In defence of Rebel Angel Theodicy

Abstract

Rebel Angel Theodicy – often called Satan Theodicy – is the thesis that horrendous evils are directly or indirectly caused by angels who disobeyed God. In this article, I defend it, developing Gary Emberger’s suggestion that they influenced the course of evolution. After defending speculative theodicy, I expound Rebel Angel Theodicy and reply to seven objections that explicate the widespread judgement of implausibility:

1. 1. That the existence of angels is metaphysically problematic.

2. 2. That God has no good reason to create angels.

3. 3. That angels have no power to harm human beings.

4. 4. That God, foreknowing the possibility of rebellion, would not delegate to angels the power to guide evolution.

5. 5. That even if there was a good reason for God to delegate this power to angels it is metaphysically impossible for an omnipotent God to do so.

6. 6. That God, knowing of the angels’ rebellion, would subsequently intervene to put evolution back onto the preferred divine plan.

7. 7. That there is no plausible motive for angels to rebel.

Rebel Angel Theodicy (RATh) is the thesis, often called Satan Theodicy, that ‘horrendous’ evils are directly or indirectly caused by angels who disobeyed God. It was taken seriously by Terence Penelhum (1971) and recently defended by Gregory Boyd (2001) and by Emberger (2022). It is, however, widely considered implausible, being criticized for instance by Richard Swinburne (1978) by Robert Adams (1985) and by Kent Dunnington (2018).¹

Boyd relates how prior to Augustine RATh was accepted by many theologians, noting the teaching of Athenagoras and of Origen (Boyd 2001: 294). But, as Boyd admits, a scientific understanding of diseases, earthquakes, and violent weather undercuts the thesis of direct intervention by rebel angels (Boyd 2001, 296). It is for that reason that Penelhum had asserted that ‘to cause natural evils with a scientific explanation, Satan would have to be able to help determine the character of the natural laws’ (1971, 237). Although this radical theodicy could be defended, I develop Emberger’s suggestion that it is not the laws of nature as such but the course of evolution that has been affected by the activity of rebel angels.² One version might be that the obedient angels oversee evolution but have to struggle against rebel interference. The version I defend, though, is based on a divine decision to delegate to angels the course of evolution, taking the risk that those angels would rebel. This version is distinctive in that the proposed motive excuses the rebels, who were not evil. (In the final section I speculate how rebel angels might have subsequently become evil.) Positing rebels who are not evil is, I concede, contrary to tradition, as shown by the way Origen was widely condemned because he was said to have allowed the possibility of Satan’s salvation. Nonetheless, my proposal is closer to the tradition than the rejection by sophisticated theists of the whole idea of angelic rebellion.³

RATh is a species of Free Will Theodicy motivated by two gaps in the explanation of evil in terms of purely human freewill, namely animal suffering and human moral weakness. After defending speculative theodicy, I expound RATh and reply to seven objections that explicate the widespread judgement of implausibility:

1. 1. That the existence of angels is metaphysically problematic. This includes Swinburne’s objection that RATh is ad hoc, as well as the Ockhamist objection that we should be reluctant to posit angels because they are neither of the same kind as us nor of the same kind as God.⁴

2. 2. That God has no good reason to create angels.

3. 3. That angels have no power to harm human beings. This includes Dunnington’s critique.

4. 4. That God, foreknowing the possibility of rebellion, would not delegate to angels the power to guide evolution.

5. 5. That even if there was a good reason for God to delegate this power to angels it is metaphysically impossible for an omnipotent God to do so.⁵

6. 6. That God, knowing of the angels’ rebellion, would subsequently intervene to put evolution back onto the preferred divine plan.

7. 7. That there is no plausible motive for angels to rebel, because it is absurd to posit angels who do not love God.

Terminology

Horrendous evils

I use Marilyn McCord Adams’ phrase ‘horrendous evil’ (1999) to refer to the excess of natural evil and suffering beyond any which a loving God would inflict upon us for own good (‘medicine for the soul’ as Aquinas put it), and beyond any that was a predictable and acceptable side-effect of human freedom.⁶ It seems evident to many of us that there is much excess suffering.⁷ The excess of moral evil is more subtle, however, consisting of the ways in which our freedom is constrained by the ease with which we give in to temptation and by our confusion when thinking about what we should do.⁸ To be sure, a degree of moral weakness is an acceptable aspect of human nature, and maybe an inevitable consequence of material embodiment, but the excess hampers the very freedom that theodicists prize.

