why we should believe in a good God rather than evil-god.
This thesis has attracted considerable sympathy in the history of philosophy, though it has only recently yielded to formalization by Law, among others. Hume (Reference Hume1779, part X), following Epicurus, asks whether God might be malevolent for being able but unwilling to prevent evil, but did not seriously entertain the idea of an evil-god having evidential parity with its benevolent counterpart. Perhaps this is because the problem of evil was thought to be decisive, rendering an evil-god challenge redundant. But with the advent of more promising theodicies advanced in the last few decades,Footnote 2 and with a renewed interest in theism as a quasi- (or actual) scientific hypothesis (and the theoretical and explanatory virtues pertinent to such discussionsFootnote 3), the prospect of advancing parodical god theories with ostensibly equal evidential support to theism has charmed some philosophers into fleshing out apologias for such parodical gods. Though the evil-god thesis has been posed by various philosophers in recent decades, I shall primarily be concerned with responding to the most recent incarnations and defences of it found in Law (Reference Law2010) and Collins (Reference Collins2019).
In this article I accept the challenge, describing a number of reasons to think that theism is more reasonable than Law's parodical alternative. I leave it an open question whether or not the problems of evil or good are significant problems, since the point of Law's article is to draw the symmetry between the two, and hence my main point is to reject it.
Law and Collins both survey a number of potential asymmetries. Law discusses a variety of theodicies, then briefly discusses the arguments from miracles and religious experience, the ontological and moral arguments, and arguments purporting to show that maltheism is impossible. He also talks very briefly about the subject of this article, the comparative simplicity of the hypotheses.
Collins develops Law's article, discussing a few more theodicies and then defending Law from various criticisms based on pragmatic encroachment and impossibility arguments. He also develops the simplicity considerations a little further.
Thus we may delineate three broad kinds of asymmetry which may give theism an advantage over maltheism: asymmetries between theodicies and demonadicies,Footnote 4 asymmetries between the positive evidence for theism and maltheism, and asymmetries between certain theoretical or conceptual concerns (e.g. simplicity, the natures of good and evil, etc.).
While I will briefly discuss this broader variety of ways in which the theist may have the upper hand, I shall concentrate my attention on the simplicity of the respective theses. Despite the importance and ubiquity of simplicity considerations in theory choice (Baker, Reference Baker2016), there has been a surprising lack of simplicity-based reasoning in response to the evil-god objection. There are some arguments which have claimed that maltheism is impossible, and other arguments that suggest that maltheism does not explain the evidence as well. But there are very few responses suggesting that maltheism is possible, yet still less intrinsically probable than theism.Footnote 5 This article will primarily take this line.
In the rest of this article, I explain the probabilistic apparatus for determining the relative plausibility of theism and maltheism, indicating what it would take for theism to be significantly more plausible. I then argue that there are powerful theoretical considerations in favour of theism arising from powerful asymmetries between theism and maltheism. I first consider asymmetries between evil and good themselves, before looking at a number of asymmetries between theism and maltheism. Specifically, I argue that maltheism itself admits of several varieties, and it is difficult to find a variety which is both simple and parallel to theism. I then argue that maltheism plausibly posits a mixture of positive and negative properties, while theism only posits positive properties, and hence theism is more coherent. I then turn to theories of moral motivation, arguing that plausible theories linking moral motivation with practical reason and action make theism simpler than maltheism. Finally, I suggest that ordinary common-sense epistemic intuitions regarding the basic ‘believability’ of the world support theism more than maltheism.
A final preliminary note: this article makes significant use of intuitions. I do not think this is a weakness of the article, for several reasons. First, and most saliently, most philosophy relies heavily on intuitions, and indeed almost all of our beliefs about anything at all depend on at least one intuition (consider the standard sceptical problems in the philosophy of science). If beliefs based on intuitions are thereby impugned, then virtually all our knowledge is thereby impugned. Second, intuitions are not necessarily nebulous, arational ‘gut instincts’; as Huemer (Reference Huemer2005) has persuasively shown, there are significantly more robust characterizations of ‘intuition’ available, and they can be significantly informed by intellectual reflection, as indeed are the intuitions contained herein. Third, many of the intuitions cited in this article have been defended at greater length elsewhere: given the breadth of this article, it is simply impossible to rehearse all those defences here. Finally, given the sheer number of intuitions which generate reasons to prefer theism to maltheism cited in this article, it is not at all necessary that the reader endorse all the intuitions. Even if the reader shares only one or two of the intuitions I endorse here, these alone may be enough to render theism more plausible than maltheism.
