which is regarded as undesirable and which, therefore, we want to prevent. We imply also that it is possible to change the dynamic system in question and replace it by another dynamic system in which the unwanted event does not occur. Thus, suppose we find ourselves driving towards a railroad crossing and suddenly we see the red lights flashing and a train approaching. Our dynamic system at the moment consists simply of velocity and direction. We are proceeding, say at 50 miles per hour, towards the crossing. The distant early warning system of our eyes informs us the crossing is dangerous. The knowledge which we have of our existing dynamic system informs us that if it continues we will arrive at the crossing at the precise moment when the train is there. The combination of a distant information system coupled with the simple dynamics of automobiles enables us, however, to prevent the disaster. We do this by puttington the brakes long before we get to the crossing. This in effect changes the dynamic system under which we have been operating. It introduces a new variable into it, indeed a new dimension, deceleration. Because of this, we are able to prevent the disaster, as we are able to avoid simultaneous occupancy of the crossing by ourselves and the train.
We must be careful, of course, in applying the analogy of a simple psycho-mechanical system like a man driving a car to the enormous complexities and uncertainties of the international system. However, the international system is still a system, even though it has important random elements in it. Because it is not entirely random, it has elements of predictability. One of the greatest difficulties lies precisely in the stochastic nature of the system. We are driving a car, as it were, that may or may not respond to brakes according to whether dice held by the driver indicate "respond" or "fail." The situation is made all the more difficult by the fact that we face here a stochastic system with a very small universe, that is, a very small number of cases. Stochastic systems with a large number of cases can be treated by the theory of probability. We have a pretty fair idea, for instance, how many people are going to die in automobile accidents next year, although we do not know exactly who they are.