This is a very rough draft of a proposed opening chapter for a book on the problem of evil,

to be co-written with Derk Pereboom. Comments are very welcome. –Keith DeRose

Might God Have Reasons for Not Preventing Evils?

1. Theism’s Problems with Evil

Virtually all monotheistic religions profess that there is a divine being who is extremely

powerful, knowledgeable, and good. The evils of this world present various challenges for such

religions. The starkest challenge is directed toward views that posit a being whose power,

knowledge, and goodness are not just immense, but are as great as can be: an omnipotent,

omniscient, and perfectly good being (for short, an oopg being). For it would seem that such a

being would have the power, the knowledge, and the moral disposition to prevent any evil

whatsoever, and from this one might readily conclude that if there were such a being, there

would be no evil at all. On one version of this challenge, the coexistence of evil with a God

defined in this way is claimed to be logically impossible. This has come to be called the logical

problem of evil. Another is that the existence of such a God is improbable given the evils of this

world, or at least that the existence of these evils significantly lowers the probability that such an

oopg God exists. The concern expressed is that these evils provide good evidence against the

existence of such a God. This version is known as the evidential problem of evil.

The project of defending theistic belief from these problems is known as theodicy, and

1
the heart of most traditional theodicies is to provide reasons why God would or at least might

produce or allow evil. Prominent among such attempts are the free will theodicy, according to

which evils are not due to God but rather to the free choices of other agents; the soul-building

theodicy, in which God allows or brings about evil in order to elicit virtue and to build character;

the punishment theodicy, by which God allows or brings about evil as punishment for sin; and

the best of all possible worlds theodicy, according to which God caused or allowed evils because

the best of all possible worlds contains evil, and God wanted to produce the best of all possible

worlds.

This focus on God’s possible reasons for not preventing evils suggests that the crucial

question on which the problem of evil hinges is whether God might have such reasons for not

preventing some evils. That focus is well-placed. It is natural and fitting that proposals as to

why God would or might allow or cause evils play a dominant role in theodicies, as even a fairly

quick look at the problem of evil, in its versions mentioned above, reveals.

2. The Logical Problem of Evil

Let’s begin with the logical problem of evil. Why might one think it is logically impossible for

an oopg God to coexist with any evil at all? Briefly, it’s because God’s omniscience and

omnipotence ensure God’s ability to prevent all evils, while God’s perfect goodness can seem to

guarantee that God will want to prevent any evil God has the ability to prevent. David Hume
2
seems to be pressing just this problem when he puts this succinct formulation of it in the mouth

of his character Philo in his Dialogues Concerning Natural Religion:

Epicurus’ old questions are yet unanswered.

Is he [God] willing to prevent evil, but not able? then he is impotent. Is he able,

but not willing? then he is malevolent. Is he both able and willing? whence then is

evil?1

Based on such thinking, it has seemed to many that there is a logical inconsistency or

contradiction between the existence of an oopg God and any evil at all. Writing in 1974, Alvin

Plantinga reported that “a multitude of philosophers have held that the existence of evil is at least

an embarrassment for those who accept belief in God,” and, getting a bit more specific about the

exact nature of the embarrassment alleged, that “most contemporary philosophers who hold that

evil constitutes a difficulty for theistic belief claim to detect logical inconsistency” (p. [83])

between the existence of an oopg God and evil.2 Since then, the work of Plantinga and others

against the problem may have given rise to a situation where more philosophers are at least a bit

more hesitant to declare that there is a contradiction here, but, at least for many, the logical

problem of evil has not gone away.

It will be helpful to our discussion to state the matter just a bit more formally. The issue

is whether the following are logically incompatible with one another:

1
David Hume, Dialogues Concerning Natural Religion, (Indianapolis, Hackett Publishing Company, 1980), Part
X, p. 63.
2
Plantinga, The Nature of Necessity, p. [83]. See Plantinga’s references on pp. 83-84 for some thinkers who have
alleged the contradiction in question.
3
 (1) God exists, and is omnipotent, omniscient, and perfectly good

(2) Evils exist

And two propositions, like (1) and (2) above, are logically incompatible if you can derive an

explicit contradiction from them together with other propositions, if the other propositions

needed to derive the contradiction are all logically necessary. And it can seem that (1) and (2)

are incompatible with one another in virtue of the following two propositions:

(P1) A perfectly good being prevents every evil it can prevent

(P2) An omnipotent and omniscient being can prevent all evils

It can seem that (P1) and (P2) are both necessarily true, since their truth can seem to be

guaranteed by the meanings of the terms “omnipotent,” “omniscient,” and “perfectly good.”

