Friday, March 21, 2008

Polgar Error

I was going over some of the basic endgames from L. Polgar's Chess: 5334 Problems, Combinations, and Games. The book's analysis for Position #5113 is incorrect: Black surprisingly can draw!

White To Play

The provided solution is 1.e8=Q, followed by 1...e1=Q(?) 2.Kf6+ and White wins.

However, Black has a surprising draw if instead they play 1...Ke3!:

At this point, I'd like to quote from Muller and Lamprecht's Fundamental Chess Endings: "If {a single pawn} has advanced to the seventh rank and is threatening to promote, everything depends on whether the attacking king can assist the queen. With a central pawn, this is almost always possible, no matter how far away the king is." I briefly touched on these endgames in this earlier post.

Here we have an exception to this "almost always" rule. It turns out that White has an unfortunate placement of king and queen. For example, 2.Kf5+ Kf2!

Black's king hides in the shadow of White's, and supports the pawn's promotion. The queen lacks any move that prevents Black from obtaining their own queen (e.g. by delivering check or by pinning the pawn to the king).

I've been entering endgame positions for books into databases. One reason is that I can later play these positions vs. the computer and practice my endgame technique. Another reason is that I can explore variations not mentioned in the books to make sure I understand the positions thoroughly. However, here we can see one final benefit, especially if you have endgame tablebases at your disposal: you can find errors in old analysis and learn new tricks.


Brendan said...

Very nice. When I read through books I often find myself asking 'But, what if x?' 99% of the time it's because x loses the game immediately for such an obvious reason that it didn't need to be mentioned, but that other 1% sure makes you feel good! :)

transformation said...

outstanding. thank you. just goes to show you.

in seirawans winning chess endings, it is said on certain counts that white wins, etc, but gone over in chessBase9 and fritz8, turns out NOT to be a win, or, if it is a win, far from easy or clear...

checking! regards, dk

barb michelen said...
This comment has been removed by a blog administrator.
likesforests said...

Nice find. Believe it or not, this position was the focus of a famous endgame study by Troitsky in 1935. Here are the notes I wrote when studying it (Chessbase rocks!).

"Folks often remember that Q vs P is drawn only if there's a rook- or knight- pawn on the 7th supported by their king. That's an oversimplification, and Troitzky's study reveals the lie. If a queen has no good checks or pins--because her own King blocks her path--the position is drawn even if the weaker side has a central pawn.

So before evaluating a Q vs P endgame with a pawn on the 7th as won or lost, do some calculation to determine whether the stronger side can actually make time-gaining checks."