Author: Bruce Moreland
Date: 17:30:06 03/18/98
Go up one level in this thread
On March 18, 1998 at 18:03:15, Ed Schröder wrote: >So everything is allowed? > >So maybe we should look in a different way to the SSDF. Quite confusing >for people I fear. > >Introduce a new second "error margin"? >An error margin for learning? :)) Pretend that these are professional chess players playing matches. After each game they may make changes. They might not play the same thing again, or they might play it over and over. It is obvious that an honest way to be stronger could involve better quality of adjustments between games, and that there can be no realistic suggestion that changing your match plan between games is unethical. If I beat you with 1. e4 it is up to me to decide to play 1. e4 again or not against you. There can be no criticism of me if I do it and you do not adapt. This is something I noticed on ICC. When a human would repeat lines against a computer, the programmers would get mad, but the human chess players would just be puzzled that another player (the computer) would play the same lost line twice in a row, and their solution was to fix the program. The humans were right. The problem I have is with pre-match preparation that involves reverse-engineering the opponent. I don't think it is alright to play a thousand games against a specific opponent, then give them a drug so they forget the whole thing. If there is a way to get around this, things change. The way to avoid repeated games is to implement machine learning. Perhaps there is a way to avoid killer books, too, perhaps by shipping a book that is huge, or by randomizing evaluation weights slightly before each game, or by tweaking the book somehow on the customer's machine, so that every copy of your program has slightly different opening preferences. The goal is to avoid deterministic behavior, which is a prominent feature of machine play. Perhaps we can get rid of it, producing something that is more human-like, which will help out on the SSDF list and will also benefit our human opponents, who we always want to amuse as thoroughly as possible, from the winning side, of course. bruce
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.