Computer Chess Corner, March 2020

Computer chess analysis using Stockfish or similar programs have had a profound impact on the game of chess. It is awe inspiring to think that we are thirty seconds away from downloading a piece of chess software that can beat the world champion. Professional computer software programs like Stockfish, Komodo, Houdini continue to grow in playing strength too. The latest Computer Chess Rating List (CCRL) puts the #1 rated Stockfish at an ELO of 3497! But playing against these machines is not why we download them. We use them as a personal coach. We want to know if the moves we played hold up to the scrutiny of a silicon super grandmaster.

A few months ago, a friend of mine who had been playing in a tournament told me that, “Komodo reviewed my game and gave me its stamp of approval.” He obviously felt good about that as well he should. Using chess software to help us improve our game is something we should all be doing but we should do so with caution. Sometimes a computer will show you the “best” move but it involves a specific set of move orders that only a computer can find let alone play.

Case and point. A few days ago, I played a series of blitz games with my friend Doron at a local mall. I was on the Black side of a Smith Morra Gambit. We have played this opening hundreds of times. Each of us try out new lines to confuse the other. In this game, Doron played a move that I had not seen before. He played 8. Bf4 (see diagram below). Typically I am used to him playing 8. Bg5 so this was a surprise.

Smith Morra Gambit with 8. Bf4!? and the resulting computer chess analysis using Stockfish was quite enlightening.
Smith Morra Gambit after 8. Bf4

Smith Morra Gambit computer chess analysis using Stockfish

I replied by playing 8…e5!? and he countered with 9. Ng5. I was shocked to see that both his bishop and knight attack f7 and that I would be losing either my queen or a rook. Clearly I wasn’t paying attention. I wound up losing the game mainly out of disgust with myself for allowing such an obvious move.

So, I did what so many of us do, I put the position on Stockfish 11, hoping to see some ideas I might have missed. I saw that the machine instantly liked 8… e6 which is a typical book move and one I had expected it to like. So I walked away for a few minutes to see if the engine would find anything else of interest. I came back 15 minutes later and saw this output on the chess analysis screen:

Fifteen minutes of Stockfish analysis showed that e5 was the best move after all.
Fifteen minutes of Stockfish computer chess analysis showed that e5 was the best move after all.

Not only did it reject the book line but it preferred my move! Normally this is something to get excited about but if you look at the line of play, it’s clear that this could only be played by a machine. In all the 15 million games I have in my chess database, there isn’t a single game where either player goes into this line.

The moral of the story – 8…e6 is the move I should have played. Yes, it’s good to know that e5 was an interesting alternative but it simply isn’t playable at the 1800 level. Computers are great at showing us the best sequence of moves but they can’t relay to us the technical complexity of a particular move order. Below is the computer’s chess analysis that follows after 8…e5.

So remember, a computer’s winning move has to be a playable move. Navigating the complexities of this kind of position is not something most humans would be comfortable with. In the future, I will play 8…e6, knowing that e5 is a reasonable but unrealistic alternative.