Author: Vincent Diepeveen
Date: 15:47:44 12/14/01
Go up one level in this thread
On December 14, 2001 at 18:30:00, Otello Gnaramori wrote: >On December 14, 2001 at 17:48:54, Vincent Diepeveen wrote: > >> >>Nah if you consider i have 3.5 out of 4 now against tiger at levels >>from 15 15 to 30 30 and that i play these games usually >>in the middle of the night, then 90 30 is like heaven. >> > >I'm curious if you have already tried the Fritz7a at tournament level, since >Frans introduced some anti-gm tricks. >I think I've already asked this to you, but I get no answer... Fritz plays real active, hard to fight for me. bu tpositional it's a major beginner. it knows shit from all kind of simple positional things which is logical as it is bitboards. it's impossible to make positional knowledge easily in bitboards. The hard thing to beat for me would be its book, not the program itself. Any stupid move i would try in opening to get it out of book is still IN book. But give me a week of preparement and it'll look like a beginner, sure. the only problem if I play an engine is that my default openings give me a 0% chance versus computers, so i have to play other lines both with white but especially with black. Just search at the icc after how little i score with caro-kann old classical lines against the computer. These lines are just TOO easy for the computer to play. It just needs a few active moves and even exchanging to endgames is good for white there. v/d Wiel has the exact same problem there... ...what more do i need to say? So what i do is real pathetic but it works somehow. Obvoiusly for a human player my tactics are not that bad compared to others at my rating strength, and my technique is more than ok, which is real important against the computer. In short my favourite program to play a match against would be fritz, but it's positoinal that bad and most important it hardly improves in playing strength. YES its endgame has improved, but who cares if you consider that it plays nowadays weaker in opening (as it can assume against other programs to be in book during that stage) then it only has extra weaknesses against the human player. The only reason why computers are that high rated at icc is because of their stupid formula's. I tried to play 15 15 several times against hiarcs8x there. It's higher rated there than i am (i'm 2410 standard rated at icc and that's basically ONLY against computers) but it doesn't accept 15 15 againts me because it only does for example 5 5 and it does that with rating-400 which is a mean formula. I would like to try to play hiarcs8 because i do not know how it reacts on things. Fritz is not ineresting for me to play as it reacts not much different from fritz5.32. don't take me wrong, against computers fritz7 is better than fritz5.32 NO DOUBTS here. But play on the internet with fritz5.32 and fritz7 book and see how close the ratings will be when playing humans! It's NOT interesting to see how newer versions of fritz play for me. It hardly improves that engine. Just book gets better, better learn files against any program are prepared, things lke that. That'll score better in computer - computer games but for me it doesn't make it more interesting. I can test that at home too. Tiger in this respect is hell more interesting. tiger1 to 2 was a big step forward obviously. Of course i won from tiger2 because of the same things as how i won from tiger1, but that doesn't take away that it's more interesting to see it play. Of course it makes huge stupid positional errors, which i won't post here of course fearing next version will fix them. Like i posted once some things which are in half of all engines now. So that's why i played Tiger at the internet a few games. Even crafty is no longer interesting for me now that i have kicked it some standard games after Bob said its king safety progressed (like 2 source code lines or so and 2 variable modifications?). But chessbase must sell new versions of their products, i understand that, and i blame them nowhere, users can be easily fooled with some good selling arguments which simply in practice are not true. Suppose they were right. Every new version is claimed: "70 points more" then sometimes even 80 points more etcetera. SSDF was at like 2900 or so when they lowered the ratings? Of course a newer version from fritz at faster hardware is going to kick old versions of fritz silly at SSDF and therefore be higher rated there. But suppose all that PR is true, then if you add it all up, the programs would be like 3300 FIDE rated by now at 40 in 2 level. That's obviously not true. Yet every year a new version must get sold. >Thanks for your reply (if any). >Otello
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.