Author: Uri Blass
Date: 13:38:32 10/18/02
Go up one level in this thread
On October 18, 2002 at 15:46:25, Jonas Cohonas wrote: >On October 18, 2002 at 15:19:00, Uri Blass wrote: > >>On October 18, 2002 at 14:53:17, Jonas Cohonas wrote: >> >>>On October 18, 2002 at 05:05:56, Uri Blass wrote: >>> >>>>On October 18, 2002 at 03:55:01, Jonas Cohonas wrote: >>>> >>>>>On October 17, 2002 at 17:55:20, Louis Fagliano wrote: >>>>> >>>>>>On October 17, 2002 at 16:18:40, Jonas Cohonas wrote: >>>>>> >>>>>>>On October 17, 2002 at 15:18:48, stuart taylor wrote: >>>>>>> >>>>>>>>Why? Because even if Kramnik wins the last game, It doesn't make it look like >>>>>>>>Deeper Blue was really any better than Deep Fritz. And also, it shows computers >>>>>>>>to be up at the top, and also gives Kasparov a big incentive to beat that result >>>>>>>>vs. Deep Junior. >>>>>>>> If the end result were 3-5 to Kramnik, it wouldn't look so close. Now it's it's >>>>>>>>almost a question of luck what happens in the one last game. >>>>>>>> The last game of Kasp.vs DB, the luck was on the DB side, because kasparov had >>>>>>>>just gone to the end of his nerves. >>>>>>>> >>>>>>>>Don't forget Kramnik had the program, long before, and Kramnik is probably a bit >>>>>>>>stronger than Kasparov VS computers. So DB already doesn't look like it was >>>>>>>>stronger. >>>>>>>>S.Taylor >>>>>>> >>>>>>>So far DF7 has not won a game, Kramnik has lost both. I have yet to see a game >>>>>>>where DF7 controled the game, i am not impressed. >>>>>>> >>>>>>>Regards >>>>>>>Jonas >>>>>> >>>>>>Well, just plain outplaying a super GM is not one of a computer's strengths. DF >>>>>>got it's wins by alertly pouncing on Kramnik's tactical errors instead of >>>>>>outplaying him and that's because that's where it's strength lies. >>>>>> >>>>>>But alertly pouncing on your opponent's tactical errors is still part of the >>>>>>game of chess and is certainly a legitimate way to win. >>>>>> >>>>>>Thus, the last hurdle for computers to jump would be to win by outplaying a >>>>>>super GM. They are not able to do that yet. But few people think that that day >>>>>>will never come. >>>>> >>>>>Well DF7 would not have had to have outplayed Kramnik, like Rebel did in one of >>>>>it's games against Van Wely, but simply keeping an advantage and winning the >>>>>endgame, that would have been enough to impress me. >>>>> >>>>>Kramnik made an outright blunder in his first loss and in his second loss, the >>>>>concrete solid Kramnik all of a sudden wants to win a beauty price, with a >>>>>knight sac that is as speculative as spectacular. >>>>> >>>>>No matter where the computers strenght lies (and there are a lot of games that >>>>>shows they can even outplay the strongest human players) it should be able to >>>>>atleast win on it's own, without "help" from Kramnik. >>>> >>>>If the theoretical result of the game is a draw then >>>>it means that it is impossible to win without help of the opponent. >>> >>>The two games Kramnik lost, where not your typical drawn positions, so your >>>logic (however true in a perfect world it might be) does not apply here. >>>When i say "help" i don't mean the normal small inaccuracies that both human and >>>computers make troughout any given game it/they play, when i say "help" it is a >>>nice way of saying "giving away". >>> >>>>If you talk about stupid mistakes then Deeper blue also won because of stupid >>>>mistakes of the opponent. >>> >>>And your point is? i mean if that is infact the case, then how come the two (on >>>the paper) strongest computers to date can not do what other weaker programs >>>have done before, namely outplay the opponent without any "help" from the human. >> >>When did weaker programs outplay kramnik or kasparov in 120/40? > >I never said that a weaker program outplayed kasp or kram under those >conditions... >I just pointed out that in the past weaker programs have done exactly that >against almost as strong human opposition... > >>> >>>>resigning in a draw position and playing an opening that kasparov was not >>>>prepared to play. >>> >>>Just because a position is proved to be drawn in the aftermath, does not mean >>>that kramnik could have drawn the actual game. >> >>I was talking about kasparov's case but I believe that kramnik could draw the >>game without hints except the hint that it is a draw. > >You also said qoute: "resigning in a drawn position" which could apply to both >players and since you did not specify which of the two had to be one in >question, i had the nerve to apply it to the most recent case... I said *resigning in a draw position and playing an opening that kasparov was not prepared to play.