Author: Bruce Moreland
Date: 13:15:36 04/04/00
Go up one level in this thread
On April 04, 2000 at 09:27:43, guy haworth wrote: > >When Eugene Nalimov started generating 6-man tables, he asked "what should we do >about the 50-move rule?" > >Some said "nothing - ignore it - give us the absolute truth" and others said "do >something - we would like to win winnable games within the 50-move rule." > >I have therefore proposed what should be done given that there is a 50-move >rule. If the ICGA's current plan is achieved, this will be in the ICGA Jrnl., >v23.1 (2000), probably being published in April. > >The proposal does not stop the composition and 'absolute truth' communities (to >whom the 50-move rule does not apply) exploring the absolute capabilities of the >chess men. > >The proposal DOES include a suggestion that a k-move rule should not apply to a >player who can clearly, effectively win a deep ending. Since a rule along the >lines "the k-move rule does not apply to X if X has the endgame table" would be >technology-oriented and exclude other ways of winning effectively, there is an >interesting (and at this stage theoretical) discussion to be had about what >'sufficiently clearly' and 'sufficiently effectively' might be. > >Is a player who takes 10% more moves than the optimal number (whatever that >might be, M-optimal, C-optimal ....) being 'sufficiently effective'? > >The article goes on to show that players who play M-optimally or C-optimally may >not win games that should be won under the 50-move rule. KNNKP provides one of >the examples. Even Z-optimal play (minimising moves until the move-count is >zeroed by FIDE Article 9.3) may 'blow' a win. > >The first examples of 'difficult' positions for which all the obvious minimax >strategies fail are likely to be found in KBBKNN and KQP(a6)KQ (and similar >pre-long-KQPKQ-phase) positions. > >The article therefore defines the metric DTR or 'Depth By the Rule', this being >the smallest 'k' for which the (theoretically won) position would NOT be a >'draw' under a k-move rule. Tables to the DTR metric can be generated and used >to effect, preferably in conjunction with (currently non-existent) tables to the >DTZ metric. > >Guy The rules of a chess game should be the same for both opponents. They should not be different for an opponent who has downloaded data that has bugs in it. I should be able to write a chess program that plays according to FIDE rules, and have the game played according to FIDE rules. Furthermore, I should not have to force my program to play on in a position where it has claimed a 50-move draw, because the opponent claims that the win would in fact take longer. Perhaps the opponent played into this gigantic monstrosity voluntarily. They should be rewarded for this? A rather obvious hypothetical case is one where my opponent has to choose between an an obviously won ending where their score is +8, and one where they can sacrifice material to achieve a case that contains a mate in 200. What possible reason could there be for rewarding the decision to force the opponent to submit to an incredibly monstrous multi-hour ordeal? I'm sorry, but if your program plays into some crap like that you deserve to take a draw. 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.