The Argument from Evil, Internal Critique, and Error Theory

 A few days ago, Greg Koukl tweeted:

There is no problem of evil unless morality is objective.

— Greg Koukl (@gregkoukl) September 14, 2022

Inevitably, the atheist-apologetics folks responded. Most of that is the usual hackery of atheist apologetics on the internet, but an interesting side issue kept arising that is worth noting, since it holds a lesson for arguments well outside this particular topic. A common response was that a person arguing against theism can engaged in an internal critique. This is true; one can typically modify an argument from common principles, ad iudicium argument in Locke's terminology, into an ad hominem argument in Locke's (rather than the fallacy manual) sense, arguing for something specifically on the interlocutor's premises. In this case we would shift the argument from saying that God cannot exist because there is evil to saying that the theist cannot say both that God exists and there is evil. However, the responses also regularly showed the reasons for thinking that this could be done by an error theorist or a similarly strong skeptical position were really based on the false assumption that internal critiques work like non-internal critiques. This is a common mistake made by amateurs and even occasionally by professional philosophers, but it is generally not a minor mistake in terms of how badly it can mess up an argument.

The argument given was amateurish, so I'll leave it in graceful anonymity, but essentially the claim made was that there is general agreement among atheistic and theistic arguments that you can be an error theorist and make the argument from evil. The empirical claim about the agreement of philosophers is in fact less sure than the internet atheist was assuming, but set this aside. Since everyone in the discussion is assuming that we are talking specifically about moral evil, and the error theorist doesn't think 'There is evil' is an objective truth (indeed, thinks it is false), the error theorist has to be engaging in an internal critique or Lockean ad hominem. However, as noted above, you cannot transfer features of a non-internal critique directly over to an internal critique. The argument from evil as a non-internal critique begins with establishing 'There is evil' as true and concludes (by some path) 'There is no God'. This is straightforwardly a problem for any theist. However, the internal critique cannot be structured the same way. As an internal critique, it is a critique of coherence, and therefore the internal critique is not getting the conclusion that 'There is no God', but the conclusion that 'There is evil and God exists' is false. Now if we also assume that the interlocutor is right in accepting 'There is evil' we can conclude from this that 'There is no God', just by the definition of conjunction. However, the error theorist is not making such an assumption; the error theorist already holds that the interlocutor is mistaken in accepting 'There is evil'. From 'There is evil and God exists' being false and 'There is evil' being false, you cannot conclude anything whatsoever about 'God exists'; 'There is evil' being false already suffices to make the conjunction false. Therefore the error theorist is not in fact in a position to make an argument from evil against God's existence by internal critique or Lockean ad hominem, because the error theorist is not in a position to say that the theist's best option is not to keep God's existence and reject 'There is evil'. To put it in other words: the error theorist could parrot the argument, but the error theorist is not in fact in a position actually to press one of the necessary points in it. When you are giving an internal critique, you don't get some kind of magical immunity from your interlocutor turning the tables and responding with an internal critique ("But you yourself don't think there is evil, so you aren't in a position to say that accepting the atheistic conclusion is the right thing to do, even if you are right").