Rebels

There are three reasons for referring to rebel angels even though there might just be the one rebel, Lucifer, alias Satan, who appears in the Bible. First, use of the plural avoids awkward pronouns. Second, Satan is often thought of as enjoying our sin and suffering, whereas the rebels might either not care or even have some compassion. Finally, Satan tests and tempts individuals, notably Jesus. Such demons of temptation are not the topic of this article, and, if they exist, only relevant in that their power to tempt us is itself part of the excess of moral evil caused by the rebels, whom I shall hypothesize to be concerned not with individuals so much as with a collective, humanity, or as I shall call us, the hominids.

Hominids and humanoids

The word ‘hominid’ is used instead of the phrase ‘human being’ because the horrendous evils may well precede the evolution of homo sapiens. I use a different term ‘humanoid’ to mean any organism capable of critical reflection and morally significant free choices, and which has powers comparable to ours, based either on brains or on complex material brain-analogs. Maybe the term could be extended to include AI, but that need not concern us here.

Angels

Angels are not humanoids because they do not have bodies made of ordinary matter, although they may be thought of as physical. In popular imagination, angels, like ‘souls’, are made of ghost-stuff, ours being like fog, the good angels like sun-bright mist, and the bad ones smokey. Ignoring that popular representation, they differ from humanoids in that angels do not depend on an Earth-like planet, they pre-existed our evolution, and they have some power over what happens on Earth, notably (mis-)guiding evolution.

In addition, I explicate the traditional idea that angels are immaterial as asserting that, unlike humanoids, they are directly aware of what they perceive and directly act on it. This contrasts with the way humanoids depend on brains or brain-analogs, because we perceive and act on things, including our own bodies, only via neural processes that represent them.

Theodicies

By a theodicy I mean a hypothesis Thdthat when conjoined with the creation thesis, Ctn, provides a hypothetical explanation for Evl, namely the occurrence of evils of the types and abundance we know of. The thesis Ctn states that God exists and caused there to be other agents, such as humanoids.

I stipulate that God, with an upper case ‘G’, is a being worthy of worship and, if that is not implied, I further stipulate that God loves what is known to exist and so seeks the flourishing of what is known to be actual. This may be contrasted with a utilitarian god, namely one who intends an exceedingly good outcome, perhaps one than which there is none better, but for which the flourishing of that which is already known to exist might be sacrificed. Reconciling horrendous evils with a utilitarian god might be as easy as noting that such a god could allow aesthetic values, including the drama of humanoids’ use and abuse of freedom, to outweigh immense suffering. I assume readers share my reaction that horrendous evils would then point to a horrendous god. If readers would like a quick summary of RATh it is that one or more utilitarian gods – of limited power – do exist, but they are creatures, namely the rebel angels. Be that as it may, I take the task of theodicy to be that of explaining how a loving God permits horrendous evils.

This stipulated distinction between a loving God and a utilitarian god is complicated by divine foreknowledge. I assume it is as if God lacks foreknowledge of what is neither determined nor predestined, because either God does not have foreknowledge or cannot, on pain of circular practical reasoning, use it when deciding what to do.

Neither naturalism, Nat, nor Ctn&Thd need to provide completely detailed explanations, but the requirement for a theodicy to succeed is that it is not significantly worse than naturalism as a hypothetical explanation. I stipulate that only those hypotheses that succeed in this sense count as genuine theodicies.

To illustrate this, let us make the Bayesian idealization of our reasoning and suppose the theodicists’ aim is to provide a hypothesis Thd such that Prob(Evl|Ctn&Thd) = Prob(Evl|Nat), ignoring both the ambitious project of showing Evl is more likely given the theodicy Thd and the concession that it might be a little less likely.

Note 1: These probabilities are relative to various background beliefs shared by naturalists and theists but excluding the description of evils Evl.

Note 2: The requirement that the hypothetical explanation succeed was explicated as Prob(Evl|Ctn&Thd) = Prob(Evl|Nat), but I do not require that Prob(Ctn|Thd) ≥ ½. If Prob(Ctn|Thd) < ½ then the theodicy Thd should be considered speculative.