Probabilistic considerations and the power of simplicity
Law and Collins both contend that, by and large, the positive arguments for God's existence can be paralleled by arguments for evil-god's existence. So, for example, the major premise in Bayesian fine-tuning arguments – that there is a moderate probability that God would create embodied life if he existed (in order to allow moral interaction) – can be mirrored by saying that there is a similar probability that evil-god would do the same – perhaps so he can inflict physical torture, or so that people can harm each other. Likewise, they suggest that the various theodicies can be easily paralleled.
While I do not have space here to appraise the validity of such parallels (except for a couple of examples I shall come back to at the end),Footnote 6 I can at least demonstrate what is necessary to render theism more believable than maltheism, and leave most of the discussion of whether theism fares better on individual arguments and theodicies to more comprehensive works.
Assuming belief is a matter of degree which can be measured probabilistically,Footnote 7 all that we need to reject the symmetry thesis is for P(theism|evidence) to exceed P(maltheism|evidence), that is, for the probability of theism to exceed that of maltheism, given the evidence we have. But the greater the ratio between the two, the greater confidence with which theism could be asserted vis-à-vis maltheism. We might say that for a reasonable degree of assertive confidence, the ratio P(theism|evidence)/P(maltheism|evidence) should be at least 2. Assuming this partition were exhaustive (or near-enough – perhaps by the alternatives being rendered extremely improbable by the evidence), P(theism|evidence) would approximate to 0.66 and P(maltheism|evidence) 0.33. But the greater the ratio (i.e. the more probable theism is relative to maltheism), the better for the theist.
P(theism|evidence) = P(theism) x P(evidence|theism)
P(maltheism|evidence) P(maltheism) P(evidence|maltheism)
The actual ratio, as demonstrated above, will be the product of the two factors on the right-hand side. But from this it can be seen that even very small differences between the numerators and denominators can generate the required ratio. For example, if:
P(theism) = 0.002
P(maltheism) = 0.001
P(evidence|theism) = 0.002
P(evidence|maltheism) = 0.001
then P(theism|evidence)/P(maltheism|evidence) = 4, and P(theism|evidence) would approximate to 0.8. So it may be that only a minuscule discrepancy in the prior probability (e.g. from 0.001 to 0.002) or the predictive power of theism over maltheism is sufficient to warrant belief in theism.
Positive arguments for theism usually work by arguing that P(evidence|theism) > P(evidence|atheism). In this case, if God has more reason to bring about some actual state of affairs (or is more likely for some other reason – perhaps if evil-god could not possibly bring about that state of affairs) than evil-god has, then P(evidence|theism) should exceed P(evidence|maltheism). So if one thinks, for example, that although evil-god has some reason to create embodied life (for torture and for moral evil) but also has considerable reason to create nothing at all (because the existence of relational life is an intrinsic good), and that God has a more robust reason to create embodied life (because he is essentially loving),Footnote 8 then this will be a reason to give theism epistemic preference. Mutatis mutandis for other arguments.Footnote 9
There are a few further points to take home from this brief discussion. First, differences (including very small ones) in the intrinsic plausibility or probability of the hypotheses can be very powerful so long as the evidence doesn't swing significantly the other way. If theism and maltheism explain the evidence roughly equally well, but theism is somewhat simpler than maltheism, theism will be overall probable, given the evidence. And it is worth noting just how powerful simplicity considerations can be: as shown in Miller (Reference Miller2016), for any empirical data, one can construct an enormous number of hypotheses explaining that data (some explaining it better than even the best scientific hypothesis). And yet usually the hypothesis we end up choosing is chosen on the grounds of simplicity. But we do not just think that the hypothesis is more probable than each other hypothesis individually: we think it is more probable than all the others put together. So the advantage it has over each individual complex hypothesis is enormous. This suggests that simplicity may have enormous weight probabilistically, such that even a small difference in simplicity may suffice to distinguish two hypotheses by quite some way epistemically.