Here we run into another problem that has to be navigated in converting from Lockean ad iudicium to Lockean ad hominem. There is no 'the theist'. There are lots of different theists, whose views form a family, not a monolith. Now, this is not an issue for Lockean ad iudicium; that's about what's actually true, and not about what follows from principles of interlocutors. But Lockean ad hominem is necessarily relative to interlocutor; internal critique is necessarily relative to the particular version of the system to which it is internal. Thus by internal critique you are not proving a given position false; you are proving a particular attempt to hold the position untenable. This is one of the foundational points in philosophy. In Book I of The Republic, Socrates argues against Thrasymachus's position that might makes right. But as Glaucon and Adeimantus point out when they leave the party, all Socrates showed was that Thrasymachus couldn't consistently maintain it; he hadn't shown the position false, much less that nobody could consistently maintain it. An internal critique just shows that something went wrong somewhere in a given set of claims in a position; it doesn't show where, and it definitely doesn't show that the problem is unfixable or that other positions in the family might not be right. If the positions in a family are sufficiently similar in a relevant way, then one can rule out tenability for the entire family, but this has to be established, and it can't be established by internal critique but by looking at what the actual positions in the family are. As it happens there are theistic positions that hold that there is no evil, and that is something that we know even without looking at potential limitations in generalizability of a given attempt to make the argument from evil as an internal critique. In the non-internal case, there would be no problem; the one putting forward the critique can simply argue that the no-evil theist took the wrong horn of the dilemma. But an error theorist doing an internal critique cannot make that argument. If you, not believing in elves, say, "This is my argument against X. At least one of these two views you hold is wrong: X, or elves don't exist," you've given an internal critique, but we might question whether you understand what an argument against X is. You can have all sorts of internal critiques, but an argument against X actually has to reach X. In fact, this is (while not impossible) usually difficult even under ideal circumstances with internal critique, precisely because it has to be an argument about coherence and pinning down possible causes of incoherence to that one X you are targeting is not always easy. But, as I'll note below, the error theorist seems to lack the resources by which one would solve this problem even in ideal circumstances.

Ah, Brandon, but you are just quibbling; obviously we are only talking about 'theism' in a sense that accepts both 'There is evil' and 'God exists'. Well, it's not quibbling to point out that the Lockean ad hominem version can only even apply to a subset of the positions to which the Lockean ad iudicium version can; that is something one always has to take into account when dealing with internal critiques. But let's assume that we are only dealing with such positions as hold both 'There is evil' and 'God exists'. But here we run into a third problem that has to be navigated in internal critique: the interlocutor sets the terms, and they can set them however they deem most appropriate. So suppose a theist whose position is this: God exists and there is evil, where 'evil' is taken in some sense consistent with the existence of God. On a Lockean ad iudicium approach, this is not a problem; you would just argue that the theist's account of evil is wrong. But with in an internal critique or Lockean ad hominem, you can't do that; you have to accept the theist's account of the terms. You can't appeal to common usage; the theist is not committed to common usage but their own, and internal critique has to build on the interlocutor's commitments. (And, in fact, I think a look at a lot of theists would show that it's quite common to have a view of evil such that how 'evil' is used in common usage is only broadly in the vicinity of how the theist understands it.) Someone who accepts the truth of 'There is evil' can point to actual evils to calibrate meanings ('Look at this evil, it's evil for this and that reason and in this and that way, right?'), and then can run an internal critique based on that calibrated meaning. They've checked that the understanding of 'evil' that they are using in the critique is at least close to the understanding of 'evil' that the interlocutor is using. If the interlocutor then says, "No, while I don't know yet exactly where it's diverging, I don't think you have my account of evil right because you keep getting a different answer than I do on this point", then the person putting forward the critique can go back to the calibration and potentially show that, in fact, they do have it right, so the problem really does exist. If we keep getting different answers to a reading comprehension quiz, we can go back to the text we share, calibrate our interpretations by the evidence of the text, and then we can go through your reasoning process and I can show you where your process of interpretation went wrong. But the error theorist can't calibrate meanings in this way; the error theorist has no way of calibrating in such a way as to show that they are not getting the interlocutor's own account of 'evil' subtly wrong. The interlocutor purportedly has the text and the error theorist doesn't think there's any text at all, so how is the error theorist going to show that his interpretation of the interlocutor's supposed text is correct in order to show that there is a problem with the interlocutor's interpretation of it? This is a problem, since going subtly wrong is a common hazard; you can spend years thinking about an argument before you realize that you were misclassifying an example or confusing one property with an associated property or being misled by the form of an expression into thinking that two things were more similar than they are. The error theorist could try to pin down the interlocutor's interpretation with other commitments, but the same problem arises as noted above: all you are doing, at best, is showing that the interlocutor is making a mistake somewhere. But if your argument is that theists make mistakes somewhere, this is, besides being obvious already, not an argument from evil; you only have an argument from evil if the conclusion is that the theist is making a mistake in accepting 'God exists', not if your argument is compatible with accepting that God exists and holding that the theist would have been better to classify or analyze evil in a somewhat different way. And in reality, what you've really shown, if you can't calibrate meanings, is that the interlocutor is making some mistake somewhere or else you don't adequately understand their view.

