Om Prakash Mohanty <ommprakasshhgmail.com> Correspondence chess is dead but not dead yet!!

When J.R. Capablanca talked about draw death, no one believed then since chess is a game of human errors. This aspect of the game makes it interesting. Long time ago when chess was played, the aim was who hunts the king first. Now it is about who does not make a mistake. With computer-based analysis and opening preparation there is less scope of errors in OTB chess while the correspondence chess scenario is more dreaded now. With evolution of NN based computer engines Stockfish and Leela, draw death is now a reality. Someone who copy- pastes a Stockfish suggested move can achieve a 2300+ rating easily. But going past 2450 is too difficult.

Here I am giving you such positions where top engines like Stockfish have a hard time trying to find a win. Let us start ...

When it comes to a benchmark position, nothing beats this position. This game between GM Gusev and GM Averbakh is legendary. The move was easy for GM Gusev as he planned the series of moves. But when you run Stockfish 14 (latest), we will find it gets stuck at 1.Qa3 Rxe6 [0.00]. Leela with certain NN does find the star move within some time even on slow hardware. The key move here is 1.Qxe5!!. Isn't it amazing that Black is going to be in zugzwang? The game will continue as:

Here a human mind may find the solution instantly but some top engines do struggle to find the solution which is as follows:

Another position where Stockfish 14 fails for long time. It just misses the key move in SPV.

It is simply stuck at 1.Nxf2?? Here the key move is very positional. The position will continue as …

Van Breukelen, Gijs-White to play and win.
This is the famous truck driver puzzle game and has a historical significance. (This endgame composition was given by GM Plaskett to a team of GMS headed by Ljubojevic and Kasparov (co-winners) of Brussels in 1987. It is believed that Misha Tal solved it after returning from a walk in a park! —Ed.)

It is interesting to see what engines suggest, especially the top official ones.


My old Rybka 3 Dynamic gets it right ...

Here goes the solution according to composer Gijs Van Breukelen.

Here the top engines get a hard time finding the winning stroke.


And here is the solution:

Here is another position which is hard for Stockfish 14 at least. A possible variation is given below. Although Stockfish 14 finds it eventually if given proper time it does not get it within depth 30 where there are engines which can get it within depth 26.

Key move 1. Rd6!! And white wins in all variations. A possible line is given below.

Here is another position where human intervention is needed to solve the position. Whenever position has fortress human intuition works better than engines.

Here is the analysis done by me some time ago:

Without human intervention it is difficult for many engines to find that the move 3.Bd2 that wins while Ng5 leads to a draw. Looking at positions like these increase analysis skills.

Now am going to give you some positions and the best move without any comments. We leave readers to find out for themselves.

Best move – 1.Nh4!!

Best move – 1...Rf6!!

There are many more positions in my collection which I use to test chess programs. But the way chess programs are developing, there is nothing much left in correspondence chess anymore at least at the top. It is dead with 99% draws.