|
|
Computer CC Challenge Match |
| The following game is being played by Volker Jeschonnek vs. a computer to shed light on the strength of a chess engine in correspondence chess played under tournament conditions. |
|
White: Volker Jeschonnek
1.Nf3 The coin toss gave me the white pieces and this is definitely a good start. I like the knight move a lot since play can transpose easily in one of the main lines after 1 d4 or 1 c4. By delaying the push of either pawn I get a peek at my opponent's intentions. 1...d5 2.d4 The program moved the central pawn and hence made a commitment. I do like the Queen's Gambit Accepted, the Tarrasch Defense, or the Catalan for White so my move is a rather obvious choice. 2...e6 3.c4 So far I get my way. By playing 2... e6 the program excluded the Slav Defense with which I have very little experience. I met it only during blitz and rapid games. The program still can choose from a large variety of moves: 3... dxc4, 3... c5, 3... c6, and 3... Nf6. But all of these moves are familiar. 3...Nf6 4.g3 Be7 The computer seems to join the trend set by my human CC opponents which would make this game my fourth Catalan within a year. On move 4, Black had interesting alternatives. 4... dxc4 5 Bg2 c5 6 0-0 Nc6 7 Ne5 leads to a well-known position which my recent opponents chose twice. In the above variation the move 5... b5 may look like the typical greedy computer move but the list of players arguing the black side represents a little "who is who" in world class chess. To name just a few: Beliavsky, Lautier, Oll, Karpov, and Korchnoi (maybe not that much of a surprise). A very popular choice in grandmaster play is 4... Bb4+ 5 Bd2 Be7. The philosophy behind this is that the white bishop on d2 is not well-placed. The same maneuver is played frequently in the Queen's Indian Defense. 5.Bg2 0-0 6.0-0 The computer still has the option to transpose to the main line of the Tarrasch Defense by playing 6... c5 7 cxd5 exd5. This opening is an old favorite of mine as Black. Recently, I looked at Nunn's Chess Openings and was surprised to see that some of the lines that were considered dangerous for Black seem to be repaired by now. In the Catalan Black can either play 6... dxc4 or 6... c6 (6... Nbd7 followed by 7...c6 is basically the same). I have the feeling that the computer will go for the latter option. 6...Nbd7 7.Qc2 c6 As I predicted WChess goes for a closed position. 8.b3 b5 This is definitely a surprise. I thought of the possibility that the computer would avoid the main line of the Closed Catalan. However, my guess in this case was 8... Ne4 with transposition to the Stonewall variation of the Dutch, e.g., 9 Bb2 f5 10 Ne1. As one can see there still were some transposition tricks left. Now it is officially a Closed Catalan (ECO code E08). Of course, Black may play f7-f5 later and we will get a hybrid here. For the spectators I wished for the main variation of the Closed Catalan because 8... b6 9 Rd1 Ba6 (9... Bb7 is a solid move) 10 Nbd2 Rc8 11 e4 c5 12 exd5 exd5 13 Bb2 would result in a very interesting position with tension in the center and some room for piece-play. 9.c5 Ne4 10.Bb2 That makes it a double fianchetto which is supposed to be a good anti-computer strategy. Let's see. I spent some time on analyzing the more direct approach 10. Ne1. White's idea is to kick out the black knight as soon as possible, e.g., 10... f5 11. f3. Another advantage of f2-f3 is that it also prepares for the central thrust e2-e4. This is White's natural plan here. However, with 10... e5!? Black can try to take advantage of White's lack of development. My main variation runs 11. Bxe4 dxe4 12. Qxe4 Bb7! 13. e3 Bf6! 14. Nc2 Re8 and it seems that Black has (just) enough compensation for the pawn. Having the white pieces I decided that there is no need to allow obscure complications this early. 10...f5 Note that 10... e5 does not work. 11.Ne1 Two of my chess heros, Ulf Andersson and Vasily Smyslov, played 11.Nbd2 in this position. So I looked very carefully at this move. However, recent practice indicates that Black can and should go (unashamed) for the exchange of all knights. With knights gone it is difficult for White to get something going. 11...Bg5 This move prevents White from challenging the knight immediately. 12. f3? is punished by 12... Be3+ 13. Kh1 Nxg3+! 14. hxg3 Qg5. Alexander Bangiev thinks that Black has a clear advantage and, indeed, White's position looks terrible. 12.Nd3 Qa5
The computer goes its own way and now we are out of the books.
