Author: blass uri
Date: 05:28:36 05/20/98
Go up one level in this thread
On May 20, 1998 at 01:17:10, Christophe Theron wrote: >On May 19, 1998 at 11:36:30, blass uri wrote: > >>I think the nunn match benefits fritz because there is a repeating >>motiv of king's danger that both programs do not understand >>and in this case the faster program is better. > >I don't understand your point. > >Is king safety the only strength you recognize in Junior? > >If both programs don't understand king's danger, well, the best program >wins. As usual in fact. That's all... my point is that there are positions when there is no danger to the kings and in these positions I believe Junior has more chances. if there is a danger to the kings and the programs do not understand it there is a good chance that the faster program will find how to get adventage of it. I think also that fritz5(like all programs I know) is weak at evaluating positions when the king is not safe but stronger than Junior4.6 at finding good moves in positions that the king's opponent is not safe. >Or you are just saying that speed is the only thing that counts in a >chess program. You are going to start a huge thread, you know! :) no, I say speed is important in positions programs do not understand. > > > >>this is not a fair match because Junior is slower by 20% >>as a fritz engine > >Ok, but 20% is not that much. > >You can take it into account in the final result. > >If Fritz wins by a small margin, we will say the match was undecided. > >If Fritz wins 7-3 or 8-2, it will be harder to say both are equal, even >considering the 20% speed difference... There is a way to check how much 20% is affective I can give Genius3 to play the nunn match against itself when one side has 1000 seconds per move and the other side has 1200 seconds per move and I will see what is the expected adventage of fritz5. the result now in fritz-junior is 5:1 for fritz5 with 2 draws (I did not finish the 7th game but I am sure fritz will win the ending) ( it has 2 pawns more than junior and a rook and bishop against a rook and knight and the 2 pawns are united passed pawns) if I tell a computer to choose 6 numbers all 0 or 1 when 0 and 1 have equal probabilty the probability at least 5 will be in the same side is 7/32 so if I say by results like that that the programs are equal I make a mistake in 7/32 of the times. I believe there will not be a convincing result to reject the conjecture both programs are equal because 20 games are not enough. after the paris results ( 9.5 out of 11 for Junior) I thought Junior's rating should be at least 100 elo above other programs. and I see that it is not truth and probably Junior was lucky and there is not a big difference between Junior and other top programs. so we should be carful not to be sure about nothing after 20 games. the only thing we can say is which program win the match if fritz5 has a better result than the result genius has against itself we can say that fritz won this match. > >But keep on playing... I had more time for doing this match the last month because I took a vacation from the first stage of the championship of israel in correspondence chess. now Junior and fritz most of the time will analyze positions in my games so the nunn match games will take more time. > > > >>I cannot do a fair match because chessbase is for fritz5 > >So ChessBase is against Junior??? otherwise I could do a fair match (not with equal time). I think they are against all the other programs because they want to prove fritz5 is number 1. fritz5 is a very good program but they want to be number 1 by unfair means. Uri
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.