This is the position after black has played 15...Rd7 apparently intending to double his rooks on the e file behind his mobile e pawn. So far, black has held his own against our own Caballero. Peter plays 16.Na5 ... and black misreads the danger signals. He plays the natural but weak 16...Bc7 instead of challenging the invading Knight with the far better, but not obvious, 16...Ne4 which holds off the invasion of the white knight into c5 and also hits the backward white d-pawn. In fact, white's best reply to 16...Ne4 might actually be the passive looking 17.Qc2.
After the innocuous 16...Bc7? white gains a lot of ground in a hurry. With 17.Nc5 the black rook on d7 and the b pawn are both threatened. Further, don't miss the fact that the b pawn is the only protector of the Knight on c3 which will be under attack from the white rook on c1 when the black Knight on c5 moves. In this position there is another surprise: it is actually better for black to lose the exchange (rook on d7 for knight on c5 ) then to try to preserve the rook. It is natural to think that the rook has been misplaced and needs to move -- "accepting" the loss of a pawn and a tempo -- but again the natural move is wrong!
Black is actually astute enough to avoid 17...Rd8 but he does not find an adequate defense, playing 17...e5? instead [17...Rae8 is his best chance] . After Peter's 18.Nxd7 Qxd7 19.Bxf6 gxf6 white is doing very well, indeed, and the game is really over.
At this point, Caballero's combination had won the game and the rest was anti-climactic.
Our analysis does not end there, however. You might have considered first taking the Knight with the Bishop, 18.Bxf6 , noticing (1) that the Queen cannot recapture on f6 (it is protecting the rook) yet (2) it would seem that after Peter's 18.Nxd7, black has the option of ...Nxd7 trying to avoid the further harm to his Kingside pawns. But after 18.Nxd7 Nxd7 white has 19.Qxd5 which is even worse for black!
Strangely, if 18.Bxf6 is worthy of consideration (as an alternative to the direct 18.Nxd7), it is not for these natural reasons (for the capture gxf6 is forced in either line, as we have just seen). The argument in favor of this alternative to Peter's winning 18.Nxd7 is much more subtle! The best move for white after 18.Bxf6 gxf6 is not actually 19.Nxd7, believe it or not! White can instead go aggressively after the d pawn on d5 with 19.Ne2-c3! If black then runs with the rook, 19 ...R7d8 white has 20.Ncxd5 and the black Queen is starting to worry about safety! If 20...Qe8, of course, 21.Nxf6+. But even the apparently safer 20...Qd6 leads to 21.Nxe4 Qe6 22.Nd5xf6+ Kg7 23.Qxe6 fxe6 24.b5! and either the Knight or Bishop on the c file falls! Black's alternative 20...Qe5 faces 21.Nxe4! Rxd5 22.Qxd5! Qxd5 23.Nxf6+ Kf8 24.Nd5 ... and white is winning easily.
The lessons here, it seems to me, are (1) the natural move is sometimes wrong -- we must learn to check our own and our opponents moves for tactical considerations such as these. Peter exploited his opponents lazy analysis of 16...Bc7; (2) sometimes there is something about a position that makes natural moves fail move after move. This is just such a position. (3) Sometimes even the winning line can be improved upon but for reasons more subtle than may at first appear. We can reject a move such as 18.Bxf6 for very good reasons yet it may be worthwhile for other reasons hidden from view.
Hi, Lemma. I watched the "Sometimes the Natural Move is Wrong". I am in disagree with you when you say "Black is actually astute enough to avoid 17...Rd8". What is wrong with the following line:
ReplyDelete17...Tdd8 18. Nxb7 Tdb8 19. Txc6 Txb7 20. Tfc1
I prefer to lose a pawn and not the exchange, I think that black may holding the position despite a pawn less.For me this line gives black more chances to defense than if it lose the exchange.
Thank you for your analisis, they are very instructive and interesting.
Have a good day
Hola, Julio! Perhaps you are right -- for some players! That is, I realize that my judgment about the value of 17...Rd8 versus 17...Rae8 is really a matter of playing style and preferences, and not actually a matter of clear calculation. If black wishes to preserve the rook, as you suggest, he can indeed play as you suggest: 17...Rdd8 18.Nxb7 Rdb8 19.Rxc6 Rxc7 20.Rfc1 ...(Como se dice "rook" in espanol; por que "T"?) This leaves black down only the single pawn but he is also weak on the open c file now dominated by white's doubled rooks. So black has to choose between either (a) playing down the rook plus weakness on the c file with NO compensation OR (b) playing down the exchange (rook for knight). You are of course right that some players would prefer to play down the pawn than the exchange. Others I would argue would prefer to have some asymmetry in the position (the knight after all can do some things a rook cannot do) -- this provides SOME compensation, at least.
ReplyDeleteLet's put the question to a few chess engines; perhaps our teammates will try this position in chess engines available to them?
Comet B68: Prefers your 17...Rdd8 until it reaches 9 ply, then it switches to 17...Rb8 at 10 ply. At 10 ply, it says white is +1.36 against 17...Rb8 (with the exchange) but is up +1.51 against 17...Rdd8 (up the pawn and c file). So CometB68 prefers my move but it is very close!
Crafty 23.01: Prefers your 17...Rdd8 until at least 17 ply! Says white is up +1.80 with the pawn and c file. Says white is up +1.89 with the exchange. So Crafty 23.01 agrees with you, not me.
Fritz 8: Also agrees with your 17...Rdd8 until at least 14 ply, and it only evaluates black as being down 0.97 -- so it does not attach any value to white's c file rooks! If black gives up the exchange, however, Fritz 8 thinks white is winning by 1.18.
Stockfish 1.9.1 64 bit on the other hand agrees with me, thinking 17...Rae8 is correct, evaluating this as being -1.85 for black, whereas 17...Rdd8 is assessed as being -2.06 for black.
Rybka 2.3.2a likes 17...Rab8!? assessing black as being only -.97 down the exchange while it thinks 17...Rdd8 is -1.18 at 15 ply.
So what can we make of all this? It turns out that some chess engines think you are right and some agree with me. What we must conclude is that the intangibles involving the c-file and other compensations are not at all clear in this position! Some engines would prefer, like me, to be down the exchange, while others, like you, would prefer to be down the pawn and the c-file weakness!
Gracias for pointing this out to me and to the team!
Does anyone else have any thoughts?
Let me rephrase the summary of the chess engine evaluations:
ReplyDeleteI thought 17...Rae8 (giving up the exchange) was preferred, while Julio prefers 17...Rdd8 giving up the pawn and c file instead.
Comet: agrees with me to give up the exchange, but does NOT agree with my 17...Rae8, preferring 17...Rab8.
Crafty: agrees with Julio's move and plan.
Fritz 8: agrees with Julio's move and plan.
Stockfish 1.9.1 agrees with my move and plan.
Rybka agrees with my plan but not my move (preferring, like Comet, 17...Rb8).
So the position is not clear at all.