10...Bb7

It took some time to check 11. dxc5. A possible continuation is 11... Qc7 12. cxb6 axb6 13. Qb3 after which Black has a choice between 13... Bxe4 (14. Ng5 Bc2!) and 13... Qxc3 14. Qxc3 Bxc3 15. Bxe7 Re8 16. Rac1 Rxe7 17. Rxc3 Rxa2 18. Bc4 Rb2!? I like Black here but the position may be drawish.

11.e5

I expected 11. Qd3 which keeps White's options in the center open (d4-d5 is a thematical move also) and would enable White to see what Black is going to do about the knight. After (11. Qd3) 11... Nd7 12. e5 is more effective than before since it denies the knight the only acceptable square. Moreover, the exchanges initiated by (11. Qd3) 11... Ba6 12. Qe3 cxd4 13. cxd4 Bxe2 14. Qxe2 are certainly in White's favor. Note that 14... Bxd4?? is a severe mistake due to 15. Rad1 Nc6 16. Qc4 and White wins. I toyed around with the idea 11... Qc7 which looks interesting. Well, maybe another time.

11...cxd4 12.cxd4

This is, of course, the natural move. I also analyzed 12. Qxd4 (with computers one never knows :) and it seems that Black has no other choice than to exchange Queens. However, Black is fine in the endgame after, e.g., 12... Qxd4 13. cxd4 Nc6 14. Rfd1 Rfd8 15. Bb5 e6. Keeping the Queens on the board can easily lead to a total disaster for Black as the following cute line shows. (12. Qxd4) 12... Nc6 13. Qh4! Qd5 14. Bc4! Qa5 (Diagram) (that's what I counted on in my first analysis session but it does not work.)

15. Bb4! Nxb4 16. cxb4 Qa3 17. Ng5! h6 18. Nxf7 and Black can throw the towel because 18... Rxf7 is met with 19. Qxe7!. (By the way, it doesn't matter if Black had played 16... Qa4 and/or 17... h5 in the above variation.)

12... Qd7

Black is lining up against the d4-pawn.

13.Qd2

Again, I analyzed the less obvious, but more forcing move 13. d5. Black must play accurately but, again, is doing well. It took me some time to find the necessary tactical tricks. After 13. d5 Qxd5! 14. Qxd5 (14. Bxe7 Re8 15. Bd6 Qxd1 16. Rfxd1 Nc6!) 14... Bxd5 15. Bxe7 Re8 16. Bd6 Nc6 17. Bb5 (on 17. Rfd1 the amazing move 17... Nxe5! is possible because of the unfortunate placement of the a1-rook) 17... a6! 18. Bxc6 Bxc6 19. Nd4 Black has no problems.

13... Rd8 14. Rfc1

The computer continues to play active, however double-edged, moves (see 10. Ba3, 11. e5) and I am not sure that I understand this. Wchess takes the only open file under control but leaves little maneuver space for the a1-rook which often allows tactical motifs against White's central pawns. In my database I found the correspondence game Wagner - R. Weierstrass, EU-chT4 1988-96, in which White played 14. Rad1. However, his setup was too slow (and too tame) and Black achieved a dream position: (14. Rad1) 14... Bd5! 15. Qb2 Nc6 16. Bb5 Qb7 17. Bxc6 Bxc6 18. Nh4 Rd7 19. f4 Rad8 20. Nf3 Bxf3 21. Rxf3 Qe4 (Diagram)

and White resigned here. The big question at this point is whether 14... Bd5 is also playable in this game. Of course, I can play 14... Nc6, which is certainly a good move, but it blocks the b7-bishop.

14... Bd5

Following the saying "No pain, no gain." After an involved calculation of variations, some of which are very beautiful (see the following note below), I concluded that Black's setup from the game Wagner-Weierstrass (see previous note) should work here, too. At the cost of a tempo Black keeps up the pressure on the d4-pawn. The alternative 14... Nc6 is certainly rock-solid but, in my opinion, Black will hardly be able to achieve more than a draw.

