The Limits of the Chess Engine for Improvement

Chess Engines: Tool or Trap?

Using chess engines for chess analysis and practice has become very commonplace. Whether you are on an online chess server, internet forum, or even at the tournament hall, discussion of which chess engine one uses or what the latest version of Stockfish is seems to pop up frequently.

However, I think we must be very careful in using our silicon friend too much lest our conversations start to sound like this:

Peter Patzer: Move 23 in the Najdorf Sicilian Sozin line you played is busted. Why do you play this trash?

Wilson Woodpusher: What are you talking about? Blockfish 44.18 says that it’s only +0.32.

Peter Patzer: Oh, my friend, don’t you know? Blockfish is terrible at assessing lines played by Nakamura. You need to use Slomodo 96.32a especially calibrated to analyzing such complex lines. It says it’s +1.2 if you let it run to at least 33 ply!

Wilson Woodpusher: Ah…you are so wise, my friend. What is your rating again?

Peter Patzer: 1500, but Slomodo 96.32a has a rating of 3700!

Wilson Woodpusher: Oh yeah? Let’s have our chess engines play each other.

Peter Patzer: Challenge accepted!

Am I saying that you should never use a chess engine to assist you with chess analysis or practice? No. They can be very useful to check for blunders as well as practicing positions – such as endgames – when you don’t have a partner. However, we should be aware of their limitations and what they can’t do to help you improve your chess.

Chess Engines Don’t Teach

When you look at chess engine analysis, you see a lot of move variations and evaluations of the resulting positions. However, for amateur players, much of this analysis is near useless unless the position involves a tactical blunder or forced win or material.

For example, the engine doesn’t tell you:

  • What pawn levers you are preparing.
  • What positional elements should be prioritized in the position.
  • Why you should trade a specific minor piece or keep another.
  • What the typical plans are in a particular pawn structure.

This knowledge is much more important for you to understand and learn than to know that 23.Bxf6 is +0.13 better than 23.Nxf6 at 33 ply.

Chess Engines Pick Moves THEY Like

When looking at the suggested moves from the output of a chess engine, we must remember a couple things.

The computer doesn’t take into account the sharpness of the position, only the resulting position at the end of its horizon. So although a top move for a computer may end up in an advantageous position with best play, the types of moves that are required to get there may be incredibly difficult for a player – particularly an amateur player – to handle.

As an example, let’s say there are two paths to an advantage:

  • One move leads to an intensely sharp position requiring 5-6 “best” moves that results in a +3.5 evaluation. This might involve
  • One move leads to a comfortable position with several variations that lead to similar positions resulting in a +2.0 evaluation.

Both moves should be sufficient for victory, but which one should you choose? If you’re a computer (or GM Hikaru Nakamura), you might opt for the first choice. If you’re not a computer , you might choose the second move.

Also remember that the computer engine won’t tell you which path is which!

The Engine Won’t Be At Your Side

Finally, if you are mainly an OTB or tournament player – as opposed to a correspondence player – the chess engine won’t be sitting next to you when you are playing. Using the engines too much in analysis will dull your own analytical muscles.

Your ability to calculate and evaluate positions is a perishable skill. If you don’t use it, you will lose it. This is a skill that must be continually developed and practiced – especially for amateurs. However, this is often hard work which is why clicking the chess engine button is so tempting. I urge you not to give in – at least too early.

As I’ve heard taught in many endeavors including music, martial arts, and academics – you will perform the way you practice. If you practice analyzing positions with high effort, you will be able to put out that effort in your games. If you constantly rely on the engines to analyze your positions, you won’t have the habits and technique when it comes time for your most challenging positions.

A Place for Engines

As I said earlier in this article, chess engines have their place. To finish up, where are a few ways I think chess engines can be very helpful.

  • They can be tireless defenders when practicing specific positions, such as technical endgames. Positions that require more practice than memorization, such as King, Bishop, and Knight vs. King (KBN vs. K) work especially well.
  • After you have analyzed a position or game on your own and feel like you’ve exhausted your own abilities, having the chess engine check for obvious errors can be very helpful. Of course, you need to understand why it is an error as well and how you can improve your analysis next time.
  • In discussing this topic with GM Nigel Davies, he noted that he and other strong players use engines for high level opening preparation. However, I would caution amateurs in doing too much of this work, as masters like Nigel have an incredible foundation of knowledge and skill to assist them in understanding and evaluating chess engine output.

The chess engines are here to stay for better or for worse. They can useful in your chess training if used properly, but when in doubt, try not using it and think on your own. Your chess skill will be the better for it.

Bryan Castro

This entry was posted in Articles, Bryan Castro on by .

About Bryan Castro

Bryan Castro is a businessman and writer from Buffalo, NY. When he's not spending time with his family or working, he can be found playing chess or practicing martial arts. He combines his interests of personal development and chess on his site Better Chess Training (