|
< Earlier Kibitzing · PAGE 2 OF 11 ·
Later Kibitzing> |
| Nov-17-03 | | Spitecheck: f5 is definitely the pawn break for black in this game, with White's pieces massing on the queenside, f5 is the pawn closest to the centre (not blockaded), Black is obliged to use it especially after ...a5 as you say <B Lau>. Indeed the computer could get probably get away with an entire pawn storm on the King's wing, Kasparov foresaw this and moved his King towards his own attacking wing, that way taking the calculation of a beneficial pawn storm out of the computers ply threshold. Kasparov's tactics of mass stacking all his pieces (including King) on the same flank, were the same in game 2 actually, it's just that he blundered which clouded the tactic somewhat. This time he did not blunder (helped having the better position) and I think it prove's he is still on the right track. Kasparov's results against the computers in general don't bare the fruit of his actual strength over these chippy, wired, electric freaks. Spitecheck |
|
Nov-17-03
 | | Sneaky: I can't wait until Tuesday. This is going to be great. |
|
| Nov-17-03 | | doktortarrasch: I was there at the NY Athletic Club today. amd the whole time, my friends and I were wondering, what the heck was the computer waiting for during the entire middlegame not playing f5? If the computer played f5 <at all> there would be serious attacking chances on the kingside. |
|
| Nov-17-03 | | Benjamin Lau: It's too bad I'm going to be so busy on Tuesday, I'd like to see the 4th game live. So far this match has been pretty good. I'm glad that Kasparov decided to be flexible and not play g4!? in the Semi-Slav this time. It's obvious from game 1 that the Fritz team gleefully anticipated and exploited the opening preparation they saw in his match with Junior. Good job Kaspy, good luck for your next game. |
|
| Nov-17-03 | | Shadout Mapes: <patzer2> Another important fact about 5..a6 is that's what Kasparov played against Huzman in his recent horrible loss. I think 7...a5 was just fine. If the computer moved it's f pawn down the board and targeted the white king, it could have gotten compensation. Kasparov might have even held off the pawn capture so as not to waste too much time. Afte Bd6 Be7 and Nb8 however, it was black who was wasting time. |
|
| Nov-17-03 | | brulla: What opening will Kasparov play??
Against 1.e4: Caro-Kann?
Against 1.d4: Nimzo-Indian? Tarrasch? |
|
| Nov-17-03 | | aulero: <Benjamin Lau> I was out, so only today I read your critic in the "X3D Fritz vs Kasparov, 2003" page. The present game (and the previous one) simply shows that Kasparov understood that he must keep closed the position and avoid wild tactical openings. Jeff Sonas statistics are interesting, but in my opinion they are very far to provide guideline on how to play against a computer programs. My evaluations are the result of my personal experiments that were in some way confirmed in both the matches "Deep Junior - Kasparov" and "Kramnik - Fritz". In particular I expresses strongly my disappointment with the Kasparov play during the match "Deep Junior - Kasparov" (see Kasparov vs Deep Junior, 2003). In my games - against crafty and weaker programs - I often gained some advantages but I was never able to mantain such superior positions if they was open and especially if queens was present. Obviously I'm a limited player, but not so limited to miss the real reasons why I won or lost a game. The worst positions for the programs was always the closed ones with queens! By the way, look again to this and the previous game! |
|
| Nov-17-03 | | AirForceOne: Do you believe in conspiracy theory?