Second, modelling the debate probabilistically helps to show how even some uncertain considerations favouring theism over maltheism can still be tremendously powerful. It is common for defenders of maltheism to respond to a theist's alleged asymmetry by saying that it is possible that the grounds for that asymmetry do not hold. To take one example which I will not further discuss, a divine command theory of ethics may or may not be plausible. But it is surely more plausible and simple than its relevant converse: that what is good is determined by what is contrary to God's nature.Footnote 10 Or, for an example we will later revisit, Collins says that it is possible that knowledge and power are neutral in themselves, in response to the theist's suggestion that knowledge and power are positive (and thus that theism is simpler than maltheism, because it involves only positive properties). But while we might not be fully convinced that knowledge and power are positive (call this theory ‘pos’), we think that it is at least much more likely that they are positive than that they are negative (call this theory ‘neg’, and that they are morally neutral ‘neut’).Footnote 11 Consider:
P(theism) = P(theism|pos) × P(pos) + P(theism|neg) × P(neg) + P(theism|neut)
× P(neut)
P(maltheism) = P(maltheism|pos) × P(pos) + P(maltheism|neg) × P(neg) + P(maltheism|neut) × P(neut)
Suppose we think that P(pos) is only 0.3. But neg is very implausible, so let us say P(neg) = 0.001. Then all we need to do is set a simplicity advantage – to make P(theism|pos) > P(maltheism|pos) and P(maltheism|neg) > P(theism|neg). Let us suppose that this simplicity advantage is such that P(theism|pos) = P(maltheism|neg) = 0.1, and P(theism|neg) = P(maltheism|pos) = 0.01.Footnote 12 Suppose the probability of each on neut is likewise 0.01.
On these assignments, P(theism) = 0.03702 and P(maltheism) = 0.01011. So theism is considerably more intrinsically probable than maltheism, and will retain this advantage as long as the evidence is equivocal (or favourable to theism). Of course, one might disagree with the assignments (though they are quite modest). But the point is a simple proof that we need not show decisively (or even with probability) that a point of asymmetry is the case: we need not show that knowledge and power are positive in order for that thesis to confer even a significant advantage on theism over maltheism. The asymmetry only needs to be more plausible than the reverse asymmetry. And of course, once these possible asymmetries add up, the cumulative case for theism's plausibility can become quite substantial.Footnote 13
The simplicity of theism over maltheism
As I have indicated, one of the most promising loci of asymmetry is likely to lie in the simplicity of the hypotheses. We might think of simplicity as one of the primary determinants of the prior probability of theism and maltheism.Footnote 14 In this section I briefly describe the importance and elements of simplicity, before applying the discussion to the case of theism and maltheism, looking at both the simplicity of their characterizations of evil, and the simplicity of their characterizations of theism and maltheism.
What is simplicity?
As demonstrated in Miller (Reference Miller2016), some version of a simplicity or parsimony principle is crucial for any broad scientific or general realism.Footnote 15 It has been common to characterize this in terms of Occam's Razor – ‘entities must not be multiplied beyond necessity’. This itself has been divided into various forms: the foremost distinction being between qualitative and quantitative parsimony,Footnote 16 which prioritize a low number of principles/concepts/kinds and a low number of raw entities, respectively. Although qualitative parsimony is more readily accepted among philosophers, quantitative parsimony has its defenders.Footnote 17 Even those who defend quantitative parsimony, however, typically grant that qualitative parsimony is more important.
Aside from this, along with a few other trivial principles (e.g. that P(X&Y) < P(Y)), it has proven difficult to characterize many other constraints on prior probability, especially where simplicity is concerned. Principles of indifference can be applied but (a) only in the rare case that a roughly symmetrical partition can be identified, and (b) even those cases often succumb to Bertrand paradoxes. One of the primary problems of theoretical choice, however – the grue paradox – does not seem to be simpler on any of these grounds. Here Swinburne (Reference Swinburne2004) suggests that more readily observable properties should be preferred (‘green’ being preferable to ‘grue’ since an instantiation of grueness requires more observations to verify).
Some have suggested that syntactic or semantic simplicity, particularly as measured by length of strings in computational models of theories, is our best measure of simplicity.Footnote 18 Less formal versions of these principles suggest that more succinctly formulable theses have greater prior probability – with the assumption that natural language is good at identifying natural kinds. It is possible to posit a further thesis on similar grounds: the relationships between qualitative kinds and properties should be as simple and few as possible.
Finally, I have suggested (Miller, Reference Miller2016) that maximal degrees of properties (and other kinds of completeness) are simpler than large, finite degrees and less complete systems, and that exceptionless and wholly regular laws are simpler than laws with exceptions or irregularities. Uniformity is simpler than heterogeneity.Footnote 19 In that article I argue, in particular, that since theism involves the positing of maximal degrees of power, knowledge, and goodness, it is considerably simpler than a parody positing only large, finite (and therefore exception-admitting) degrees of these properties. Since it is possible that some of these properties follow from others (Swinburne suggests that all follow from omnipotenceFootnote 20 – which, of course, would undermine maltheism all on its own), theism may be even simpler on the grounds of positing even fewer fundamental properties.Footnote 21
Conceptions of evil
This in mind, let us consider how asymmetries between good and evil themselves render theism or maltheism more credible.
It is common to understand evil as the same sort of thing as good, ontologically speaking – just the opposite. One might imagine a scale where moral/axiological neutrality sits in the middle with positive units of the properties ‘good’ or ‘evil’ existing on either side, where both appellations represent the same sort of thing. Law's failure to consider the ontological morphology of good and evil, along with his commitment to the symmetry of god and evil-god, suggest that he probably has this sort of conception in mind.Footnote 22
Turning to accounts of evil simpliciter (which can apply to evil action), we find a history of diverse theorizing about the ontology of evil itself.Footnote 23 Most notably, evil has been seen as a privation of good – not necessarily as the entire absence of good, but a defect, corruption, or slight absence of it. This perhaps implies a simpler ontology: it requires only ‘good’ as a primitive natural kind. Collins points us to Anglin and Goetz's (Reference Anglin and Goetz1982) argument:
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.
Collins parodies this argument by suggesting that good can be conceptualized as the privation of evil and that an exactly parallel argument can be made. But in any case, (3) is not immediately convincing, and Anglin and Goetz fail to motivate it.
It is possible, however, that there are arguments in the vicinity which aid the theist. For given a corruption theory of evil, a maximally evil being will be entirely corruptive of good, but such a being is problematic in two respects. First, it involves positing a positive property and then positing corruption in addition (multiplying kinds and thereby being more complex than theism, which posits only the former), and second, it suggests that there is some temporally or modally prior agent or object which is corrupted: but nothing can be temporally prior to any god, and it is difficult to conceive of an object being modally prior to God.
This is still vulnerable to the possibility of a privation theory of good. But I suspect that a privation theory of evil is more intuitively appealing to most than a privation theory of good. Rape and adultery are corruptions of the good of sex, abuse and slander are corruptions of the good of language, idolatry is the corruption of affection. It is more difficult to see how the goods of sex, language, and affection could be mere ‘corruptions’ of evil. There are other cases where goodness seems conceptually prior: although Collins suggests that health can be conceived of as lacking any infirmity, more primitively it seems as though bodily functions are goods which are compromised by corruptions of that function. It would be odd – and certainly more complex – to conceive of health as the corruption of, or lack of, corruptions of function!Footnote 24 And one might think that the fact that indifference is morally wrong is some further evidence that evil is a privation of good. Suppose someone has no ill will or malice towards someone else – no feelings at all, perhaps – but is utterly indifferent to their immense suffering. This case seems to be an absence of both good and evil, at least in their straightforward forms. And yet it seems that it is clearly wrong. So the absence of evil is not obviously sufficient for something being good. But the absence of goodness does seem to make something evil in this case. And hence on this view of good and evil there is some reason to think that theism is simpler than maltheism. And, importantly, as with all the asymmetries I discuss in this article, one need not have a decisive case for, in this case, the privation view of evil. One only need note that there is some asymmetry between the plausibility of the relevant theses: as long as the privation theory of evil is more plausible than the privation theory of good, theism will be more probable than atheism.
Another possibility is to conceive of evil in the same way Augustine conceives of sin: as disordered love. If wrongdoing consists in misdiagnosing the worth of something or in loving it out of place, it may be either inconsistent with omniscience or with pure, unadulterated evil. Even if not, however, one might think that a being with rightly ordered loves is simpler than one with disordered (or strictly inverted, if the being is to be maximally evil) loves, since uniformity is simpler than heterogeneity.Footnote 25 This may find an interesting correlate in the very earliest religious ideas: Longman and Walton (Reference Longman and Walton2018) describe how in the ancient Near Eastern context in which the Judaeo-Christian tradition first developed, evil was closely related to the ideas of chaos and disorder (represented frequently by the sea), while in Genesis God's good creation was essentially a project of him creating order from disorder. And certainly there is an intuitive pull to the idea that human flourishing consists in our lives and characters being rightly ordered: that order is better than disorder. But if so, and if order can be tied to uniformity, then there are some senses in which the goodness of order and of ordered loves may be simpler than the chaos and disorder of evil.
