The title is a question that comes up in every single chess game. While we tend to pay more attention in our studies to brilliant tactics — sacrifices, forks, pins, removal of the guard, etc. — many games are won or lost by the humble skill of exchanging wisely.
Here are a couple recent games I played against Shredder (last night and this morning) that have to do with this question of when to exchange pieces. These games showed me that I have a thought pattern that can lead to bad mistakes. The time control was game in ten minutes, which didn’t affect the first game at all but did affect my decision in the second game.
INCORRECT “RULES” FOR TRADING PIECES
(which most human players follow)
1. When you have an attack, don’t trade pieces.
2. When you’re defending, do trade pieces.
The first game will demonstrate several examples where Rule 1 is wrong. I was White in an Alekhine’s Defense, and played a pawn sacrifice that I invented: 1. e4 Nf6 2. e6 Nd5 3. d4 d6 4. f4!? Bf5 5. Bd3 Bxd3 6. Qxd3 de 7. fe! (sacs a pawn) 7. … Nb4 8. Qe2! (there is no point defending d4 because Black simply takes the pawn and forks on c2) 8. … Qxd4 9. Nf3 Qc5 10. Nc3 e6 11. a3 Nd5 (diagram)
Position after 11. ... Nd5. White to move.
FEN: rn2kb1r/ppp2ppp/4p3/2qnP3/8/P1N2N2/1PP1Q1PP/R1B1K2R w KQkq – 0 12
To exchange or not to exchange?
Applying Incorrect Rule 1, I chose not to exchange, and I played 12. Ne4. Seems obvious, right? You keep the piece on the board and you even gain a tempo with the attack on the queen. I had no doubt that this move was best.
Well, I still don’t think that 12. Ne4 is a bad move. My real mistakes came later. But I was surprised to see after the game that Shredder considered 12. Nxd5 the best move. Why is that? Well, let’s see. First of all, if 12. … ed then White strikes immediately with 13. e6! fe 14. Ng5! and Black’s king is suddenly feeling quite uncomfortable. I see no reason to doubt the computer’s assessment that White has full compensation for the sacrificed pawns.
The other option is 12. … Qxd5, which Shredder likes better. But now we see the first point of White’s trade — he has deflected the queen, so now he can castle. So the trade facilitates the deployment of White’s remaining pieces. It also eliminates a strong defender and leaves Black’s queen in an exposed position; White has at least one tempo “in the bank,” so to speak.
Let’s go further with what Shredder considers best play for both sides: 13. O-O Be7 14. c4 Qb6 and now another interesting move.
Position after 14. ... Qc6. White to move.
FEN: rn2k2r/ppp1bppp/2q1p3/4P3/2P5/P4N2/1P2Q1PP/R1B2RK1 w kq – 0 15
Where should White develop his bishop?
Shredder’s answer is 15. Bg5! Again White is not afraid of trading pieces even in a position where he is down a pawn. The main point of the move is to interfere with Black’s main task at the moment, which is to castle. If Black leaves the bishop on e7, he can’t castle because the bishop hangs. If 15. … Bxg5 16. Nxg5, Black will either have to give up the f7 pawn or expose himself to a tremendous kingside attack.
So Black plays 15. … Bc5+ 16. Kh1 a5 (to discourage b2-b4) 17. Rad1 O-O. Again this is computer analysis, and humans might play differently, but the moves seem reasonable to me.
Position after 17. ... O-O. White to move.
FEN: rn3rk1/1pp2ppp/2q1p3/p1b1P1B1/2P5/P4N2/1P2Q1PP/3R1R1K w – – 0 18
How does White make progress?
The correct move may again come as a surprise to players who are slaves of Incorrect Rule #1. The answer is 18. Nd4! Once again, White is not afraid of trading pieces even when down a pawn. The reason is the tremendous scope this gives to his other pieces. First, the knight move opens up the f-file, and if Black captures with 18. … Bxd4 19. Rxd4, the other rook also has tremendous attacking possibilities on g4 or h4.
