The Epicurean paradox holds that evil cannot coexist with an all-powerful, all-good God, and the three standard replies are free will, soul-making, and skeptical theism. Skeptical theism is the strongest of the three, because it denies that humans are in a position to know what reasons God might have for permitting evil.
The paradox is usually attributed to Epicurus, but he never wrote it down in any surviving text. The version everyone quotes comes from Lactantius, a Christian writer who recorded it around 300 CE, roughly six centuries after Epicurus died in 270 BCE. Whether Epicurus actually said it, or Lactantius assembled it from his arguments, is still disputed.
The argument did not become a formal deductive problem until J.L. Mackie published "Evil and Omnipotence" in 1955, claiming that the three divine attributes are logically incompatible with any evil at all. That version took real damage from Alvin Plantinga's free will defense in 1974, which argued that even an all-powerful God cannot actualise a world of genuinely free creatures who never choose wrongly. Most philosophers now treat the deductive form as dead, which is why the debate moved to William Rowe's 1979 evidential version: not "evil is logically impossible with God" but "the sheer amount and distribution of suffering makes God's existence unlikely."
None of the three replies is free. Free will only covers moral evil, and it strains badly against a tsunami that kills 230,000 people. Soul-making, from John Hick's 1966 Evil and the God of Love, needs suffering to be roughly proportionate to the character it builds, and most of us have seen cases where it plainly is not.
- Not Epicurus's own words: The paradox as commonly quoted appears in Lactantius around 300 CE, not in any of Epicurus's surviving writings from the fourth century BCE.
- Deductive version largely abandoned: After Alvin Plantinga's 1974 free will defense, most philosophers accepted that Mackie's 1955 logical problem of evil fails as a strict contradiction.
- Evidential version replaced it: William Rowe's 1979 formulation reframed the argument as probabilistic, resting on whether pointless suffering provides evidence against God.
- Skeptical theism's core move: Stephen Wykstra argued in 1984 that "we see no reason for God to permit this" does not entail "God has no reason," which undercuts the inference both problems rely on.
- Soul-making's origin: John Hick's 1966 book argued that suffering is a necessary condition for moral and spiritual development, a claim he inherited from Irenaeus in the second century.
What exactly is the Epicurean paradox?
The paradox comes down to four possibilities, and the earliest surviving version lays them out as a dilemma with no comfortable exit. If God is willing to prevent evil but not able, then God is impotent. If God is able but not willing, then God is malevolent. If God is both able and willing, then where does evil come from? And if God is neither able nor willing, why call him God at all? Each horn forces the same conclusion: an all-powerful, all-knowing, all-good being and the existence of evil cannot both be true.
Stated formally, the argument runs as a reductio. Premise one: if an omnipotent, omniscient, omnibenevolent God exists, he would know about evil, be able to prevent it, and want to prevent it. Premise two: such a being would eliminate evil entirely. Premise three: evil exists. The three cannot stand together, so at least one must give way. This is the logical problem of evil, and it differs from the later evidential problem of evil, which concedes the logical point and argues instead that the sheer amount and distribution of suffering makes God's existence unlikely rather than impossible. The distinction matters because the two arguments require completely different defences.
One detail that circulates online almost every time the paradox does: Epicurus (341–270 BCE) probably never wrote it in this form. The four-horn formulation is a reconstruction, and the earliest verbatim version we have comes from Lactantius, a Christian apologist writing around 300 CE in De Ira Dei ("On the Anger of God"), who attributes it to Epicurus while arguing against him. Whether Epicurus held the view is contested. What is not contested is that the argument as stated—the version pasted into debate forums—is formally valid. If the premises are true, the conclusion follows. The whole fight among philosophers is over whether premise one is true, and that is where free will defences, soul-making theodicies, and skeptical theism all take their swing.
Is the Epicurean paradox logically valid?
Validity and soundness are separate tests. An argument is valid if the conclusion follows from the premises — if you accept the premises, you cannot deny the conclusion without contradicting yourself. It is sound only if the premises are also true. The Epicurean paradox, as Lactantius recorded it around 300 CE in De Ira Dei and as J.L. Mackie restated it in "Evil and Omnipotence" (Mind, 1955), passes the first test cleanly. If God is all-powerful, all-knowing, and all-good, and evil exists, then at least one of those attributes must fail. Nothing in that inference is broken.
