The basic issue, of course, is what we take success and failure to be when we are speaking of philosophical arguments. Van Inwagen proposes that an argument fails if it cannot pass a particular test:
The test is the ability of the argument to win assent from the members of a neutral audience who have listened to an ideal presentation of the argument. That is: the argument is presented by an ideal proponent of the argument to an ideal audience whose members, initially, have no tendency either to accept or to reject its conclusion; the proponent lays out the argument in the presence of an ideal critic whose brief it is to point out any weaknesses it may have to the audience of 'ideal agnostics'. If--given world enough and time--the proponent of the argument is unable to use the argument to convince the audience that they should accept its conclusion, the argument is a failure. [from the Detailed Contents* ]
There are a number of notable features of this suggested test.
(1) The success of the argument is detached from the question of whether the argument persuades anyone in the real world. (This is probably shares with most accounts of philosophical success and failure.) But it depends crucially on being able to persuade.
(2) It is an impartial spectator theory. In order to determine whether an argument is successful, one has to somehow get into or at least approximate the mindset of impartial spectators, looking at the argument without prior bias one way or another.
(3) Not only that, in order to determine whether an argument is successful, one has to somehow get into or at least approximate the mindset of an ideal critic pointing out any weaknesses of the argument.
Now, it's not exactly clear to me why we need to bring in the ideal critic at all. Why do we need to go so far as imagining the brief of an 'ideal critic'? We don't need our critic to be ideal, just competent. Indeed, we don't need to imagine a critic at all, except perhaps as an imaginative crutch; we just need to develop the discipline and habits required for honestly assessing the potential weaknesses of the argument. Indeed, everything we could get out of our imaginary ideal critic will, in fact, come from this and nothing else; it's the only way we can have any clue whatsoever what an ideal critic would say.
The ideal critic comes into the picture in part because van Inwagen has an excessive attachment to thinking of philosophical debates on the model of court cases. I think that thinking of philosophical debates on the model of court cases is extraordinarily dangerous. Courts have specific ends in view, are confined to specific means in light of those ends, and are heavily influenced by historical contingencies, that don't necessarily have any parallel in real philosophical discussion. These can lead to distortions, as I think we see when people try to apply the concept of 'burden of proof' to philosophical discussion in general. The fact of the matter is that philosophical discussion is more fundamental and basic than anything we get in a court; everything in a court is designed to keep the distinction within sharp confines, but philosophical discussion is free. What is more, it is not as if these worries about thinking of philosophical discussions in terms of court cases are new things; the worry is built into the very Socratic impulse that even today is a major part of what we do. Plato's Socrates remarks in a number of places on the difference between philosophical discussion and the sort of discussion one gets in courts; the court conventions have changed since his day, but the remarks are still salutary. If nothing else, thinking of philosophical discussions as if they were idealized court cases limits the genuinely dialectical aspects of philosophical discussion; and, as for Socratic midwivery, the model ignores that entirely.
The emphasis on persuasion is itself somewhat odd. It turns philosophy into a form of idealized rhetoric: rhetoric at the limit case. I've sometimes joked that analytic philosophy, of the sort that van Inwagen does, is less a form of inquiry than it is a rhetorical toolbox, but even I don't think that even analytic philosophers are simply engaging in idealized rhetoric. There are too many things that they do that have nothing to do with persuasion one way or another. And this fact alone suggests that any attempt to judge whether an argument is a failure is simplistic if it does not specify reasonable, appropriate, and relevant ends. And persuasion doesn't seem always, or even usually, to be among those ends. Perhaps there are some people who assume that there are argumentative equivalents of crucial experiments, lines of reasoning that will be decisive for everyone just on their own, but it seems to me that most people usually in practice do not assume any such thing.
As far as I can see, while van Inwagen assumes throughout that persuasion is the primary end of argument, he doesn't argue for it; the closest he comes is saying that a mathematical proof, "whatever else it may be, is an argument that should convince anyone who can follow it of the truth of its conclusion" (p. 37). I'm inclined to think that the history of mathematics shows this to be false, unless you are packing a lot of controversial things into the "anyone who can follow it" clause. But setting that aside, no one, as far as I can tell, takes persuasion to be a necessary or fundamental characteristic of successful mathematical proof -- nobody thinks of a mathematical proof that, whatever else it may be, it is persuasive. Indeed, despite a lack of agreement about the general requirements for something to be a mathematical proof, almost everyone thinks that persuasion is quite a secondary issue when it comes to whether something is a mathematical proof. Nobody thinks that the criterion for the success of a mathematical argument is its literal "indisputability". This is to put the emphasis in the wrong place: it's not that knowledgeable and reasonable people can't dispute it but that there's good objective reason to think that any objections they raise will exhibit objectively discernible flaws or beg the question or at most qualify the conclusion of an argument. The grounds of mathematical success are not, as van Inwagen would have it, psychological; at least, they are not usually thought to be so.
