I think white would just take the bishop with 21. Rxd1, going up the exchange.

Or have I missed something?> I'm afraid so. :-(

After 20. Qh6 Bxc2 21. Rxe7 Rxe7 22. Bf6, it looks like white has sacrificed material to get a mating attack.

But black can counter with: 22...Re1+! 23. Kh2 <forced> Qb8+! <putting queen on 8th rank with tempo> 24. g3 Qf8!

This is frequently called "technique".>

Thanks for the very succinct, yet lucid explanation.

So today, If 20... Qc6, black should keep going, I would imagine.

So after a relatively long think for Rybka:

19 Bxg7 Bxd1
20 Bh6 Qc6 (not Bxc2?)
21 Rxd1 f6
22 Rd4 Rac8
23 a4 bxa4
24 Rxa4 Rb8
25 Qc1 Kf7
26 Qa1 Ra8
27 Be3 Red8

Rybka values this position as +1.80

I don't see anything particularly strong or forcing yet.

Whatever happened to "fighting for the draw"? This position is not like white is grinding out some winning attack, at least not such that I can see. I'll keep looking, but keep in mind that the chess engines might be their weakest in this sort of game, where both sides have lots of moves of nearly equal value, there are still quite a few pieces and pawns on the board and no simplifying exchanges are worth forcing.

There might be a win by calculation, but it's not jumping off the silicon at me.

However 19...Bxg2 20. Bh6 Bh1 is nicely blocked by the interference move: 21. Nd5!, and white still goes up a piece and simplifies the ending: 21...Bxd5 22. Qe5 f6 23. Qxd5+ Qxd5 24. Rxd5.>

White also has the prosaic 21.Qg3, when black has nothing on the light squares to compensate for the piece.

Asessing the material balance first, whilst elementary, is necessary. For example, a tactic that wins a bishop may be winning with material equality but not necessarily if you are already behind on material.

A related and important theme if the attacker is behind on material is to burn material as part of an effective defense. If you calculate options without starting with an awareness of the material balance, a player might not consider for example a defense where the defender loses a piece, when in fact this might lead to material equality and a complete rebuttal of the attack.

Finally, top class players do advocate for the assessment of material imbalance before doing concrete analysis. For example, outstanding writers on chess thinking such as Kotov and Silman argue that a range of strategical factors need to be assessed first.

<YouRang in response: I think white would just take the bishop with 21. Rxd1, going up the exchange.

Also, if white does play 21. Qe5?, then instead of 21...f5, black should play 21...f6!, attacking the queen and giving black time to extract his bishop.>

YouRang's analysis about 21...f6 is a good point that I missed in a previous analysis.

It would have been better than the text after 20...Bxc2?? 21.Qe5, but it is bad because after 19...Bxg2 20.Bh6 Bxh3 21.Qxh3, black only has two pawns for two pieces. The open white king does not offer compensation in this position.

Why resign?>

I don't recall many kibitzers arguing that players should resign in technically lost but not crushing positions. I think the debates tend to revolve much more around whether or not the position IS technically lost or not (and hence, whether the puzzle solution is theoretically winning or just leading to advantage). Certainly in this game no one has argued that white should automatically resign if he concedes the two pieces for rook. Kibitzers like myself and <alphastar> have only argued that it is highly likely that white will go on to win.

I agree with you that for many of the positions and subvariations lately, the "losing" side in a technically lost position could definitely play on and make the winner prove it with good, strong chess and no blunders.

There might be a win by calculation, but it's not jumping off the silicon at me.>

I'd judge the advantage for white at around at least +2.0. I do think white has a strong attack here, but it is based on strategical grounds not calculation.

Black's kingside dark squares are weak (particularly h6), and whilst 21...f6 guards one of them it creates a hole on e6 and weakens g6. Then the kingside can be further weakened with the plan of h3-h4-h5. If black plays e5 at some point to deny white some space and make occupation of e6 very difficult, this weakens f6.

White has four pieces to maneuver on the kingside, and the rooks are cumbersome defenders for this type of position. Assessing the relative strength of the rooks as kingside defenders, versus the minor pieces as attackers, I think the rooks are not much better (if at all) than the knight or bishop. So in a deep strategical sense, with regard to the kingside attack white is kind of a piece up.

Based on this strategical analysis, I think the idea that white has only a slight material and positional advantage is an underestimation.

I also think this reasoning leads to the conclusion that white's best way to try and win is to attack the king, not to try and get to the endgame. The latter would diminish white's strategical edge, since the kingside structure is weak in the middlegame but no problem in the endgame.

19 Bxg7 Bxd1
20 Bh6 Qc6 (not Bxc2?)

I used many hours of extensive calculations from Rybka 2.3.2a and Shredder 11. It is a little tricky to get a computer to look for a draw, but I tried to direct the analysis for black toward that objective.

The answer is probably white does win, but against good play it seems to takes about 30 moves before the advantage really bears fruit. White can get the queens off reasonably quickly, but black has good play for his rooks until white has a serious promotion threat.

There are some good tricks for black that come up in the variations - like whether black can trap the bishop at h6 by pushing f6 and g5 (depends on the fate of white's h pawn). But there seem to be no variations where white can force any major increase in advantage in less than about 30 moves.

Over the board, no way I would resign the position for black until about 35 moves later, if white maintains the advantage.