Problem of Evil

The basic structure

Classical theism holds three commitments about God: omnipotence (all-powerful), omniscience (all-knowing), and omnibenevolence (perfectly good). The world also contains, uncontroversially, vast quantities of suffering — disease, predation, natural disaster, cruelty, child mortality, prolonged dying. The problem of evil is the family of arguments that this combination is in trouble.

The earliest preserved form is the so-called "Epicurean paradox", quoted by the Christian apologist Lactantius around 300 AD and attributed (probably loosely) to Epicurus:

Is God willing to prevent evil but not able? Then he is not omnipotent.
Is he able but not willing? Then he is malevolent.
Is he both able and willing? Whence then is evil?
Is he neither able nor willing? Then why call him God?

The modern debate has refined the paradox into two distinct arguments — the logical and the evidential — with different burdens of proof and different replies. Distinguishing them is the first move every serious treatment makes.

Logical vs evidential — the two arguments

Mackie's 1955 paper "Evil and Omnipotence" sharpened the logical version. He argued that the conjunction of three theistic claims — God is omnipotent, God is wholly good, evil exists — is internally inconsistent, given two further "quasi-logical" rules: a good thing always eliminates evil so far as it can, and there are no limits to what an omnipotent thing can do. From these, Mackie thought, contradiction followed: a wholly good and omnipotent being would eliminate all evil, so any evil shows there is no such being.

Rowe's 1979 paper shifted the strategy. Instead of demanding strict logical inconsistency, Rowe granted (for argument's sake) that some evil might be necessary for some greater good. He then argued that the actual distribution of suffering — vast, unevenly distributed, frequently apparently gratuitous — is much better explained by atheism than by theism. The argument is probabilistic: it doesn't show theism is impossible, only that the evidence weighs against it.

	Logical Problem (Mackie 1955)	Evidential Problem (Rowe 1979)
Type of claim	Deductive: theism is contradictory	Inductive: theism is improbable given evil
Strength required	Even one instance of evil refutes theism	The kind and quantity of evil is what matters
Canonical form	"God is omnipotent and wholly good and evil exists" is inconsistent	Probably some suffering is gratuitous; gratuitous suffering rules out a perfect being
Standard reply	Plantinga's free-will defense (1974)	Skeptical theism (Wykstra, Alston, Bergmann)
Burden of proof	The atheist must show inconsistency	The theist must show suffering is not gratuitous, or block the inference from "looks gratuitous" to "is gratuitous"
Where the consensus is	Plantinga is widely accepted to have answered the logical version (Alston, Draper, even some atheists agree)	Live debate; no consensus

One way to see the shift: Mackie's argument requires an a priori claim about what God must do; Rowe's requires only that we trust our ordinary judgments that some suffering serves no purpose. Theists find the second version harder to dodge.

Worked example: the free-will defense

Alvin Plantinga's God, Freedom, and Evil (1974) is the standard text. The book elaborates a defense (rather than a theodicy) — a defense only needs to show theism is consistent with evil, not give the actual reason God permits evil.

The skeleton runs as follows:

1. A world with libertarianly free creatures is more valuable than a world without.
2. God cannot guarantee that genuinely free creatures always choose well — that would just be programming, not free choice.
3. So even an omnipotent God may face a tradeoff: free creatures (sometimes choosing wrongly) or no free creatures.
4. If God created free creatures and some of them chose wrongly, the resulting moral evil is the price of the higher-value world.

Plantinga adds a more contested claim — "transworld depravity" — to handle a worry: couldn't God have created free beings whose particular essences happened never to go wrong? Plantinga argues it is at least possible that every creaturely essence God could have actualized has, in every world where that creature exists and is free, some occasion on which it freely goes wrong. If that is even possible, then it is possibly the case that God could not have created free creatures who never sin. That is enough to block Mackie's claim of strict contradiction.

The free-will defense is generally acknowledged, even by leading atheist philosophers (William Alston, Paul Draper), to have answered the logical problem in its standard form. The remaining live debate is about the evidential version and about natural evil — suffering caused by impersonal processes that no creaturely free agent caused.

Other theodicies — Augustine, Hick, and skeptical theism

Augustine's privation theory. Augustine, in Confessions (Book VII) and City of God, argues that evil is not a positive substance but a privation — a lack of good in something that ought to have it. A blind eye is not the presence of an evil substance; it is the absence of the good of sight in something built for seeing. On this account God did not create evil because evil is not a thing to be created. Augustine combines the privation theory with a strong free-will account: human evil arose when free wills turned away from the highest good toward lower goods. The privation theory is metaphysically deep but does little to explain why so much good is missing from the world; critics treat it as renaming the problem.