There are also peculiarities in the idea of appealing to a jury, even of "neutral agnostics," as van Inwagen calls them. Suppose we assume that all neutral agnostics will vote the same way on any issue. Then it follows that we don't need a jury at all: a single individual will suffice to give us all the information we need. The only reason to appeal to a jury of impartial spectators rather than a single impartial spectator is if the impartial spectators can still disagree. That is, we have no reason to appeal to a jury of neutral agnostics except insofar as the neutral agnostics have the capacity of voting differently even with the same information. But if this is the case then we can't simply appeal to a fact about what persuades a jury: we must define what persuading a jury is. With one and only one person there is no problem: that person is the one and only who needs to be persuaded. But if we have multiple individuals, and they can vote differently with the same information, things are not so easy. To count as having persuaded the jury does the vote need to be unanimous? Some of van Inwagen's arguments seem to suggest it must be. But that seems bizarrely harsh; if neutral agnostics can vote differently even with the same information, then that suggests that either there is no fact of the matter about what counts as success and failure beyond the vote -- and obviously van Inwagen must assume that this is false, because he has to assume that success and failure are not simply decided by legislative fiat -- or neutral agnostics are not infallible and therefore can sometimes vote incorrectly. Would we really count an argument as unsuccessful if, out of a jury of 500 impartial and unprejudiced neutral agnostics, one could expect a single Nay vote? Would we not rather consider it a mere statistical fluctuation? And assuming that neutral agnostics can vote differently with the same information, how would we know how all of them would vote? If the criterion is not unanimity, however, it would seem to be plurality or majority. If majority, is it bare majority? But then it seems we must specify the number of jurors: a bare majority in a jury of three is not the same sort of thing as a bare majority in a jury of ten million. Is it overwhelming majority? But then we have to say how overwhelming the majority must be, and it doesn't seem that we can give a nonarbitrary number (69%? 70%? 92.3%?). And similar questions are raised if we go with the weaker claim that we need a plurality rather than a majority (along with the question of why we don't need a majority, i.e., why a mere plurality suffices). Needless to say, van Inwagen considers none of this (although mere plurality seems inconsistent with one or two of van Inwagen's arguments).
And we also need to know what question, precisely, the jurors are being invited to assess. Is it just a vote of success or failure? Van Inwagen doesn't seem to think so; at one point he refers to the jurors as rendering a "Scotch verdict", i.e., a verdict of "not proven". Now, he might be speaking inaccurately here, and simply means that they evaluate it as a failure. But if he doesn't, a real Scotch verdict is a tertium quid verdict: you get the verdict of 'not proven' when the evidence fails to allow a verdict of 'innocent' or 'not guilty' but the case for your guilt is inadequate for a verdict of 'guilty'. These verdicts that indicate that cases are neither successes in the proper sense nor failures in the proper sense open a whole cans of worms. And yet if the assumption is that the jurors have only two options, 'success' and 'failure', we need to answer why there is no tertium quid. Why can't jurors consider verdicts like, "minimally successful" and "partially successful" and "highly probable but not certain" and "close enough to proof for almost all practical purposes" and "proven, but with the exact extent of application still unknown"? These are part of our usual repertoire of assessment.
Van Inwagen responds to one possible objection to his argument that his suggestion is at least interesting and useful. As you might expect from the above arguments, I don't think it is useful -- in particular, I think it is obviously not useful because it is not defined well enough to be useful. But the most egregious argument van Inwagen makes in his whole discussion is this one (p. 52):
And all philosophical arguments, or at any rate all philosophical arguments that have attracted the attention of the philosophical community, have been tested under circumstances that approximate sufficiently to the circumstances of an ideal debate, that it is reasonable to conclude that they would fail the 'ideal debate' test.
When I first read this I had to read it several times over to make sure I was not misreading it. But no -- he gives just before this the claim that certain arguments (his own arguments on the question of determinism) would fail the ideal debate test because they have been presented to several generations of graduate students at various universities, and the graduate students were split on whether they were successful, and this is just a more general version of this point. Notice two things about this:
(1) It answers the question of how we can know that something would pass the 'ideal debate' test, but only by throwing it over completely. It means that the real way we determine success and failure of philosophical arguments is looking at what gets consensus in the philosophical community, perhaps, at most, ignoring extreme cases that can be traced to deeply ingrained biases. If we're going to assess things by whether or not the philosophical community has achieved near-unanimity on a topic, why this whole convoluted debate set-up?