15.Rc2

I spent some hours on the consequences of the forced variation 15.Qc3 Nc6 16.Bb5 Rac8 17.Ba6 Rc7 (Diagram)

where White has a nasty pressure on the c-file. I can't really claim that it is my analysis which makes me think Black should be fine here. It is more correct to say that the combination of facts that 1) White has no direct way to use its advantage in development and 2) that Black has indeed very active pieces make me think that Black is going strong. For example, Black threatens to chase away the rook by means of Bh6 and Be4. Suddenly Black's bishops wake up. A variation that illustrates Black's chances (the beautiful variation I promised you) is the following: White loses after 18.Bb5? Bxf3 19.gxf3 Nxd4! 20.Qxc7 (everything else loses material) 20... Nxf3+ (Diagram)

21.Kg2 (21. Kf1 leads to a forced checkmate after 21... Qh3+ 22.Ke2 Rd2+ 23.Ke3 Bh6+ 24.Ke4 and now Black can choose between three mates, namely 24... Qg4, 24... Qf5, 24... Rd4. 21. Kh1 Qh3 leads to the same position as the main variation.) 21... Qg4+ 22.Kh1 (22.Kf1 Qh3+ see last note) 22... Qh3! 23.e6 (23.Qxd8+? Bf8 and the point h2 cannot be defended anymore) 23... Be5! 24. exf7+ Kg7! 25.Qxe5 (25.f8Q+ Rxf8 26.Qxe7+ Rf7 27.Qxe5+ Nxe5 is no better) 25... Nxe5 and Black will win on material.

15... Nc6 16.Rd1

Note that both 16.Rac1 and 16.Bb5 lose material as the following variations show: 16.Rac1? Bxf3 17.Bxf3 Nxd4 18.Bxa8 Nxc2 19.Qxd7 Rxd7 (20.Rxc2?? Rd1 mate); 16.Bb5? Bxf3 17.gxf3 Nxe5! 18.Bxd7 Nxf3+ 19.Kg2 Nxd2 20.Bc6 Rac8 21.Rd1 Bh6. Hence, it can be said that White is handicapped by the unfortunate central pawn formation.

16...a6

The knight is a cornerstone of Black's opening strategy (the rook-queen-knight-bishop battery puts quite some pressure on the d4- pawn - see next note) and the text protects it from being pinned by the bishop later. It can, of course, also be seen as a first step to push Black's queenside pawns.

17.Kh1

Speaking in human terms, Wchess probably wants to play the natural move 17.Qe3. At the moment, however, this is not possible because the move would hang the d4-pawn. Part of White's problem, that makes Black's rook-queen-knight-bishop battery work, is the unfortunate king position. Or, speaking in variations, 17.Qe3 Bxf3 18.Bxf3 (On 18.gxf3 Nxd4! works as well) 18... Nxd4! (D)

When I saw 17.Kh1 it was clear that Wchess had found a clever idea to reduce the pressure on the center. The move has backsides too but it is probably a good trade. Hopefully, new combinatorial motifs may arise that involve the facts that (after a knight move) the g2-pawn can be captured with check and that the h3-square is somewhat weak (one can dream of a pawn march h7-h5-h4-h3 or Qd7-h3). Moreover, a move of the f-pawn doesn't solve the potential back rank problem anymore.

17...Rab8

It is clear that 17... Bxf3 18.Bxf3 is simply bad for Black. The text move, to a certain degree, revives the old combinatorial idea by removing the rook from the h1-a8 diagonal (hence "prophilactically unpinning" the knight). It's main idea, however, is to find a new setup that keeps the pressure on White's position and preparse the advance of the queenside pawns.

18.Qe3

Of course not 18.Bxa6? Bxf3 19.gxf3 Nxd4 and White is in deep trouble.

18...Qb7 19.Rdc1

It is obvious now that Wchess has a good position and that I was wrong when I thought the backward d-pawn would give the computer trouble. The text move restricts Black's options and, moreover, introduces the possibility e5-e6. With the knight being attacked twice, f7xe6 no longer works automatically. E.g., 19... Rd7 20. e6 fxe6?? 21. Rxc6 Bxc6 22. Qxe6+ Kh8 23. Rxc6 and White gets two minor pieces for the rook.

