Test position

Another way to indicate how silly engines really are.

Try setup this position and see if the engine can find the correct path.

Ke1?? Nd3+

White needs to lose a tempo with Kg7-h7-g6 then go down to f1

Isn't it little weird why that top engine and high end comp allows such miss.

qrb5/rk1p1K2/p2P4/Pp6/1N2n3/6p1/5nB1/6b1 w - - 0 1

Analysis by Houdini 3 Pro x64:

1.Kg6 Bh2 2.Kf7 Bg1 3.Kg6 

=  (0.00)    Depth: 51/89   00:00:46  640mN

(Doe,  17.06.2013)

New game

qrb5/rk1p4/p2P4/Pp2K3/1N2n3/6p1/5nBb/8 b - - 0 1

Analysis by Houdini 3 Pro x64:

4...Bg1 5.Kd4 Bh2 6.Ke3 Bg1 7.Ke2 Bh2 8.Kf1 Bg1 9.Kxg1 Nh3+ 10.Kf1 Nhg5 11.Ke1 Nf3+ 12.Bxf3 g2 13.Bxe4# 

+-  (#9)    Depth: 59/56   00:00:29  555mN

(Doe,  17.06.2013)

Chess engines often use something called the "null move heuristic". This basically says "if White/Black does nothing, does Black/White have any real threats?".

The problem with this approach is that chess actually doesn't have a "pass" move that does nothing, so using the null move heuristic will make the engine misjudge zugswangs.

To avoid this problem, engines that use the null move heuristic will turn it off in endgame positions,where spotting zugzwangs may be very important.

Here, the position doesn't look like an endgame (Black has far too much material) so the engines may have missed the zugwang.

http://www.anders.thulin.name/SUBJECTS/CHESS_PROBLEMS/

Problem is found here:

Otto T. Bláthy: Vielzügige Schachaufgaben (1890)

Stockfish 5 said at the beginning that it was a draw by the king alternating between g6 & f5 & the bishop between g1 & h2 and along the little path once I reached e5 it said black had counter play via after Ke5 playing down the line Nd3+ Nxd3 Bd4... but then after about 5 seconds realised the line ....Bd4+, Kd5, Bg1, Bxe4, b4 Kc4# now with the king back on g7 it's saying Kg6, Bh2, Kf5, Bg1, Ke5.... but rating it as a dead draw