Soul-making theodicy. John Hick's Evil and the God of Love (1966) draws on Irenaeus (2nd century AD) rather than Augustine. The world is a "vale of soul-making": a hard environment in which finite persons can develop genuine moral character — courage, compassion, perseverance — through free response to challenge. A pain-free paradise would produce only conditioned creatures, not virtuous agents. Suffering, even severe suffering, is the medium in which character forms. Hick argues this requires a post-mortem continuation in which the soul-making process completes for everyone; otherwise the price of the unfinished is too high. Critics (Rowe, D.Z. Phillips) reply that the kinds of suffering Hick must explain — fawns in forest fires, infants with bone cancer, mass extinction — bear no plausible relation to character formation.

Skeptical theism. The newest major reply is structural rather than substantive. Wykstra's "On Keeping the Mystery in 'Mystery'" (1984) introduced the CORNEA principle: it's reasonable to infer "X has no Y" from "no Y is detectable" only if we have antecedent reason to think we'd detect Y if it were there. Applied to evil: from "I see no good served by this fawn's suffering" we can infer "no good is served" only if we'd expect to be able to see God's reasons. But God, if real, is omniscient and we are not — the cognitive gap is enormous. The expected probability of our detecting such reasons is low. So our failure to see them is weak evidence at best.

Skeptical theism is influential because it doesn't require a positive theodicy; the theist just blocks the inductive inference. The standard objection (Draper; Rowe in later work) is that the same skepticism, taken consistently, would undermine our confidence in apparently positive evidence for God too — answered prayers, religious experiences, perceived providence. You can't be skeptical about your moral perceptions when they tell against God and credulous when they tell for him.

Worked example: Rowe's fawn

Rowe's 1979 article asks us to consider a single piece of suffering in detail. A lightning strike ignites a forest. A fawn is trapped, badly burned, and lies in pain for several days before dying. No human knows about it; no human is morally improved by it; the fawn is incapable of soul-making. Rowe's argument is structured tightly:

1. Premise 1. There exist instances of intense suffering that an omnipotent, omniscient being could have prevented without thereby losing some greater good or permitting some evil equally bad or worse.
2. Premise 2. An omnipotent, omniscient, wholly good being would prevent the occurrence of any intense suffering it could, unless preventing it would result in the loss of some greater good or the permission of some evil equally bad or worse.
3. Conclusion. There does not exist an omnipotent, omniscient, wholly good being.

Premise 2 is the moral premise — most theists accept some version of it. Premise 1 is the inductive premise: the fawn's suffering is offered as a representative case for the claim that some suffering looks gratuitous and probably is.

The theist replies focus on Premise 1. Skeptical theism says we can't tell. Soul-making theodicy says even animal suffering may have a role we don't see (an extension Hick himself was careful with). Free-will defenses note that natural events involve no morally responsible agent; some theists invoke Plantinga's speculation about fallen angelic agency, others bite the bullet and accept that theism is committed to denying that any case of suffering is truly gratuitous.

Hume's Dialogues — the modern ancestor

The clearest modern statement of the evidential problem predates Rowe by two centuries. Hume's Dialogues Concerning Natural Religion (published posthumously in 1779) stages a conversation among Cleanthes (a natural theologian), Demea (an orthodox theist), and Philo (a sceptic, generally taken to speak for Hume). Part X opens with Demea conceding that this life is full of misery, expecting Philo to use that to push toward heavenly compensation. Philo instead turns it against natural theology:

"Epicurus's old questions are yet unanswered. Is he willing to prevent evil but not able? Then is he impotent. Is he able but not willing? Then is he malevolent. Is he both able and willing? Whence then is evil?"

Philo's deeper move is evidential rather than logical. Even granting that God's existence is consistent with the world we see, he argues, no one observing this world without prior commitment would infer a benevolent designer; the data fit at best an indifferent or limited designer, and more naturally fit no designer at all. Hume's Dialogues is the textual ancestor of Rowe's argument and of every subsequent evidential treatment.

Counterarguments to the problem

The greater-good defense. Many evils are necessary conditions for goods that outweigh them. Without the possibility of pain there is no compassion; without the possibility of injustice there is no courage; without the possibility of loss there is no genuine love. The reply has weight for some kinds of suffering but is widely thought too weak for others — animal suffering before humans existed; suffering that produces no growth in anyone.

Felix culpa. Augustine, Aquinas, and others suggest the world's actual story (fall, redemption, atonement) is more valuable than a never-fallen world would have been. The view requires accepting heavy theological commitments and gets harder as the suffering scaled by it grows.

