Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!linus!philabs!seismo!hao!hplabs!sri-unix!edmond@bbn-unix From: edmond%bbn-unix@sri-unix.UUCP Newsgroups: net.sf-lovers Subject: WAR GAMES Message-ID: <2970@sri-arpa.UUCP> Date: Sun, 10-Jul-83 20:15:03 EDT Article-I.D.: sri-arpa.2970 Posted: Sun Jul 10 20:15:03 1983 Date-Received: Wed, 13-Jul-83 05:58:56 EDT Lines: 35 From: Winston EdmondRSaunders seemed to feel that the WOPR's missle attack was accidental. I didn't get that impression at all. WOPR had studied how to fight and win a war. Part of the problem was how to start the war. Although the firing was changed to be under computer control once the signal had been given, the order to launch was still in the hands of people. The whole purpose of the fake attacks was to provoke the decision makers to order a "retaliatory" strike. That would give WOPR a first strike. By forcing the U.S. into a series of alerts, the USSR was forced to go on alert itself in response, and it became difficult then for either side to back off. This was obviously part of WOPR's strategy. It happens that the details of the attack were tied to the game being played, which no one else had played since the time WOPR had been put in charge of running things. WOPR is quite happy to play with real missles. In the film, when told it should only be a game, not real, WOPR says "What's the difference?" Indeed, since the whole point is that a good "game" strategy may someday become a real war, the "game" must be considered as quite real to be sure of accounting for all factors and ensuring the highest probability of "success." To WOPR, by design, it isn't just a game. From a more mundane viewpoint, if you write a program to solve a heuristic problem, is it likely your program will stop to wonder if what it is doing is "just a game" or will it just take what information it gets and apply it to analyzing the problem at hand until it meets your built-in criteria for winning? Part of the message, if you will, is that if you build a machine to win against all odds, remember that trying to stop it makes you one of the odds it may try to overcome. In this film, the problem was finally solved, not by defeating the computer, but only by convincing it that it no longer wanted to do what it had been doing. -WBE