Author: James B. Shearer
Date: 23:05:10 11/23/97
Go up one level in this thread
I would suggest the following contest. Select a set of initial positions. Associate with each initial position rebel settings so that rebel will play deterministically. The object of the contest is to find the fastest win (mate) against rebel from each position. Prizes could be given for the best result for each position and/or the best results for the entire set. Ties would be broken by date received. I would suggest choosing rebel settings such that rebel will play reasonably quickly on average hardware so that contestants on slower machines have a chance. This also makes competing more entertaining. Difficulty could be increased by choosing initial positions where rebel is ahead. Of course some of the winning entries might be embarrassing. On the other hand they might help identify fixable problems in the program. A variation would be to ask for quickest mates from the point rebel leaves book. Again the results might help make the program stronger. James B. Shearer PS: I tried this some time ago (for my own amusement) against the chess program included in os/2 (which is not very good) searching for the quickest wins for white after each of the 20 possible opening moves and also the quickest win as black. My best results were mates in 11-15 moves. Many were similar so more variation in the initial positions might be desirable. I have not tried this against rebel.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.