Introduction
In philosophy of religion in the Western world, arguments about whether or not there is a god tend to focus on classical theism.1 In classical theism, God is understood to be perfect in every way. Importantly, this perfection is supposed to include at least three characteristics: omnipotence (being all-powerful), omniscience (being all-knowing), and omnibenevolence (being all-good).
“The problem of evil” is the name given to a particular set of arguments against the existence of the God of classical theism. This set of arguments generally is divided into subsets: there are variations on what is known as the logical problem of evil, and variations on what is known as the evidential problem of evil.
The tendency to refer to these arguments as “problems” may have come from the fact that the canonical presentations, such as those in Epicurus and Hume, were posed in the form of a series of questions rather than as arguments. Today, many philosophers of religion prefer to drop the traditional label and instead speak of “arguments from evil.” I will go back and forth between the different usages.
One last point about terminology before we continue: when you see the word “evil” in an argument from evil, it generally refers not only to literal evil (like evil people or actions), but more broadly (and somewhat archaically) to bad things like suffering and injustice.
The logical problem of evil
A typical example of the logical problem of evil runs as follows:
(L1) If God exists, then God is omnipotent, omniscient, and omnibenevolent.
(L2) If God is omnibenevolent, then God wants to prevent all evil.
(L3) If God is omnipotent and omniscient, then God can prevent all evil.
(L4) If God wants to prevent all evil and can prevent all evil, then evil does not exist.
(L5) Evil exists.
Therefore,
(L6) God does not exist.
Statements L1 through L5 above are the assumptions that the argument starts with (the so-called premises of the argument). Statement L6 is the conclusion of the argument. This argument is what is known as a deductive argument: an argument which purports that its premises alone entail its conclusion with absolute certainty.
You probably can see how the premises are supposed to entail the conclusion: from L5 and L4, it follows that God either does not want to prevent all evil or cannot do so. If God does not want to prevent all evil, then from that and L2 we know that God is not omnibenevolent. If God cannot prevent all evil, then from that and L3 we know that God either is not omnipotent or not omniscient. So, God either is not omnibenevolent, or not omnipotent, or not omniscient. But from this and L1, we get the conclusion that God does not exist.
There is no problem with the logic of the argument: if the premises are true, then the conclusion must be true. (When this happens—when an argument is structured so that the truth of the premises alone would guarantee the truth of the conclusion—logicians say that the argument is valid.) But for an argument to actually establish its conclusion, the argument must also have true premises. (When a deductive argument is valid and has true premises, we say that the argument is sound. Sound arguments are the gold standard in philosophy: if you have a sound argument for some conclusion, then you have conclusively proved the conclusion.) So, are the premises true?
Some objections to the logical problem of evil
Most objections to the logical problem of evil focus on premise L2.2 The general strategy of such objections is to argue that it is possible that (a) there are goods whose goodness outweigh the badness of evil, and (b) that closing off the possibility of evil would also close off the possibility of those goods, so even an omnibenevolent god would permit the evils in order to make the goods possible.
Some such objections try to name a specific good that has the needed characteristics. For instance, a popular version of the free will defense holds that the goodness of having free will outweighs the badness of any evil that people might freely do, and that one can guarantee that there is no evil only by removing people’s free will altogether. Similarly, the soul-making theodicy holds that there are certain characteristics such as courage and compassion that we never would be able to develop in a world without evil, and that having the opportunity to develop these characteristics outweighs the badness of the evils that make such opportunities possible. Responses to the two contend either that the alleged good cannot outweigh the required evil, or else that the alleged good can be guaranteed even without the supposedly required evil.
At least one objection, the unknown purpose defense, does not try to specify a good that has the needed characteristics. It simply argues that for all we know there is such a good, and shifts the burden of proof back to proponents of the logical problem of evil to show that there can’t be any such good.
Remember that the logical problem of evil purports to show that based on the premises alone there is no chance at all of the conclusion being false. This means that if any of the scenarios above is possible, no matter how remote the possibility, then the argument fails. Because of this, most philosophers today think the logical problem of evil has not been shown to work, and many think it has been refuted outright. Given this sentiment, attention has tended to turn to the evidential problem of evil.