The simplicity of theism and the complexity of maltheism
Difficulties in characterizing evil-god
There are further problems when we actually try to characterize evil-god.
In the first place, it is not immediately obvious with which evil-god the concept of evil should be attributed. There is ambiguity on a number of levels: we have already discussed different accounts of evil itself, but there may also be different normative accounts of evil-god and his preferences, character, and actions. So even supposing we had narrowed this down to, for example, action, we might still ask whether evil-god would be anti-utilitarian (maximizing pain/suffering and minimizing pleasure), or maximally selfish, or maximally destructive, or maximally irrational, or pluralistically evil, and so on. Presumably the problem of good rules out an anti-utilitarian god in the same way that the problem of evil rules out a utilitarian God.Footnote 26 But the other conceptions are themselves difficult to flesh out in a way that fits the data. Let me consider a couple of examples.
There are a few conceptions of maltheism on which it is far from guaranteed that evil-god would create anything at all. It is not clear that a maximally selfish god would create anything else at all, nor is it clear that a maximally destructive god would create anything (or allow it to continue existing). Nor is it clear that a god who wanted to maximize average evil would create anything at all (indeed, it seems unlikely!). While there is a parallel response to the theist on the last of these, the theistic parallel theory to the first two theories does not seem so easily assailable. So it appears that varieties of theism are, on average, more able to explain the (initial and continued) existence of a world than varieties of maltheism are.
Let us consider in a bit more detail some more specific hypotheses. One of the primary constituents of an evil-god's character must surely be selfishness. But this leads to a number of paradoxes. For an omniscient god would know that it is in one's own best interests to enjoy relationships with others. But then he would be in many or most ways indistinguishable from God (by always following the good or pursuing healthy relationships, even if perhaps for the wrong reasons). This is not to say that God is wholly self-interested or that self-interest is not frequently immoral – but genuine, knowledge-suffused self-interest devoid of irrationality will at many times have material consequences similar to genuine benevolence. And although this would not make evil-god wholly good, it is hard to see how he would thereby be maximally evil, as maltheism requires.
But an even greater difficulty arises if we think that it is in one's best interests to be moral.Footnote 27 There is considerable plausibility to this. It is difficult to work out a plausible account of personal flourishing that doesn't include the possession of certain moral virtues. And it is overwhelmingly implausible that human flourishing consists in possessing certain vices. But if personal flourishing involves possessing virtues, then a maximally selfish god would want to possess virtues. And this seems impossible.
One might think that maybe God just does what is bad for himself, and so is not really selfish. But then he lacks one of the primary constituents of evil, and in any case is probably irrational. As I will argue later, there is a certain complexity in positing a practically irrational omniscient being.
Along similar lines, one might think that a maximally hateful God would hate himself – but then he would probably want the worst for himself, which would be to cease existing, or to have his desires frustrated. But if his desires were good, then it is hard to see how he could be maximally evil, and if his desires were malicious, the frustration of them would presumably involve acting for good. And note that the theistic parallel is far from as problematic: according to Christianity, for example, God is all loving and loves himself. But with God's love there is no such conflict or complexity.
One might also think that hating oneself or wanting the worst for oneself is irrational (and perhaps impossible). If so, we again have the problem of a practically irrational omniscient being (for which, see ‘Evil-god and rationality’).
I shall return to some of these suggestions later, particularly examining the concept of rationality and the implications of conceiving of evil-god as irrational. But we can see that it is hard to parse evil action in a way that renders maltheism both coherent and simple. This is one reason for giving theism a greater prior probability than maltheism. But a proper assessment would require maltheism to be fleshed out fully first.
Positive and negative properties
Law himself suggests a further way in which theism might be simpler than maltheism. He writes: ‘a good god can be defined in a simple way, e.g. as possessing every positive attribute . . . The concept of an evil-god, by contrast, is more complex, for he possesses both positive attributes (omniscience and omnipotence) and negative attributes (evil)’ (Law (Reference Law2010), 372). Yet his response is relatively vacant: he simply acknowledges the asymmetry and says that this is irrelevant when both options are improbable. But he gives no reason to think that either option is improbable, other than the problems of evil and good. But suppose – as I do – that the problem of evil is uncompelling and the problem of good nearly thus. Then Law has no reason to reject this asymmetry as either spurious or irrelevant.
Collins improves on this suggestion, however, by suggesting – following Kant – that the only unconditional good is a good will – and consequently that omnipotence and omniscience are intrinsically neutral. Though they may be goods in the context of theism, they may also contribute to evil-god's evilness.