19...Rdc8

White's pieces are very active and fully compensate for the structural disadvantage. The obvious strategy for me is to reduce pressure by exchanging material. After the text move I will be able to play Na7 and to challenge White's possession of the c-file. Note that it is much too soon (for positional reasons alone) to play b6-b5. White would get the c5-square for the bishop and will double rooks on the b-file. Having established a firm blockade of Black's queenside pawns the thrust of White's a-pawn will put enormous pressure on them (probably more than Black can handle). This plan explains why I chose the d-rook to oppose in the c-file. If Wchess would double rooks in the b-file I have the option to let the b-pawn stand where it is.

20.Kg1

Wchess is probably thinking that the other pieces are on very good squares already. It is hard to argue about that. However, I expected 20. Rb1 with the idea to double rooks on the b-file. This plan is more ambitious and it took me some time to work out a defense. I hesitate to show the variations here because they are very long and involve various tactical ideas. Or in short: I am not sure that I got it right and the discussion is not complete. I stopped analysis when I felt that Black has enough resources to equalize.

20...Na7

Black would be comfortable after 21. Rxc8+ Rxc8 22. Rxc8+ Nxc8 23. Qd2 (the a-pawn needs protection).

21.Bc4

This logical move allows the program to fight for the control of the c-file. Because of the pressure on the black pawns on a6, e7, and f7, Black is forced into a series of exchanges. Note, for example, that 21. ... Be4?? is refuted by 22. Bxf7+. However, I hesitate to award an exclamation mark to the move. It may well be the best try here but for the aspiring player it is important to recognize that in certain positions simplification gives the opponent "automatic" drawing chances. Or, in other words, the "accumulation of small advantages" (Steinitz) is sometimes not enough for a win if the position loses its dynamic tension in the process. Note that it is possible that one side may have a clear advantage (piece mobility, space advantage, better pawn structure, or other) but the game may also be a clear draw nonetheless. In endgames this phenomenon is well-known. Jan Timman wrote in one of his books that Gennady Sosonko explained to him that in endgames with little material there was no such thing as a clear advantage. Analysis should focus on determining the result of best play: win for one side or a draw. I don't like to say this but chess is a nasty game at times. In the late middle game the same observation as above applies rather often. Some positions are just draws even though one side may have an "advantage."

21...Bxc4 22.Rxc4 Rxc4 23.Rxc4 Nb5

In my preparation I thought that 23... b5 was satisfactory here. But this is not true. After 24. Rc5! White has a big advantage (and, in the light of what I said above, the winning chances are real).

24.Qc1!

I simply missed the strength of this move because I overlooked that Wchess has an in-between move after the capture of the bishop (24... Nxa3). I was mostly concerned about 24. e6 but my analysis showed that Black is not worse after 24... f5! Other tries I looked at for White were 24. Bb2 and 24. h4. Here, too, Black has little to fear.

24...Nxa3

This move is absolutely forced since Wchess threatens to retreat the bishop and follow up with a2-a4. The poor knight would have no good place to go.

25.Rc7

As I said, I missed this in-between move when I analyzed 23... Nb5. But, from a theoretical point of view, the position on the board is very interesting. Wchess' heavy pieces are active, which should be worth something, but on the other hand there are only few fighting units left. After the weaker move 25. Qxa3 Black would have obtained equal chances (even a slight edge maybe) with 25...Rc8!

25...Qa8

I considered active defense (25... Qe4) first but during analysis it became evident that the absence of real threats in connection with the pawn weaknesses on a6 and c7 will eventually force Black to passive defense. In a position like this, where Wchess has the superior piece play by far, I cannot risk a tactical hand-to-hand fight. The attackers can become targets and, obviously, if there are too many targets I will be unable to defend. Losing a tempo in this position may easily cost the game. The text move aims at simplification by keeping the option of opposing in the c-file alive.

26.Qxa3 Bf8

Not 26... e6 because the move severely weakens the squares around the black king (e7, f7, and f6) and the important d6-square.

27.Qb3!

Wchess is playing another fine move in this game! White has several threats, e.g., 28. e6, 28. d5, and 28. Ng5. The big question is now if Black can go for the counter tactic 27... Rc8. This will probably lead to the sharp variation 28. Ng5 (not 28. Qxb6?? Rb8) 28... e6 29. Qc3 (not 29. Nxe6?? Rxc7 30. Nxc7 Qc6 winning the knight) 29... Rxc7 30. Qxc7 Qe8.