The evidential problem of evil
The key difference between logical and evidential arguments from evil is that the latter do not try to demonstrate with certainty that God does not exist. Some, such as the one we will look at, try to show that the existence of God is improbable. Others try to show only that theism is less probable than some competing metaphysical view like naturalism. Yet others try to show simply that the probability that God exists is lower than one might initially think.
Here is a basic evidential problem of evil:
(Def) Gratuitous evil is defined as evil that has no good purpose—evil that is not necessary for any overriding good.
(E1) If God exists, then God is omnipotent, omniscient, and omnibenevolent.
(E2) If God is omnibenevolent, then God wants to prevent all gratuitous evil.
(E3) If God is omnipotent and omniscient, then God can prevent all gratuitous evil.
(E4) If God wants to prevent all gratuitous evil and can prevent all gratuitous evil, then gratuitous evil does not exist.
(E5) There are evils which as far as we can tell are gratuitous.
Therefore, probably,
(E6) God does not exist.
The logic of this argument is not too difficult: from E5 it presumably follows that some evils probably actually are gratuitous. From that fact and E1 through E4 it follows, in a manner parallel to the logic of the logical argument from evil above, that God probably lacks one of his essential characteristics, and therefore does not exist.
An objection to the basic evidential problem of evil
The responses that generally are thought to work for the logical problem of evil do not work against the evidential problem of evil sketched above. One cannot respond by saying that God may have a good reason for allowing gratuitous evil, because gratuitous evil is by definition a kind of evil for which there can be no good reason. One cannot respond simply by pointing out that there may be some good reason for any apparent example of gratuitous evil, because the argument already concedes this. So how does one respond?
One popular strategy, called skeptical theism, calls into question the implicit move from for all we can tell gratuitous to probably actually gratuitous. It does so in a way that is reminiscent of the unknown purpose defense, but goes further. According to skeptical theism:
(S1) We are justified in inferring probably really X from for all we can tell X only if we can reasonably expect that we would be able to tell if X were not the case.
(S2) Because of our finitude compared to God’s, we cannot reasonably expect that we would be able to tell if there were some good purpose for which God permits seemingly gratuitous evil.
From S2 it follows that we cannot reasonably expect that we would be able to tell if the evils we were looking at actually were not gratuitous. From this and S1 it follows that we are not justified in inferring that any evil actually is gratuitous simply because for all we can tell it is. But if that inference is not justified, then it would seem that the truth of premise E5 provides no evidence at all for the truth of the conclusion.
The most natural worry about skeptical theism is that if one accepts S1, then it is difficult to see how one avoids wider skepticism about making inferences from appearances. For instance, there would seem to be nothing a god possibly could do that would count as evidence of his goodness, since our finitude means that we could not reasonably expect to be aware of any evil purpose a god might have for seeming to do good things. Nothing we observe could count as evidence for an external physical world, for other human minds, or for the continuation of basic patterns into the future, since our finitude means we could not reasonably expect to be aware of any good purpose for which God might systematically deceive us about such appearances. Even worse, there actually seems to be no essential role played by God in skeptical theism: all of the heavy lifting is done simply by the fact that our knowledge is finite enough that if appearances were sufficiently and systematically enough mismatched to the way things are behind the scenes, there is no way we would be able to tell. In short, if S1 is true, then it is unclear whether we can justifiably infer anything from the way things appear: to invoke S1 to defeat the evidential problem of evil might amount to a nuclear option that leaves nothing standing, including most of one’s own position.
Is there a way around these concerns that does not involve special pleading? Proponents say there is; critics are not so sure. I leave it to you to further consider and investigate for yourself.
Notes
1 This is pronounced THEE-ism, with a hard “th” (as in “thought” or “thin”).
2 Not all. Augustine, for instance, attacks premise L5 by claiming that only good things really exist, and that evil is just the absence of good.