This looked like a pay-back to Kasparov's game two. Nothing is against Kasparov; but X3D's team. It seemed so STRANGE since the move number 1. How well did the Fritz team prepare to meet Kasparov? It took almost 5 (FIVE) minutes for Fritz to decide on the very first move (?)!!! How long does it take for a super computer to search a library of 2-3 millions of moves? Then Fritz isolated its a-pawn. Finally it can't think about expanding its K-side for counter-play. After the first two points, I could guess the result. Only at one point, I thought Fritz played the grand mastery game when it sac'ed the a-pawn and went for the kill as in Botvinnik-Capablanca AVRO 1938. If we could believe this was true, then we must pray for Kasparov. I can't bear to see another (Deep-) Fritz-Kasparov, as the final game in 1997 (!) Hope my suspicion is total wrong.BTW, I need some help. I read somewhere it wrote (in the meaning): "The Soviet masters prefer the bishop pair over the knight pair; but if they have the knight pair, they prefer them too!" Please point me to which book it is in. Thank you. |
|
| Nov-17-03 | | refutor: <conspiracy theory...it took almost 5 minutes for fritz to decide on the very first move> mig said that was because there was a big opening ceremony involving miss new york. the fritz team gave him 5 minutes or so to recollect his thoughts and focus on the game. and, for the record, the semi-slav with ...a6 is ugly as all sin |
|
| Nov-17-03 | | AdrianP: <AirForceOne> Re your query about the quote, it sounds to me that it might come from Botvinnik's essay on the "The Soviet Chess School". There is also an extensive discussion on B pairs v N pairs in IM Watsons "Secrets of Modern Chess Strategy" which might point to the source (or even be the source). Chigorin's games are often cited as examples of a player preferring the N pair to the B pair, and some of Petrosian's games indicate a similar preference. Re the <conspiracy theory> ... I'd be very surprised. For one thing, Kasparov looked genuinely devastated by his blunder in game 2. I also wouldn't be surprised if Kasparov would forfeit all his appearance fee for a 4-0 victory over the computer! The only reason why results against computers sometimes look curious is that in some positions they play at 3000 ELO and in others they seem to play at 1600. The shortcomings of computers are that, unless they can see a threat 18 or 20 moves apart, all positions look the "same" to them subject to their "evaluation algorithm". Formulating a plan and executing it is something which is completely absent from the way computers play chess. Because a computer can see so deep into a position, their moves sometimes have the appearance of genuine understanding of the strategic aspects of a position, but that's an illusion. |
|
| Nov-17-03 | | mauk: This is a great example of how AI has not made much progress in the last 50 years. It is all computer power. Consider: 1) Who says that computers cannot blunder? 10. ..,e4? is a huge blunder for the computer. It gets lost inmediately afterwords. It does not know how to play closed games and yet, it closes the game! 2) Computers do not understand long term plans. 23. Kd1 is a great example. Everybody, but the computer, knows where the king is heading. To the computer, the positions before and after the K move look virtually identical. Similarly for the positions before and after 24. Kc1. Kasparov can see much deeper than the computer because, for him, taking the king to b2 takes just one move! 3) It took way to long for the computer to resign. The game was really lost after 23 Kd1. Nothing much happens after this. It is clear that white's King goes to the queen side, that an a4-a5-a6 push follows and that black cannot prevent this nor does it have enough space in the King side to do much damage there. Again, how can the computer resign when it probably things a draw is easy. |
|
| Nov-17-03 | | AdrianP: <mauk> Agreed on all points! I was following the game with Fritz analysing as I: before Kd1 Fritz thought the position was level after Kd1 Fritz thought it was winning by 0.5 pawn. Stupid mutt! It had no idea where the K was going, and simply thought that it was to its advantage that the K was coming into the "open", not knowing that it was going to walk all the way over to the Q-side, where it would be untouchable and Fritz's pieces would all be misplaced...! |
|
| Nov-17-03 | | courts: Only 5 total captures, wow! X3D seemed "confused" by those wasted King moves between the h and g files. X3D tried to give up the dark bishop twice. I did not see the compensation. |
|
Nov-17-03
| | Sneaky: <Computers do not understand long term plans> that's the bottom line. It's amazing how you can win a chess game without a long-range plan, but you can, just by following the short-term plan of maximizing material. Even if you are like Capablanca, and you see on move 5 that White should go on to create a passed e-pawn and queen it in 50-60 more moves, you still could make a tactical error along the way and lose a whole piece to the short-sighted materialistic machine. And as we know, that's easy to do--just ask Kasparov! |
|
| Nov-17-03 | | Alyosha Karamazov: "How to beat the living daylights out of a computer," by Garry Kasparov.