The above statement of what I mean by a theodicy assumes a debate between naturalists and theists. The reason for this restriction of atheism to naturalism is that we have some intuitive grasp of how probable Evl is relative to Nat, namely thar the naturalist could not have predicted them in detail but nonetheless does not find them too puzzling. The probability of Evl relative to non-naturalist non-theistic hypotheses or to a god unworthy of worship is, however, inscrutable. Readers who disagree may replace Nat by not-Ctn.

I have stipulated that theodicies are about successful as hypothetical explanations as naturalism. An example of a failure in this regard would be a hypothesis BLk that explains how God took a very small risk of horrendous evils, where BLk treats our state as just bad luck. In that case Prob(Evl|Ctn&BLk) is a small percentage of Prob(Evl|Nat).

One type of criticism of a proposed theodicy Thd, then, is that it fails as a hypothetical explanation and, hence, is not a genuine theodicy in the stipulated sense. The other is that Prob(Thd|Ctn) is too low. Perhaps the greatest difficulty for theodicists is that a genuine theodicy requires some detail, but detail tends to reduce its probability. For instance, RATh is a more detailed version of FreeTh, the Free Will theodicy that much horrendous evil is the result of creatures’ wrong-doing. Now, Prob(FreeTh|Ctn) is fairly high provided we grant that if God intervenes too soon or too often then that would both reduce the significance of creaturely freedom and lessen the opportunity for our individual and collective repentance. (That proviso is a matter of further debate, which is beyond the scope of this article.) FreeTh is not a genuine theodicy, though, if it leaves some kinds of evil improbable, that is, if Prob(Evl|FreeTh) is too low. If we add the detail required to provide a genuine theodicy, as in RATh, then the theodicy becomes somewhat speculative. Just how speculative depends on how well we can respond to various objections.

In defence of speculative theodicy

If the Argument from Evil is analyzed as a deductively valid argument from various premises to the conclusion that God does not exist, then, as Alvin Plantinga pointed out, the argument is defeated merely by providing any coherent hypothesis on which the premises are all true but God does exist (e.g., Plantinga 1974). He calls such hypotheses defences and contrasts them with those theodicies that aim to explain why a good, loving God allows so many grievous evils.

A common reaction, however, to the failure of a deductively valid argument from evil is to propose a probabilistic argument (Tooley 2021, §§1, 2, 3). To compare theism with naturalism, therefore, we may consider the quotient of probabilities q = Prob(Evl|Ctn) ÷Prob(Evl|Nat). Using the Bayesian idealization of our reasoning, Prob(Ctn|Evl))÷ Prob(Nat|Evl) = q × Prob(Ctn)÷ Prob(Nat). That is, if, bracketing off our knowledge of evils, Ctn was k times as likely as Nat, then, taking these evils inro consideration, Ctn is q × k times as likely as Nat. The theodicists’ project, therefore, is to show that q is not too low, ideally that q ≥ 1.

RATh is one version of FreeTh.⁹ But there are other versions, notably Original Sin Theodicy. Ignoring mythical details, it amounts to Peter van Inwagen’s story of ancestral hominids abusing their freedom with ‘horrific’ results (Van Inwagen 2006, 86–87). Even if we believe FreeTh to be true, we should be reluctant to believe either RATh or Original Sin Theodicy, or for that matter some Genesis-inspired combination. Nor does van Inwagen ask us to. All he requires is that the theodicy be an epistemic possibility: ‘Given that God exists, the rest of the story might well be true. I can’t see any reason to rule it out’ (van Inwagen: 2006, 66.) This raises the question of just how probable a hypothesis must be for it to count as epistemically possible. I would use the legal criterion that it has not been shown to be ‘guilty’ (i.e., false) beyond reasonable doubt. A speculative theodicy that fails to meet that standard is merely a Plantingan defense, undercutting the Deductive Argument from Evil. I grant that for that purpose there is no need to multiply speculations, but to undercut the Probabilistic Argument from Evil, even partially, we require at least an epistemic possibility. It is worth multiplying such speculative theodicies because the aim is to show that the disjunction of speculative theodicies, including those we have not thought of, is probable relative to Ctn.