There seems to be a tendency among amateurs (and, indeed, among some professional philosophers) to think that internal critique is, because internal, hermetically sealed. But in reality internal critique requires meeting people on their own ground, which requires building interpretive bridges to get them correct. When you are doing this with someone whose relevant views are very different from your own, this is very difficult. We have ways of dealing with this, which involve calibrating interpretations based on shared reality. These don't make the internal critique less internal; they involve using our normal communicative means to make sure that your internal critique is getting the 'internal' right. Sometimes this is fairly easy, sometimes it's quite difficult. But if we can't directly calibrate meanings based on agreements about how concepts apply to shared reality, things get massively harder than even the difficult calibration cases. It's like arguing with a storyteller about how the storyteller's story actually goes when you think the storyteller just made it up (so you aren't catching them out on a factual mistake); the question then becomes, How do you prove that the storyteller is getting their own story wrong when they tell you that they don't think you are getting their story right?

We see these sorts of problems a lot. For instance, Chris Daly, in arguing against Nicholas Sturgeon's argument that the error theorist has problems with the argument from evil, considers in passing a case of an error theorist arguing against a theist who is a moral expressivist rather than a moral realist:

The theist who is an expressivist about ethics asserts that God has a strong con-attitude to suffering whether this suffering is brought about by agents or by nature, and so would prevent such suffering occurring if he had the intelligence and power to do so, and if he did not have a correspondingly strong pro-attitude to any of the consequences of that suffering. But, as the theist presumably concedes, there is such suffering.

'Presumably concedes' is doing all of the work here. Yes, if your interlocutor baldly concedes that they hold an obvious inconsistency, you have caught them in an inconsistency; this is not an internal critique, this is your interlocutor doing your work for you. Get better interlocutors. In reality, someone doing an internal critique can't presume; they need to know what the con-attitudes and pro-attitudes are supposed to be, and how their strength is measured, and to be able to assess whether their expressivist just forgot to mention one, or hasn't yet communicated a qualification, or some such, and need to be able to determine that they are getting the expressivist account sufficiently precisely right, and also that their interlocutor is not just making a mistake somewhere out of confusion rather than out of their relevant commitments. They need to make sure that the interlocutor has to concede, and what is more, has to concede it in the right way.

The ultimate problem with respect to calibration, I think, is that people (correctly) recognize that you could make an internal critique version of the argument of evil, thus not deriving the premises from your own moral view, and (incorrectly) assume that this means that setting up the internal critique does not involve your own view at all. Since you are not a telepath capable of assessing people's views directly, you have to establish a relevant common ground between your view and theirs so that you can assess whether you are getting their view correct. If you are trying to build on internal critique of an Icelander's view that elves are causing such-and-such phenomena, and you don't believe that any of the alleged phenomena really happen, you are going to run into difficulties, you are going to run into difficulties assessing whether you are correctly pinning down what these phenomena you don't believe exist would be if they existed and what the implications of their existing would or would not be for the question of whether they could be caused by the particular cause as understood not by yourself but by your interlocutor. Your interlocutor's own judgment is obviously already that the phenomena as he understands them are caused in the way he understands them to be caused by the cause as he understands the cause; you are literally trying to prove that the interlocutor is making a mistake about his own view. To do this, you have to be able to rule out, to at least a reasonable and relevant degree, that your understanding of all these things is not off, and since you can't directly inspect your interlocutor's understanding, you have to establish common ground that can help you determine that your interpretation of your interlocutor's position is right and that when your interlocutor tells you that their view does or does not have a particular implication that they are wrong on their own interpretation (or at least something close enough), rather than, say, the interlocutor's disagreement about the implications being evidence that you don't quite understand their meaning yet, which would often be the case. 

