Author: José Berdiñas Bonefua
Date: 19:04:11 03/19/98
Go up one level in this thread
On March 17, 1998 at 17:02:16, Scott Carmichael wrote: > For the past several weeks I have been playing Fritz 5.0 vs. Hiarcs >6.0 on identical computers, with the same amount of ram allowed for each >(32Mb). I have been testing them manually at various preset time >controls, and have added no powerbooks, or additional data to either >program. The only modification that has been done is to take the >database which is included with Fritz (out of the box, as you say), and >use the import tree feature to include these games in the programs >opening tree. Both programs are installed upon a hard drive, and neither >are interupted unless Fritz 5.0 asks for the endgame tablebases, which >come with the software. > In these games, Fritz 5.0 has scored an amazing 60% of wins, with >only 7% of all games being drawn. In fact the only program I own which >has scored well against Fritz 5.0 is MChess Pro 5.0, and then only at >games played at G/30 or less. > Can anyone offer me a plausible explanation as to these results? I >have trusted the SSDF for years, and have maded my purchases according >to the combination of their evaluations and the results of various well >known Comp. vs. Comp. tournaments. It saddens me to think that they have >allowed any manufacturer to ruin the well deserved faith that so many of >us have placed in them over these years. > >Thank you for your input, >Scott Carmichael This result is must to book tuning!!! Some of the games that these playing now are games already played by chessbase. with auto 232, and this provoked that elo to be inflated. How much??. That is the question of these days. To obtain elo "real" you have to use 1, 2 or more unknown programs. To play fritz 5 against these and later Hiarcs against these same.( more games =best results) The relation between the rating of Fritz and of Hiarcs will give you a best idea of the true power of Fritz. As "unknow programs" (= no top SSDF) could use : Virtual, CM 5000 , Crafty, Comet, Kallisto , CM 4000 and others. This might avoid book tuning. To avoid effects of book learn , you must screen out duplicate games or to disconnnect this feature. If Fritz obtain an elo similar , is almost sure that i am mistaken!!!! 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.