Using the Bayesian idealization, consider a range of pairwise incompatible theodicies Thd₁, Thd₂ etc. Then:

Prob(Evl|Ctn) ≥ Prob(Evl&Thd₁|Ctn) + Prob(Evl&Thd₂||Ctn) + etc. =

Prob(Evl|Ctn&Thd₁) × Prob(Thd₁|Ctn) + Prob(Evl|Ctn&Thd₂) × Prob(Thd₂|Ctn) + etc.

Given the idealization that for any theodicy Thd_j, Prob(Evl|Ctn&Thd_j) = Prob(Evl|Nat), it follows that:

q ≥ Prob(Thd₁|Ctn) + Prob(Thd₂|Ctn) + etc.

Among the theodicies Thd₁, Thd₂, Thd₃, and so on, we should include all those versions of and alternatives to FreeTh that we have not thought of. It could be suggested these alternatives contribute enough to the sum Prob(Thd₁|Ctn) + Prob(Thd₂|Ctn) + etc. to make speculative theodicy irrelevant. But that requires the skeptical theists’ thesis that the detailed divine plans for creatures are ‘beyond our ken’.¹⁰ My response is that we need to distinguish the plans of God, who is loving, from those of a powerful utilitarian, a rebel angel perhaps, who is prepared to sacrifice the flourishing of some for the overall good. The details of this overall good probably concern the future, and maybe what is far away. So, they are not the sort of thing that we should expect to know. But a loving God is concerned with, among other creatures, the actual hominids who are suffering, and we have a good understanding of the conditions of their flourishing, even given belief in an after-life. The probability of some relevant details of the divine plan in this case is not high but inscrutable, in the sense that we have no way of assigning even an approximate value to it. Now, if we thought theodicy was merely intended to remove the intellectual scruples of those otherwise committed to a loving God, then the inscrutable probability of unknown theodicies would suffice. But for those still seeking the truth, however, reliance on Skeptical Theism would result in the provisional agnostic judgement that the probability of theism is itself inscrutable. For them, the theodicies we have formulated support theism over agnosticism.

Two reasons for dissatisfaction with hominid freedom as an explanation of horrendous evils

Horrendous evil is older than the hominids

The abuse of freedom by our ancestors may well have multiplied horrendous evils, but there are two reasons why we should extend Free Will Theodicy to include the activity of rebel angels. The first is that horrendous evil would seem to precede ancestral sin. Even those skeptical of the degree to which non-hominid animals suffer should grant that near-humanoid hominids must have had nervous systems so like ours that they would suffer extreme and dysfunctional pain. Otherwise, it would have to be hypothesized that the evolution of hominids over the last million years has resulted in greater sensitivity to pain. It seems more plausible that violence would select for those who feel less pain, not for those who feel more. It is likely, therefore, that hominids suffered horrendously before they were fully humanoid, something explained by rebel interference millions of years prior to the evolution of animals subject to severe and dysfunctional pain.

The other reason for dissatisfaction with hominid versions of Free Will theodicy is that horrendous evils are not required to achieve the divine purpose of us making free decisions with grave consequences that we can come to know of and repent. For this purpose, it suffices that we can harm each other in ways that are serious enough to warrant reflection. For normal human beings, although selfish and weak, have sufficient fellow-feeling not callously to trivialize the moderate harms we inflict on others, on the grounds that these are good for them, correcting sinful tendencies (‘medicine of the soul’), providing them with obstacles to overcome, and so on. God would only allow agents to inflict horrendous evils so they could appreciate the gravity of their decisions if those agents lacked this fellow-feeling. That there are human psychopaths who indeed lack such feeling is itself a horrendous evil and so part of the problem rather than a solution.

If it is replied that in fact the risk of horrendous evils was rather low, then my rejoinder would be that this is BLk, which I have rejected. RATh provides a better reply, because, I claim, lesser evils would not be enough to ensure the gravity of the choice made by rebel angels. This claim will be supported when their motive is explained below. Here it suffices to note that we would not suppose angels to have a sense of fellowship with humans prior to a decision to obey God, who plans a community of all creatures capable of love. Hence angels might well inflict suffering on us for our long-term good.

We are considering, then, the hypothesis that there are angels with the power to influence evolution on Earth-like planets. Assuming adequate replies to objections to RATh, that hypothesis is sufficiently probable relative to Ctn for RATh to be a significant way of providing FreeTh with greater detail.