And you can derive a contradiction from (1), (2), (P1), and (P2): for (1), (P1), and (P2) together

imply that there is no evil, which contradicts (2).

So, the thought that God’s existence is logically incompatible with the existence of evil

seems, at least at first blush, to depend on (P1) and (P2) both being necessary truths. But as

Nelson Pike points out, (P1) isn’t really necessarily true: it doesn’t follow simply from a being’s

perfect goodness that it prevents every evil it can prevent. 3 Pike argues:

Consider this case. A parent forces a child to take a spoonful of bitter medicine. The
3
Pike, “Hume on Evil,” p. {40}.
4
 parent thus brings about an instance of discomfort – suffering. The parent could have
refrained from administering the medicine; and he knew that the child would suffer
discomfort if he did administer it. Yet, when we are assured that the parent acted in the
interest of the child’s health and happiness, the fact that he knowingly caused
discomfort is not sufficient to remove the parent from the class of perfectly good
beings. If the parent fails to fit into this class, it is not because he caused this instance
of suffering. (p. {40})

We can make the same point with another example, one in which an agent allows, rather

than causes, an evil. So, for instance, if a doctor has a choice between saving the life of an

accident victim or relieving the mild pain of another potential patient’s slight scrape, but cannot

treat both people, and chooses to save the accident victim’s life, her allowing the pain from the

slight scrape does not count at all against her goodness, even though she really could have

relieved the pain of scrape had she chosen to do that instead of saving the accident victim. If

this doctor is less than perfectly good, it is not because she allowed this evil (the mild pain) that

she could have stopped.

Some readers may be growing impatient with these examples: “Sure, the limited humans

in these examples can get off the hook for allowing or even causing some evils that they could

have prevented. But we’re supposed to be talking about an omniscient and omnipotent God

here! And such a being could not similarly get off the hook for causing or preventing evil. For

God could get all the good effects of the medicine without having to cause any discomfort, and

God would never have to choose among multiple recipients of help; God could always help them

all.”

But Pike’s point here – and ours following him – is just that it doesn’t follow from God’s

5
perfect goodness alone that God prevents all the evils God can prevent. Insofar as the logical

problem of evil depends on the premise that a perfectly good being eliminates every evil it can,

it is badly based. But as impatient readers may be sensing, the logical problem hasn’t been

completely debilitated; perhaps it can have a somewhat different basis. Just a bit of patience

here in finding a more secure basis for the logical problem of evil will be rewarded by revealing

an important insight into the nature of the problem.

Following Pike fairly closely, we can say that the lesson of such examples is that to be

necessarily true, (P1) has to be weakened to:

(P3) A perfectly good being prevents every evil it can prevent, unless it has a
sufficient reason for causing or allowing an evil,

where we understand “a sufficient reason for causing or allowing an evil” to mean a reason or

excuse that is sufficient to result in the agent’s causing or allowing the evil in question to not

count at all against her goodness.4 (P3) is not refuted by the examples we have considered that

4
Here we modify Pike’s formulation slightly. What Pike concludes is that “a being who permits (or brings about)
an instance of suffering might be perfectly good provided that there is a morally sufficient reason for his action”
(p. [41]), where the phrase “morally sufficient reason” is understood in this way: “To say that there is a morally
sufficient reason for his action is simply to say that there is a circumstance or condition which, when known,
renders blame (though, of course, not responsibility) for the action inappropriate” (pp. [40-41]). But to be a
perfectly good being requires more than avoiding courses of action for which one might be appropriately blamed.
At least arguably, there are possible courses of action which, though they don’t rise to level of being morally
wrong or being the objects of appropriate blame, nevertheless are such as to render an agent who chooses them
less than perfectly good. Some cases of failing to give aid may give intuitive examples of this. There may be
circumstances in which an agent has no obligation to give aid to another, in which it would not be morally wrong
to fail to give aid, and in which blame for failing to give aid would be inappropriate, but in which it is nonetheless
better to give aid, so that a failure to give aid does render the agent less than perfectly good, despite her doing
nothing morally wrong. Thus, since the issue before us is whether a perfectly good being can fail to prevent evils,
it seems best to formulate the issue in terms of an agent’s having a “sufficient reason,” as we are using that phrase,
rather than in terms of Pike’s notion of a “morally sufficient reason.”
Our reformulation may also help to sidestep a certain problem that Pike’s formulation faces. Marilyn
Adams writes that Pike’s presentation embodies the assumption that God is “a moral agent whose creative
activities are governed by moral obligations” (HEGG, p. 10), an assumption denied by some theists: “Whereas
6
refuted (P1), for, though the agents in the examples are indeed blameless for causing or allowing