The trouble sits in a premise nobody states out loud. To get from "God is all-powerful and all-good" to "evil should not exist," you need a bridging claim: that such a God would have no reason — none at all — to permit evil. That is not a logical consequence of omnipotence or omnibenevolence. It is an independent assertion about what God would do, smuggled in as though it were definitional. Mackie made the bridge explicit by claiming there is no "satisfactory solution" that reconciles the three attributes with evil. The argument is valid. Whether it is sound turns entirely on that bridge.
This is why Plantinga's God, Freedom, and Evil (1974) attacked the premise rather than the deduction. He did not argue that the logic was faulty; he argued that God might have a sufficient reason for permitting evil that no human has ever considered — a reason connected to free will, which requires the real possibility of moral evil if it is to mean anything. Under that reading, "God would eliminate evil" is simply false as stated, and a valid argument with a false premise yields nothing. Alvin Plantinga's version carried enough weight that by the 1980s most philosophers had stopped defending the logical problem of evil and shifted to the evidential version, which William Rowe framed in 1979 in the American Philosophical Quarterly: not that evil is logically incompatible with God, but that some particular suffering looks pointless.
That retreat matters for how you read the paradox. The Epicurean formulation is not refuted — it is disarmed at the premise level. The strongest remaining objection comes from Stephen Wykstra's 1984 paper in the International Journal for Philosophy of Religion, which introduced what he called the Humean obstacle: we are in no position to judge that God has no reason for permitting a given evil, because our cognitive grasp of the relevant goods is roughly the grasp of a one-day-old infant assessing adult finances. If that is right, then the bridging premise fails not because we have found God's reason, but because we cannot rule out that one exists. John Hick's soul-making theodicy (Evil and the God of Love, 1966) and the free will defense each offer a candidate reason; skeptical theism goes further and denies we are equipped to audit the list.
The free will defense: evil as the cost of choice
Alvin Plantinga's 1974 book God, Freedom, and Evil is the reason most philosophers of religion stopped treating the Epicurean paradox as a live threat to theism. The argument in outline: a world containing creatures who freely perform both good and evil is more valuable than a world containing creatures who never do wrong because they were built incapable of it. If that is true, then an all-powerful, all-knowing, all-good God has a morally sufficient reason to permit evil — namely, that eliminating it would mean eliminating the freedom that makes the good worth anything. Free will is not a bonus feature bolted onto a deterministic creature; it is the capacity to choose, and a capacity that cannot be exercised in one direction only is not a capacity at all.
Plantinga's decisive move concerns what "all-powerful" actually covers. It does not mean God can do logically impossible things — make a square circle, or make a being who is free with respect to an action and simultaneously guarantee that the being performs it. So the premise that God could have actualized a world in which free creatures always choose the good is not a straightforward truth about power. It may be that every possible world containing significantly free creatures contains at least some who go wrong. If so, God faces a constrained set of options, and the honest question is not "why didn't he pick the sinless world?" but "was there a feasible world with more good and less evil on balance?" Plantinga's answer, developed across the 1960s and consolidated in 1974, is that we have no reason to think there is. That is a narrow claim, and it is enough. J.L. Mackie's 1955 Mind paper had framed the problem as a strict contradiction, and a defense only has to show the contradiction does not hold.
Natural evil is where the defense gets thinner, and Plantinga knew it. Earthquakes, childhood leukemia and the 2004 Indian Ocean tsunami are not the product of any human choice, so the free will defense as stated covers moral evil directly and natural evil only indirectly — via the claim that a world with free creatures may require regular, predictable natural laws, and that such laws will sometimes produce suffering as a byproduct. That is a real argument, but it is doing heavier lifting than the moral-evil case, and critics from William Rowe's 1979 American Philosophical Quarterly paper onward have pressed exactly here.