If Black doesn’t trade but plays 18. … Qb6, White still has a great attack after 19. Rf4. It was quite funny to see Shredder’s evaluation at this point, because it swung wildly between -0.2 and +0.8 several times (i.e. from 0.2 in favor of Black to 0.8 in favor of White). What this means is that it tried one line after another and each time it would initially think that Black was a little better, but end up with White having a huge advantage. Here’s my favorite line: 19. … Nd7 20. Nf3 (back we go!) Nb8 (back we go! I would already stop analyzing here) 21. Rg4 Re8 22. Bh6 Bf8 23. Ng5!!
Position after 23. Ng5!! Black to move.
FEN: rn2rbk1/1pp2ppp/1q2p2B/p3P1N1/2P3R1/P7/1P2Q1PP/3R3K b – – 0 23
Well, we really didn’t need to analyze this far to conclude that White has an overwhelming advantage, but it’s always fun to see the fanciful stuff that the computer comes up with. After 23. … gh 24. Nxf7+!! leads to mate in all lines.
Okay, now let’s go to my second Shredder game, which shows Incorrect Rule #2 in action. In this game I was Black in a Scotch Four Knights Game, a dry, technical variation that humans virtually never play. But Shredder plays literally every opening in the book, so I’m getting lots of practice against variations that I don’t know very well.
After 25 moves of mostly trading pieces, we got to this position.
Position after 26. Qc6. Black to move.
FEN: 8/3q1ppk/2Q3p1/p2p4/1b1Nn3/1P2B2P/P4PP1/6K1 b – – 0 25
As I said, most of my strategy for the last 15 moves has been to trade material. So, I thought, what could be wrong with trading queens? It has to be right, because according to Incorrect Rule #2, the defender is always glad to trade pieces.
Except that after 26. … Qxc6?? 27. Nxc6 Nc3 28. a4 I realized that I had absolutely no way to keep from losing the a-pawn! As if the threat of Nxb4 wasn’t bad enough, there’s also Bb6 to worry about. I resigned one move later.
Okay, admittedly this is an extreme example. I was playing fast because it was a 10-minute game, and if I had taken just a little more time I would have seen that 26. … Qxc6 is simply a losing move. But still, I can’t rely on that as an excuse. If I were playing this position in a tournament game, I might be in time trouble there, too. The purpose of training is to recognize thought patterns that could hurt you in a tournament. This trade-first, ask-questions-later mentality is certainly one that I need to watch out for.
So what should Black have done? The first thing to realize is that Black must keep queens on the board. That’s the only way he can keep from losing his queenside. By process of elimination, the best move has to be 25. … Qd8, which keeps the queen on the board and defends both d5 and a5. Then Shredder recommends 26. Qb7 (eyeing the f7 pawn, and vacating c6 for the knight) 26. … Kg8 27. Nc6 Qd6. In this position we see two more reasons why Black wanted to keep queens on the board. First, if White continues 28. Nxb4 ab, Black is happy because he has a Q+N against Q+B and also his one pawn on the queenside dominates White’s two. But what if White plays 28. Bb6 instead? (diagram)
Position after 28. Bb6. Black to move.
FEN: 6k1/1Q3pp1/1BNq2p1/p2p4/1b2n3/1P5P/P4PP1/6K1 b – – 0 28
This position is another beautiful example of why trading is not necessarily good for the defender. If you just took the queens off the board, White would be winning here. But with the queens on the board, the defender can become a counterattacker! After 28. … Bc5! 29. Bxa5?? Bxf2+ White is probably going to get mated. After 28. … Bc5! 29. Bxc5 Qxc5 White can win either the d-pawn or the a-pawn with various checks, but Black will easily get a perpetual after … Qxf2+.
Conclusions:
- Evaluate every trade on its own merits. Trades don’t inherently favor either the attacker or the defender.
- When attacking, a trade may be a good idea if it (a) gets rid of a good defender; (b) moves your opponent’s pieces to awkward positions; or (c) improves the position or scope of your remaining pieces.
- When defending, a trade may be a bad idea if it (a) gives up a good defensive piece or (b) forfeits any chances for a counterattack.