This position looks great for White but Black has hidden resources (like back-rank mate threats and in some variations perpetual check). On 31. Qxb6 Qa4! the black queen exposes the weakness of the white queenside. Also, many queen endings are obvious draws (except in the case where the exchange of minor pieces on d6 results in the creation of a powerful passed pawn). Hence Black can often play Bf8-e7 with win of an important tempo.

27...Rc8 28.Ng5

Another in-between move by the computer which threatens checkmate in two moves (29.Qxf7+ Kh8 30.Qxh7 mate) and hence forces e7-e6.

28...e6 29.Qc4

Not 29.Nxe6?? Rxc7 30.Nxc7 Qc6 and Black wins.

29...Rxc7 30.Qxc7 Qe8

The computer is going for the position from the last analysis diagram and it is clear that Black has problems. For example, White would obtain a winning position if it were possible to force an exchange of minor pieces on d6 (knight vs. bishop) and then to capture with the e5-pawn. However, this is easier said then done. In many variations Black protects the vulnerable f7-square by placing the bishop on e7. After a possible h2-h3 Black will transfer the queen to e1, e2, or f1 with the idea to capture the f2-pawn. Most of the times this would result in a perpetual check and hence Black can igonre an attack on the e7-bishop.

31.Qxb6

When I first saw the possibility of 31.d5?? I feared that I was lost. However, the move (speculating on 31... exd5 32.e6 unclear) would be a terrible blunder. After 31...Qe7 White loses material without any compensation. I also looked at 31.h4 Be7 32.Qxb6 (32.Ne4 Qd8! and Black, who has everything covered, is better) 32... Bxg5 33.hxg5 Qa4 with equality.

31...Qa4 32.h3 Be7

Black needs to drive away the white knight before the a2-pawn can be captured (32... Qxa2?? 33.Qb7 and White wins). On 32... h6 I feared 33.Ne4 but I am unable to produce any kind of refutation. Maybe it is a valid move and I was just too eager to bring the bishop into play.

33.Nxe6

I have analyzed 33.Ne4 for quite a while and think that Black, using the ideas explained in the commentary to move 30, can achieve a draw. It seems that the computer is thinking the same.

It is also possible that Wchess just doesn't want me to have an outside passed pawn. The trade knight vs. 3 pawns also saves White's a-pawn

33...fxe6 34.Qxe6+ Kf8

First I thought that White has a small advantage in the position but after some analysis I think that if Black acts bold and fearless (not necessarily my best attributes but I try to be) then White may have no more than comfortable equality. The pawns seem to do a good job at blocking the black bishop. If White keeps the queen active I don't see a way for me to make progress.

35.Qc8+ Kg7

I thought about 35... Kf7 but did not like it. My goal in this position is to secure a draw and I invite Wchess to repeat moves by 36.Qe6 Kf8 37.Qc8+ etc

36.Qb7

Wchess ignores my indirect draw offer.

36...Qb4 37.Qxb4

It is true that 37.Qxa6 Qxd4 has little appeal for White. I thought that Wchess would play 37.Qd7! which seems to keep the balance. The queen exchange seems wrong to me and I am at a loss why the computer is playing this.

ADDITION: I am thankful to Anthony Sanders for the information, that although the queen exchange makes Wchess' life harder it is most certainly still good enough to achieve a draw (see next note for details). Anthony asked himself: "Is this position a computer killer?" He ran Fritz 5.32 under ChessBase 7.0 on a computer comparable to Steve's (Pentium processor, 233 MHz) and found that even at a depth of 15 ply the engine preferred the queen exchange Qxb4 (evaluation +0.09) over Qd7 (evaluation 0.00). It needed a depth of 16 ply for Fritz 5.32 to come to the more realistic conclusion that Qd7 is equal (0.00) and Qxb4 is trouble (-0.29). So one can say that, given enough running time, Fritz 5.32 would not have fallen for the dubious queen exchange. However, the necessary running time was about 2 hours 30 minutes to discover that 37.Qxb4 is not equal and 7 hours 33 minutes to complete work on the 16 ply. That is more than I expected

37...Bxb4

The bishop is now in the back of the white pawns with the immediate threat 38... Bc3.