Nunn's Chess Openings cites Dolmatov's assessment 12... a5 13. e3
Qf6 unclear. From the games I looked at the thematic moves for
Black seem to be 12... a5 and 12... Qe8. Usually Black plays both
moves as in the exciting correspondence game Bouwmeester, H. -
Javelle, G. from the Europe Echecs 20 years tournament (1981-1987)
which went 12... a5 13. a3 Qe8 14. e3 Qh5 15. Bc1 Qh6 16. Re1 Bf6
17. f3 Ng5 18. Bb2 Nf7 19. e4 fxe4 20. fxe4 e5 21. Nd2 exd4
22. exd5 Nde5 23. Nf4 Ng4 24. Nf1 d3 25. Qd2 Bf5 26. dxc6 Rad8
27. Rad1 Kh8 28. h3 Nge5 29. Ne3 Qg5 30. Kh2 Bc8 31. c7 Rde8
32. Ned5 (Diagram)
32... Bxh3 33. Rxe5 Nxe5 34. Bxe5 Bg4 35. Bxf6 gxf6 36. Qxd3 Bxd1 37. Qxd1 Rg8 38. Qf3 1-0 What do I do now? From the examples of GM play it seems that the Black Queen belongs to the kingside. I analyzed the position quite a bit and feel that White should stay away from forced lines for the moment. The position is closed and that means that the loss of a tempo may not be that significant. |
|
|
13.a3
This is a useful move if Black, now or later, returns to the main scheme. White takes the b4-square under control and hence fights Black's strategic dream b5-b4 and Bc8-a6. After that I wish to follow the "cookbook", i.e., I prepare for e2-e4. White's idea is to play e3, Re1, f3, Nbd2. Let's see what happens. 13...Ba6 14. Re1 One must be alert. The natural move 14. e3 allows 14... b4 (which is the idea of Ba6). After the sequence 15. axb4 Bxd3 16. Qxd3 Qxb4 17. f3 Nef6 18. f4 Bh6 (Diagram).
It seems that White is somewhat better however a draw is the most probable outcome. So, to prepare f2-f3, White has to transpose moves. 14... Rab8 I expected the computer to go for something more aggressive. 14... f4 comes to mind. Here, White must be careful because of the vulnerability of the kingside which gives Black several tactical motifs. I spent some time on this position and finally I decided that the following is a good (and safe) continuation: 15. f3 (protects the f2-square) Nef6 16. b4 Qc7 17. gxf4 Bxf4 18. Nxf4 Qxf4 19. e3 Qf5!? White seems to have a small advantage but Black can be satisfied also. 15.e3 Rf7 Black prepares doubling rooks on the f-file which may become open after a later e3-e4. 16.b4
Poor necessity if I want to stick with my plans! It is
unfortunate to play this move since b4 might have been a
good square for a knight. However, the "logical" move 16.f3 (Diagram)
is met with the queen sacrifice 16... Qxe1+. After 17.Nxe1 Bxe3+ 18.Kh1 (not 18.Kf1?? b4+ 19.Nd3 Nf2 and Black wins) Nf2+ 19.Kg1 Ne4+ Black has an easy seesaw draw. 16... Qc7 17.f3 Nef6 18.Nd2 This was not an easy decision since 18.Nc3 is more natural in regard to promote e3-e4. However, experience has taught me that in the early stages of the game drastic measures seldom lead to the desired results. Joachim Breideneichen, a very strong and experienced OTB player, once explained it to me that way: "You shouldn't try to punish each mistake of your opponent right away. It is often better just to make good moves and keep your options open. When enough mistakes are on the board THEN it is time to take decisive action." I chose 18.Nd2 over 18.Nc3 because it is more in harmony with the other pieces (Nc3 blocks the b2-bishop) and, indeed, it gives me another option. Since the e5-square is still in White's hands there is the possibility of a later Nd3-e5. If Black exchanges knights on e5 then White gets the great d4-square for the second knight. 18...Bh6 Wchess is probably hoping to achieve e6-e5 later. The retreat of the bishop is a necessary preparation for this purpose since after the "dream scenario," i.e. a multiple exchange of pieces on the e5-square where Black captures with a piece last, f3-f4 would simply win a piece. 19.Ne5 After a long think of ten days I decided to play this move. My problem simply was to choose a plan. Since Wchess has no threats White can (and probably should) look at several ideas. The pawn structure suggests to prepare g3-g4, e3-e4, or a3-a4. However, even after much thought it remains difficult to evaluate how promising each of these possibilities are. Although Wchess' position is somewhat passive it is quite flexible still. In the end I made my decision by instinct. With the text move I (am willing to) drop the e3-e4 idea and go for a later a3-a4 |
|
|
19...Nxe5 Wchess is obviously not buying into the "There will be a strong knight on d4" philosophy. 20.dxe5 Nd7 21.f4 Although a pair of knights has been exchanged Black is lacking space for maneuvering. 21...Rd8 This move simply loses a tempo since the rook's position is not improved. 21... g5 was worth a look. 22.Nf3 The text move serves several purposes and, by the way, discourages 22... g5. It is, however, difficult to prove that this would actually be a mistake. 22...Nf8 It seems that Wchess is opting for passive defense. 23.Bf1 This pins the black b-pawn and threatens 24. a4. 23...Bb7 24.a4 a6 25.Ra3
This move takes care of all tricks involving a7-a5, which would either intend to completely block the queenside by a5-a4 or lead to massive exchanges on this wing. Wchess cannot capture on a4 because the pawns on a7 and c6 will become terribly weak. Moreover, the capture would open ways for White to infiltrate the black queenside.
White will capture on b5 when the time is right. Note that the pawn structure contains hidden dangers for Black. After a later Nf3-d4 and a4xb5 White may be able to sacrifice the bishop on b5. In return for the piece White would get two connected passed pawns and a powerful knight. This sacrifice would be even more effective if the other white pieces would be ready to support the advance of the passed pawns.
|
|
|
25...g6 This move appears purposeless to me. White has a clear plan in the position, intrusion along the a-file and/or a bishop sacrifice on b5 (after Nd4 and axb5). Since the key to successful defense is anticipation Wchess should have done something about that. If, after the sketched piece sacrifice, the white knight arrived on b5 it will attack the black queen and the jump to d6 would again threaten a piece of higher value. It is rather obvious that the defending side cannot allow the opponent to have too many ways of winning a tempo. So, 25... Re7 would have been a more reasonable choice. Also, Black may be forced to play g5 at some point to seek counterplay or a complete blockade of the kingside. In this case Black would have spent two moves on what could be done in one. 26.Nd4 Ba8? This move is even worse than it looks because the computer is practically waiving its right to make a move. It is hard to believe that the computer overlooked that the (thematic and hence foreseeable) move 27. Rea1 absolutely forces the return 27... Bb7 (see the next note for more details). 27.Rea1 This comes with the threat 28. axb5. The alternative to the return of the bishop, 27... Qb7, is demolished easily. After 28.axb5 axb5 White wins with 29. Bxb5. For example, 29... cxb5 (29... Rc7 30. Bxc6 Rxc6 31. Nxc6 Qxc6 32. b5 Qb7 33. c6 and wins) 30. c6 and wins (30... Rc7 31. cxb7 Rxc2 32. Rxa8). 27...Bb7 28.axb5 axb5 I don't believe in 28... cxb5 because 29.Rxa6 seems very strong. The b5-pawn will fall and White gets fantastic compensation for the exchange. 29.Ra7 White is in no hurry to sacrifice the bishop on b5. Note that it is not enough to create connected passed pawns on the queenside. One must also make sure that Black won't be able to block the pawns and that Black won't get counterplay because of unprotected white pieces. In this context, the b5-square is very important. It is the job of the second rook to control this square a second time. 29...Bg7 30.R1a5 As I explained above, White needs to control the important b5- square. |
