What can Law's Evil God Challenge Do?

Stephen Law debated William Lane Craig on Monday night. I didn’t go, though I would have liked to. I did, however, get the chance to read Law’s contributions to the debate (which he posted on his blog) as well as some other internet reviews and discussions of the event. Public debates of this sort are never particularly illuminating — if you want to be illuminated there’s no substitute for reading the primary academic literature and thinking about it for yourself — but they may serve some purpose. They may encourage people to view something from a different perspective, to take their first steps toward self-learning, and to appreciate different viewpoints.

As might be expected, there is much bias affecting the internet reviews and discussions of Monday night’s event. Still, from what I’ve read it seems that Law did relatively well in the debate, at least when compared to Craig’s former opponents. He had a strategy, he stuck to it, and didn’t get bogged down in peripheral issues (for instance, he studiously avoided any discussion of Craig’s Kalam argument). That said, there is some confusion about the precise nature of his strategy and what it was intended to do.

The strategy in question centred on the Evil God Challenge (the EGC, for short). This is something Law has written about in the past, and his output on it includes a rather good article published a couple of years back in the journal Religious Studies. The problem with Law’s use of the EGC is that the EGC is open to several interpretations. In this post, I want to review three plausible interpretations of the EGC and pass some comments on which I think is the strongest. Before getting to those three interpretations though, I want to briefly outline the main features of the EGC.


1. The Main Features of the Evil God Challenge
The EGC is, in effect, a clever inversion of the traditional evidential problem of evil. The evidential problem of evil runs something like this:

• (1) If an all-powerful, all-knowing, all-good being existed, that being would not allow gratuitous suffering to occur.

• (2) Gratuitous suffering probably does occur.

• (3) Therefore, (probably) an all-powerful, all-knowing, all-good being does not exist.

The argument is evidential in that it does not claim that there is a logical inconsistency between God and evil, but, rather, that there is strong evidence against the theistic hypothesis. There are two key elements to the argument. The first is the meaning of the term “gratuitous suffering”, which is simply “suffering that does not help to secure some greater good or prevent some greater evil”. The second is the truth or falsity of premise (2).

In responding to premise (2), theists can adopt one of two strategies: (i) they can argue that every apparently gratuitous instance of suffering does, in fact, serve some greater good (or prevent some greater evil); or (ii) they can argue that we have no way of knowing whether or not they serve some greater good. The first of these strategies can be called constructing a theodicy and the second can be called donning the skeptical theist’s hat. In responding to the theist, the proponent of the evidential argument will either: (a) argue that none of the profferred theodicies do account for observed instances of suffering (there’s so much of it, after all); or (b) point to the worrying implications of skeptical theism. No doubt theists will remain unconvinced.

This is where the EGC comes into play.

As I mentioned above, the EGC inverts the evidential problem of evil. That is to say: it presents us with an evidential problem of good. It does so by first asking us to imagine that an all-powerful, all-knowing, all-evil being exists (call it “Evil God”). It then highlights all the wonderful, joyous events that occur in the world, events that don’t seem connected to any greater evils (call them “gratuitous joys”). And finally, it suggests that these gratuitous joys provide evidence against the existence of Evil God. As follows:

• (4) If an all-powerful, all-knowing, all-evil being existed, that being would not allow gratuitous joy to occur.

• (5) Gratuitous joy probably does occur.

• (6) Therefore, (probably) an all-powerful, all-knowing, all-evil being does not exist.

Again, the argument can be responded to in one of two ways. First, the believer in Evil God can argue that every apparently gratuitous instance of joy does, in fact, serve some greater evil (call this constructing an e-theodicy). Or second, the believer in Evil God can argue that we have no way of knowing whether or not apparently gratuitous joys serve some greater evil (call this donning the skeptical e-theist’s hat). The Evil God skeptic will respond by either (a) arguing that none of the profferred e-theodicies do account for observed instances of joy (there’s so much of it, after all); or (b) pointing to the worrying implications of skepticial e-theism.

At a first pass, the EGC is simply the presentation of this inversion in as clear and detailed a manner as possible (Law’s original article certainly does this). It is then left to the theist to work out the implications of the inversion for their own belief system. But this is unsatisfactory. It would be better if we could spell out in more detail exactly how the EGC should affect the theist’s belief system. So let’s look at three possibilities.


2. The EGC as a Reductio of Theism
The first possibility is to present the EGC as a reductio of classical theism. Classical theists believe in the existence of an all-powerful, all-knowing and all-good being. The problem is that while their standard arguments (cosmological, design etc.) may establish the existence of a creator God, they do nothing to establish that the creator is all-good. Hence they must do something more to show that the creator is all-good. This brings the EGC challenge into play since it suggests that the case for an Evil God is just as strong as the case for a Good God.

This is where things get interesting. When confronted with the idea of an Evil God, most classical theists will be inclined to simply dismiss it as being absurd. But if all other arguments fail to support the existence of a Good God, and if the evidential problem of good is just as strong as the evidential problem of evil, shouldn’t they then accept that idea of a Good God is absurd too? In other words, shouldn’t they accept the following argument:

• (7) The EGC shows that the evidence in favour of the existence of a Good God is roughly equal to the evidence in favour of the existence of an Evil-God.

• (8) The case for Evil God is absurd.

• (9) Therefore, (probably) the case for Good God is equally absurd.

This seems like a reasonable enough way in which to run the EGC. The problem is that classical theists have some responses. They can argue against (7) by suggesting that the evidence is not perfectly equal (or that there are other arguments for the goodness of the creator); or they can argue that (9) does not follow from (8) because the concept of a Good God is crucially disanalogous from the concept of an Evil God; or, they can even argue that the concept of an Evil God is not absurd.