Two things are worth separating. Plantinga's defense is a logical rebuttal: it shows the Epicurean paradox is valid but not sound, because its first premise — that a good, all-powerful God would eliminate evil — is the weakest link. It does not claim to explain why God actually permits this particular earthquake or this particular diagnosis. That explanatory project is a theodicy, and the stronger candidates there are John Hick's soul-making theodicy (1966) and skeptical theism, which concedes the evidential problem of evil raised by Rowe but denies we are positioned to judge what reasons God might have. Those come next.
Soul-making theodicy: suffering as a tool for growth
John Hick's Evil and the God of Love (1966) opens with a distinction that most online summaries skip: he was not answering Mackie's 1955 logical problem, because he was not trying to show that evil and God are logically compatible. He was answering the question of why a good God would permit suffering on the scale we actually observe. That places him squarely in the evidential problem of evil, the version William Rowe laid out in American Philosophical Quarterly in 1979, where the claim is not that evil disproves God but that its sheer quantity and distribution make belief in God unreasonable.
Hick's move is to reject the premise that human beings arrive finished. On his account we are not created perfect and then broken by sin, as Augustine had it. We are created as raw material, and virtue is something we have to build. Nobody is born patient, forgiving, or courageous, and no one becomes those things in a world without friction. Hick's image, borrowed from Irenaeus in the second century, is that we are children of God in the making rather than children already made. A universe engineered to eliminate all possible pain would also eliminate the conditions under which compassion, endurance, and moral seriousness exist at all. You cannot be brave in a world with nothing to fear. You cannot forgive an injury that never happened. The trade-off is deliberate: less suffering, less character.
The most quoted piece of the theory is Hick's doctrine of epistemic distance. God, on this view, deliberately keeps his existence ambiguous. If the divine were as obvious as the sun, trust would be pointless and obedience would be coercion. Real relationship requires the possibility of the alternative. Hick argues that the world's religious ambiguity is not a design flaw but the point: the gap between us and God is what makes freely chosen faith even coherent. This is also where the theory draws its sharpest criticism. Critics reply that 2004 Indian Ocean tsunami casualties were not evenly distributed as a character-building curriculum, and that an infant who dies at eight months has no soul to make. Hick's defenders usually concede the infant case and retreat to a general theodicy, one that explains why a suffering-permitting world exists rather than why this particular child suffered this particular death.
Skeptical theism: maybe we just can't see the reasons
By 1979 William Rowe had reframed the debate. His "The Problem of Evil and Some Varieties of Atheism" in American Philosophical Quarterly conceded that Mackie's logical version fails if free will is possible, then argued that the sheer quantity of apparently pointless suffering still makes God's existence unlikely. Rowe called a case "pointless" if no good we can identify justifies it — a fawn burned in a forest fire, say. From "we see no reason for this" he inferred "probably no reason exists." That inference is the target of skeptical theism.
Stephen Wykstra's 1984 paper in the International Journal for Philosophy of Religion made the reply with an analogy that has been cited ever since. A one-month-old infant undergoing a painful medical procedure cannot discern any reason the parent permits the pain. The infant's cognitive equipment is simply not suited to the task; the parent's reasons involve long-term health, germ theory, the necessity of treatment. In Wykstra's terms, the infant is not in a position to draw the inference. If God's mind is to ours as ours is to the infant's, then our failure to see a justifying reason is exactly what we should expect regardless of whether one exists. The slogan: the absence of evidence, under conditions where we would not expect evidence, is not evidence of absence.
Why this blocks the evidential argument, not the logical one
Skeptical theism is a narrower tool than it first appears. It was never needed against Mackie's logical problem, which Plantinga's free will defense already defeated in 1974 by showing that God and evil are not logically incompatible. Skeptical theism instead attacks Rowe's evidential version at its load-bearing premise — the move from "no reason is apparent" to "no reason exists." If human cognitive reach is limited relative to a mind that could have reasons extending across all of history, then that inference is not reliable, and Rowe's probable conclusion is not established. The debate after 1984 turned on whether this cuts too deep: critics like William Alston's respondents and later Michael Bergmann's interlocutors argued that if we cannot judge God's reasons, we also cannot trust our moral judgments about anything.
