Author: Vincent Diepeveen
Date: 08:39:49 02/01/00
Go up one level in this thread
On February 01, 2000 at 10:05:48, blass uri wrote: >On February 01, 2000 at 09:21:23, Robert Hyatt wrote: > >>On February 01, 2000 at 08:48:31, Vincent Diepeveen wrote: >> >>>Hello, >>> >>>For years i've been wondering why Kasparov played h6?? in the last >>>game against deep blue. It is for sure that Kasparov knew that h6?? was >>>a bad line after Nxe6! >>> >>>Kasparov has said several things about this game, all crap what he says >>>in my eyes. Latest explanation of him i heart being said on the internet >>>was that Kasparov with his excellent memory and as a professional and >>>young player had forgotten it and hastely played h6 thinking it would be >>>transposition anyway. >>> >>>Now i found that already very unlikely. See here my big surprises when >>>checkingout the log files of Deep Blue. >>> >>>I figured out that Deep Blue had thought already about moves before this! >>>Even though some moves were from book, the moves Nxe4 and Bd3 especially >>>there deep blue lost like 6 minutes of time somehow. Kasparov must have >>>thought deep blue was out of book! >>> >>>Some moves that were made quickly by deep blue Kasparov must have thought >>>there that they were predicted correctly by the computer, which was in fact >>>true. the move e6 was in fact predicted correctly. >>> >>>Now first we can clearly deny that kasparov hastely played h6 by accident. >>>The log file shows clearly that kasparov has thought nearly 93 seconds about >>>making h6. >>> >>>Now for the average player 93 seconds in opening might mean he just drank >>>a bit of his tea, but Kasparov who plays theorem always within a second, >>>has given this position over a minute of thought. >>> >>>That can mean only one thing! >>> >>>Kasparov OBVIOUSLY knew both lines. He KNEW h6 was bad. He KNEW Bd6 was good. >>> >>>He MIGHT have considered deep blue out of book. It is very unlikely that >>>deep blue would have played Nxe6. The score is -0.20 after the sacrafice >>>and before that it is about +0.20, so over 0.40 pawn difference between Nxe6. >>> >>>Of course Nxe6 was in book. Log file clearly show that. >>> >>>Kasparov made Qe7 directly. Within 20 seconds. If Kasparov had forgotten >>>the Nxe6 line he would have been in shock for at least quite some time. >>> >>>Kasparov was not. So kasparov WILLINGLY risked the Nxe6 line. He very likely >>>must have thought in that 1 and a half minute before deciding to move h6 >>>about that he already had a chance that deep blue would NOT capture at e6, >>>after which black is fine. If deep blue WOULD capture, nah then kasparov was >>>a piece up, like in previous games deep blue might simply allow queen exchange >>>after which kasparov was a piece up and would win chanceless. >>> >>>Let's not call Kasparov too many words, but he sure was arrogant underestimating >>>the computer in an open position. He sure was ignorant about the aggressive >>>tuning of deep blue and he sure was naive risking Nxe6 and not even blinking >>>when the program played it. >>> >>>Vincent >> >> >>I have always been convinced that he played h6 intentionally. At least one >>member of his "team" was quoted as saying "we made a bad _decision_ in the game >>in round 6." "decision" != "mistake". > >It is possible that he thought that deeper blue was out of book so there is no >chance that deeper blue is going to play Nxe6 but it does not change the fact >that he was not prepared to the line Nxe6(I believe that this line is not good >for white). First of all i'm convinced Nxe6 is winning. Secondly aggressively playing programs that don't want to exchange their queen there (with exception of tiger, what program likes to exchange the queens there, i can't imagine a single program wanting to do that there!), will smoke any human as has been proven. >If this is the reason that it was clearly a wrong decision because even if the >computer is out of book it can be back in book and he could guess that the >computer may use the book without trusting it in 100%(the same way that Ed use >Eoc) >no computer in this time used Eoc but it is logical to assume that the computer >can use knowledge about theory without trusting it with no doubt(humans also do >the same). From simple minded perspective one should think that no program plays nxe6, this is not true (DIEP for example takes at e6 at big depths) but the principle is very valid for the large majority of programs. In principle no program likes to take at e6. I think that Kasparov took the risk, yes even prepared it at home. He must have seen that in 1997 by far not all programs had Nxe6 in their book. So he just took the risk, if deep blue would not take then kasparov would have a relaxed game. So there was already a chance he would win without much efforts. Then if it would take he would be a piece up, very tough play, but with a piece up, so if you're selfconfident then and underestimate your opponent incredible then you already in advance plan to risk it all as you think you win anyway! Kasparov obviously has played all moves on purpose and KNEW theory after Nxe6. So we talk about a perhaps prepared line from kasparov viewpoint. He clearly took the decision however to play that line. Not in a single database the line is considered good. In his own co-authored books kasparov give h6 a big question mark. Black is incredible underdeveloped. Usually we see kasparov play such positions with white! He never plays such things with black! He played it on purpose against Deep Blue however. Kasparov normally bets on horses himself and now suddenly he goes riding the horse himself. I'm sure later Kasparov realized how incredible stupid he would look if he would tell he prepared it, as we know that Kasparov knew in advance that this line was lost for black! It would show his incredible underestimation and disrespect for his computer opponent. If IBM wouldn't have put away deep blue in the best guarded safe of the states, then i would have been happy with this selfhumiliation of Kasparov! Vincent >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.