This is certainly more credible than Law's response, but it is still far from decisive. For one thing, it is extremely intuitive – particularly to those who hold to one of the asymmetrical conceptions of evil I identified above – that goods could be used in the service of evil. Indeed, this has been relatively standard within the Christian tradition: the abuse of sex results in betrayal, breakdown of relationships, and rape; the abuse of language results in judgement and slander; and the abuse of affection results in idolatry. Even acts of kindness can be used in the service of self-righteousness and self-salvation. If so, then we have a ready explanation of how things which are intrinsically good can contribute to evil when exploited.
I am not suggesting that all these are uncontroversially intrinsically good. But it is certainly intuitive to think that at least one of sex, language, affection, and acts of kindness is intrinsically good. And it is very implausible that goods cannot be used in the service of evil. So the fact that evil-god being omniscient and omnipotent would make him capable of greater evil should bear no evidential weight for the thesis that they are intrinsically neutral or evil. But then it is hard to see what evidence there is in favour of it.
But this leads us to the second point: since there is no evidence that knowledge and power are not intrinsically good, we remain with conceptual analysis and intuition. But these seem to lean in favour of knowledge and power being intrinsically good. For it is plausible that the scientific thirst for knowledge is good in and of itself, and that ignorance is bad.Footnote 28 It is also plausible that aiming at truth is intrinsically and not just instrumentally good. Indeed, this seems to be one of the key reasons why deceit is wrong, even when relatively innocuous.Footnote 29 And presumably it is part of the reason why most of us would prefer to be slightly worse off and know the truth, as our intuitions about films like The Truman Show and The Matrix, and thought experiments like Nozick's experience machine,Footnote 30 suggest. Truth and knowledge are sufficiently intimately related that they both carry similar evaluative weight.
Moreover, even if one does not think that knowledge is intrinsically good, it seems even more plausible, at least, that rationality is intrinsically good. Rationality is a matter of following epistemic norms and believing what one ought to believe. Even if rationality is not a moral norm (though insofar as it is in our control, one might very reasonably think it is), it still seems clear that following epistemic norms is good and disregarding them bad. But an omniscient being presumably follows all epistemic norms, for they know all the facts, all the laws of logic, all material conditionals, and so on. So if omniscience entails rationality, and rationality is an intrinsic good, then maltheism mixes a positive property (rationality) with a negative one (malevolence) and hence is more complex than theism. We will discuss more ways in which the concept of rationality is significant in the next section.
Similarly, there are reasons to think that power is intrinsically good – for one, there are arguments that omnipotence itself entails goodness, since such an agent would follow only reasons (and thereby morality, since what is moral is what one has most reason to do) and would not be causally affected by irrational desires, for example.Footnote 31
Alternatively, one might take a leaf from perfect being theology and suggest that greatness is primitive – then we need only posit one natural kind, from which the derivative kinds of omnibenevolence, omniscience, and (most saliently here) omnipotence follow. Even if omnipotence is morally neutral, it seems clear that a maximally great being would have it. And this seems simpler than any parallel maltheistic concept.
If even one of these is intrinsically good, then theism is simpler than maltheism. And we do tend to think that power, knowledge, and rationality are good – and that is why it is especially bad when they are abused. If so, then maltheism combines positive and negative properties, while theism combines only good properties, and so is more uniform and hence simpler. But even if we have no compelling argument that one is intrinsically good, it is surely more probable that they are – or at least one of them is – intrinsically good than that they are intrinsically bad. And this itself creates an epistemic asymmetry, which is all that is required to justify belief in theism over maltheism.
Evil-god and rationality
There are a number of theses connecting moral facts or beliefs with practical reasons for action and, in turn, with moral motivation and desires. The number of theses and the fact that many of them are given the same or similar names make the discussion a complex one. And several authorsFootnote 32 have developed arguments based on these kinds of positions to the conclusion that a God cannot be evil. There is no need to rehearse the exact arguments here, but we note that if they succeed, or even if they are more likely to succeed than their evil-god parallels (as is surely the case), then theism will be more probable than maltheism, ceteris paribus.