The metaphysics of angels

If angels are metaphysically possible, then we should expect that God will create them in addition to humanoids. For although there might be no great value in creating a larger number of humanoids it is good to create all possible kinds, at least at a coarse level of classification into kinds.¹¹ Hence God would create angels if they are possible. This answers the second of the list of six objections. I now argue for their possibility, which answers the first.

Angels are conceivable and conceivability is a pro tanto case for metaphysical possibility: Hume’s Razor tells us not to multiply necessities (Forrest 2024). There is, however, a serious objection to that pro tanto case, one which starts from the metaphysics of humanoids. For even if each one of us had a simple immaterial soul, it would not, I say, be a res cogitans. Hence any post-mortem soul would be in an altered state of consciousness: Brahman or Nirvana maybe, or an unstructured beatific vision. The reason for rejecting that res cogitans notion of a psychologically complex soul is that it fails to cohere with neurophysiology. For instance, brain injury often results in personality changes, not just changes to our capacity to remember and communicate with others (e.g., Golden and Golden 2003). This sets up the problem of angelic metaphysics, for I am supposing angels not to be embodied in some complexities that extend across vast regions of space, as in some sci-fi scenario. Instead, I assume a more traditional idea of angels as immaterial and lacking complex structure. The humanoid dependence on such complexity prompts, therefore, the question, ‘How are angels possible?’

Ad hoc and anti-Ockham?

Any difficulty in answering the ‘How are angels possible?’ question would show the ad hoc character of RATh, something Swinburne complains about (1978, 296). Now, to be worthy of the same exalted status as science, a hypothesis should satisfy three criteria. RATh’s partial satisfaction of the criteria suffices to show it has a status somewhere between the genuinely scientific and the ad hoc. The criteria are:

1. 1. It should not be too improbable given the rest of science

2. 2. It should explain several different otherwise puzzling beliefs, and

3. 3. It should have direct empirical support.

RATh satisfies an analog of (1) provided we can answer the ‘How are angels possible?’ question. It satisfies (2) quite well, explaining natural evils as well as naturalism, I say, and offering an explanation of our moral weakness as well as would Original Sin Theodicy. In addition, it preserves something of the tradition of evil angels, which is found in Judaism, Islam, and Christianity. The empirical support is, I concede, weaker. For it is not so much of the rebels but of angels generally and is anecdotal (see Heathcote-Jones 2009).¹²

On the hypothesis that I shall state, it is human consciousness and power that is complicated, with the angelic case being more straightforward. Angels who had no experience of us could complain that positing material humanoids was ad hoc. Hence the Ockhamist objection will fail.

How angels are possible

The objection to the metaphysical possibility of angels relies on the Representation Theory of Consciousness, as I call it, namely the thesis that beings can only be conscious if they have complex internal structures that represent that of which they are aware and on which they act. This should be rejected as a parochial extrapolation from an ontological nook, because one straightforward explanation of representation is that it is indirect awareness and agency, with the conscious agent being directly aware of, and directly affecting, that which represents.¹³ Hence, conscious agents may be classified into three kinds:

1. 1. That which is not restricted in space-time

2. 2. Those that although having a restricted location know and act directly rather than via representations, and

3. 3. Those who are not merely restricted in space and time but further restricted by operating via representations, that, as far as we know, require complex material structures such as brains.

These three kinds are the divine, the angelic, and the humanoid, respectively.

That there are these three kinds, not just two (the divine and the humanoid) is not, I say, open to any Ockhamist objection. For if we appeal to simplicity or elegance when choosing between theories, then the application to the more fundamental trumps that to the less fundamental; and I have provided a more fundamental, unifying, theory of persons.

The objection to the pro tanto argument for the metaphysical possibility of angels has been explicated as the ontologically parochial Representation Theory of Consciousness. The defeat of the objection restores the pro tanto argument from conceivability to possibility. Such arguments are strengthened by giving more detail (see Chalmers 2010, Ch 6). Here I rely on the thesis of the ubiquity of consciousness, as in Absolute Idealism (Bradley 1897) or in Panpsychism. (Chalmers 2015; Goff, Seager, and Allen-Hermanson 2022). For the sake of definiteness, I shall use a version of Ernst Mach’s and Bertrand Russell’s Neutral Monism (Stubenberg and Wishon 2023). I suppose that conscious things are composed of elements each of which may be described as being aware of an object from a perspective. For each element there is a region, a, the perspective, an object, b, and a dyadic perspectival relation Rxy such that the element is Rab. For some elements the perspectival relation is co-location, in which case there is awareness of the object here now. Otherwise, it is a relation of being at some, perhaps indeterminate, distance and spatio-temporal orientation, for example, being about five metres to the left of here in the immediate past. Moreover, agency requires that the objects include non-actual possibilities, in which case the perspectival relation is future-oriented.¹⁴ There might be restrictions on the basic kinds to which objects can belong, but there are no further constraints. (In this article, we neither need to decide whether the objects are universals or particulars, nor whether the elements are fundamental, nor whether despite their complex description they are simple.) The ubiquity of consciousness amounts, then, to saying that there is awareness of everything of the right kind to be an object, and awareness of it from every perspective.

