17 May 2020


One of the things I do routinely with Chess Informant is to race through batches of games in openings that I favor. Sometimes the end of the game--when a Grand Master resigned--offers an opportunity to practice my technique finishing a won position against the Silicon Beast (Stockfish 10 in this instance). Yesterday I failed to win from a position where I had a two pawn advantage.

In Krysa,L. -- Kobalia, M., Gibraltar 2020, Informant 143/122, Kobalia resigned after White's move 23. I took White against Stockfish.

Black to move
After 23.bxa4 1-0

Stripes,J -- Stockfish 10

23...Qxe2 24.a5 Qxa2 25.Bb6

These moves were simple enough, especially as they were provided in Branko Tadic's annotations to the game.


According to the engine, my advantage is 2.38 pawns. I should be able to finish this, but how?

White to move

First Attempt

26.a6 Rab8 27.Rb1

This was the only move that did not lose, later analysis revealed.

27...Bb2 28.Qxb8

I found this move after trying a couple of losing moves. Again, the only move that maintains equality.

28...Rxb8 29.Bd4 Rg8 30.Rxb2 Qa4 31.a7 Qxd4

White to move


Once again, White had a single non-losing move.

32...Qxb2 33.a8Q =

Second Attempt

26.Qc6 h6 27.Qc4 Qxc4 28.Rxc4

Black to move

It turns out that the bishops do not change much. This is still a rook ending with a single pawn advantage and that pawn on the flank. Perhaps computer vs. computer gives White winning chances, but all I can do is draw (except when I manage to lose).

28...Rfe8 29.Rfc1 Kh7 30.R1c2 Re1+ 31.Kg2 Ra6 32.Rc6 Ra1

White to move

33.Bc7 Ra8 34.Re6 Bd4 35.Re4

Not a particularly effective strategy, chasing the bishop, but I was feeling rather clueless, despite an apparent advantage.

35...Rd1 36.Bb6 Bxb6 37.axb5 Rb1 38.Rc6 Rb8 39.Ree6

White to move

I played another 30 moves against the beast because a rook ending against a computer is not terrible practice.

There may have been other efforts starting at move 26, but I never managed to find a way to victory.

Stockfish suggests 26.Rfd1 Rxa5

White to move

What would you play? There is a single winning move.

07 May 2020

Chess Informant 143

Chess Informant 143 arrived yesterday as book and CD. After a few more hours of post-dinner work, I installed the CD on my notebook and started working through the first article in the magazine. Ivan Sokolov presents some games and game fragments from the Corus Chess tournament that was held in January at Wijk aan Zee with a one-round visit to Eindhoven.
I was well-familiar with Eindhoven when I was reading about the 1975 Wijk aan Zee tournament in high school because my other passion at the time was the military history of World War II. I read Cornelius Ryan, A Bridge Too Far, watched the film based on the book, and even bought and played an Avalon Hill board game based on the battle.

I usually follow this tournament, although this year that amounted to know more than watching Danny King's PowerPlay videos on YouTube. In January 2000, I followed this tournament live on the Internet Chess Club for the first time. No other Grandmaster chess event consistently draws my attention year after year, but the past couple of years, it seems, I have not found the time to follow it closely.

Hence, Ivan Sokolov's highlights, like Danny King's, are most welcome. I recall watching Sokolov play in this event a few years ago. Sokolov presents more than a dozen games and several fragments. His commentary is instructive and entertaining. The second entry is "seemingly 'simple position'" (16) where Anish Giri blundered.

Black to move

When Stockfish thinks for less than half a minute, there is very little difference between the three moves that Sokolov analyzes here. But, then, the engine finds one move that is equal and a slight advantage for White in the other two. This slight advantage, it turns out, is a technical win that is just deep enough that the engine's evaluation is less than reliable. I played one of them against Stockfish last night. The computer diverged from the game's variation, but finding the winning line still was easy.

The candidate moves:

a) 39...Rd7 (played in the game)
b) 39...Kf5
c) 39...h5

The second choice leads down the same road as the game with a winning pawn ending for White. 39...h5 draws.

The game, Firouzja,A. -- Giri,A., continued 39...Re7 40.Kf3 Kf5 41.Rxh4 Kxe5 42.Re4+ Kf6 43.Rxe7 Kxe7

White to move

44.Kf4 Kf6 45.g4 a5 46.a4 h6 47.h4 Ke6 48.g5 hxg5

And this position caught my interest, thanks to Sokolov's commentary. In blitz, I would have made the wrong move without a thought, but there is an important principle here that should be understood. It involves counting.

White to move

How would you finish?