In addition to the obvious dangers, there is another subtle but by no means minor disadvantage to a pure deterrence posture: the threat of mutual sui… - Herman Kahn

We must not look too dangerous to neutrals. While this is probably less important for us than not looking too dangerous to friends and allies, it is still too important to be ignored. Here again, we must not look excessively destructive. It may be all right to promise the Soviets that if they attack us we will destroy every Soviet citizen in retaliation (though personally I think this far too destructive a proposal), but we should not threaten nonbelligerents with near annihilation because of our quarrel with the Russians. Many of the world's inhabitant perhaps two-thirds of them, do not feel it is their quarrel but feel it is their world. The more destructive we look, the less they like us and our program. To the extent that some in our midst talk and threaten potential world annihilation as a U.S. defense measure, we focus undeserved attention on ourselves as being dangerous and even irresponsible—appearing to be willing to risk uncounted hundreds of millions or billions of bystanders as to our selfish ambitions and desires. Neutrals and bystanders will inevitably suffer heavily in any thermonuclear War. But there is a difference between damage and annihilation.

Few people differentiate between having 10 million dead, 50 million dead, or 100 million dead. It all seems too horrible.

The unacceptability of the Doomsday Machine raises awkward, unpleasant, and complicated questions that must be considered by both policy maker and technician. If it is not acceptable to risk the lives of the three billion inhabitants of the earth in order to protect ourselves from surprise attack, then how many people would we be willing to risk? I believe that both the United States and NATO would reluctantly be willing to envisage the possibility of one or two hundred million people (i.e., about five times more than World War II deaths) dying from the immediate effects, even if one does not include deferred long-term effects due to radiation, if an all-out thermonuclear war results from a failure of Type I Deterrence. With somewhat more controversy, similar numbers would apply to Type II Deterrence. (For example, some experts would concede the statement for an all-out Soviet nuclear attack on Europe, but not if the Soviets restricted themselves to the use of conventional weapons.) We are willing to live with the possibility partly because we think of it as a remote possibility. We do not expect either kind of deterrence to fail, and we do not expect the results to be that cataclysmic if deterrence does fail.