Panpsychism and its variants, including Neutral Monism, must solve the Subject Summation Problem (Goff et al. 2022, §4.3.) That is, the minds constituted by sums of elements must be distinct.¹⁵ The one exception to this might be the divine mind which is either the whole manifold of elements or that part not otherwise integrated.

A person, I hypothesize, is – or is constituted by, if you prefer – a maximal coherent sum of elements separated from other coherent sums of elements that are maximal in the sense that their sum is not a proper part of any coherent sum. (I require maximality because otherwise a stage, such as an angel before rebelling, would be considered a whole person.)

To say a sum of elements is coherent is stipulated to mean:

1. 1. The sum of the perspectives is a continuous spatio-temporal region without fission (future branching) or fusion (branching if we reverse temporal orientation)

2. 2. The past oriented elements provide the motives for the choice among future oriented ones

3. 3. The chosen possibility then becomes a past-oriented element.

Separation is satisfied if the sequence of chosen possibilities from nearby perspectives either lacks coherence or else coheres significantly less well.

In abstraction, this criterion for personhood is straightforward, but the details are difficult because there is a problem of salience. The array of elements is not divided into discrete parts, but rather varies continuously. Salience arises if there are some good-making features of the future possibilities that appear with greater frequency or density from some perspective than from neighbouring ones, with a sequence of choices resulting in a coherent narrative. Those features that are no more densely represented from perspectives in a given region than from those in neighbouring ones will not be part of the mind of the person in question. I speculate that only some features are attended to, namely those that are required for the ongoing narrative. The others form a background.¹⁶

This speculative theory of persons should suffice to reply to the objection that RATh is ad hoc and contrary to Ockhamist requirements of theoretical simplicity.

How can angels affect nature?

To answer the third objection, we need to speculate as to how angels can affect us. Now, Gregory Boyd (2001, 304) asserts that ‘The mystery of how demonic forces influence nature is no different than the mystery of how the Spirit of God influences nature.’ This assertion is unsatisfactory, for angelic activity is restricted in space and time whereas the divine is not. Hence, RATh requires a speculation about localized agency, one that generalizes the special humanoid case. My proposed metaphysics of agency is that being aware of future possibilities from a given perspective, we may select between them. In the humanoid case these possibilities differ only in the immediate future states in the agent’s body and mind, but angels are not restricted in this way.¹⁷

Possible futures correspond to the ways the salient elements can change, constrained but not determined by the past state of the universe nearby together with the divinely ordained natural order. The perspectival relation in an element must be along a straight line – or, if space-time is curved, along a geodesic. For if we say that an object appears, say, over to the left we mean directly to the left, not along some twisting or zig-zag path. The geometry of space–time could specify (connected) regions from which certain features relevant to evolution stand out compared to how they seem in neighbouring regions.¹⁸ Presumably many other features could also stand out, but I am supposing those others are not salient to evolution-guiding angels. It would, however, be improperly anthropomorphic to think of them viewing the wonderful variety of life as we might in some documentary on, say, the Cambrian Explosion. There is, to be sure, a problem of the qualia that vexes all philosophers of mind, except maybe eliminativists, whom I don’t pretend to understand. This is a known problem of how things appear and is not exacerbated by admitting ignorance about what it is like to be an angel.