38.d5

Forced since 38.f4 Bd2 39.g3 Be3+ simply drops a pawn.

ADDITION: Actually, as my friend and critical reader Anthony Sanders discovered, I am completely wrong here! 38.f4 is the right move and needs to be played. I overlooked that capturing the pawn takes time that White can use to win the king race towards the center. White can execute this plan in two ways: 1) 38.f4 Bd2 39.Kf2 Bxf4 40.Kf3 or 2) 38.f4 Bd2 39.g3 Be3+ 40.Kg2 Bxd4 41.Kf3 Kf7 42.Ke4. The advanced pawns together with the active white king would establish a blockade and because the white king controls d5 and e4 there will be no zugzwang. Anthony looked if Fritz 5.32 (details as above) would have found that 38.f4 needed to be played. The answer was yes. However, at a depth of 18 ply the computer was undecided. There was a tie between the game continuation 38.d5 (which seems to lose) and 38.f4 (which seems to draw) at an evaluation of -0.31. Depth 19 ply breaks the tie after a running time of 15 hours and 12 minutes. Finally the computer seems to realize part of the problems. 38.d5 falls to -0.5 (15h 12m) while 38.f4 roughly stands still at -0.34 (11h 27m). For Anthony this proves that for CC one actually needs the long running times and a state-of-the-art computer. Finally, Anthony believes that I won from a drawn position by (unintentionally) setting a trap for the computer.

38...g5

This move prevents White from protecting the e5-pawn with f2-f4 and also threatens the maneuver Kg6-f5. I am curious if the computer has some surprising tactical resource here or if White is plain lost. Note that White is vulnerable to zugzwang. With Bc3 Black can force a move of the e-pawn. After Bb4 the pawns would be fixed and the Black king could walk into the center (Kg7-f6-e5). White has to do something against this. Let's see.

39.Kf1 Kg6 40.g4 Bc3 41.e6 Kf6 42.Ke2 Bb4 43.Kd3 Ke5 44.Kc4 Bd6 45.Kb3 0-1.

Summary: This game showed the computer in a good light and is, I suppose, the game that should be looked at when one is interested in what Wchess can do in CC. The engine chose a rather unambitious opening variation and I thought that by pressuring White's pawn center I could force Wchess into passivity. Unfortunately, the computer found ways to overprotect d4 with natural moves. I am impressed by its discovery that all Black's tactical ideas in the opening are based on the unfortunate position of the white king. 17.Kh1 also made clear that these tactics were within the computer's range. Having stopped my "attack" the computer took over the initiative and nicely followed up with conservative strategy. It doubled rooks on the c-file and challenged my strong bishop on d5. After the opening the computer had a clear advantage. My "idea" to save the game was to seek simplification and to look for positions (maybe with material down) that could be defended passively. At a point Wchess could have avoided exchanges but it decided to enter a deal "piece activity vs. simplification." The critical position of this trade-off was reached on move 26. White had indeed a very good position. I have no idea why the computer allowed/forced further simplifications for more piece activity since the absence of a strong tactical blow must have been evident to the computer. (Maybe I am wrong here.) Humans certainly fall for optical impressive positions where the opponent stays in the game just by one tempo or tactical trick (Wchess' back rank problem in the game) hoping that the defense will eventually run out of ideas. It seems that the position after move 32 reveals flaws in the positional evaluation of Wchess. On move 33 Wchess sacrifices a piece for 3 pawns which is neither good nor bad. After this the game was roughly equal with material K+Q+6P (Wchess) vs. K+Q+B+3P (VJ). On move 36 I offered an exchange of queens and expected the computer to decline. I thought that the resulting endgame could easily be calculated as losing for White. Actually, I was wrong here twice. First, as analysis by my good friend Anthony Sanders indicates, White can still save the game after the queen exchange and, second, the endgame needs much more computing time than I expected. Wchess misjudged the situation and eventually lost decisive material due to zugzwang.

Last Updated: 2000.03.06