< Earlier Kibitzing · PAGE 2 OF 2 ·
|Dec-31-07|| ||Peligroso Patzer: Great pun, but it might have been even better for this game: Deep Blue vs Kasparov, 1997, in which Kasparov, in essence, accused Deep Blue of cheating. Of course, the pun actually used for the 1997 game ("Deep Blue See [sic]", implicitly rejecting Kasparov's accusation of human intervention) was also terrific.|
|Dec-31-07|| ||ajile: The anti-computer strategy I've seen is to keep the position closed and locked preferably like in this game with a space advantage. The computer failed to see that it's pieces were being slowly restricted and strangled to death. Kasparov correctly played to close the position and then reduce Black's counterplay to zero. After this happens the computer is basically lost without any pawn breaks or space to move.|
|Dec-31-07|| ||MadBishop: The pun "Rage Against The Machine" is actually the name of a heavy/death metal band.|
|Dec-31-07|| ||CapablancaFan: This is actually the 2nd time <chessgames> has used this pun. Lasker vs Capablanca, 1914 Regarding this game though, Deep Blue begins to go "wrong" IMO with 11...Nh5? A totally time wasting move that does nothing but result in the program spending more tempos moving it around everytime it's attacked. A better continuation was 11...Rc8! to shore up some strength on the semi-open c-file. Had Blue made this simple change, it would have changed the whole course of this game. The proof? Please look at the final position.|
|Dec-31-07|| ||MostlyAverageJoe: This game is a testimony of how far the software & hardware have progressed since 1996. A mere 3 minutes/move analysis (averaging less time per move than would be available in a regulation game after getting out of the book) shows many improvements for the black (and very few for white). Here's what Hiarcs 11.2MP picked with 0.10 threshold for reported improvements:
13... Rc8 (+0.12, a 0.11 improvement)
15... dxc4 (+0.28, a 0.19 improvement) -- this move would've prevented the later lockup of black's DSB.
17.cxd5 (+0.50, a 0.33 improvement) -- the only move found where Hiarcs considers white to have lost some of the advantage.
19...Nf8 (+0.39, a 0.20 improvement)
Now Kasparov grinds out small improvements that do not register with the 0.10 threshold.
33...Re8 (+1.12, a 0.52 improvement)
There are more alleged inaccuracies after that, black's game deteriorates rapidly, and, by move 42, white's advantage is about (+4.25).
I think that this deterioration could be caused by bad time management and the approaching time control, but this does not explain Deep Blue playing 40...bxc6 instead of Rxc6 (a 1.22 inaccuracy), or 41...Kh6 instead of Kg8 (a 0.71 inaccuracy).
|Dec-31-07|| ||amadeus: <MAJ: 17.cxd5 (+0.50, a 0.33 improvement) -- the only move found where Hiarcs considers white to have lost some of the advantage.>|
Rybka, at 15 ply, evals Nf3 as 2 centipawns better than cxd5.
|Dec-31-07|| ||fm avari viraf: Gary the great, systematically outplays Deep Blue as if the Computer went into Deep Sleep. Simply a wonderful game.|
|Dec-31-07|| ||Jimfromprovidence: Has anyone evaluated 17... Bc7 for black instead of 17...Bb4? The bishop ends up there on move 21 anyway at the cost of several tempi of development.|
|Dec-31-07|| ||MostlyAverageJoe: <amadeus: Rybka, at 15 ply, evals Nf3 as 2 centipawns better than cxd5>|
Give it 3 minutes on a fast CPU with dual core and it should get deeper than 15 plies. Hiarcs got to 18 plies in 3 minutes (after spending 3 minutes apiece on preceding moves, so it had some of the hash pre-computed).