But we can give an indication of how other such arguments – which may not argue for the impossibility of maltheism but only the relative complexity of it – work by looking at the kinds of connections there might be between moral facts or judgements, reasons to act, and desires to act. Our discussion is simplified to some extent by the fact that for an omniscient being, one's moral judgements will be the moral facts.Footnote 33 But when collapsing these two, it is important to bear in mind that either moral facts or moral judgements may generate reasons or desires. So if even one of them does the job, that will aid the theist. The maltheist has to hold the stronger thesis that neither of them stand in the relevant relation to reasons or desires. So what sorts of relations might there be?
1. Moral facts generate reasons to act.
2. Moral judgements generate reasons to act.
3. Moral judgements generate motivations or desires to act.Footnote 34
These do not need to be relations of necessity: Weaver (Reference Weaver2015) has shown that even weak versions of these theses can generate the impossibility of certain kinds of evil-god. But the necessity relations are not in themselves implausible. At the very least, they are orders of magnitude more probable than their converses: that moral facts necessarily generate reasons to act against the moral order, etc. An easy way to see this is that if we ask someone why they did something and they respond, ‘because it was good to do so’, we could reasonably leave it that. But the response ‘because it was bad to do so’ appears considerably less intelligible. That may be all the theist needs to show that theism is more probable than maltheism.
To make a comprehensive case for any of these theses would take us too far afield: each thesis has its own substantive literature. If both (1) and (2) are false, it is hard to see how we could have any objective reasons for acting morally at all. But most of us would, I think, want to say that we have objective reason to act morally. Moreover, when we reason morally with people to try to influence their behaviour, we presumably think that by giving an argument that something is immoral we will thereby give them a reason not to do it. And if they do not appreciate that reason at all, despite conceding the action's immorality, we would be likely to think that there is something wrong with that person – pathological, perhaps.
Along similar lines, if moral reasons are only practical reasons when there is a desire to act morally, then it seems that the only practical reasons are desires. But most of us would not say that someone is acting irrationally if they act morally when their only desires are to act immorally. This suggests that there are objective reasons for acting morally. If so, then probably (1) or (2) is true (or both).
In support of (2) and (3), there are conceptual arguments: perhaps what it is to say that an action is good is to say that the reasons in favour of it outweigh the reasons against it.Footnote 35 Or, to look at it another way, we might think that someone who judges that something is wrong but sees no reason not to do that thing, or has no desire whatsoever to do that thing, has simply not understood what it means to say that something is wrong. The propositions are certainly inherently plausible, and are much weaker than the less plausible Socratic thesis that no one ever knowingly does wrong.
A further conceptual argument is available when considering what it means to have a reason to do something. No doubt this is somewhat controversial, but it is possible that to say that we have reason to do something just is to say that, in some sense, we ought to do it. If I have compelling reason to accept an argument, then I ought to accept the argument. Likewise, if I ought to accept the argument, then I have a reason to accept the argument. One might even be able to characterize moral beliefs this way: ‘I ought to φ’. If so, then the connection to practical reason seems plausible enough.
To reiterate: I cannot go into the complex discussion surrounding these theses, but there are weak versions available which are, at the very least, much more plausible than the maltheistic converses.
But if any of these are true then maltheism is more complex than theism in a variety of ways. Why? For evil-god, all moral facts are moral judgements of his. So if he is omniscient and knows all moral facts, then in choosing to perform evil, he must be acting in perfect contradiction to many of the practical reasons he has or is acting in perfect contradiction to many of the desires he has (or both).
Why is this problematic for the maltheist? First, as Swinburne argues, if irrational, immoral desires have to be posited in addition to moral judgements to explain evil-god's behaviour, there is an ontological addition. This is exacerbated if the immoral desires have to be ubiquitous and overwhelming. Second, evil-god would have conflicting reasons or desires, some of which were good and some evil, which seems more complex than having only reasons and desires to act morally. Third, if (3) is true, evil-god would have some good desires, which would impugn his evilness. Fourth, if (1) or (2) are true, it would make evil-god practically irrational but epistemically rational – and hence a complex mixture of positive and negative rational properties. Fifth, if someone is so wholly irrational and controlled by overwhelming evil desires as evil-god must be, this may impugn evil-god's omnipotence. Sixth, if someone is wholly irrational, it is far from clear that we could predict only evil desires and consequences. Humans who suffer from severe thought disorder do not only act in evil ways – they act erratically and in a variety of ways. But that would not characterize evil-god well. Seventh, it is not clear how a maximally irrational evil-god could explain the order of the universe.
