Sebastion Holsclaw has an interesting, thoughtful post on empathy and assumptions we make about those we argue issues with:

Proponents of the war suspect that their opponents don't believe that war is ever good (or believe that there is never too much negotiation or believe that the US can never really do good in the world) and that other arguments are just a way of hiding their true feelings from the public. Opponents of the war suspect that the US wanted to go into this war no matter what evidence was available (for reasons which are ususally left unsaid lest the speaker sound silly) which supposedly explains the multiplicity of reasons for the war as well as explaining the intelligence failures.

I think we often ascribe irrationality to our opponents because it means that we don't have to deal with their arguments.

He concludes by suggesting that we always make the assumption that those we're debating are rational, which is generally sound advice. I'm a big fan of hashing out issues, following Hitchens' advice in Letters to a Young Contrarian that such exchanges, even if they seem fruitless, challenge both sides to revise and consider their points and positions, even if they don't end up changing each other's minds.

It wouldn't actually make much sense to ascribe irrationality to your opponent from the get-go...what would be the point, then? You'd be banging your head against the wall at that point. And some exchanges end up that way. If one party is not obviously interested in engaging in the points the other is making, and either wildly flies off on tangents, makes personal attacks, or engages in other actions that indicate they aren't rational (in the context of the discussion), then it's best to withdraw from the debate.

But I agree with Sebastian that until the other party demonstrates that lack of intellectual honesty, a presumption of rationality is always the best policy.

No matter which side you're on with a given issue.