The awareness of various future possibilities will include different courses for evolution, which may be the result of the chaos-theoretic amplification of quantum-theoretic indeterminacy (see Emberger 2022). For example, minute variations in the locations of comets in the Oort cloud could result, millions of years later, in the asteroid hitting Earth 65,000,000 years ago instead of a near miss. That does not imply that rebels aim comets to hit asteroids in a celestial game of snooker. Rather, the rebels were aware of possible future courses for evolution and chose some rather than another. I use the asteroid example because, for all I know, God might have preferred dinosaur folk to hominids, but presumably there have been many less dramatic ‘chance’ events that altered the course of evolution.

Dunnington’s objection

Dunnington has criticized Satan Theodicy, that is to say, RATh, on the grounds that there is no place in the natural order for direct angelic action over the details of our lives. And he is surely right about individual natural disasters. We know the Black Death, for example, was not directly due to malevolent angels. It was a bubonic plague, caused by Yersinia pestis. Even given the ignorance of the cause, some prudent restrictions on shipping and some minimal hygiene would have greatly lessened its impact. Or consider the Lisbon earthquake and tsunami of 1755. We know its cause: plate tectonics. We also know that there had been a severe earthquake in 1531 and that after the 1755 disaster the city was rebuilt with greater strength. Likewise, in some earthquake-prone countries, such as New Zealand and Taiwan, planning prevents the worst destruction. Hence the 1755 death and destruction should not have been so great. That it was catastrophic was not due to ignorance, but neither, I say, was it due to direct malevolent diabolic intervention. Instead, we should blame our moral weakness caused by the rebels’ decisions about evolution.

My response to Dunnington, then, is that, first, the rebels are not malevolent and second the evils explained by RATh are those that afflict hominids and perhaps other animals that could have been avoided if evolution had proceeded differently. By the time humanoids had come to live in cities or densely populated farmland they should have understood certain hazards well enough to prevent catastrophe, and their, perhaps inevitable, quarrels should not have resulted in warfare. It is the underlying moral weakness of hominids that explain catastrophes, which weakness is in turn explained by RATh.

On the divine motive, and that of rebel angels¹⁹

Why did God permit the rebels to harm us?

The first three objections have been answered. The next objection is that God has no reason to permit angels to influence our history (Adams 1985, 236.) My answer has three parts. The first is that God did not intend sentient beings to live in solitary bliss but to belong to a community, one that extends to all creatures, but especially the angelic and humanoid, who are invited freely to obey God as an expression of their love. For the sake of this community angels can interact with humanoids. The second part of the answer is that God did not have in mind any precise kinds of humanoids, leaving that up to the angels’ guiding evolution on Earth and perhaps elsewhere. The third part is that billions of years ago God loved angels but not the humanoids, who do not have foreknown status either as future individuals or as future kinds. If the angels guiding evolution had obeyed, then there would have been a different kind from hominids, one that was happier, better, and more innocent. It is not unfair of God to allow us to come to exist rather than the more innocent humanoids, because there were no kinds foreknown to come to exist. Nor need we suppose that obedience to God would have resulted in some specific kind that God had in mind, dinosaur folk maybe. For obedient angels would still have had considerable freedom as to the kind of humanoid that evolves. Hence God can, therefore, just leave it up to the angels to choose among various courses of evolution all of which result in humanoids and all of which have a good outcome eventually.

The next objection is that omnipotence is an essential divine attribute, so God cannot cease to have it, as required if angels are to guide evolution. The short reply is that this is an objection to Free Will theodicy more generally, and this article is concerned with defending an unfashionable version of that widely accepted theodicy. The longer reply is the thesis of divine quasi-kenosis: either God has abdicated the power to interfere with the natural order, or has good reasons not to, reasons which are independent of the further consequences of such interference. (For a discussion of quasi-kenosis see Forrest 2024.)

Why does God not intervene?

Quasi-kenosis does not assert that God never intervenes, merely that such intervention is constrained by the laws of nature. Theists may hold that God has and will intervene; and Christians will point to the Incarnation. And maybe but for intervention there would have been an atomic war in the 1960s. Who knows? The problem remains that however grateful we might be for providence, we should ask why there is not more intervention.²⁰ RATh provides a partial answer. Both the gravity of creatures’ choices and the opportunity for repentance on observing the consequences prevent too much swift divine intervention. In the human case this prompts the psalmists’ lament, ‘How long?’ (e.g., Psalm 13.) Angels may well operate on a different time scale from us, being somewhat removed in space–time, unlike God who is omnipresent. Hence much of our suffering could be due to delayed divine intervention, itself the result of God’s interactions with angels, whom God loved before we existed. This answers the sixth objection.