(2) An answer immediately presents itself: Because van Inwagen really does think that the philosophical community closely approximates a disinterested jury of people without no significant biases on the topics they are considering, concerned only with the question of whether an argument is persuasive or not. He must be assuming something like this to make the arguments he makes toward the end of his chapter on philosophical failure. And I am thoroughly flabbergasted by it. It's an assumption that makes no sense. Philosophers are not, in fact, that unbiased and disinterested on most of the subjects they discuss; they are not jury but already prosecution or defense. They can and do sometimes switch sides, but philosophers are trained to be not objective jurors but partisan arguers. Anyone who does history of philosophy in any extensive way comes up against this quite seriously at some point or other. I did my thesis on Malebranche, and it is truly extraordinary that whenever I'd say that, I was immediately pigeon-holed as a Malebranchean. Since Malebranche has views that are widely considered to be strange, this is not a good thing to be pigeon-holed as if you don't agree with Malebranche's arguments. But if I noted that I didn't agree with Malebranche's arguments, it was equally astonishing how many people would then be puzzled as to why I would study them. They can't have been thinking that through very carefully and consistently, because everyone recognizes arguments that they think are brilliant but wrong. But much of my dissertation was involved with showing that all standard and obvious objections to one of Malebranche arguments for the claim that we see all things in God failed -- and I repeatedly had to point out to people that this did not mean that I accepted the argument. I think there are independent reasons for thinking the conclusion is false, as Malebranche understood the conclusion; my point was merely that nobody in the early modern period or in the secondary literature had given a non-question-begging argument against it. But the assumption even philosophers make about philosophers is that they argue as already convinced. The assumption is common enough even in history of philosophy, where we take impartial and objective evaluation far more seriously than analytic or continental philosophers in general do.
And even if we set aside the question of personal bias and neutrality, even if philosophers were, as some historians of philosophy think ideal, impartial and reasonable evaluators of the strengths and weaknesses of arguments, they are not the right sort of jury. Sometimes you hear philosophers bemoaning the fact that philosophers tend not to form consensuses like certain other disciplines do (sciences in particular). But there is no great mystery to this. The sciences reward consensus-forming as long as certain procedures are followed: agreements through experimental verification, processes of peer review, etc. Philosophy has nothing like this. Philosophers are rewarded for coming up with creative reasons not to agree with other people. The whole thrust of professional philosophy is toward inventing ways to regard opposing arguments as failure, as long as those ways don't exhibit any obvious flaws. However much philosophers are interested in the truth, philosophy as a profession is not structured so as to converge on it; it is structured so as to have the maximal possible divergence that can be sustained given common conventions. We are not trained to find ways to come to agree with each other; we are trained to find ways to disagree with each other. We are not trained to create certainty; we are trained to create doubt or possibility of doubt even about such apparently obvious conclusions as 'There is a mind-independent world' or 'There are other minds than my own'. This has a genuine value -- you learn much more about even obvious conclusions when you do this, just as you learn much more about a city if you deliberately allow yourself to get lost in it, just as you often learn more about science or mathematics if you don't simply accept the argument for the right answer but also demand to know the answer to the question of exactly why other apparently good answers are wrong. But it makes for very poor judgments about success and failure of arguments: if an argument meets a standard of success, philosophers are trained to find a standard of success according to which it will fail, just to see what such a standard would be. This is why, incidentally, philosophers often make awful arguments outside their specialization: as philosophical training and the philosophical profession are structured today, philosophers aren't trained to accept standards of good argument, but to question even those. In other words, our best philosophical training today, while it produces excellently trained philosophers, also produces excellently trained sophists. It's OK to have sophistical arguments, as long as they are brilliant sophistical arguments. An entire jury of such people is the worst possible jury for determining what impartial people find persuasive; their whole training is often devoted to improving the tools that reduce their susceptibility to persuasion, and being perversely difficult to persuade is something a sufficiently clever person can spin into a successful career. Van Inwagen argues that some mathematical proofs have the feature that anyone who understands them is persuaded by them; for any such proof you can give, I very much expect that there is some philosopher who, if they understood the proof, would nonetheless reject it by noting some prior supposition that has not yet been proven. No one plays the game of What the Tortoise Said to Achilles better than philosophers do.
Likewise, van Inwagen claims that there are no philosophical theses that are both substantial and uncontroversial; but, as far as I can see, this is in context trivial: if you are talking about the hosts of philosophers, nothing counts as substantial for them unless they can have a controversy over it. And if you allow that there are uncontroversial theses that are not substantial in this sense, this raises the real possibility that there are successful philosophical arguments.
In short, I don't think any of van Inwagen's account of philosophical success and failure stands much chance of being right: it is shot through and through with assumptions so problematic that they sometimes border on absurd. Where van Inwagen is going with this, of course, is toward the claim that arguments from evil are failures because all philosophical arguments are failures; if I'm right, this is exactly the sort of sleight of hand it looks like it will be. There are many different kinds of arguments from evil; many of them are absurdly crude, but there are arguments from evil that meet plenty of entirely reasonable standards of success: they raise interesting and stimulating ideas, they are logically valid inferences from widely accepted premises, they can be persuasive to a reasonable person under the right circumstances, they genuinely show that some theistic positions are untenable (even though they are never as all-purpose as some people want to make them) and so forth. Despite van Inwagen's claim that his suggestion is a useful way of looking at success and failure of arguments, I see no reason whatsoever to think it more useful than these more common criterion. Indeed, I see no reasons to think it useful at all.
---
