The third option states that the relevant difference between logical and evidential formulations of the problem of evil is that the logical argument from evil arrives at a
‘stronger’ conclusion, and the extra strength of its conclusion is due to the fact that it makes a stronger claim to the truth of its premises. Specifically, according to this option, logical formulations claim that the premise stating the incompatibility between God and (for the sake of argument) a certain type of evil is necessarily true, whereas evidential formulations only say that it is likely. So, to put it another way, logical formulations claim that ‘given the existence of this type of evil, God definitely doesn’t exist’, whereas evidential formulations limit themselves to the claim that ‘given the existence of this type of evil, God probably doesn’t exist’. That is, precisely inverting the kind of scepticism that Rowe and Tooley demonstrated regarding the judgement that any one instance of evil be of a kind that is incompatible with God’s existence (whilst being certain enough that the underlying incompatibility premise is true), these evidential formulations assert a scepticism as to the truth of the incompatibility premise whilst remaining certain enough of their judgement that such a type of evil exists.
Given the evident similarities between this option and what I said in response to Rowe and Tooley, it is clear that a lot of my arguments given there apply here also. That is, being sceptical as to the certainty of the truth of a premise in an argument does not make that argument invalid, nor threaten the validity of any underlying problem. There is still a logical
29 Michael Tooley, ‘The Problem of Evil’ (emphasis added).
inconsistency between the underlying propositions. Further, the claim that it is necessarily true that God’s existence is incompatible with a certain kind of evil amounts to the same thing as saying that it is DVATWD, and so any argument that we should shift to evidential formulations on these grounds becomes subject to the same counter-arguments outlined against that position; it is unreasonable to require any premise to be necessarily true in order for an argument to qualify as logical. I will not repeat myself here.
Similarly, much of what I said in response to option 2 applies here also. I do not see any evidence of understanding the relevant difference between logical and evidential formulations in this way in Rowe or Tooley, nor Plantinga, nor anyone else who has contributed significantly to the literature. Neither is it a burning issue for those who are not involved in the philosophical debate, yet consider the problem of evil to be a threat to religious belief. There is, therefore, a lack of evidence for the claim that this option is widely adopted. Should it be taken nonetheless, I reiterate that it would make nothing beyond a semantic difference to my overall thesis; the problem of evil would remain logically binding, relying upon a notion of logical inconsistency, and therefore remain unavoidable, even if we were to settle upon labelling it an ‘evidential’ formulation (since it would not claim any component of the problem to be necessarily true).
1.4.1 A specific response to option 3
However, there is a criticism that we can add as a specific response to this option, and it begins by noting the argumentative move made in my discussion of option 2. There, we noted that (what some people perhaps mistakenly believe to be) logical formulations are (erroneously) reported to claim that God is incompatible with evil per se, whilst evidential formulations weaken this claim to only extend the incompatibility to a certain kind of evil.
The argumentative move to go from absurd logical formulations to reasonable evidential formulations is to simply specify a certain type or class of evils, such that they are more obviously incompatible with the existence of God. Option 3 now understands logical formulations to claim this incompatibility premise to be necessarily true, whereas evidential formulations only claim that it is probably true that God’s existence is incompatible with certain types of evil. Should we wish to preserve logical formulations under this understanding, we would need to find a way to make that incompatibility premise necessarily true. And we can achieve this, simply by repeating the same argumentative step that we saw
in option 2. That is, we further specify which class of evils are incompatible with God’s existence.
The first move of that kind, seen in the discussion of option 2, was to move from an unreasonable ‘all evil/evil per se’ to a more reasonable ‘some evil’, probably where that
‘some evil’ is understood as being the type or class of evil that is particularly bad. Moving on from this foundation, option 3 now seems to say something to the effect that logical formulations look like this:
P1. There exists a maximally-good, maximally-powerful creator of the universe.
P2. A maximally-good, maximally-powerful creator of the universe would not create or permit any particularly bad evil in its creation.
P3. Some particularly bad evil exists.
Whereas evidential formulations look like this:
E1. There exists a maximally-good, maximally-powerful creator of the universe.