* It is clear that I was talking about kasparov in the sentence. > >You say quote: "I believe that kramnik could draw the game without hints except >the hint that it is a draw." except the hint it is a draw, well that might tell >you how hard it is to draw that game, not everything applies to logic Uri. > >>> >>>>I did not expect Fritz to win on it's own. >>>>I expected kramnik to win almost every game after knoeing the conditions >>>>of the match. >>> >>>That does not explain why the DF team allowed an exchange of queens 3 games in a >>>row, thereby further minimizing their chances of actually allowing fritz the >>>chance of winning on it's own. >> >>Preventing the opponent to trade queens in book means that you need to change >>a lot of book lines and you take the risk of getting inferior position. > >Yes, but since computers play MUCH better (against humans) with the queens ON >the board, then that risk is minimal, plus i don't think it would be that hard >with 1 year of preparation to make a good and solid no Q exchange book. > >>Criticizing the decision about it is easy after you know the result. > >Oh i would have critziced it before had i known they where going to do that... > >>Fritz could also lose by not trading queens and chessbase had no way to know >>what is the best chance before the match. > >Yes if they set the parameters wrong, but it could easily be tuned to avoid Q >exchange unless it would lose if it didn't. > >>> >>>Furthermore the two games it "won" was due to kramnik's shortcomings (in lack of >>>a better word) not a real effort from the DF team or fritz by itself. >>> >>>>Hsu said in the chat that if the Fritz team were competitive they should not >>>>agree to the conditions of the match and I do not agree with him. >>> >>>Still i have not seen any attempt from the DF team to actually play for the win >>>in any of the games so far. >> >>Fritz played for the win in game 5. > >Well that is your opinion, not mine. > >>It is not the fault of chessbase that kramnik blundered but the position was >>better for Fritz even before kramnik blundered. > >Why do you think i blame chessbase?? if anyone i would blame the DF team, >however i am actually blaming Kramnik and NOT awarding DF, saying DF has not won >a game yet, Kramnik has lost both. > >>> >>>>It is the opposite. >>>> >>>>If a program is really good it can beat everybody even when the opponent gets >>>>the machine before the match. >>> >>>Only if it get's the chance. >>> >>>>If the program is not deterministic in the opening stage the human has no way to >>>>prepare a killer line because even if he gets the program out of book by means >>>>of 1.a3 then he cannot know if the program is going to choose 1...e5 or 1...d5 >>>>and even if he guess correctly he cannot be sure about the next not forced move. >>> >>>I am not sure what your point with this is... >> >>My point is that even if the player has the program and the book he cannot >>prepare a killer line against it in case that the program is not deterministic. >> > >You just have to make sure that the comp plays openings that gives it the best >winning chances, which includes no exchange of queens, avoiding closed >positions, keeping as many pieces on the board etc. > >Regards >Jonas Kramnik already beated Junior before the match. Queens were in the board and Kramnik won by king attack. There are ways to win without trading queens and I do not think that it is possible to prevent exchange of queens and closed positions by book because there are a lot of ways to play moves that were not played in the past. 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.