The honest limitation is that skeptical theism succeeds by making itself unfalsifiable. Nothing could ever count as evidence that God has no reason, since the reply is always that the reason lies beyond our grasp. That is a real cost, and it is why the position is usually defended as a defeater — a reason to withhold the atheist's conclusion — rather than as a positive theodicy. It tells you why you cannot infer God's nonexistence from suffering. It does not tell you why the suffering is there.
How do the three theodicies compare?
The three responses do not attack the same target. The free will defense and soul-making theodicy both concede the premise that God would eliminate evil if he could, then argue he has a good reason not to. Alvin Plantinga's 1974 book God, Freedom, and Evil runs the first route; John Hick's 1966 Evil and the God of Love runs the second. Skeptical theism, set out by Stephen Wykstra in a 1984 paper in the International Journal for Philosophy of Religion, refuses the concession altogether. It says the premise itself — "an all-good God would eliminate evil" — is a claim about God's reasons that human beings are in no position to make.
That difference determines what each one can survive. The free will defense is aimed at the logical problem of evil, the version J.L. Mackie pushed in a 1955 Mind article, and most philosophers of religion now treat it as having done that job: it shows the existence of evil and an all-powerful, all-knowing, all-good God are not formally contradictory. It does much less against the evidential version William Rowe published in 1979, which claims not that evil is logically incompatible with God but that its sheer amount and distribution makes God's existence unlikely.
| Response | Key proponent | Core claim | Main strength | Main objection |
|---|---|---|---|---|
| Free will defense | Alvin Plantinga, 1974 | A world containing free creatures who may choose good is more valuable than one where they cannot, so God cannot remove moral evil without removing the greater good | Widely credited with defeating the logical problem of evil outright by showing no formal contradiction | Leaves natural evil unexplained: earthquakes, childhood cancer and the 2004 Indian Ocean tsunami killed roughly 230,000 people with no free choice involved |
| Soul-making theodicy | John Hick, 1966 | Suffering is a necessary instrument for developing virtue and character, and a world with no hardship would produce no courage, compassion or patience | Handles natural evil directly, since a landslide or illness can shape character as readily as a moral crime can | Implies God is complicit in each instance of suffering, and the scale looks wrong: Hick has to justify why soul-making requires the death of one child rather than a scraped knee |
| Skeptical theism | Stephen Wykstra, 1984 | Our cognitive grasp of possible goods is too limited to judge whether God has a reason for permitting any particular evil | Attacks the premise the other two grant, so it answers both the logical and evidential versions at once | Cuts both ways: the same limits it invokes make confidence in God's goodness, revelation or moral commands harder to sustain, and it can read as an all-purpose excuse |
For most readers arriving from a debate or an introductory course, skeptical theism is the row that survives best, and it is the one doing the real work against Rowe's evidential problem rather than Mackie's logical one. It wins because it targets the weak premise instead of negotiating over it. The flip case is the reader who came in asking whether the paradox is valid: there, Plantinga 1974 is the relevant answer, and skeptical theism is beside the point, because validity is a question about logical form and Wykstra offers no formal argument. If the question is "does the paradox prove too much?", use skeptical theism. If the question is "does the paradox even hold together?", use Plantinga.
Which response actually works?
No answer defeats the Epicurean paradox outright, because the paradox is a valid argument. What each response actually does is attack a different premise. Free will and soul-making attack the inference from "evil exists" to "God would eliminate it." Skeptical theism attacks something more basic: our competence to make that inference in the first place. Ranking them means asking which premise is weakest, and the answer changes depending on whether you are facing the logical version of the problem or the evidential one.
- Against the logical problem, skeptical theism wins. J.L. Mackie's 1955 "Evil and Omnipotence" claimed the existence of evil was logically incompatible with an all-powerful, all-good God. Alvin Plantinga's 1974 God, Freedom, and Evil answered with the free will defense, and by the early 1980s most atheist philosophers, including Mackie himself, conceded the logical version had failed. Skeptical theism arrived as cleanup: it explains not just why God might permit evil, but why we would be unable to see the reason.
