In his Dialogues Concerning Natural Religion, David Hume alludes to an ancient version of the argument from evil, due to the Roman philosopher Lucretius:
The main defect in this argument lies in the second premise. It is not at all obvious that a perfectly good being must create a world utterly devoid of evil. This might follow if we thought that evil was a positive quality that had to be caused to exist, since it would presumably in that case take a partially evil cause to produce a partially evil effect. However, if we accept the privative theory of evil, that route to premise two is denied us.
An alternative to Lucretius's argument seeks to spell out why a good and omnipotent being would create a flawlessly good world. This alternative is so widely used that it deserves to be called the Standard Argument from Evil.
There are a couple variations on the standard argument that we will encounter later. These variations concern premise 4, 5 and 6. A probabilistic version of the argument would replace premise 4 and 5 with something like this:
4/5p. It is very probable that the best possible world an omnipotent being could actualize would be a world devoid of evil.
There is also a quantitative version of the standard argument, in which the crucial fact is not simply the existence of some evil, but the existence of a certain quantity of evil. In this version, premises 4, 5 and 6 would be replaced by:
4/6q. The best possible world an omnipotent being could actualize would have fewer or lesser evils than the actual world does.
It would also be possible to combine the quantitative and probabilistic variations. However, as we shall see, theistic defenses that are successful against the standard argument tend to be successful against these variations. If the standard argument fails, it is difficult to find convincing reasons to accept 4/5p or 4/6q.
I am going to assume that theists will accept premises 1 and 6. I am assuming that we are dealing with a eutheist who believes in an omnipotent God and who accepts the real existence of evil. Such a theist can respond to the standard argument in either of two ways: the tough-minded approach and the tender-hearted approach. The tough-minded response involves challenging premises 2 or 3. The tender-hearted response instead challenges premises 4 or 5. It is also possible for the theist to employ both responses, challenging 2, 3, 4, and 5.
A. Simplest version: for any declarative sentence p, God can bring it about that p. This would entail that God can bring about true contradictions, change the past, bring about His own eternal non-existence, etc. Radical possibilism of Descartes.
B. Lewis rejects the simplest version: God can do anything, but an intrinsic impossibility is a non-entity, not a "thing" at all.
C. The paradoxes of omnipotence. Can God create an absolutely immovable object, or an absolutely indestructible one (i.e., one that even He cannot move/destroy)? If we say, Yes, then it seems that God could lose His omnipotence, that omnipotence is not essential to God. If we say, No, we seem to be denying God's omnipotence in the actual world. Perhaps the best answer: God has essentially the greatest power that it is possible for anything to have essentially. This makes it difficult to resolve the paradoxes -- would it be greater to be able to create an indestructible object, or not to be able to do so?
D. An alternative account -- in terms of the power to create situations, states of affairs: x is omnipotent iff, for all ç, ff ç is a possibly-instantiated, intrinsic situation-type, then x can actualize a situation-token of type ç.
Examples of non-intrinsic types: a rock too heavy for God to lift, a rock that God did not create, an action freely chosen by a human creature.
To show that God can instantiate a non-intrinsic type T, we must do the following:
Find an intrinsic type T', and actual circumstances C, such that if God were actualize a token x of type T' in C, we can prove that x would also be of type T.This is a non-trivial task. Merely invoking omnipotence is not enough.
A. Both Swinburne and Lewis appeal to means/ends reasoning in explaining why God permits evil: God permits suffering in order to provide humans with the knowledge they need to exercise responsible freedom (Swinburne), God permits evil in order to transform us int a state of true lovableness (Lewis).
B. An obvious objection: couldn't an omnipotent God simply produce the desired end-state directly, by simple fiat? An omnipotent God is not limited to ordinary means, so why does He have to pay the price of permitting evil and suffering?
C. This overlooks God's aim. God is not interested merely in the final state of the universe, but in how the universe moves toward that final state. It is important to God that the process by which His desired end-state is reached be one which is appropriate to the created natures of the creatures involved.
A. If we assume that utilitarianism is true -- that a morally good agent cares only about maximizing pleasure and minimizing pain -- then the arguments from evil are unanswerable.
B. To be successful, the argument from evil must do one of the two following things: