Chess Games, Problems, Ideas

White to draw

This is one of my favorite problems. It's not super hard, strong chess engines will find the correct line relatively quickly. However, carrying it out to a draw is difficult. One misstep by white can lead to slow death!

Black is so close to promoting the h3 pawn, but white can defend. The endgame is rather exotic and uncertain, to the point that it inspired me to write a stockfish endgame scalefactor function to properly analyze if black can win. Assuming perfect play from both sides, it's a draw. However, without endgame tablebases even a strong chess engine such as Houdini playing white looses against the modified stockfish playing black! more...

The white rook can keep on the attack with impunity because if the black king captures, the game is a draw by stalemate! Note that after 23... Nxd8 the KNNKP endgame tablebases show it as a draw. Not a surprise since there isn't any way for black to keep any white pawn behind the first or second Troitzky line.

White to win

A neat little problem where black can defend tenaciously but can only prolong the inevitable for 48 moves, so perhaps the heading should be Black to defend for 48 moves!

White to win

This gem is after move 23 in Yuri S Gusev vs Yuri Averbakh, 1946. Chess engines have no clue what's going on until several moves into the attack. Maybe by 2030 we'll have computers with enough horsepower to “see” the winning sequence (2022 update: stockfish 15 can now instantly “see” the winning sequence). Respect to Gusev for calculating this beautiful attack back in 1946! For the record, there are six main variations all of which have been explored to the point that it's clear white can force a mate. Email me of you think you found some variation where black can defend, I am convinced none exists.

White to win

No joke, white can win!

White to win

White can win, but computers do not understand how (credit: chess.com).