The motive for rebellion

The final objection in the list is that there is no plausible motive for the rebellion. For we should suppose the rebels acted rationally and with knowledge of objective values. Otherwise, God created angels in an irrational state precisely to test their obedience. I find that implausible because we would expect God to value the sort of love expressed by a rational choice to obey. Hence irrationality is one of the evils theodicists should seek to explain. On pain of circularity, therefore, we should not attribute such irrationality to the rebels. But how can it be rational to ‘defy Power, which seems omnipotent’? (Shelley 1820). And why would any angels desire to inflict suffering upon us? Now Aquinas asserts that pride and envy are not ‘sins of the flesh’. Even if he is right, we should not follow him in treating these sins as the explanation of Lucifer’s disobedience (Summa Theologiae 1.109). For such sins should not be attributed to angels before they rebelled if it is irrational to disobey God; but, if attributed at all, judged to be consequences of their disobedience.

The angelic dilemma

I speculate that the angels were entirely rational but subject to a moral dilemma: to obey God is good as an expression of love, but it is also good to act as a utilitarian for the greater glory of God even if that is contrary to the divine command. The dilemma arises because there are two incommensurable values: static perfection and drama, that is, a narrative with meaning. If we encountered innocent paradise-dwellers we too might be tempted to ‘stir the possum’ for the sake of the consequent drama, and I speculate that the rebels disobeyed God to bring one about, but one that they might have expected would include swift divine victory and their subsequent obedience. If so, the rebels not merely acted rationally but righteously, for we are under no obligation to obey God, which is an act of love (see Bell and Renz 2024).

Two contrary objections

It might be objected that it is absurd not to love God. I agree but that absurdity arises only when and because God has exhibited the divine love in various ways: to the angels who had not rebelled perhaps, but also to us. Jews might note their liberation from slavery in Egypt and return from exile in Babylon; Christians will note the Incarnation. Coming to be aware of these events there would be no further angelic rebellion. Initially, though, it was not absurd to disobey.

The other objection is from the tradition that the rebels are thoroughly evil, contrary to my favourable description of them, almost as teasing God. More precisely, although disobedient they acted rationally and with good intentions. I reply to this objection by showing how the rebels might have become wicked and subsequently evil, in the sense of slaves to their own wickedness. This could happen if for every divine plan that might have a glorious conclusion there could be some new obstacle, overcoming which would require a revised divine plan, with further divine glory. God would not intend such a sequence but something like it might result from God’s interaction with the rebels, assuming both divine and angelic knowledge of relevant future possibilities. We may suppose that God’s plan A is for happy innocence but the rebels force God to choose some plan in which evils are gloriously overcome. They might have anticipated a decisive end to their rebellion. God’s Plan B, however, is one that the rebel angels can again interfere with. So, for the sake of even greater glory, they force divine plan C, and so on. The rebels would, then, give a resounding affirmative to St Paul’s question, ‘What shall we say then? Shall we continue in sin, that grace may abound?’ (Romans 6.1). The rebels’ decision repeatedly to delay the completion of the drama should be judged wicked. It presents the paradox of a sequence of decisions each of which is rational, but with an overall narrative that could not be rationally chosen and that we should condemn as wicked. Compare an authoritarian government that condemns one generation to poverty to secure rapid economic development purportedly to favour the next generation, but then sacrifices that one to the one after and so on.

I say more: stepwise utilitarian thinking can lead to an evil state, by which I mean a de-personalized slavery to wickedness. I call this slavery by analogy, because, even if Y does not legally own X we would say that X is Y’s slave if Y can punish X by death for disobedience. Repeated wicked acts can lead to the state in which repentance is like death in totally disrupting the narrative. Therefore, decisions are not made for the sake of the agent’s flourishing, nor of anyone else’s, but for mere continuance of the wretched state, which is why I consider the evil state to be de-personalized.

Conclusion

Some may have their preferred detailed version of Free Will theodicy that they take to be the best, perhaps the only, way of explaining horrendous evils. I propose, however, a smorgasbord from which one or more may be selected depending on subjective assessments of overall plausibility. More objectively, it is the disjunction of theodicies that has apologetic force. We should put RATh on the smorgasbord, because it can be defended from the various objections that explicate its widespread dismissal.