Thursday, November 24, 2011

Validity and Goodness of Argument

Edward Ockham has been having a discussion with Bill Vallicella on the subject of truthmakers. That's not an issue I'm particularly interested in (I think that once you get rid of either of Armstrong's two major theses of necessitarianism and maximalism, as most people do or at least consider doing, truthmaker theory is entirely a waste of time and we are better served handling truth in other ways). But I was interested in an incidental comment on a side issue:

This wrongly implies there can be degrees of goodness or badness in arguments. Not true: an argument is either valid, or it is not. All invalid arguments are equally bad, and all valid arguments equally good.

This is certainly much too strong a claim, since it is not true by any reasonable standard of goodness in arguments. In order to get the "All invalid arguments are equally bad" clause even to the level of plausibility we have to include enthymemes -- arguments with suppressed premises -- as valid arguments. Certainly an invalid argument that can easily be made valid by adding reasonable assumptions is better than an invalid argument that cannot be made valid at all. But most invalid arguments can be made valid by adding in the assumptions that make them valid, for the obvious logical reason; the only invalid arguments that can't are those that are invalid regardless of what assumption one adds, and thus the claim would have to be reduced to "All unsalvageable invalid arguments are equally bad". Which may well be true -- I don't in fact think it is, because things equally bad have to be equally bad means to their ends -- but is a far cry from including every invalid argument. As I always tell my students, it is always useful to know that an argument is valid, but knowing that an argument is invalid is only useful if you are doing certain kinds of things.

And, further, there are purely logical tasks in which invalid arguments are not equally bad; in certain approaches to defeasibility, for instance, some invalid arguments are better than others as being more validisimilar, as we might say (i.e., preserving truth to a greater extent).

Likewise, the claim "All valid arguments are equally good" would require us to say that sound and unsound arguments are equally good, as long as the unsound argument is valid. Since evaluating the truth of claims is an important part of evaluating arguments, holding soundness or unsoundness irrelevant would clearly be doing violence to any reasonable conception of 'equally good' when it comes to arguments.

Moreover, all arguments are means to ends, and means that are appropriate for their ends are always better as means than means that are not appropriate for their ends. Thus arguments that are better suited to their ends are better arguments. This is why we do not in most cases consider circular arguments good arguments. Why not? Circular arguments are not just valid but necessarily so. Yet it is certainly true that nobody considers "God exists, therefore God exists" as an equally good argument for God's existence when compared with a valid argument that doesn't have "God exists" as a premise. The reason must not lie in validity but in something else; and, in fact, it seems clear enough that the reason we don't usually think circular arguments good arguments is that they are poorly suited for most of the things you want an argument for. Thus not all valid arguments are equally good. (This same point could be adapated to showing that not all invalid arguments are equally bad as well; because enthymemes are arguments that, as they stand, are invalid, but are nonetheless still able to be well-suited to the usual ends of argument because they are not unsalvageably invalid.) For essentially the same reasons, we can't ignore relevance and irrelevance in the evaluation of whether arguments are good and bad. If you and I are arguing about whether God exists and I 'refute' your argument by arguing, completely validly, that you are a Manchester United fan, there is no reason why anyone would have to regard this as equally good in context as any other valid argument, even a valid argument that God does not exist.

Thus goodness in argument, as in other things, is said in many ways. I think Ed would have been fine, given the rest of his argument, had he settled for the more modest claim that any argument that is valid is good to at least some extent. But it's certainly not true that "All invalid arguments are equally bad and all valid arguments are equally good."