MOST decisions are reactions to what has already happened. When evils are perceived, we attempt to mitigate them. Then our security depends on our capacity to cope with changing circumstances. Suppose, however, that collective confidence wanes, either because of decreasing trust internally or increasing menace from without. The institutions through which people relate may be in such disrepute or the environment perceived as so precarious that the slightest error could ramify throughout the system to cause catastrophe. In this situation we would need to anticipate difficulties from which we could not recover, rather than just reacting to those that can be overcome.  Now, the evils that do appear are limited, if by nothing else, by our capacity to recognize them. But the nightmares that might occur are potentially limitless. Shall we err by omission, then, taking the chance that avoidable evils might overwhelm us? Or shall we err by exhaustion, using up our resources to anticipate evils that might never happen, in order to forestall those few disasters that might actually do us in?