There are a number of reasons, therefore, why theses connecting moral facts or judgements with practical reasons or desires to act would make maltheism more complex than theism. If even one of these reasons succeeds, that will be enough. And note that we have not had to commit ourselves to a very strong connection here. If there is any theoretical connection between these concepts – as we have briefly argued there is – then theism is more probable than maltheism, ceteris paribus.
Phenomenal conservatism and anti-scepticism
The final respect in which maltheism is more complex than theism is in the epistemic commitments of each. It seems simpler to suppose that the world is fundamentally more reliable and truthful than not. Hence most people are not sceptics, Cartesian diabolists, and so on, even though the empirical evidence is underdetermining. Why do most people reject solipsism? Why do they reject brain in a vat theories? Why do they reject Descartes's evil demon? Presumably because we think that it is simpler to suppose that appearances are a reasonable guide to reality. Likewise in the perhaps more directly analogous moral case: most people are moral realists and optimists about moral knowledge even without any account of why their moral knowledge would be reliable, and even when they believe that their moral knowledge is determined purely by evolution, neurophysiology, and so on – that is, causes which are surely not related to moral facts. And yet most of us think that we are not unreasonable to be optimists about moral knowledge. In the same way, most people would put a greater credence in theism than maltheism, even if they are not theists, and it seems simpler to go with that appearance.
Religious experience also confirms this impression. For although people of different religions have religious experiences, these experiences are rarely so specific as to confirm particular doctrines with each religion. And they overwhelmingly confirm the intuition that God is good. It is extremely rare that people have a religious experience of God as perfectly evil.
But there is an extra consideration open to theists of certain stripes. Christians, for example, hold that God revealed himself through the incarnation in Jesus of Nazareth. For many Christians, this is in fact the best evidence they have of their belief in God, and they would refer to it as one of the primary reasons for their beliefs. And most Christians think that the evidence for God revealing himself in Jesus far surpasses the evidence for any other religion. But then they are essentially confronted with the message that God is essentially good. For the opposite to be the case, the convincing evidence for Jesus’ message (from a Christian's perspective) would have to be essentially one grand hoax on the part of evil-god. But this would violate certain principles of testimony and phenomenal conservatism. So it is simpler to suppose that Jesus was telling the truth.
There are a couple of points worth mentioning here: first, this preference for simplicity is not merely instrumental. Rather, we really do think (most of us, at least) that a kind of naïve realism is true, even though the evidence underdetermines the theories. Second, this simplicity is difficult to characterize: it is probably not mere parsimony, in the sense of positing fewer entities or kinds of entities. In fact, the precise reason for preferring realist theories over these sceptical theories is difficult to work out, and hence these problems still remain in philosophy! But, in fact, we need not delineate the precise constituent of simplicity at work here. All we need to note is that there is something which makes belief in realism justifiable vis-à-vis sceptical hypotheses, and that the dispute between theism and maltheism appears superficially to be structurally parallel to, for example, the dispute between realism and Cartesian diabolism. In both cases, the sceptical hypothesis can be made to account for all the evidence we observe. But our intuition – which we are reasonably entitled to believe – is that realism is more plausible than that we are all subject to a ‘grand deception’. Likewise we are entitled to believe the ‘realist’ hypothesis of theism, since theism seems to many people more intuitively plausible, and is the prima facie simplest interpretation of most religious experience and, arguably, the ministries of certain religious figures. There is more that could be said about this. But for now I leave it as a challenge: if there is a relevant disanalogy, it behoves the maltheist to identify one.
Of course, one might object – and some evil-god theorists do object – that there are competing religions, so that some people are being tricked either way. But this depends on the claim that there are roughly equally plausible competing religious claims. And maybe evil-god theorists think that all the main religions are equally plausible. But most Christians, at least, think that this is not so.Footnote 36 So if the evil-god theorist wants to argue that the Christian is being unreasonable, they will have to actually argue that other religions are equally plausible. And it is difficult to see how they could do so.
Conclusion
Although I have not been able to comprehensively examine the asymmetries between theism and maltheism, I have given a sense of the direction in which that study should go. I have specified conditions under which natural theological arguments may confirm theism over maltheism and suggested that the most promising source of asymmetry is in conceptual analysis, particularly with respect to the parsimony of the hypotheses. I have offered a number of reasons for thinking that theism is simpler. These include the difficulty of coming up with a simple and coherent conception of a perfectly evil-god, the mixture of positive and negative properties that an evil-god would have to have, the various problems of practical irrationality that evil-god would have, and the fact that theism fits better with our basic intuitions and evidence of God through revealed theology.Footnote 37
