Ginger GM Logo

Computers vs Experts
GM Danny Gormally

30 January 2019

How have computers influenced analysis? Over the last twenty, maybe even thirty years, there is little question that the single biggest influence on chess has been computers, in particular powerful chess engines. This has completely transformed the chess scene, so that post-mortems are almost a thing of the past, or seem irrelevant. Why bother to analyse with your equally weak flesh and blood opponent, when the machine waiting in your hotel room can give you the answers to the game in seconds?

In truth, there have been a number of ways that computers have transformed the game of chess, not least in opening preparation. But for this article I thought it would be an interesting experiment to compare some of the analysis from the past, with the computer analysis that we have access to now. Part of the inspiration for this article was scanning through an old book of mine, Jan Timman’s “The art of chess analysis.” It seemed to me that even without the computer, it was possible to identify some mistakes. Nevertheless, there is also some excellent analysis in there as well, that the computer often agrees with.

Let’s look at one game from that book. This was an important game as it was part of the 1974 match between Karpov and Korchnoi, and as Timman suggests in his analysis, there were plenty of other experts chipping in with their view on this game as well.

Could not fetch PGN data. Sorry.

The very next game in the book is Timman’s own game with Boris Gulko. The position soon becomes very complex as Gulko feels obliged to sacrifice a piece. Unsuprisingly, the engine thrives in such positions, which rely heavily on calculation.

Could not fetch PGN data. Sorry.

The dangers of over-depending on computers

While it seem that computers have no limitations in analysis, there is a danger in using them too much. Eventually, you can neglect your own brain. I went through a period some years ago when I got very lazy. Before every game in a tournament I would simply turn on the computer and think “what does the engine think here?” and didn’t do any analysis myself. Perhaps not surprisingly, my results weren’t great.

I think there needs to be a combination. It’s fine to check stuff with an engine, but you need to do your own analysis as well. One of the ways you can do your own analysis is to use training positions. This can be a useful tool, to sharpen your calculation. If you check out the books of Mark Dvoretsky, he has many training positions that he has tested on some of the strongest players in the world, who went through his training school. it was interesting to run these games through an engine as well, to check any holes in analysis, especially bearing in mind that these books were often written without any reference to a computer program.

Could not fetch PGN data. Sorry.

Computers can also show us the key to a position

Let’s look at the following game that I played against Stockfish online. I’d like to say it was a “training game”, but in reality it was more like a “game I played because I was bored and had nothing else to do”. anyway, we reached the following position:

Could not fetch PGN data. Sorry.

Why computer vs computer matches will never be popular.

I made the point on social media recently that the possibility exists that the top 10,000 chess games ever played, in terms of pure quality of chess ever, have been computer vs computer games. In fact you might even enlarge that, to the top million or above, so strong are engines now. I also made the point that in some sense it’s a little bit strange that such games don’t recieve greater attention from the wider chess public. Why are we bothering to look at games between 2700s, when there are games between 3500 players available?

And when you see some of the recent games of Alpha Zero, with some of the astonishing sacrificial play in it’s games against Stockfish, then you can’t even argue that this lack of attention is due to aesthetic reasons. Just as it’s inevitable that machines will eventually produce masterpieces of literature and art, the chess machines of today are already producing games that can be counted as great creative achievements.

Therefore the single reason why machine vs machine games aren’t as popular as human vs human ones is obvious. Machines don’t have a personality that we can identify with. They don’t make mistakes, they don’t get tired, they aren’t human like we are. They don’t grimace at the board like Nakamura, or tell us how they were winning at the press conference like Kramnik. There’s an emotion and feeling to chess matches that machines may never experience.

I was reminded of this a few days ago when I took part in the Harrogate congress. In the last round I played Alan Merry, and we were tied for the lead with 3.5/4. Although now I’ve slagged off computers, I’ll also put the other side of the argument. It was fascinating to go over the game afterwards with an engine, not just to understand what I missed, but to bury some misconceptions about the evaluation of the position, that I experienced during the game. For example I believed towards the end of the game that my opponent had a multitude of wins, but as we’ll see, although he missed a clear win at one point, it wasn’t quite as straightforward as I thought…

Could not fetch PGN data. Sorry.

About the Author
GM Danny Gormally

Danny Gormally is a talented English Grandmaster. He lives in the bustling market town of Alnwick, somewhere near Scotland.

Danny’s first DVD for Ginger GM, Improve Your Practical Play was released in September 2013.