E2. Probably, a maximally-good, maximally-powerful creator of the universe would not create or permit any particularly bad evil in its creation.
E3. Some particularly bad evil exists.
That is, the relevant difference between them is the degree to which we claim P2/E2 to be true. Logical formulations assert it as true (claim it as necessarily true), whereas evidential formulations only assert it as likely (contingently true). To save the logical formulation according to this option, we must make P2/E2 necessarily true, and we can do this quite easily by simply repeating the move made in the discussion of option 2. That is, we further specify the type of evil that is incompatible with God’s existence. (People discussing the problem of evil tend to do this fairly intuitively, given that the usual practice is to set the incompatibility premise in stone, and then discuss the ‘evil exists’ premise.)30
So, for example, we can further specify that God’s existence is not necessarily incompatible with ‘particularly bad’ evil, but remains necessarily incompatible with
‘pointless’ or ‘gratuitous’ evil (that is, evil that the permission of which is not justified by a morally sufficient reason). Should the evidential formulator still claim that this is not necessarily true, then we can repeat the move and further specify that God’s existence,
30 As mentioned earlier, William Rowe, for example, seemed to take his incompatibility premise to be uncontroversially true.
though not necessarily incompatible with pointless evil, remains incompatible with
‘unconscionable’ evil (that is, evil that is beyond justification by morally sufficient reasons, and should under no circumstances be permitted by any being who has the power to do so).
Either of these moves, should the evidential formulator accept the resulting incompatibility premise as being necessarily true, will push the debate back to the issue of whether or not such a type of evil occurs in the world, and this will leave us back in the position of Rowe and Tooley, and, according to this way of understanding the difference between evidential and logical formulations, back dealing with a logical formulation once again. Should the evidential formulator continue to reject the necessary truth of the incompatibility premise (which by now will be becoming very difficult), then we can repeat the move ad infinitum, until we eventually get to a merely structural definition of the type of evil that we are interested in. That is, something that looks like this:
P1. There exists a maximally-good, maximally-powerful creator of the universe.
P2. A maximally-good, maximally-powerful creator of the universe would not create or permit any evil (x) in its creation, where (x) is such that is incompatible with God’s existence.
P3. Some evil (x) exists.
P2, in this instance, becomes simply tautologous, and therefore necessarily true. This would render the problem of evil a valid ‘logical’ problem, according to option 3. The crucial premise of P3 would now be stating something akin to ‘some evil exists that a maximally-good, maximally-powerful creator of the universe wouldn’t allow’, which is a perhaps roundabout way of saying things, but remains a meaningful proposition and is arguably the underlying intuition in any formulation of the problem of evil.
So even if we take option 3 to be a serious option, it is easy to save a logical formulation under this understanding, simply by further specifying the type or class of evils that God’s existence is taken to be incompatible with.31 But I do not think that we should take
31 This is very similar to the line of reasoning that Richard Swinburne follows in the opening chapter of his Providence and the Problem of Evil (Oxford: Clarendon Press, 1998), pp. 3-29. He begins by stating that the premise that there is at least one morally bad state (his ‘premiss four’) is ‘obviously true’ (p. 8), and so he discusses instead the incompatibility premise. Once suitably modified, however, the incompatibility premise is then fixed as being true (pp. 13-14), whereas: ‘the theist now denies the new fourth premiss. He claims there are no morally bad states of the kind specified there. [...] The atheist denies this.’ (p. 14) And eventually: ‘premise 2* [the modified incompatibility premise] would seem ... to be necessarily true. [...] Hence whether the atheist has a sound argument from his premisses to the non-existence of God turns solely on whether premise 4* [the modified ‘evil exists’ premise] is true.’ (p. 19) In general, people discussing the problem of evil have shown a tendency to fix the incompatibility premise by further specifying which type or class of evils God is taken to be
option 3 to be a serious option, since it falls foul of the criticisms that I levelled at both option 2 and option 1. I do not think many people see this as being the crucial difference between logical and evidential formulations, and it is a mistake to think that any component in a logical argument must be deductively valid all the way down or otherwise necessarily true.