In my earlier article (Kavka, 1980), I examined a simplified version of the bilateral nuclear balance of terror from the point of view of utilitarian moral theory. I suggested that the leaders of the U.S. face a choice between one policy (i.e., disarming their country) that appears to them to pose a more probable risk of a lesser utilitarian disaster for mankind (i.e., world domination by the U.S.S.R.), and another policy (i.e., continuing nuclear deterrence) that appears to pose a less'probable risk of a greater utilitarian disaster (i.e., nuclear war). Soviet leaders face a similar dilemma. The choice must be made under conditions in which policy-makers have little confidence in precise estimates of the probabilities or utilities of the various possible outcomes. (A main source of uncertainty is each side's lack of information about how the other side would respond to a given policy, e.g., how the U.S. would react if the Soviets disarmed.) Using the illness case (that Dr Bernard discusses in his comment) as an illustration, I proposed that when choosing between potentially disastrous alternatives under such uncertainty, it is not irrational to risk a greater disaster, in order to maximize one's chances of avoiding all disasters. This strategy of disaster avoidance, as I call it, is most plausible under certain special conditions. I suggested that these conditions may hold in the nuclear deterrence situation (at least for the U.S.), and concluded that practicing deterrence, because it maximizes the likelihood of disaster avoidance, may be a reasonable course of action for leaders seeking to maximize utility for mankind.
Dr Bernard, in his comment, supplies numerical values to fill out my descriptions of the illness and nuclear deterrence cases, and applies his novel method of utility analysis to these cases. To this 1 have no objection. However, he indicates that his filled out versions of these problems, and perhaps his method of solving them as well, represent "completing and formalizing Kavka's approach". Here I must demur. There are fundamental differences between our methods for dealing with these problems, and between our views Theory and Decision 14 (1982) 99-102. 0040-5833/82/0141-O099500.40.