It’s a common device in philosophy (one I use excessively) to turn philosophical investigation into an imagined debate between advocates for the salient positions on the table. Sometimes this is a helpful stylistic device, because it lets the reader keep track of the dialectic more easily than the alternatives. But sometimes it has costs.

Consider a philosopher trying to find out whether there are representations who thinks to herself as follows.

bq. Let’s imagine a debate between proponents of the two positions, the representationalist who says that there are representations, and the anti-representationalist who says that there are not. The anti-representationalist’s position is obviously incoherent, since she cannot state her position if it is true. So the representationalist has to win this debate. So there are representations.

Now I think there are representations (it wouldn’t be worth writing this blog if there weren’t) but I think this is a lousy way of arguing for them. It’s a lousy argument because it’s obviously a _contingent_ fact about the world that it contains representations, but no step of the argument relies on contingent premises. So the argument proves more than we could hope to prove, since it seems to prove that representationalism is true everywhere.

From this I infer (perhaps incorrectly) that we should be very careful in arguments from the loss of the anti-p-ist in a debate to the truth of p. David Lewis noted long ago that by many debate rules the anti-dialethist would lose debates because they begged questions against the dialethist. He denied that implied anything about whether dialethism is actually true, in fact it is false though there are no non-question-begging arguments to that effect.

More pertinently, a large chunk of Tim Williamson’s paper on “Everything”:http://users.ox.ac.uk/~sfop0009/files/everything.pdf seems to turn on an argument from the incoherence of generality-relativism to the truth of generality-absolutism, with the argument going via the embarrassment of the generality-relativist at being unable to state her position. I don’t think this is much help to the generality-absolutist, because it isn’t a conclusive argument that her position is true, even if her opponent’s position is unstatable. To be fair, Williamson acknowledges that the difficulty for the relativist isn’t conclusive, the absolutist still needs to quantify over everything, but I think he still puts too much weight on the incoherence objection.