The Problem of Evil

The problem of evil is the most powerful of the arguments against theism and the central difficulty of the philosophy of religion: the existence of suffering and wickedness seems flatly incompatible with — or at least strong evidence against — the existence of the omnipotent, omniscient, perfectly good God of classical theism. Its force is unique among atheological arguments because it proceeds entirely from premises the theist already accepts (God's nature) together with a datum no one can deny (the world contains evil). It is not a sceptic's external challenge but an internal tension in theism itself, and it has been felt as such by believers from the author of Job to Dostoevsky.

This page states the problem in its two principal forms — the logical (deductive) problem and the evidential (inductive/probabilistic) problem — distinguishes the kinds of evil and the key notion of gratuitous evil, and clarifies the crucial methodological distinction between a defense and a theodicy. The responses themselves — free-will, soul-making, skeptical theism, and the rest — are developed in theodicies and defenses; the related challenge from God's hiddenness (rather than suffering) is taken up in divine hiddenness.

The inconsistent triad

The problem in its starkest form, as Epicurus is reputed to have posed it and Hume sharpened it (through Philo in the Dialogues):

Is 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?

This presents theism with an apparently inconsistent triad — three claims that cannot all be true:

1. God is omnipotent (able to prevent any evil).
2. God is omnibenevolent (wills to prevent every evil he can).
3. Evil exists.

The atheist argues that (3) is undeniable, so one of (1) or (2) must be surrendered — and a being lacking either omnipotence or perfect goodness is not the God of classical theism. The theist must show that the triad is not genuinely inconsistent: that (1), (2), and (3) can all be true together. Everything turns on the hidden assumptions needed to derive a contradiction from the three.

The logical problem of evil

The logical (or deductive or a priori) problem, pressed most forcefully by J.L. Mackie ("Evil and Omnipotence," 1955), claims that the triad is strictly logically inconsistent — that the mere existence of any evil whatever entails that the classical God does not exist. The claim is very strong: not that evil makes God improbable, but that the propositions "an omnipotent, omnibenevolent God exists" and "evil exists" are contradictory, like "this figure is a square" and "this figure has three sides."

But the triad is not explicitly contradictory; to derive a contradiction one needs bridging premises. Mackie supplies:

• (P1) A good being always eliminates evil as far as it can.
• (P2) An omnipotent being can do anything (logically possible).

From these plus (1) and (2) it follows that an omnipotent, perfectly good being would eliminate evil entirely — so if such a being exists, no evil does, contradicting (3).

Plantinga's free-will defense and the collapse of the logical problem

The logical problem is now widely regarded as a failure, and the reason is Alvin Plantinga's Free Will Defense (God, Freedom, and Evil, 1974). Plantinga's strategy is modest but decisive: to defeat a claim of strict inconsistency, he need not show why God permits evil, nor even that his story is true — he need only exhibit a logically possible proposition $p$ such that $(1),(2),p,$ and (3) are jointly consistent. Any such $p$ shows the set is not contradictory.

His candidate: it is possible that God could not create a world containing moral good (the great good of free creatures freely choosing the good) without also permitting moral evil (those creatures sometimes freely choosing wrongly), because whether a free creature goes right or wrong is up to the creature, not to God — even an omnipotent being cannot make someone freely do anything. If, possibly, every creature God could create would go wrong at least once (Plantinga's notorious notion of transworld depravity), then a world with free creatures and some moral evil may be the price of a world with moral good — and a perfectly good, omnipotent God might have sufficient reason to permit it.

The key amendment is to the bridging premises: a good being eliminates evil as far as it can unless it has a morally sufficient reason to permit it, and an omnipotent being can do anything logically possible — but making someone freely choose rightly is not logically possible. With these corrections the deductive contradiction cannot be derived. Mackie himself effectively conceded that the logical argument, as a proof of inconsistency, does not succeed — a notable moment in the field. (Whether the free-will defense extends to a full theodicy, and whether it covers natural evil, is the subject of theodicies.)

The collapse of the logical problem is why the action has decisively shifted to the evidential problem.

The evidential problem of evil

The evidential (or inductive, probabilistic, a posteriori) problem concedes Plantinga's point — that evil is not strictly inconsistent with God — but argues that the kinds, quantities, and distribution of evil we actually observe constitute strong evidence against God's existence, making theism improbable even if not impossible. It is the far more serious challenge, because bare logical consistency is cold comfort: a hypothesis can be consistent with the data yet rendered wildly improbable by them.