evil, they have sufficient reasons for doing so. Unlike (P1), the weaker (P3) seems like it could

be necessarily true. But, even so, (P3) can’t simply replace (P1) in the reasoning we’ve been

considering for the logical impossibility of God’s co-existing with evil, for the weaker (P3)

when combined with (P2) just doesn’t give you enough ammunition to derive that there is no

evil from the existence of an oopg being.

But we can still derive a contradiction if we add the following proposition to our set:

(P4) An omnipotent and omniscient being does not have a sufficient reason for

causing or allowing any evils,

where “sufficient reason” is taken in the same way that we understood it in (P3). If (P4) is

necessarily true, as (P2) and (P3) also seem to be, then it follows from the existence of an oopg

being that there exists no evil whatsoever, and (1) and (2) are logically incompatible, after all.

And if you were impatiently objecting that our examples of limited agents who were

justified in causing or allowing evils were irrelevant to the case of God, it may well have been

the thought that (P4) is necessarily true that motivated your reaction. Since those agents’

excuses for causing or allowing evil were essentially rooted in their limitations, and since the

same is so of many other examples of justified causing or allowing of evil that might naturally

nontheistic value theories assign all persons a common ontological status as humans and find it natural to see all
persons as woven into a common web of rights and mutual obligations, many Christian thinkers regard God as of
the wrong ontological category for such entanglements” (p. 12). Our formulation makes no assumption about
whether or not God has moral obligations, or whether God can be appropriately blamed for courses of action. In
requiring that God have a “sufficient reason” for not preventing evils, in our sense, what we are requiring is that
God have a reason such that God’s not preventing the evil does not count at all against God’s goodness, whether
or not that goodness partly consists in meeting moral obligations and avoiding moral wrongdoing.
7
spring to mind, one may well begin to suspect that whenever an agent is justified in causing or

allowing an evil, it will have to be due to such a limitation in the agent’s abilities. But that in

turn suggests that a being with no such limitations – an omnipotent and omniscient being – just

could not have such an excuse for not preventing evils. In other words, it leads to the suspicion

that (P4) may be necessarily true.

The important insight derived from seeing that (P1) is not necessarily true involves the

importance of the matter of whether God might have a sufficient reason for allowing evils. For

when (P1) is replaced by the weaker (P3), we have to add (P4) to restore the contradiction. We

can then see that the whole issue of whether the existence of an oopg God is logically compatible

with the existence of evil comes down to the crucial question of whether it’s possible for such a

God to have a sufficient reason for not preventing some evils. For if that is possible, then (P4) is

not necessarily true, and the logical problem of evil again crumbles – and this time, with no

apparent repair in sight. But if, on the other hand, the suspicion that all sufficient reasons agents

might have for not preventing evils must be grounded in limitations that an oopg being would

not have proves true, then (P4) is necessarily true, and the existence of an oopg God turns out to

be incompatible with the existence of any evil at all.

It’s no wonder then, that proposing sufficient reasons that might apply even to an oopg

God has been such a focus of theodicists!

3. But Could God Have a Reason for Permitting These Evils?: The Evidential Problem of
8
Evil

Though many have made it, the claim that the existence of an oopg God is incompatible with the

existence of evil is a very strong contention – so strong that it can be refuted by what we will

call a merely logical theodicy. To defeat the logical problem of evil, all a theodicist has to come

up with is a sufficient reason God might have had to allow some possible evil in some possible

situation. The theodicist need not put the proposed reason forward as God’s actual reason for

allowing any actual evils; it need only be a sufficient reason God might have had in some

possible situation. Indeed, the reason proposed could be one we somehow know for sure isn’t

God’s actual reason for allowing evils, so long as it is a sufficient reason God might have had in

some possible situation. And the evil in question can be any possible evil at all, so the theodicist

will be free to focus on those possible evils that would be the very easiest to square with the

existence of God.