- Against the evidential problem, the gap narrows but does not close. William Rowe's 1979 paper in American Philosophical Quarterly conceded the logical version was dead and shifted ground: the sheer quantity of apparently pointless suffering makes God's existence improbable, not impossible. Stephen Wykstra's 1984 reply in the International Journal for Philosophy of Religion introduced the "parent analogy." A one-month-old receiving a vaccination sees only the pain; the parent sees the disease being prevented. Wykstra's claim is that our cognitive distance from God is at least that large.
- The free will defense cannot cover natural evil. Plantinga's defense is built for moral evil, the suffering caused by human choice. Childhood leukemia in a two-year-old in a country with no relevant medical history has no chooser behind it, and calling it the work of demons or of some free non-human agent is a stretch most philosophers will not grant. This is why Plantinga's own framing is a defense, not a full theodicy: it shows a possible reason God might allow evil, not why so much of it exists.
- Soul-making is strong on growth, weak on scale. John Hick's 1966 Evil and the God of Love argues that a world without suffering would be a world without opportunity to develop character, courage, or compassion. That carries weight for adversity a person can eventually integrate. It founders on a four-month-old with leukemia, and on the roughly 5.9 million children under five who died in 2015 according to the WHO and UN IGME estimates. Hick, a Christian, admitted that any soul-making theodicy ending at death cannot possibly justify that volume of suffering for people who never had a chance to grow from it.
- Skeptical theism is self-consistent but corrosive. If we cannot judge what a good God would permit, the move protects theism at the cost of every other moral judgment. The same limited reasoning that says "we cannot know why God allows this" also says "we cannot know that God is good," which is the position Ivan Karamazov reaches, not a defense of faith. Theologians from John Hick onward have flagged this worry: the defense works, but it may cost more than it saves.
- The evidential case is where most contemporary work sits. Mackie's logical problem is widely treated as settled; Rowe's evidential problem is not. That means the practical question for a philosophy student today is not "Does the Epicurean paradox refute God?" but "How much apparently pointless suffering would it take to make God's existence improbable?" No one has a clean answer, and the honest position is that both sides argue from incomplete information.
The item people most often get wrong is the ranking itself. They treat "which response works" as a single ladder, then get confused when Plantinga appears to win one debate and lose another. He does both: he beats Mackie's logical problem outright and offers little to Rowe's evidential one. Skeptical theism sits at the top of both lists, but for an unflattering reason. It wins by limiting what we can claim to know, and if you accept it to save theism, you have also given up your grounds for calling anything God does good. That trade-off is the argument, and it is not resolved by citing Epicurus's 341–270 BCE lifespan or Lactantius's c. 300 CE record in De Ira Dei. The paradox identifies a real tension. The responses manage it. None of them closes it.
What do critics say about these answers?
The free will defense has always had a natural-evil problem, and Plantinga knew it. His 1974 book defends the logical consistency of God and moral evil; it says much less about a tsunami, a paediatric cancer ward, or the 2004 Indian Ocean earthquake that killed roughly 227,000 people. Critics press hard on the asymmetry: if free will requires the genuine possibility of choosing wrong, that justifies a murderer's choice, but it does not justify the ground opening under a village where no agent chose anything. Plantinga's later move is to invoke fallen angels as free agents behind natural evil, which works as a logical possibility but strikes most readers as a patch rather than an explanation. Then there is quantity. Even granting that some suffering is the price of freedom, the ratios look wrong: a child who dies of bone cancer at four had no meaningful exercise of will in the matter, and the sheer volume of animal suffering across 500 million years of predation predates any human chooser.
Soul-making faces a sharper objection, and John Hick himself conceded its shape. His 1966 Evil and the God of Love argues that a world without hardship would be a world without the conditions for moral and spiritual growth, which is a plausible account of adversity in general. The problem is dysteleological suffering: pain whose severity or duration or randomness makes growth impossible rather than possible. A person tortured into psychosis, a genocide that destroys the moral community it was supposedly meant to build, an infant's suffering before any self exists to be shaped. Hick's response is that soul-making may extend into an afterlife, which rescues the coherence of the claim but removes most of its evidential force in this life. The objection, stated plainly: a theodicy that needs the next world to make this one tolerable is doing very little work in this one.
