Author: Uri Blass
Date: 15:22:20 02/26/02
Go up one level in this thread
On February 26, 2002 at 17:01:55, Janosch Zwerensky wrote: > >>Of course it would also be interesting to know, what the most used method was to >>find the moves for that game (single or multi pv, how many pvs, or more than one >>engine at once etc.), and what was the thinking time per move? Which hardware >>was used, and how long did the game last in total? How many correspondence games >>did you have to win in total, to win that championship? > >It would also be interesting to get Uri's opinion on whether and to what extent >his opponents used computer assistance too and to what extent he had to rely on >his own chess judgement. > >regards, >Janosch. My total result is 6.5 out of 8(5 wins and 3 draws) The thinking time per move was 30 days per 10 moves when the time is counted from the time the player gets the card of my opponent to the time that he send the move. I used only one line when I gave programs many hours to calculate. The thinking time per move was usually 5-24 hours per move. There were some exceptions when I used more time when one of them was when I used more than 80 hours for 10...Bd6 against har aven(Deep Fritz could find it after less than 24 hours but I decided to give it more time) 10.Nf1 of GM har aven seemed to me anti chess so I wanted to punish him for it. I suspected that he planned some trap against computers and I was happy when Deep Fritz changed it's mind to Bd6 after many hours because it was also a move that other programs did not consider. I decided to check it with other programs and after I found no problem with the move I decided to play it. I did not analyze other moves to find what was planned against the typical computer move exf4. another case when I analyzed a position for a very long time was against the female world champion Luba kristol after she played a move that was not in my databases I also gave Deep Fritz some days to calculate and Deep Fritz failed low on 14...Rc8 and chose 14...dxe4 [D]r2qkb1r/1b3pp1/p3pn1p/3pn3/1p1NP1P1/3BBPN1/PPPQ3P/2KRR3 b kq - 0 13 I did not investigate the reason that Deep Fritz failed low because 14...Rc8 was relatively a quiet move and I prefered to check what happen after 14...dxe4 and I saw that after the game line 15.Bxe4 Nxe4 16.Nxe4 Bxe4 17.fxe4 I do not have to play Nxg4 but can choose also Bc5 that part of my programs like. It was enough to convince me to play 14...dxe4 Later in the game I analyzed the position after 17.fxe4 and I found based on playing some moves of Deep Fritz against itself that inspite of the fact that Deep Fritz likes Nxg4 even after 24 hours it is not good and it is better to believe gandalf that suggested Bc5. I believe that all my opponent except one of them used computer programs to help them but they played in more games that I played(I played only in the final) so they had less computer time to use. I used human judgment and in few case I did not choose the computer move. I believe that in 2 cases it only helped me to win faster but in one case it was clear to me that the typical computer move is probably leading to a loss(I found it based on games of the comp against itself) so I played a different move and one move later offered my opponent a draw(I did not see a forced win for the opponent but I felt that the gravitation of the position is against me) My opponent did not believe that he was better and except the draw but I believe that he had better position. In another case I did not choose the computer move because I could not see a win in a game of deep Fritz against itself and I wanted to win my opponent. I did the mistake of assuming that the opponent will play a move that Deep Fritz suggested and felt that I had better chances to win in this case but my opponent did another move and I got slightly worse position(the game was finished in a draw). I think that the computer move could give me better chances in that game. The main use of human judgment was choosing which program to believe in cases that different program suggested different moves. Interesting note is that I got 4 out of 4 with black and only 2.5 out of 4 with white. I guess that my choice of the opening was not good with white. In one game I started with 1.e4 c5 2.Nc3 because of the fllowing: 1)I did not like the swesnikov based on experience in a previous game when my opponent got a drawn position and lost only thanks to a clerical error. 2)I knew that my opponent has experience in playing the sweshnikov. 3)I thought that I may have a chance to follow one of aduard nemeth's wins against Fritz that began with 1.e4 c5 2.Nc3 Unfortunately I could not get an advantage from nemeth's line and decided not to waste time about the game and agreed to a draw. Here is the relevant game that was played by email when I used only p450 for that game(this game was finished relatively fast because it was played by email and not by mail): computers said that the endgame is slightly better for me but after analyzing I decide that the position is equal and agreed to a draw. I beliebe that the opening was wrong here(maybe 2.Nc3 is not wrong but 2.Nc3 together with 3.f4 is not the way to get an advnatage in a correspondence game and it seems to work only for nemeth against Fritz) 1.e4 c5 2.Nc3 Nc6 3.f4 g6 4.Nf3 Bg7 5.Bb5 Nd4 6.0-0 Nxb5 7.Nxb5 d6 8.d3 Nf6 9.e5 dxe5 10.fxe5 Ng4 11.Bf4 a6 12.Nc3 Nxe5 13.Nxe5 Qxd4+ 14.Kh1 Bxe5 15.Bxe5 Qxe5 16.Qf3 0-0 17.Rae1 Qd6 18.Qe3 Bf5 19.Qxe7 Qxe7 20.Rxe7 b5 21.Re5 c4* The hardware that was used in my games was p450 in the beginning of the games(and also for all this game that was played by email) and p800 or p850 later. The games began 1.5 years ago(the 2 games that I played by email were short amd were finished in 3 monthes and the rest of the games were finished after an average time of about a year). 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.