Eschatological compensation. Worldly suffering is finite; eternal fellowship with God is infinite; therefore even the worst earthly suffering is outweighed by a sufficiently good afterlife. Marilyn McCord Adams's Horrendous Evils and the Goodness of God (1999) develops a sophisticated version. Critics object that some suffering ("horrendous evil") is the kind of thing that, even if compensated, leaves a permanent injustice in the moral world — and that compensation is the wrong category for some violations.

The non-classical theist move. Process theology (Whitehead, Hartshorne) and open theism (Hasker, Sanders) drop or weaken one of the classical attributes — usually omnipotence — and accept a God genuinely limited by the structure of reality. The problem of evil shrinks to manageable size, but at the cost of giving up classical theism. Most analytic philosophers of religion treat this as a separate position rather than a reply to the original problem.

Counter-counter: the asymmetry objection. Some atheist philosophers (Draper, Schellenberg) point out that theists routinely treat the world's apparent goods as evidence for God while refusing to treat the world's apparent evils as evidence against. A consistent epistemology must allow both directions of inference; theism is in trouble in any framework that does.

Variants of the problem

Hiddenness of God. J.L. Schellenberg's Divine Hiddenness and Human Reason (1993) develops a structurally similar argument from non-belief: if a perfectly loving God exists and seeks loving relationship with finite persons, no finite person who is open to that relationship would fail to believe; non-resistant non-believers exist; therefore no such God exists. The hiddenness argument is sometimes treated as a sister problem to the problem of evil.

The animal suffering problem. The hundreds of millions of years of vertebrate suffering before any human ever lived create a problem that free-will defenses cannot easily handle. Trent Dougherty's The Problem of Animal Pain (2014) explores a Christian theist response involving non-human eschatology; the topic remains active.

The horrendous-evils problem. Marilyn McCord Adams distinguishes ordinary evils from "horrors" — events such that participation in them, as victim or perpetrator, gives the participant prima facie reason to think their life as a whole is not worth living. Examples: childhood torture, genocide, certain forms of slavery. Adams argues theists owe an account on which God personally defeats horrors in the lives of their victims; standard theodicies aimed at aggregate goods are inadequate.

The doctrinal problem. For traditions with specific doctrines — original sin, hell, particular providence — the problem of evil splits into specific subproblems about why those doctrines, if true, are compatible with divine goodness. Eleonore Stump's Wandering in Darkness (2010) develops a narrative-based response.

Common confusions

• Defense vs theodicy. A defense aims only to show theism is consistent with evil; a theodicy attempts to give the actual reason God allows evil. Plantinga is explicit that the free-will defense is the former, not the latter. Critics sometimes attack it as a failed theodicy when it was never offered as one.
• "Greater good" doesn't mean "useful for someone." A theodicy that says suffering is useful for moral growth does not commit one to saying it was useful in this case. The theist position is structural — that some pattern of permitted suffering is necessary for some pattern of goods — not that each particular instance has a particular payoff.
• The problem isn't an emotional reaction. The philosophical problem of evil is a logical or evidential argument with stated premises. The pastoral problem (helping a sufferer) and the existential problem (sustaining one's own faith) are real, but they are different questions and the literature treats them separately.
• Mackie's logical argument is widely thought to have failed. Even Mackie eventually acknowledged Plantinga's defense as a serious answer to the strict-inconsistency form. The contemporary debate is about evidential versions, not logical ones.
• Theism doesn't need a complete theodicy. A theist can hold that the actual reasons for evil are partly opaque while still maintaining theism is rationally defensible — this is Plantinga's position and most contemporary analytic theist defenses.
• "God allowed it" is not "God did it." The free-will defense distinguishes God's permitting suffering (logically required by free creaturely agency) from God's causing it. Critics argue the distinction collapses when God is omnipotent enough to redesign the system; defenders argue it doesn't.

Why it matters

• Most influential atheistic argument. The problem of evil is the most cited reason for non-belief in surveys of philosophers and non-philosophers alike, ahead of arguments from naturalism or science.
• Defines modern philosophy of religion. The Mackie-Plantinga and Rowe-Wykstra exchanges of the second half of the 20th century are the central technical literature of analytic philosophy of religion.
• Stress test for moral epistemology. The evidential argument depends on our being able to judge that some suffering is gratuitous; replies depend on our cognitive limits. The exchange is also a debate about how much we can trust our moral reactions.
• Shapes pastoral theology. How a tradition handles suffering — whether it offers compensation, growth, mystery, or solidarity — depends on which theodicy it accepts.
• Connects to non-religious questions. Many secular discussions — the limits of consequentialism, the meaning of suffering, the role of moral evidence — pick up the conceptual machinery developed in this debate.