Return of the off-side queen
One of the fundamental reasons a successful mating attack occurs is that the defense does not have adequate pieces near the king. One of the chess teachers near where I live told me he first noticed in his students games that mating attacks occur when the opponent's queen is separated from the king, and that once he began looking for this theme he found it was true in the games of masters as well.
The game I'll show you today has this theme. A queen is distracted from her king by the lure of winning material -- in this case a whole rook! The result is that a successful mating attack results. This is the first lesson of the game. Unfortunately, I did not play the mating attack correctly and this brings us to the second lesson. I didn't mishandle the attack because of a lack of calculation ability, or a misunderstanding of the idea. My failure was procedural -- how do we decide on a move over the board? The errors are described in more detail below the game.
So why did I make these errors on move 19 and move 27. In neither case is the position beyond my calculation ability. In fact, I found the winning attack on move 20 that would have worked even better on move 19. And the mistake on move 27 is literally a one move mistake.
First, lets look at what happened on move 19. The story really starts at move 17. ... e3 where I sacrifice the a8 rook. My plan on move 17 was 17. ... e3 18. Qxa8 exf2+ 19. Kh1 Re8 20. Bd2 (as in the game) and then 20. ... Ba6 trying to deflect the defender of e1. For example 21. Bxa6 Re1. I played this line through move 19 without any real consideration to other ideas, but realized when I got to move 21 that I should double check before sacrificing any more material. This is when I realized that 20. ... Ba6 21. Bxa6 Re1 22. Nd2 and I don't have anything left in the tank. At this point Qh4 and Be4 struck me as promising and I analyzed the winning attack.
So the problem was that I followed a plan that was made on move 17 and didn't stop to think about it again until move 20. If I had stopped and given real consideration again on move 19, I believe I would have found the winning continuation. This is not a lack of chess skill, ability, or knowledge. This is a procedural mistake.
So what happened at move 27? This is painful, because I thought 27. ... Qxf3 was checkmate. I literally overlooked the queen coming back to life from the a8 square and defending f3. The whole idea of the attack was that the queen was out of play. If I had taken the time to look at the whole board and thought of every possibility, I easily see that I have to play 27. ... Nxf3+ and go on to finish the attack. For so many moves I simply hadn't had to consider the queen on a8 and I developed tunnel vision. It only took me a few seconds to make this blunder, which is again a procedural problem. I should take the time to double check every move.