Author: José de Jesús García Ruvalcaba
Date: 11:07:29 09/09/99
Go up one level in this thread
On September 08, 1999 at 17:36:38, José de Jesús García Ruvalcaba wrote: >Should the game Rebel-Hoffman be taken in account to calculate the performance >rating of Rebel in the GM challenge? > >-Yes >-No >-Abstain I think there are two kind of competitions for chess computers, and that their results should be clearly separated. One of them requires that a playing entity remains the same forever: same hardware, same program, same version and same settings. This is the kind of competition SSDF conducts. In this case, the goal is to test the same exact playing entity, and any move produced by external factors should be replayed. I remember that Blass Uri pointed out to SSDF that Junior made moves that could not be reproduced at the same time control. Once it was verified, the games were replayed, and I think that SSDF took the right decision. The 'strength' of the playing entity can be seen as some constant, and the goal of the testing is to measure it. The other kind of competition tests the team that is behind the playing entry. The team changes the opening book before every game, and even tweaks some parameters of the engine. Even the hardware can be changed (like Diep in the last WCCC). This is how the WCCC is played, and I think the GM challenge belongs to this category. Many games have been lost in this competitions due to mistakes of the team in the pre-game tuning. Rebel played the first game of the challenge on slower hardware than the next three, so this definitely disqualifies the series of games for the first case. The strength of a playing entry in the second case is clearly not constant, it changes with time and can increase or decrease, and still nothing prevents us from trying to measure it. Human playing strength is similar, young and disciplined players tend to become stronger, and nobody complains they are underrated. So, in Rebel-Hoffman game, the team choose a not well tested machine, took a risk and lost the game. So the game is lost, and should count as a loss. If the goal is to reproduce competition in FIDE rated tournaments, as I think they clearly belong to the second category, the game is significative. I think that if a computer loses a game this way in a FIDE tournament the game counts and is RATED. It is not a "forfeit loss", as 64 moves were played and Rebel was clearly lost after them, and the team resigned the game. Summarizing, my answer is "yes, the game Rebel-Hoffman should be taken in account to calculate a performance rating of Rebel in the GM challenge". José.
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.