Author: Tony Hedlund
Date: 10:41:02 05/31/99
Go up one level in this thread
On May 30, 1999 at 13:15:06, Melvin S. Schwartz wrote: > >On May 30, 1999 at 12:44:42, Dave Gomboc wrote: > >>Mel, >> >>Please don't take this as an insult. >> >>Do yourself a serious favour, and find a copy of Arpad Elo's "The Rating of >>Chess Players, Past and Present". If you are a serious chessplayer, you will no >>doubt recognize the author's surname. Pay particular attention to the procedure >>that was used to determine the initial list of ratings for grandmasters, because >>it is also the system that SSDF uses to generate their ratings. If I recall >>correctly, he also discusses why a large pool of players is preferred to a small >>one. > >Hello Dave > >I am interested in comparing software! And we are interested in establish a rating for a player (software+hardware). We are soon introducing a new player (Junior5 AMD K6-2 450MHz, 128RAM). To do so we let it play against the pool of players with established ratings we already have. And according to Arpad Elo it doesn't matter what ratings the players have in the pool, we still get a good estimate of the strenght of the new player. >As Dr. Hyatt put it so clearly - "If >program A on hardware B beats program D on hardware E - does that say much about >A compared to B? This belies the principles of science - you have to have a >uniform platform for all participants to make any kind of judgement". > >That was quoted word for word from his posting on "Uneven Hardware dated >5/29/99. I understand that to mean in order to get a true evaluation you must >compare both programs on the same type of proceesor. That's correct. When we have tested Fritz5.32 AMD K6-2 450MHz, 128MB against our pool you can compare it to Junior5 AMD K6-2 450 MHz, 128MB. >And that comparing programs >running at different speeds results in a hypothetical solution. Remember his >words - "you have to have a uniform platform for all participants to make any >kind of judgement". Why do you insist on refuting this most logical explanation >of how to compare programs? Perhaps you have missed the logic here. You must of cource compare Junior5 P200 with Fritz5.32 P200 and Rebel9 P90 with MCP6 P90. >By the way, my post was stated BEFORE I saw Dr. Hyatt's statement! > >Regards, >Mel> Tony >> >With the advent in the last year or so of programs that modify their play over >>time, I'm not convinced that the SSDF should retain exactly the same procedure >>that they use today, but that's another topic, and one that is completely >>orthogonal to the issue being debated. >> >>Dave
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.