Rowe's evidential argument

William Rowe's formulation (1979) is canonical. It hinges on gratuitous evil — evil that serves no greater good, that an omnipotent being could have prevented without thereby losing some greater good or permitting some equally bad or worse evil:

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. (There is gratuitous evil.)
2. An omnipotent, omniscient, wholly good being would prevent the occurrence of any gratuitous evil. (A perfectly good God permits an evil only if it is necessary to a greater good.)
3. Therefore, there does not exist an omnipotent, omniscient, wholly good being.

Premise 2 is the theist's own principle (God permits no pointless suffering). The contested premise is 1, and Rowe supports it inductively with cases of apparently pointless suffering. His famous example: a fawn trapped in a distant forest fire, burned and dying in agony over days, unseen by any human, its suffering serving no discernible purpose. Even if we cannot prove it serves no good, it appears gratuitous, and — Rowe argues — we are entitled to move from "we can see no greater good served" to "there is (probably) no greater good served," just as we are generally entitled to trust careful observation. The sheer scale of such seemingly purposeless suffering — animal pain across hundreds of millions of years of evolution before any humans existed, the Holocaust, childhood cancers — makes it probable that some of it is gratuitous, and hence that the classical God does not exist.

The "noseeum" inference and skeptical theism

The decisive question is whether premise 1's inference is legitimate: from "we can see no justifying good" to "there is no justifying good." Critics dub this a "noseeum" inference ("I no see-um, so there ain't none"), and skeptical theism (Wykstra, Alston, Bergmann) attacks it head-on. Wykstra's CORNEA principle: we are entitled to move from "I see no $X$" to "there is no $X$" only if, were there an $X$, we would likely see it. But given the vast gap between divine and human cognition, we have no reason to expect to discern God's reasons for permitting a given evil even if such reasons exist — they may lie centuries downstream or in goods we cannot conceive, much as a toddler cannot see the point of a painful injection. So our failure to see a justifying good is not good evidence that there is none; the noseeum inference is unwarranted, and premise 1 is unsupported.

The evidential debate thus largely becomes a debate over skeptical theism — whether the appeal to divine inscrutability successfully blocks Rowe's inference, and at what cost (the worry that the same skepticism, consistently applied, would undermine all our moral knowledge and even our trust that God is good). This is pursued in theodicies and defenses.

Kinds of evil and key distinctions

Several distinctions structure every discussion:

• Moral evil — suffering and wrongdoing brought about by the free choices of moral agents (murder, cruelty, injustice). This is the evil the free-will defense most directly addresses.
• Natural evil — suffering caused by natural processes with no human agency (earthquakes, disease, predation, the burning fawn). This is harder for theism, since the free-will defense does not obviously apply — what free choice explains a tsunami? (Responses: natural laws as a necessary good, soul-making, free fallen angels, or natural evil as the by-product of a law-governed world; see theodicies.)
• Gratuitous (pointless) evil — evil not necessary to any greater good. As Rowe's argument shows, the theist is committed to denying that any evil is truly gratuitous; the evidential problem presses on whether that denial is credible given what we observe.
• Horrendous evils (Marilyn McCord Adams) — evils so extreme (genocide, the torture and murder of a child) that they threaten to render the victim's life not worth living, posing a problem that abstract "greater good" theodicies seem obscene to address; Adams argues the response must be not a general theodicy but God's defeating such evils within the individual's life through intimate goods.

A further distinction concerns the aim of a response:

• Defense vs. theodicy. A defense aims only to show that God and evil are logically consistent (or that the evidential argument fails) — it offers a possible reason God might have, without claiming it is the actual one. A theodicy is more ambitious: it offers an account of the actual reasons a good God does permit evil — a justification of God's ways (Greek theos + dikē, the "justice of God"; the term is Leibniz's). Plantinga's free-will defense answers the logical problem with mere possibility; answering the evidential problem credibly seems to require something closer to a theodicy. The distinction matters because a defense can succeed while every theodicy fails: it may be that we cannot say why God permits evil (the defense and skeptical theism suffice for consistency) even though no positive justification convinces.

Where the problem of evil sits

The problem of evil is the hinge of the entire theism debate. Its logical form — the claim that evil is strictly inconsistent with God — is widely conceded to have failed, defeated by Plantinga's free-will defense, which needs only the bare possibility of a morally sufficient reason. But this is a thin victory, and the real contest is the evidential form: whether the amount, kinds, and apparent pointlessness of suffering make God improbable. That debate turns on the legitimacy of the noseeum inference — Rowe's move from seeing no justifying good to there being none — and hence on skeptical theism, with its own costs for our moral knowledge. The problem also feeds directly into the cumulative case calculus: it is the heaviest weight on the atheist's side of the scale, which any net assessment of the arguments for theism must absorb. The constructive theistic responses — free-will, soul-making, skeptical theism, and the appeal to mystery — and the distinct but parallel problem of divine hiddenness carry the argument forward.