Author: Thomas Lagershausen
Date: 14:17:26 04/08/01
Go up one level in this thread
On April 08, 2001 at 12:57:47, Uri Blass wrote: >On April 08, 2001 at 12:50:34, Chessfun wrote: > >>On April 08, 2001 at 08:07:59, Thomas Lagershausen wrote: >> >>> >>>>>>Here is my guess for the evidence. >>>>>> >>>>>>Gambit1 is probably not better than Deep Fritz on one processor based on Enrique >>>>>>results. >>>>> >>>>>Enrique played 40/40 games.The programs have their own evidence on different >>>>>timecontrols.One example.DeepFritz is on 5min/game 150 Elo Points stronger than >>>>>DeepShredder, on timecontrol 40/120 DeepShredder is 50-70 Elo points stronger >>>>>than DeepFritz.Can you explain that? >>>> >>>>I do not believe it. >>>>I believe that Deep Fritz is better than Deep Shredder in all time controls. >>>> >>>>Deep Fritz is winning Shredder5 at 2 hours per move. >>>>see http://www.icdchess.com/forums/1/message.shtml?162151 >>> >>>Uri you are a computerchessexpert.But to come with only one game is not a strong >>>argument.Is phalanx stronger than Fritz because he can win some games against >>>Fritz? Why is Shredder winning every official tournament(40/120)and DeepFritz >>>placed behind Shredder? I would say DeepFritz is the best speedchess progam of >>>the world but on tournamenttimecontrols it´s only number three. >> >> >>The 8400 lighting games 1/1 I have played suggest otherwise. >>They suggest strongly that both Gambit 2.0 and Tiger 14.0 are >>stronger at speedchess than Deep Fritz. > >I believe that Thomas did not mean to the case when Deep Fritz is using one >processor but to the case when Deep Fritz use more than one processor. Well, Uri what i mean is that it will do a better job against a human topplayer like wourldchampion kramnik as Deep Fritz could do on a 8-waymachine.Off course you wan´t to know why i think so.Let´s allow me to give you the answer christophe gave me today of this question: The reason why playing style is important against human players is that computers and humans have, at this time, very different abilities. Ask Ed. If your computer program plays quiet and passive against a strong human, then the guy is simply going to grab the initiative and the program is going to have a very hard time. In order to avoid this problem, Ed has designed in Rebel an algorithm called "anti-human". It works. It definitely works, and Rebel as an impressive record against human GMs. It is very important for a computer program to play with its strengths in order to optimize its playing strength against a strong human player. For example, it is important for the program to create sharp tactical positions, to open the game, and if possible to create king attacks. This is the best way to make the human player "crack" under the pressure. It does not even matter if the attack is 100% correct. An attack that will fail against a computer has chances to succeed against a human. This is why playing style is much more important against a human player than the speed of the computer. > >He seems to believes that tiger may be better than Deep Fritz at tournament time >control even if Deep Fritz is using 8 processors. This question is to funny for me.But let me say:"I would love to see such a match between these two monsters." Regards Thomas > >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.