Here:

The "Global Test" standard is likely to raise more questions than it answers because it is a threshold without a real specification, a probability without degree. It is analytically defective because the degree of risk one is willing to endure depends on the severity of the consequences. Most people would accept a one in six chance of losing $10 at a game of cards yet refuse the same odds at Russian roulette. The proof needed to pass a "Global Test" before preempting a suspected terrorist attack will depend on whether the threat is a gun or a suitcase nuclear bomb, and is therefore not global at all. Yet standards do have a value in this context, provided they are not the pseudo-absolute ones implied by a "Global Test". It is the test of reasonable action in the face of the best available information, the standard on which Eisenhower decided to launch Overlord in the middle of an Atlantic storm or which impelled Spruance to proceed to Point Luck in defense of Midway in ignorance of the exact whereabouts of the Japanese Fleet. It is no guaranty against mistakes. But it is a guaranty against paralysis.

Some liberals seem most comfortable in heavy fog, where nothing actually means anything. Terms like "black and white," and "absolutes" make them nervous. Hell, some get cranky when you talk about "good" or "evil." If there are no right answers, if no conclusions can ever be drawn, you can't possibly be proven wrong, can you? In a round room, nobody can corner you.