1. Make a nice closed pawn structure.
2. Make a plan.
3. Make quiet, perfect moves to develop the plan.
4. Drop the plan's conclusion on the silly but completely unemotional CPU like a ton of bricks. Wham-o!5. Do a touchdown dance (you are on ESPN, after all). |
|
| Nov-17-03 | | chessamateur: This was extremely comical to watch... But I am looking ahead. How do you think Kasparov intends to win with Black? (I'm assuming he disdains another drawn match) |
|
| Nov-17-03 | | numlock31: Does the X3d Fritz program resign itself or does one of the operators/programmers resign on behalf of Fritz? Since even at the final move, Fritz only had a -1.50 evaluation for is position, you would think the computer would still want to play. How does it know that it should resign if it calculates that white only has 1.5 advantage? I'm assuming the handlers of the program resigned for Fritz?? Anyone know? |
|
| Nov-17-03 | | MoonlitKnight: The programmer and the Chessbase team didn't want to be humiliated any further. They initially said that they wouldn't resign until Fritz gave 2.00 in favor of Kasparov, but changed their mind in this hopeless position. |
|
| Nov-17-03 | | Dick Brain: I thought the programmers resigned a little early. I don't think a computer would resign in that position. Maybe the programmers were embarassed by the ridicule they were receiving on Fritz's passive play despite the quuenside threat. |
|
| Nov-17-03 | | talchess2003: I am beginning to think that this match indeed was fixed... The computer played absolutely horrible this game. I was rolling on the floor after Bd6.. was that a waiting move (something that most strong players do not make), or did the computer actually think Kasparov would bite after game 2? |
|
Nov-17-03
 | | Eggman: <Bd6.. was that a waiting move (something that most strong players do not make), or did the computer actually think Kasparov would bite after game 2?> It was a move that resulted from the computer's number crunching. Period. |
|
| Nov-17-03 | | davewv: Can someone explain why GK took that long march with his king? |
|
| Nov-17-03 | | Benjamin Lau: <aulero>
I agree that Sonas' statistics are not conclusive, but thought that you might want to know that closed positions in anti-computer chess is not as clear cut as you might think. I think that Fritz could have possibly held on, perhaps even played for the win, if it played ...f5! at some point, attacking Kasparov's pawn chain and starting a nice kingside pawnstorm. On the Deep Junior v.s. Kasparov game you cited, I don't think it was all that bad. Kasparov had a clear advantage in that game. He tactically blundered later, but I don't see how that would have been different from say the closed Ruy Lopez Kasparov played in the 2nd game of the Fritz match. <davewv>
A classic tenet is that, if you have a clearly winning advantage and it seems like it's going to last a very long time, you should consoldiate your pieces and protect your king before making an attack. |
|
Nov-17-03
 | | jaime gallegos: 9. Qa4 ! is not new, it was played for Samuel Reshewsky against Paul Keres in a URSS-USA match, but Keres developed his K-side counterattacking ( when the center is closed we must develop the flank! is chess axioma !) and won the game... according to chessbase.com. However the computer was not able to develop a good plan ! Has Garry found the way to make the computer thinks too much ( I read X3D Fritz can see 10 movements or more at 18 Ply of deepness ! ) with too many figures and no space to move ??? Can the X3D Fritz team avoid this problem on two days ??? |
|
| Nov-17-03 | | fred lennox: <davewy> I can speculate. Black threatens 22...Qxe3 so to free the queen at d2 the king moves. the king on d, c, b, and a1 inhibits the knights mobility. In general blacks pawn structure favors kingside attack and there are threats, so GK moves his king to safety. |
|
 |
|
|
< Earlier Kibitzing · PAGE 2 OF 11 ·
Later Kibitzing> |
Kibitzing
