Phil, what puzzles me is why people are so sure that reducing procedural protections in civilian courts is superior to constructing an alternative system of military courts with lower procedural protections. We are agreed, yes? That procedural protections in civilian courts are too high for war-on-terror prosecutions? If no, then you can't think the French (who don't even use a jury and have never been famous for their generosity to criminal defendants) have something to teach us. If yes, then there is just an empirical question of whether we should demand that federal judges relax procedural protections in terrorism cases or use an alternative military-commissions system—a question that it is far too early to answer because there is so far very little evidence as to how this alternative system will perform. As you point out, we might be more inclined to trust civilian prosecutors, judges, and juries than military prosecutors, judges, and juries, but if civilian judges do in fact relax procedural protections whenever they try a suspected terrorist, then this trust will certainly erode. Dahlia now says that she is making an empirical argument, not a catch-22 argument. However, I read the evidence she and Emily discuss as showing that there is serious disagreement among government officials about the proper level of procedural protections—how little is too little?—but from the outside it is impossible to know who is right.