Daly had an answer to this problem, but not a solution. His response is, more or less, that the error theorist could just copy the homework of a moral realist (or whatever person has the relevant common ground) making an argument from evil; but this increases rather than reduces the severity of the calibration problem. What the error theorist needs to be able to do in order to make the critique himself is assess whether the moral realist atheist's internal critique of the moral realist theist is correct, so the error theorist now has to be getting two moral realists right, instead of one. Plagiarizing someone else's critique is not the same as being able to give the critique yourself.

The arguments he gives fail in any case. For instance, he suggests that the error theorist could take over an Objection from Bad Semantics if the theist held that God's goodness is different from human goodness, since it doesn't require that the error theorist make any commitments about first order moral views. "What it requires is that any proponents who are competent language users employ ‘is good’ with the same pattern of use when they apply it to humans as when they apply it to God." The problem with this is that we already know that for the objection to succeed it has to be false or else we have to take the theist simply not to be a competent language user. That is, we already know that the theist in question is not employing 'is good' with the same pattern of use, ex hypothesi, so this completely fails as an internal critique (we are attributing to the theist a position he has already flatly denied, and so are either simply failing to get his position right or are arguing on non-internal grounds that he is wrong to have denied it). A similar problem occurs with Daly's other example, the Objection from Bad Methodology; and indeed, it seems clear that we are going to run generally into the problem that the moral realist critiques that the error theorist can directly assess as effective are at least usually not going to be internal critiques. (Daly gives another possible approach for the error theorist that is infinitely more promising, but it's irrelevant to the question here, since it is not an internal critique or ad hominem at all.)

Thus (1) the error theorist doesn't seem to be in a position to press the right points to make the internal critique work without being subject to internal critique himself; (2) the error theorist doesn't seem to be able to make the internal critique of how the theist combines belief in evil and belief in God an actual argument against the existence of God; and (3) the error theorist doesn't seem to have the basic means that would be required for calibrating his critique so as to be sure that it is an actual internal critique against any given interlocutor. Someone can run the argument from evil as a Lockean ad hominem, but there are major obstacles to an error theorist doing so; the assumption that one can do so seems largely to be based on the false assumption that moving from a non-internal critique to an internal critique doesn't fundamentally change the argument or how one can even argue. When we compare the non-internal and internal versions of arguments, we will usually find that the latter, just for reasons of structure, (i) can make oneself vulnerable to a responding internal critique; (ii) does not apply to all of the same positions the former does; (iii) does not allow for all of the same rational moves and inferences; (iv) gives weaker and less certain conclusions; (v) can be effectively vulnerable to kinds of arguments that would not work for the non-internal version.

Now, conceivably there is some way to navigate these problems in the error theory case, so that we can show there is some particular way by which the error theorist can in fact make the argument from evil by internal critique. The field of possible arguments is vast, and some trail through it might be possible. But we already know enough to know that it will be a walk on a non-straightforward razor's edge. And I can guarantee you, absolutely, that no one has actually done the work to show that such a path even exists. The point is mostly of interest to those of us who like arguments and their quirks in their own right. In reality, an error theorist who puts forward an internal critique argument from evil is guaranteed either to be an idiot or an analytic philosopher playing argument games; intelligent people in serious arguments don't try to prove things by the argumentative equivalent of Rube Goldberg contraptions, and that's precisely what an error theorist's internal critique argument from evil would be. No one is honestly thinking that the error theorist can catch the theist in a game of Mousetrap, pulling the lever that releases the ball that falls on the see-saw until the trap finally after so many indirect connections descends on the mouse. The real interest is its use in showing the kinds of complications that come up with internal critiques.