To answer Knyte's original question, I think the strongest argument for attack ads is this: For every person that hears a negative "rumor," the other side must actively refute it. In other words, they have to make sure that every potential voter that hears the attack ad also hears why that claim is baseless, otherwise the target candidate is at a disadvantage.

Of course, as Knyte says, the attacker's side may lose votes for running the ad. This is why negative ads are almost always paid for and run by third parties (as with the swift boat ads mentioned earlier). As long as they are setup this way, the non-target candidate can deny involvement and might even come out and say the attack ad is unfair (which will make him look good publicly). That's why we always hear a lot of talk about a "clean" campaign from candidates, without it usually happening.