In any case, improvements for white are not that important here. It would be interesting to know what improvements would Rybka find for the black, and whether they correlate with those from Hiarcs. If you'd like to run such an experiment, make sure it does its analysis in forward scan mode, not backsliding -- the backslide does not simulate the conditions occurring during the game.
|Dec-31-07|| ||Sularus: Goodbye 2007.|
|Mar-28-08|| ||just a kid: Rage against the Machine.Great band.|
|Apr-17-08|| ||crafty: 30. b6 ♗d8 31. ♗c3 ♘c6 32. ♖ce1 ♔g8 33. ♕d2 ♗f6 (eval 1.24; depth 14 ply; 250M nodes)|
|Nov-26-08|| ||Cactus: Deep Blue had No Shelter?|
|Nov-26-08|| ||brucejavier: great game by the great garry!|
|Jan-10-09|| ||WhiteRook48: and the trapped rook and bishop! Even computers would resign in this position!|
|Feb-02-11|| ||hottyboy90: WhiteRook48 the computer did resign!|
|Feb-02-11|| ||jmactas: Holy queenside congestion batman!|
|Mar-11-11|| ||hottyboy90: Thats a very random comment!|
|May-15-14|| ||talwnbe4: Amazing how like a computer Garry plays, isn't it ? Deep Blue analysed at 100 million positions a second, but today's software has much better pruning and time management.|
Fruit prefers 11..Nb4 over 11..Nh5.
15..dxc4?! 16. Ndxc4 Bb4 17. Bxg6 fxg6 18. Bc3 Nxe5 19. Nxe5 Ba3 20. Rc2 Bb5 21. Nc4 Bxc4 22. bxc4 1.09 Fruit 2.1
15..Ngxe5 is what Fruit prefers instead of DB's 15..Rc8.
Move 19 Fruit plays 19..Nge7 after 3 minutes, same as Deep Blue.
Fruit finds 33..Bb4 best too after 3 minutes.
Fruit considers 33.. Ne5 the best after 3 min 15 sec, same as DB. depth 17 ply, but really it isn't looking ahead 17 ply.
Fruit gives 22..Nf5 23. Re2 Re8 as preferable to 22..Re8 but evaluates this position as 0.70. You can see DB's position is already not good.
Yeah, the blockage of blacks QB resulting in the rook also being locked up was horrible.
|Jun-05-14|| ||Zhbugnoimt: 20.Bxh7+! Was a very strong shot, and Kasparov would have played it had he been playing vs a human, but against deep blue...|
|Jun-05-14|| ||goldenbear: <patzer2: Deep Blue's terrible 30...Bb8? adds new meaning to the term "bad bishop." Maybe here it's like saying "bad doggie" after your normally well behaved pet inexplicably chews up and destroys your good sofa.> It would have to "add new meaning", I guess, because Deep Blue has the "good bishop" and Kasparov has the "bad bishop"...|
|Aug-11-17|| ||Toribio3: Beautiful game! Kasparov is the Superman.|
|Aug-12-17|| ||HeMateMe: what's the finishing line? white moves his king out of the way, puts a Rook on the g file and plays g5 as a battering ram?|
|Aug-13-17|| ||beatgiant: <HeMateMe>
Here's an example with your kingside attack plan: 43...Kg8 44. g4 hxg4 45. h5 gxh5 46. Qe3 Kf8 47. Qg5. I don't see how Black stops the mate threats, even with computer moves.
Besides that, Black's pieces can hardly move, and White also threatens to lift the blockade on the b-pawn (with Qe7, or if Black's rook moves, with Qxc6).
|Aug-31-17|| ||thegoodanarchist: <HeMateMe: what's the finishing line? white moves his king out of the way, puts a Rook on the g file and plays g5 as a battering ram?>|
Premium Chessgames Member beatgiant: <HeMateMe>
Here's an example with your kingside attack plan: 43...Kg8 44. g4 hxg4 45. h5 gxh5 46. Qe3 Kf8 47. Qg5. I don't see how Black stops the mate threats, even with computer moves.>
With the Black king on h6, your 43...Kg8 is not a legal move.
Monty Newborn, who wrote the book on this match (and computer chess development), gives the only move for Black as 43...Kh7, followed by 44 Qe7 Qxe7 45 Bxe7 Bxe5 (or 45...Re8) 46 b7 +-
< Earlier Kibitzing · PAGE 2 OF 2 ·