Skeptical theism draws the most uncomfortable criticism, because it seems to cut the branch it sits on. Stephen Wykstra's 1984 paper in the International Journal for Philosophy of Religion argues that we are in no position to judge whether God has reasons for permitting suffering, since a finite mind cannot survey an infinite good's implications. William Rowe had argued in 1979 that some suffering looks pointless, and Wykstra replies that "looks pointless to us" is exactly what we would expect if there were a point beyond our ken. Fair enough. But critics like Michael Bergmann's own respondents and, more combatively, William Hasker, reply that the same move undermines every moral judgment we make. If we cannot see the goods that might justify an infant's agony, on what grounds do we call anything gratuitously wrong? Skeptical theism also functions as a conversation-stopper: any counterexample you produce is met with "God may have reasons you cannot see," which is unfalsifiable and therefore explains nothing.
None of these three answers is dead, and none is universally accepted. The free will defense remains the strongest reply to the logical problem and the weakest reply to the evidential one that Rowe formulated. Soul-making does real work on ordinary adversity and stalls on the extreme cases. Skeptical theism survives the logical problem with room to spare but pays for it in moral epistemology. If you want the honest summary: the debate shifted after Mackie's 1955 paper not because anyone solved it, but because the burden moved from "is this contradictory?" to "what does the evidence suggest?" — and that second question is still open.
Frequently Asked Questions
What is the Epicurean paradox in simple terms?
If God is all-powerful, all-knowing, and all-good, then evil should not exist. Evil does exist. So either God cannot prevent it (not all-powerful), will not prevent it (not all-good), or does not know about it (not all-knowing). Each option contradicts classical theism, which holds all three attributes together. The argument is usually dated to around 300 CE, when Lactantius recorded it in De Ira Dei, though the wording most people quote today comes from David Hume's Dialogues Concerning Natural Religion (1779).
Did Epicurus actually write the paradox?
No. The paradox is attributed to Epicurus by the Christian writer Lactantius around 300 CE, roughly 570 years after Epicurus died in 270 BCE. None of Epicurus's surviving texts — the Letter to Menoeceus, the Principal Doctrines, or the Vatican Sayings — contain the argument in this form. The four-part phrasing familiar today traces to Hume, who put it in the mouth of the character Philo rather than citing a Greek source.
What is the best answer to the problem of evil?
Skeptical theism is the most defensible response, because it denies the premise that we can tell whether God lacks reasons for permitting evil. If a mind is vastly greater than ours, our failure to see a reason is weak evidence that none exists. The honest cost: the same logic blunts our ability to judge any apparent divine goodness, which critics like Michael Bergmann's respondents have pressed since the 2000s.
What is the difference between the logical and evidential problem of evil?
The logical problem claims evil is logically incompatible with God, so both cannot exist. The evidential problem claims evil makes God's existence unlikely without claiming strict contradiction. The logical version is now widely abandoned in analytic philosophy after Alvin Plantinga's free will defense (1974), which showed no formal contradiction follows. William Rowe's 1979 evidential argument, built on "noseeum" inferences, is the version still actively debated.
Does the free will defense explain natural evil?
Not directly. Plantinga's free will defense addresses moral evil — suffering caused by human choices — and says nothing about earthquakes, childhood leukemia, or a 2004 Indian Ocean tsunami that killed roughly 227,000 people. Some theists extend the defense by attributing natural evil to the free choices of fallen angels or to a fallen world, but that move is contested even among sympathetic philosophers and generally treated as weaker than the original argument.
What is skeptical theism?
It is the position that humans cannot reasonably infer God has no reason to permit evil merely because we cannot see one. Our cognitive limits, the argument runs, make such an inference unreliable. It blocks evidential arguments from evil rather than solving evil. The standard objection, raised by Rowe and others from the 1980s onward: the same skepticism would undermine ordinary moral knowledge, since we judge good and evil by the same faculties.
