Improving Chess Analysis Skills – Bishop against Knight endgame

I am a believer in the approach of analyzing one’s games as a major tool for chess improvement, and already wrote a post about it. I keep a database of all games I played, and as I am bored – look back at my old endgames from many years ago. Most of them are more complicated than they appear on the first sight, but what I find even more interesting is that my old annotations from around 1999-2001 often contain more mistakes than the actual games. Mark Dvoretsky pointed out the phenomena that players are more relaxed during analysis than during the actual games, so annotations contain even the most obvious errors. He used Shirov’s games, but then also went on to give an example of how Kasparov missed mate in one while annotating Lasker-Steinitz game. I suspect in a lot of cases, including my own – the process of finding mistakes in old analysis has to do with

• computer engines having become stronger by the time you get around to double check your analysis
• having more time to focus on a position (Alexei Shirov is probably a bit more busy playing in tournaments than Mark Dvoretsky, I also don’t play as much now as I used to)

Medalen – Jiganchine, 2000

Black to move. Does 73… Bb8 win?

I played 73…g3!?, and after 74. Nd6 Bxd6 we both promoted our pawns, I got a drawn queen endgame with an extra pawn (which my opponent quickly lost). During and after the game in my analysis I was convinced that Bb8 wins on a spot. But actually – it leads to an even more forced draw. This must be some kind of optical illusion that makes you think that a bishop is doing a superior job to a knight in blocking a pawn, but they are actually equivalent as the next diagram shows.

74. Kb5! g3 75. Ne3 h5 76. Kxa5 h4 77. Kb6
Black to move. There is no win.
77… g2 78. Nxg2 Kxg2 79. a5 h3 80. a6 h2 81. a7 Bxa7+ 82. Kxa7 h1=Q 83. b8=Q = with a dead draw