28 April 2011

Tablebase 1 - Capablanca ½

The endgame in 1921 Capablanca - Lasker, Game 5, plus the endgame from the next game in that series (game 10 of the same match), prompted me to look at other Capablanca endgames. My favorite resource for doing that is 'Capablanca's Best Chess Endings' by Irving Chernev, containing 60 complete games with an emphasis on endgame play. Janowsky is the opponent who appears most often in the book, losing every time, but he missed a draw in Chernev's game no.24, where a few moves earlier Capablanca had overlooked a theoretical win.

In the diagrammed position, White has just captured a Pawn on b4, leaving six pieces on the board and sending the game into the domain of the endgame tablebases. The players continued 80...Bd8 81.Bc3+ Kxg6, and after a further 82.b4 Kf5 83.Kd5, Janowsky resigned. Here Chernev quotes some analysis by Cheron & Averbakh showing that, at the moment of resignation, Janowsky had a draw in hand with 83...Kf4 84.Bd4 Kf3 85.b5 Ke2 86.Kc6 Kd3 87.Bb6 Bg5 88.Kb7 Kc4 89.Ka6 Kb3 90.Bf2 Bd8 91.Be1 Ka4. This is confirmed by the tablebase.

The drawing procedure involves realizing that the main battle will be to move the b-Pawn through the critical square b6. To do this, White must have the Pawn on b5 and the King on a6, then play Ba5. This forces the Black Bishop off the a5-d8 diagonal, allowing the b-Pawn to march unhindered to the promotion square b8. The drawing sequence demonstrated by Cheron & Averbakh shows that Black has just enough time to play the King to a4, thereby preventing Ba5 and securing the draw.

New York 1916
Janowsky, D.

Capablanca, J.R.
After 80.Bd2-b4(xP)

What Chernev didn't point out, but the tablebase finds immediately, is that Capablanca's 81.Bc3+ is a mistake that throws away the win. If White plays instead 81.Be1, Black lacks an important tempo attacking the Bishop while the King travels from g6 to a4. The move 81.Bd2 also wins because it prevents the Black King from passing through f4; the extra tempo is enough to make the Black King arrive one move too late on the Queenside. Correct play is thus a close counting exercise : how many moves for White to set up the winning formation vs. how many moves for Black to prevent it.

To play through the complete game, see...

Jose Raul Capablanca vs David Janowski; New York 1916
http://www.chessgames.com/perl/chessgame?gid=1064779

...on Chessgames.com, where one of the kibitzers also comments on the tablebase's discovery.

1 comment:

Patzer Sees Check said...

Another wondeful Capablanca post by you! I reposted it here: http://patzerseescheck.blogspot.com/2011/04/tablebase-1-capablanca.html Let me know if you have any objections.