Author: Uri Blass
Date: 12:21:34 04/17/02
Go up one level in this thread
On April 17, 2002 at 14:37:56, Vincent Diepeveen wrote: >On April 17, 2002 at 13:26:34, K. Burcham wrote: > >Let's start with the positions that were too difficult >for deep blue. perhaps you can create cool FEN files from it. >i already did a few years ago. lost them. > >Here what Seirawan sees as very bad moves in the report written >for icca june 97: > >many choices for computer so many mistakes: >Game 1: 10..h6? 11..Qa5? 12..Bc7? 13..g5? 22..g4? > >long bookline then kasparov makes so many bad moves that the >shuffle moves from deep blue can't be seen as bad anymore because >doing nothing was ok: >Game 2: 26. f4(?!) 28. Qf1?! 37.Be4? 44.Kf1? >some consider f4 ok and personally i consider qf1 ok. However >Qe4 is a bad move. Qb6 there wins by force the opposite bishop >endgame that happens after exchanging. Kf1 is tactical drawing >Kh1 there is winning for white. Diep finds Kh1 here quick. After >a few minutes also for the right reason. Deepblue was tactical weak >it didn't even see Qg3 (even seeing the draw is no big deal, you play >it for other tactical reasons then). > >All with all game 2 is not such a bad game from Deep Blue. Kasparov played >however around 8 real bad move. strategical especially major mistakes. >moves a 2000 player already knows doesn't fit in that position. > >However despite all that, kasparov can draw game still in the end. >The 37th and 44th move are real big bad moves in combination with the >many 'do nothing shuffle moves'. > >another idiotic setup from kasparov where if he would have wanted to would >have won 2 eyes closed: >game 3: 5..Be7? 8..Qd7?! (12.Nh3?? from kasparov, only move to not win >the game chanceless) 16..Kh8? 28..Bd8?! 32..a5? > >Kasparov played like an idiot game 3 obviously. Especially opening the position >with bxc5?! setup is pretty stupid against computer. Despite that kasparov >had to make even more bad moves to draw the game. > >One move i really can't explain is 12.Nh3?? that's so insane sick bad. >Any other move wins there for white the game in a strategic >chanceless way. Idem for bxc5 i want to give that 2 question marks because >a 1900 rated can win that game if you don't capture on c5. > >Especially the openings setup moves from deep blue we must see as >serious mistakes. It just knows so little about setting up an openings >position. In 2002 not a single program that joins world champs will fall >for that (perhaps except the lowest seeded participant which i already >can't imagine). > >Game 4: again kasparov shuffles a little without plan and without high >level. Like i do when i play a 2000 player. Again Deep Blue doesn't know >what to do: >12.Bxg6? 15.Rdg1? 20.Nd1? now 2 gross mistakes at the king >safety of deep blue which lose the game normally spoken: >22.a3? 24.b4? Lucky kasparov now plays the endgame like an idiot >so called 'time trouble' or simply because the guy thought any endgame >woudl win. 39.Rg1? is another mistake from deep blue but further it >played ok in the endgame. > >Game 5: 6..e6? (another simplistic mistake regarding the non existing >pawn structure code, this is a gross mistake) 10..Bb4+? (patzer sees >check, patzer gives check. He version 1.0 of DIEP played that too!) >(18.Bg5? from kasparov only way to blow the position a 1800 patzer >move). Kasparov blows the game then but deep blue plays the amazing >blunder 29..Qg6? which completely gives away black advantage. Now >Kasparov if he would have played 2300 level could have easily won >the game after qg6, but he didn't. 'Missing' obvious patzer Nh4 >manoeuvre. The result was that extreme simple moves from deep >blue were enough to draw the game. >Game 5 could have been a game versus an old 2100 player and a very >young and aggressive young 'talent' rated 2000 and old 12 years. > >Game 6 we know. long book line then 8 moves from deep blue and 1-0. > >> >> >>Deep Blue vs Kasparov 1997 Six game match >> >>if you have any Deep Blue positions that were played in the six game match, that >>you feel are very difficult for todays programs, could you please post them >>here. I will check to make sure I have them. >>If you have any positions you feel that cannot be found by todays programs, >>please state that with the position please. >>I am trying to compile the complete list of Deep Blue moves that anyone thinks >>cannot be found, or played by todays programs. probably have most--but you might >>have one i overlooked. >>If you have read somewhere about a Deep Blue position that someone thought was >>too difficult for todays programs, please post with position and game number. >> >>question everything, >>kburcham It may be interesting to see the right moves in these positions based on analysis of top programs and the difference between the score after the right move and the score after deep blue's moves in order to know if one of them can be described as a tactical blunder after a long search. If the difference after a long search is only 0.1 or 0.2 pawns in the evaluation then I am not convinced that these are big blunders. About the claim that 1900 rated can win game 3 then it may be interesting to do comp-comp games to check if computers gets 100% for the right side. I believe that computers usually know better than 1900 players. There are positions that computer play worse than 1900 but they almost never do not happen in game. Note that my latest israeli rating that is eqvivalent to the fide rating is 1987 so maybe I can check and try if I can beat the top programs in the relevant position(I did not do it). There are also some other players with rating of at least 1900 in this board and they can also try to do it. 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.