A Top CC Player’s Startling Move

     While looking over the games of some of the top players on Lechenicher SchachServer I came across a position from a game by one of the top players, Djordje Petrovic, that startled me.
     This Djordje Petrovic is a 48 year old programmer from Belgrade, Serbia and has been playing CC for 16 years. He is an ICCM with the ICCF and is rated 2533 on LSS which places him 3rd on the LSS rating list although he has not been active for over a year.



Position after 20…Ng6. 

     After first running an analysis at 10 seconds per move using Komodo 8 I noticed White’s next move, 21.Be4, was given a “??” which indicates a serious error. How could this be from a top CC player?
     I decided to let Houdini 2 and Komodo 8 examine the position for about 10 minutes and both engines suggested white should have played 21.Bh5 which resulted in a position that was evaluated at move 36 by H2 as being dead equal. Komodo’s analysis reached move 29 with the same conclusion.
     I actually made the move 21.Be4 on the board to see if the evaluation changed. After 15 minutes H2 thought black should capture the B and gave an evaluation of about a half P in black’s favor. K8 thought black was doing even better and evaluated the position at a little over one P in black’s favor.
     Black actually played 21…d4 which in the 10 seconds per move analysis by K8 also got two question marks. Here things got interesting because after 10 minutes both engines thought black was better with the following recommendations:

Houdini 2: 22.f5 (0.50 Pawn) and 22.Qh5 (one Pawn)
Komodo 8: 22.Qh5 (1.10 Pawns) and 22.f5 (one and 1/3 Pawns)

So while both engines thought 22.Qh5 left black a Pawn better, they disagreed over the merits of 22.f5.

     The move Black actually played (21…d4) after 10 minutes resulted in the determination that White was better by 4.5 Pawns (H2) and 3.5 Pawns (K8).
     I ran a couple of Shootouts from the position after both 21.Be4 followed by 21…dxe4 and 21…d5 and White kept winning against both moves.
     After messing around some more I gave up because I was unable to determine the true merits of whether 21…d4 or 21…dxe4 was better or to determine where, exactly, the evaluation changed from Black having the better position to White winning after 21...dxe4. Thinking that because the game was played a few years ago maybe the engines they were using weren’t as good as they are today, I switched over to the weaker Fritz 12 engine, but there were no drastic changes.
     What I’d like to know is how the players conducted their analysis and reached their conclusions that White was winning after 21.Be4. Or was he? Was 21.Be4 a mistake? How did Black (rated over 2100) conclude that 21...d4 was the best move. Did Black err in not taking the B? After 21...d4 did he booger up the game at a later move?
     I don’t know and am not willing to let my engines run for a couple of days to see if there are resources hidden deep in the position or even if my quad-core laptop is capable of finding them.  What this proves is that even with engines, there is still a little room for human input in correspondence chess...if you have the equipment and software and are a good player with a lot of patience. Equipment, software, good player, patience. That's not me so why am I playing CC?  That's another question I can't answer.