Author: Vincent Diepeveen
Date: 11:46:15 04/30/98
Go up one level in this thread
On April 29, 1998 at 15:00:32, Moritz Berger wrote: >On April 29, 1998 at 13:47:27, Thorsten Czub wrote: > > >>Exactly. And these kind of deals run for a long time by know, >>starting with kasparov (also freds best friend) having difficulties in >>beating fritz and and and. Kasparov giving his name for a company >>friedel gets paid as advisor, Kasparov using chessbase, John Nunn using >>it too and doing this and that with it and and and. Campaigns. Friends >>and deals or SUPPORT of friends. Ald always against competitors. >>Corrupt world we live in. > >... yeah, right ... Kasparov losing to Chess Genius in London ... on >behalf of ChessBase ... dream on ... > >Moritz Did any of you read the Seirawan analyzes of the horrible childish play of Kasparov against DB? Few examples: game 1: What the hell did you do Kasparov, just having 3 rows, 1 open file against the computer? Open files are deadly against computers if you don't control them. Pure lucky that you got these freepawns and won therefore the ending. Such a passive play and still winning, that's of course great, but it's not necessary. game 2: Of course playing an opening with a closed centre is generally good against computers, but why playing such an opening? if you want a closed and won position against computers, then don't advance your pawns, everyone knows that advancing pawns leads to opening files. I also personnaly don't understand 23...,c4 after this you can NEVER win this game anymore with black against a reasonable program. Wanna bet all my money for it. But then i got upset, missing Qe3, and even resign? All people who played more than once at ICC blitz against programs KNOW how bad they are in seeing perpetual check. Even if it was not a draw, then still there is a big chance that it ends up in draw against a program. I tried to evaluate that in Diep, but that leaded to more and more losses. It's hard to evaluate, where humans simply can TRY it, and draw. It's very too hard to evaluate. game 3: moves like 12.Ng5-h3? missing the obvious 12.Be3 (game 3) which wins easily. I would not even CONSIDER Ng5-h3?. When i followed this game life over the phone (didn't have internet then), i directly said "What is this for kind of horrible losers move, he doesn't want to win this game", game 4: and 35..,Rxg4? instead of the obvious and logical and only move i saw was 35...,Rff2, after which against chessprograms you get a rookending which i win BLINDFOLDED. game 5: 18. Bg5? what is this for move from a world champion? But never mind, Kasparov can still easily win the game, but then move 32.Re6? I would not even CONSIDER Re6. I'm just 2227 and playing masterclass for Utrecht, the kind of guy Kasparov usually beats blindfolded during a simultaneously exhibition. NEVER leave your king so much open, but capture these pawns with Nh4 you even win another pawn, but i would not even have looked so far. Just played it. game 6: 1-0 Deep Blue wins. Dear mr. Kasparov, i'm used to play caro-kann, just like Karpov. I learn how to play the Caro-kann out of a book written by guess who? Remember that book you wrote? Do i need to say more about h6??????????????? ? Objectively judged: Deep Blue won from a 2100 player who did great effort not to lose tactically against Deep Blue, and initially tried to close the position. ---------------------------------------------------------------------------------------------- My goodness, dream on. Kasparov headed for that 2.5-2.5 and simply didn't wish to win. He probably misgambled in the last game. His explanation in the newest ICCA is also foolish. Kasparov says: "i was playing the best chess in my life". Thats Russian for: "i was playing the most horrible chess in my life". Normal GM game after annotating the moves about 2 moves at maximum get a ? mark. Not to mention guys like Kasparov, who usually do without it. Against Deep Blue average 3 questionmarks, and at least 1 missed easy win or draw a game. Not the opening was the problem, the problem was that he didn't want to win. If anyone talks to kasparov just ask him why he played like an IM and why he says in ICCA that he played the best chess of his life. That last statement is another lying statement. From my viewpoint he just tried to give IBM 2.5 points, and it will remain unclear what he wanted to do the last game. We'll never know whether he wanted to lose that game or not. It's obvious that the didn't realize that IBM would not give him revanche, he probably remembered Chinook and the many matches that this program played, and he forgot that he was dealing with IBM here. The analyzes from chess-strength viewpoint of Seirawan are very very accurate i figured out. Note that also Seirawan doesn't understand much from computers, This is not criticism on Seirawan, because he doesn't get paid to understand how chessprograms play, he gets paid to play professional chess against humans, and in this respect his analyzes are very good. Quote of annotation of game 5 of Deep Blue's 36..,Nb5! "Impressive positional play by Deep Blue. The computer recognizes that white's d5-bishop is much more powerful than his d6-knight and seeks a trade. Years ago such positional recognition would be unthinkable for a computer". Ok, let's analyze that at my 8086 with Psion.... Nb5, within a second. Now let's study the position. After Nb5,cxb5 Txd5 a) Deep Blue exchanges a its knight for a bishop b) bishop at d5 covered by pawn c) bishop at d5 very very active getting lots of mobility points d) bishop at d5 prevents rook activity e) after c4xb5 white gets a doubled pawn at b-file. programs think this horrible. So Nb5 is a 1 ply move for a computer, and will never go off this move. Programs not playing this must be losers chess programs dear sir. Deep insight of the grandmaster how computers works... What he means to say is: for a human Nb5 is a hard to find move because the knight at d6 is seemingly stronger than the bishop for a human, and the draw margin becomes larger. Second example: Game 2, missing Qe3 leading to a draw is analyzed by Seirawan. He analyzes 2 different variations: 45....,Qe3 46. Qxd6 and 45...,Qe3 46.Qd7+ Now he concludes that Qxd6 is a draw, and he's right in this of course, Qxd6 is also not so hard for my program. But Qd7 is even harder because you first do some extra moves. But then the simple human way in which Seirawan explains Qd7+ to be draw is: "and once again we have variations that are very similar to the lines after 46.Qxd6 line. However this 'similar' Qd7 takes Diep at least 23 ply to see that 44.Kf1 is drawing i figured out, where in the future the other line Diep might get lucky and see it at tournament level that 45.Qxd6 is leading to a draw. Now the weird thing is the judgement of Seirawan about this 23 ply combination, which totally is not 23 ply but more like 40 ply, so 20 moves. Seirawan's question is: "How could Deep Blue later blunder a winning position into an elementary draw by perpetual check which was only 8 or 10 ply deep". Seirawan is of course referring to 44.Kf1? which is played by all programs because they 'search' that queens are exchanged and then Kf1 is closer to the centre than Kh1, where in fact queens don't get exchanged and then Kh1 prevents perpetual check. Greetings, Vincent.
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.