|
|
30...Kh8 31.Bxb5 The time is right now. 31...cxb5 32.Nxb5 Qb8 The most natural move which I analyzed heavily. Black has other options but none of them looked convincing to me. 33.c6 Being a materialist I seize the opportunity to make a profit. On move 31 I sacrificed my king-bishop for two black pawns and hoped that the resulting connected pawns would eventually force Black to give back the piece under unfavorable circumstances. The text move actually forces the computer to return the piece (a few variations are shown below) but the right evaluation of the position needed careful study of many variations. Although Black seems to have very passive pieces one has to make sure that (after simplification) the passed b-pawn remains an asset. |
|
|
33...Bxc6
On 33... Rc8 White wins rather easily with 34.Nd6,
for example, 34... Rfc7 [34... Qxd6? 35.exd6 Rxc6 36.Qxc6 Bxc6 37.Rxf7 Bxb2 38.Rxf8+; 34... Rxc6 35.Nxf7+ Kg8 36.Qa4!! (the reversal) because after 36... Kxf7 37.Qxc6 the bishop is pinned.] 35.Rxb7! [Not 35.Nxc8 Rxc6] 35... Rxb7 [35...Qxb7 36.cxb7 Rxc2 37.bxc8Q +-] 36.Nxb7 and wins. 34.Qxc6 Rxa7
I think that 34... Rc8
was a tougher defense. I worked out 35.Qd6!! and the following variation is truly wonderful: 35... Rxa7 36.Nxa7 Qxd6 37.exd6 Rb8 38.Nc6!! Rb6 39.Ra8!! and White is winning because neither 39... Bxb2 40.Rxf8 checkmate, nor 39... Bxb2 40.Rxf8+ Kg7 41.d7 work for Black. It is not much of a surprise that White can play with many pieces hanging but it was certainly hard to prove that there is no way for Black to take advantage of the relative vulnerability of the b-pawn. My analysis consists of several pages. Relatively best for Black seems to keep the queens on the board by playing 35... Rb7 36.Rxb7 Qxb7 37.Nd4 although I think that White should win. 35.Rxa7 Rc8 36.Qb7 Qxb7 There is no choice. 36... Bh6? 37.Nd6 Qxb7 38.Rxb7 Rc2 39.Bd4 Rc1+ 40.Kg2 Rc2+ 41.Kf3 Rxh2? 42.Nxf7+ Kg7 43.Nxh6+ Kxh6 44.Bc5 and wins. 37.Rxb7
|
|
|
White has a winning position. Note that since Black's knight and bishop are (permanently) out of play White does not have to worry about counterplay generated by the black rook. White's b-pawn is ready to run and will most probably cost Black the rook. Hence it doesn't really matter if Black wins the h2-pawn (or even the g3-pawn). Black can only hope that White will get his king into trouble. But if the king stays away from f3 I don't see what Black can do. 37...Rc2 38.Bd4 Rc1+ 39.Kg2 Kg8 Realizing the hopelessness of the position. After 39... Rc2+ 40.Kf1 Rxh2 41.Nd6 White is ready to push the b-pawn. 41... Rh3 is no threat since White has 42.Kg2. So it is better for Black to check some more: 39... Rc2+ 40.Kf1 Rc1+ 41.Ke2 Rc2+ 42.Kd3 (the king is safe here) Rxh2 43.Nd6 Rh3 44.b5 Rxg3 45.b6 h5 46.Rc7 h4 47.b7 h3 48.b8=Q h2 49.Rc1 Rg1 50.Qb1, but again White wins easily. 40.Nd6 Rc2+ 41.Kf1 Rc1+ 42.Ke2 Kh8 Giving away another tempo. Wchess could resign as well. 43.b5 Rc2+ 44.Kd3 Rc1 45.b6 h5 46.Rc7 Re1 47.b7 Rb1 48.Rc8 Kh7 49.b8=Q Rxb8 50.Rxb8 g5 1-0. Summarry: In this game Wchess chose an opening variation that resulted in a closed position. In its quest for counterplay the computer used its pieces to probe the queenside. In this kind of position, unfortunately, this is the wrong idea and the wrong side. The computer needed to advance the pawns on the kingside at some point. In the end Wchess' maneuvers simply resulted in a huge loss of time. After it became clear that White was making progress on the queenside (!) the computer also failed to foresee the need to establish a defensive position. It seemed to be surprised by the piece sacrifice that opened up the queenside and led to a more or less straightforward White win. Last Updated: 2000.29.06 |
|
Copyright © 1999 by Volker Jeschonnek, all rights reserved