BUTLER@MIT-DMS@sri-unix.UUCP (08/03/83)
I haven't seen Wargames, but I would have to say that WOPR's "misunder- standing" about whether or not it can "win the war" is based on its programmed goals. You or I might run through many scenarios of nuclear war, and then determine that there is no way to win because of our human way of looking at the concept of nuclear war. The cost in lives, materiel, and the effect of global thermonuclear war on the race as a whole are concepts that we all understand because of our "programming." WOPR's programs state that if, at the end of the scenario, certain criteria are met, the war is won. Those criteria are based upon the assumption that if They are in worse economic and military shape than us, We are the winners. The criteria in question do not concern them- selves with whether or not the United States will actually continue to exist as a political entity, or if my aunt or your sister or any other person or persons will survive. WOPR is measuring two stacks of beans (their railroads, our railroads; their aircraft, our aircraft; etc., etc., etc.), and if our stack of beans is higher at the end, we win. As a result, no number of previous games would have taught WOPR that nuclear war is not a game that can be won. I somehow doubt that the Tic-Tac-Toe analogy would have done it either, but if the system were ordered to consider the analogy valid, I suppose it might work. ---RLB