A swindle that never was
I’ve started collecting chess swindles for a new book [so if you have any good ones, please send them to me!]. Over the weekend, I visited a great little weekender on the Gold Coast and saw first-hand a fantastic swindle take place – or, rather, nearly take place.
Liu-Wohl was played on board 2 in the penultimate round, an enthralling clash between coach and former pupil. Wohl, an IM and a bit of a chess legend in Australia, was Black and steadily outplayed the young Liu from a worse endgame to take complete control. But as I watched the tables turn from next to the board, I spotted a really cute swindling motif for White. Liu, a pretty good tactician, had his head in his hands, looking dejected in playing every move as his final seconds ticked down, and I briefly thought he was pulling some sort of theatrical bluff on his older opponent. Alas, the gestures were all legitimate. As I was trying to evaluate the (I assumed) inevitable queen endgame to come, Liu let his clock run down to zero on his 38th move and resigned in the same motion.
Immediately after they shook hands, I asked him why he didn’t continue playing with 38.h7!. I think it was only here that both players realised the game was far from over, and that White has some serious self-stalemate chances: 38…b3?? 39.Rxc3! bxc3 40.g6! is immediately a draw, for example.
Another pretty line is 38…Rb5? 39.Rxc3 (anyway!) 39…bxc3 40.g6! c2 41.gxf7 and there is still no way to avoid the draw.
Wohl suggested (correctly) 38…Rc8!, which is the only try for Black to keep winning chances. I pointed out that White can still continue with 39.g6!
39…Ke7+ (or 39…fxg6 40.Rxc3 Ra8 41.Rc8+ with stalemate to follow) 40.Kg7 fxg6 (else Black even loses) 41.Re2+!
41…Kd6 42.Rf2 Ke6 43.Rf6+ Kd5 44.Rf8!,
forcing a queen endgame. Black has no better than 44…Rxf8 45.Kxf8 c2 46.h8=Q c1=Q,
but White can pick up one of the pawns with 47.Qg8+, leading to a difficult queen endgame that I’m sure Liu would have been very happy to escape to. Even more painfully, tablebases confirm that it’s a theoretical draw.
It was a shame that Liu didn’t find this sequence over the board, although it would have been difficult anyway to hold the queen endgame given the time situation. And (like all swindles?) Wohl probably deserved to win. Still, it will make a nice addition to the book…
(You can replay the variations using the board above.)
The reason the g-pawn will win, while the b-pawn will not, should be understandable from a human perspective – against a b-pawn the white king is already in the optimal corner (h8) from f8, while against the g-pawn the king would be optimally placed at a8. So Black gives up the b-pawn to obtain the winning g-pawn – apparently the position of the white king is far more important than how far the pawn has advanced. Of course I immediately realized I didn’t understand queen endings either when analyzing Qxb4, because I wanted to win with Qe3 with Black to cut the f8 king off from escaping to a8, but that is a draw, only Qf4 wins. Shoot!
That’s some really cool analysis. I thought that 39.Rf2 would just transpose to the main lines I analysed, but of course I didn’t realise the importance of the little 42…g5 nuance. I had seen 42…g5 but I thought giving up the b-pawn (which Black has to do to win, I think) would just lead to a draw as White’s king is already in the ‘optimal’ position to draw. I guess I just don’t understand queen endings at all 🙂 Nicely done!
Now after digging a bit deeper i still found a drawing line for white.
Instead of 39.g6, white plays 39.Rf2. Black has several answers
39…b3 40.g6 Ke7+ 41.Kg7 fxg6 42.Rf7+ Kd6 43.Rf6+ Kd5 44.Rf8 Rc7+ 45.Rf7 Rxf7+ 46.Kxf7 c2 47.h8(Q) c1(Q) and tablebase draw
39..Rb8 40.g6 Ke7+ 41.Kg7 fxg6 42.Rf7+ Ke6 43.Rf6+ Kd5 44.Rf8 Rxf8 45.Kxf8 c2 46.h8(Q) c1(Q) and tablebase draw
There are still some other lines but nothing very important.
tablebase win for white should of course be black
In fact Lomonosov says mate in 57 which still should be checked for the 50 moves rule at least in OTB as in correspondence chess it does not matter anymore see my article http://chess-brabo.blogspot.com/2013/10/iccf.html
I believe 42… g5 instead of 42…Ke6 still wins for black. After 43. Rf8 Rxf8 44.Kxf8 c2 45. h8(Q) c1(Q) we have a tablebase win for white.