Many who deeply feel that theistic belief is profoundly challenged by the existence of the

world’s evils would find such a merely logical theodicy far from satisfying. For many, the

greatest challenge from evil to theistic belief comes not from the bare fact that some evils or

other exist, but from more concrete facts about how much evil there is, how bad many of these

evils actually are, and concerning some of the especially problematic kinds of evils that actually

exist. For instance, some might feel that the greatest challenge to theistic belief comes from the

prevalence in our world of what Marilyn Adams calls “horrendous evils,” which she

characterizes as follows:

9
 Among the evils that infect this world, some are worse than others. I want to try to

capture the most pernicious of them within the category of horrendous evils, which I

define (for present purposes) as ‘evils the participation in which (that is, the doing or

suffering of which) constitutes prima facie reason to doubt whether the participant’s life

could (given their inclusion in it) be a great good to him/her on the whole’. The class

of paradigm horrors includes both individual and massive collective suffering . . .

[E]xamples include the rape of a woman and the axing off of her arms, psycho-physical

torture whose ultimate goal is the disintegration of personality, betrayal of one’s own

deepest loyalties, child abuse of the sort described by Ivan Karamazov, child

pornography, parental incest, slow death by starvation, the explosion of nuclear bombs

over populated areas. (HEGG, p. 26)

Those who feel that the greatest challenge to theism comes from the prevalence of these most

pernicious forms of evils in our world will find no relief in a logically consistent account of how

in some possible situation far removed from the actual world God might have had a reason for

not preventing some very minor evil of a relatively unproblematic type from occurring – though

such a merely logical theodicy would suffice to solve the logical problem of evil.

Shifting from the challenge provided by the bare fact that some evils or others occur to

more concrete facts about the amounts and kinds of evils which actually occur is naturally

accompanied by a shift in focus from the logical consistency of the existence of God with the

evils in question to the matter of whether those evils merely provide strong reasons to deny

God’s existence.5 If the meanings of “omnipotent,” “omniscient,” and “perfectly good” don’t

10
logically guarantee that if an oopg being exists there is no evil at all, it’s not easy in that case to

see how a contradiction could nevertheless be derived from the assumption that an oopg God co-

exists with a certain amount of evil, or with evils of a certain type. 6 But as we consider evils that

get harder and harder to square with the existence of God, it is natural to think that these evils

provide stronger and stronger evidence or reasons for an atheistic conclusion. So we arrive at

the evidential problem of evil: Do the evils of this world – evils in the amounts and of the types

that actually occur – count as strong evidence against the existence of God?

Again, the issue of whether God might have sufficient reasons for not preventing these

evils will loom large. For God’s perfect goodness still assures us that God would prevent every

evil for which God lacks a sufficient reason for allowing it. But as we consider the most

pernicious and massive evils that have actually occurred, it can seem they are so bad that it’s

extremely unlikely that God has a sufficient reason for not preventing them, and thus that these

evils constitute strong evidence against the thought that God has such a sufficient reason. But

since God – understood in the oopg way – does prevent evils absent sufficient reasons for

allowing them, this in turn makes it seem extremely unlikely that such a God exists.

4. Merely Logical Theodicies, Know-It-All Theodicies, and a Middle Course

As was the case with the logical problem of evil, the hope for relief from the evidential problem
6
Unless a type is specified in such a (question-begging) way as this: evil such that a oopg God could not possibly
allow it.
11
seems to be promised if theodicists can provide accounts of God’s reasons for not preventing

evils. Again, as before, these must be reasons that even an omnipotent and omniscient being

might have. But now theodicists face a tougher challenge, for merely logical theodicies will no

longer suffice. The reasons proposed must be reasons God might have, not just for allowing

some evil or other, but for allowing evils in the amounts and of the kinds – including the most

pernicious kinds – that actually exist. And since we’re now concerned not just with logical

consistency, but with evidential relations, it will no longer do to propose reasons that God might

have had in some possible situation.

But then, what will help? When she proposes reasons God might have for allowing evils,

must the theodicist put these forward as God’s actual reasons? This can seem too much. David

Lewis lays out this dilemma well, using terminology that differs from ours and that he takes

from Plantinga, on which “defence” means what we have been calling “merely logical theodicy,”

and “theodicy” means an account of God’s actual reasons:

Alvin Plantinga, our foremost modern authority on free-will theodicy, would recoil

from that name for his subject. He has taught us to distinguish ‘theodicy’ from

‘defence’. ‘Theodicy’, for Plantinga, means an audacious claim to know the truth

about why God permits evil. And not just a trivial bit of the truth – God permits evil

for the sake of some good or other – but something fairly substantive and detailed. One

who claims to know God’s mind so well (especially if he claims to know without

benefit of revelation) will seem both foolhardy and impudent.