Law’s original article on the EGC considers these kinds of responses, which suggests that he too originally understood the EGC as a reductio of classical theism. (To be fair, his approach was softer: he argued that both were equally unreasonable, not equally absurd). But in clarifying his argument from Monday night, I think he gestures toward an alternative interpretation.


3. The EGC as Support for the Evidential Problem of Evil
Go back, for a moment, to my earlier presentation of the evidential problem of evil. As you recall, the key premise there was:

• (2) Gratuitous suffering probably does occur.

This premise is supported by identifying instances of suffering that seem pointless to us, applying the principle of warranted induction (i.e. what seems to be the case probably is the case), and concluding that there probably are genuinely pointless instances of suffering. Generally, classical theists do not accept this line of reasoning. While many of them will accept that there are instances of seemingly gratuitous suffering, they will reject the idea that we are warranted in jumping from a claim about what seems to be the case to a claim about what actually is the case.

Now look at the key premise in Law’s evidential problem of good:

• (5) Gratuitous joy probably does occur.

This premise is structurally equivalent to premise (2). That is to say, we would support it in exactly the same way: by first identifying instances of joy that seem (to us) not to serve any greater evil, by applying the principle of warranted induction, and by concluding that there probably are genuinely gratuitous instances of joy. But what do theists say in response to this? Do they reject the inference in the same way that they rejected the inference in the evidential problem of evil?

Law suggests that many will accept the inference in the case of the evidential problem of good. But this means that they ought to accept the inference in the original argument too. In other words, it implies that they ought to be persuaded by the evidential problem of evil. In more long-winded fashion, we say the following:

• (10) When confronted with the EGC, classical theists accept the principle of warranted induction.

• (11) The principle of warranted induction states that if a defined reference class of events in the world seems, after a reasonable search, to have a certain quality Q, then it probably does have that quality Q.

• (12) A reference class of events in the world seems, after a reasonable search, to have the quality of gratuitous suffering.

• (5*) Therefore, (classical theists ought to accept that) gratuitous suffering probably does occur.

The idea that the EGC can be used to reinforce the evidential problem of evil is an attractive one, but it is also somewhat suspicious. After all, part of the strength of the EGC was its suggestion that the evidential case for an Evil God is just as good as the evidential case for a Good God. To turn around and then use the EGC to support the evidential case against a Good God is downright odd. This is does, however, open up a pathway to a third possible interpretation of the EGC.


4. The EGC as Part of the Argument for Indifference
Paul Draper’s 1989 article “Pain and Pleasure: an Evidential Problem for Theists” suggested that the debate between theists and non-theists could be construed as a debate between those who think the universe has some morally good creator behind it (theists), and those who think it it is morally indifferent (non-theists). In other words, between those who think the Hypothesis of Indifference (HI) is, on the available evidence, more likely to be true than the Hypothesis of Theism (HT).

I think Law’s EGC could be employed as part of the evidential case for HI. How so? Well, go back to the idea of the principle of warranted induction. This principle states that we can make warranted inferences about the actual nature of a defined reference class of events by performing a reasonable survey of the members of that reference class. So we must ask: When trying to make warranted inferences about the moral character of the universe, what is the relevant reference class of events? I would argue that the relevant reference class of events is all events affecting sentient beings. Why only those events? Because under most conceptions of morality, it is only such beings that have moral significance.

What Law’s EGC does (when read in its totality) is show us that the reference class of events involving sentient beings appears to have a balanced set of moral qualities. Yes, it is true that some events have the quality of gratuitous suffering, but it is also true that a roughly equal number have the quality of gratuitous joy. In other words, the events in the relevant reference class seem to have equal amounts of the quality Q (goodness) and ~Q (badness).

But wouldn’t that imply that the universe has no discernible moral character? And isn’t this exactly what we would expect if the hypothesis of indifference were true?

I’m not entirely sure how to move from these rhetorical questions to a proper argument, but maybe something like this could work.

• (11) The principle of warranted induction states that if a defined reference class of events in the world seems, after a reasonable search, to have a certain quality Q, then it probably does have that quality Q.

• (13) The moral character of the universe should be observable in the class of events affecting sentient beings.

• (14) If HT is true, then we would expect there to be significantly more goodness than badness in the class of events affecting sentient beings.

• (15) If HI is true, then we would expect the class of events affecting sentient beings to have roughly equal amounts of goodness and badness in it.

• (16) If HI is probably true, then HT is probably false.

• (17) The EGC suggests that the reference class of events affecting sentient beings, after a reasonable search, seems to have roughly equal amounts of goodness and badness.

• (18) Therefore, the reference class of events affecting sentient beings probably does have equal amounts of goodness and badness in it (from 11 and 17).

• (19) Therefore, HI is probably true (from 15 and 18).

• (20) Therefore, HT is probably false (from 16 and 19).

I haven’t mapped this out because it’s incomplete (for instance, premises 13 and 14 don’t feature in any of the argument’s derivations so I’m wondering what they’re doing there). But if it could be completed, I think it would be the most promising and most persuasive way in which to employ the EGC. That said, it would certainly still be open to critique. For instance, (15) is particularly problematic. The claim it makes is that roughly equal amounts of goodness and badness imply indifference, but that isn’t necessarily true. After all, we would expect the same if Manicheanism were true. So how can I say HI is supported by that evidence more than Manicheanism (or even some forms of polytheism)?

To sum up, I think there are at least three different interpretations of the EGC. Each has its strengths and weaknesses. The reductio interpretation could be rhetorically effective but there are some well-known responses. The “supporting the evidential PoE-interpretation is intriguing but, ultimately, suspicious. And the indifference-interpretation is, to my mind, the most promising but also the most incomplete. More work needs to be done on the concept of moral indifference, and the expectations we would have from a morally indifferent universe, for it to become persuasive.