‘Defence’, on the other hand, means just any hypothesis about why omniscient,

12
 omnipotent, benevolent God permits evil. Its sole purpose is to rebut the contention

that there is no possible way that such a thing could happen. To serve that purpose, the

hypothesis need not be put forward as true. It need not be at all plausible. Mere

possibility is enough. . . .

And not epistemic possibility, or ‘real’ possibility given the actual circumstances

and laws of nature; just ‘broadly logical’ possibility. That’s an easy standard. If

somehow it could be made to explain why God permits evil, the hypothesis that pigs

fly would be good enough for mere defence. . . .

Defence is too easy; knowing God’s mind is too hard. 7

Lewis may be too hard on the project of what he calls “defence” and what we are calling

“merely logical theodicy.” For though the hypothesis that pigs fly would suffice for a merely

logical theodicy if it could be made to explain why God permits evil, it’s not so easy to come up

with possible reasons God might have that really do explain, consistent with God’s omnipotence,

omniscience, and perfect goodness, why God might possibly permit evil. As we have noted,

many have thought that there is a contradiction between the existence of an oopg God and the

existence of any evil at all, and a merely logical theodicy would be enough to solve that logical

problem of evil.

But it is true that merely logical theodicies are not enough to solve the evidential problem

of evil, and Lewis certainly seems right that claims to know God’s actual reasons seem unwise.

Lewis wisely charts a middle course between these two options:

7
David Lewis, “Evil for Freedom’s Sake,” Philosophical Papers 22 (1993): 149-172; pp. 151-152.

13
 I think the topic worth pursuing falls in between, and has no place in Plantinga’s

scheme of theodicy versus defence. Pace Plantinga, I’ll call that topic ‘theodicy’, but I

don’t mean the know-it-all theodicy that he wisely disowns. Rather I mean tentative

theodicy, even speculative theodicy. The Christian needn’t hope to end by knowing for

sure why God permits evil. But he can hope to advance from a predicament of not

having a clue to a predicament of indecision between several not-too-unbelievable

hypotheses (maybe still including the hypothesis: ‘none of the above’). The job is to

devise hypotheses that are at least somewhat plausible, at least to the Christian, and to

find considerations that make them more plausible or less. (p. 152)

This book will consist mainly in evaluating possible reasons God might have for

allowing evils, including great evils of the most pernicious kinds, that seek to steer this middle

course. Accordingly, we will evaluate them not as know-it-all attempts to know at any

substantive level of detail the mind of God, but nor, when we are dealing with the evidential

problem of evil, as mere attempts to provide reasons an oopg logically might have had in some

possible situations for allowing some possible evils. Rather, we will seek to determine whether

any of them might provide plausible – or in Lewis’s words, “not-too-unbelievable” – accounts of

reasons an oopg God really might have for not preventing the evils in question. If some of the

proposed reasons meet this standard, they may provide some relief from the problem of evil for

those who antecedently can’t even begin to see any reasons that God might possibly have and

think that it’s extremely unlikely that God does have sufficient reasons. As Lewis writes above,

this may constitute an “advance from a predicament of not having a clue to a predicament of

indecision between several not-too-unbelievable hypotheses (maybe still including the

might make belief in the existence of God (in the face of evil) more attractive primarily by

making more believable the hypothesis that God does in fact have a sufficient reason that one

does not understand. For the predicament many start in goes beyond not having any clue as to

what God’s reasons might be, to thinking, based on the perniciousness of some of the evils in

question, that there’s good reason to think there just isn’t any reason an oopg being has for not

preventing such things. Seeing some plausible accounts of reasons that even an oopg being

might have for allowing these evils may make it seem more likely that God can have a reason

that one hasn’t considered.

But first, we will consider, in the next chapter, some intriguing recent attempts to defuse

the problem of evil by arguing that, since we shouldn’t expect to be able to begin to discern

God’s reasons, even in the absence of our having any accounts of God’s possible reasons that are

at all plausible, the evils of this world don’t constitute a good objection to theistic belief.

15