Puzzle Trouble

The most popular form of chess training has a huge problem

If there’s one thing every chess player does, apart from playing, it’s puzzles. The idea of a chess puzzle is so commonplace that it’s easy to think of them as somehow preordained, but in fact they have specific and somewhat arbitrary requirements, especially that one and only one move leads to a big advantage.

Most chess positions don’t fit this description. There are about 40 moves in an average chess game. Working by hand, I’ve found I average about one “good puzzle” per game. Computer algorithms can take ten games or more to find a single good puzzle, depending on how strict their filtering is. When it comes to lichess generating puzzles this is no problem - they host around 100 million games per month, so there’s no shortage of candidates.

But it does present an enormous problem for puzzles as a means of improvement: Can exclusively drilling one specific, unusual type of position prepare you to handle every position?

This problem doesn’t get much attention in discussions on chess improvement. It’s often taken as a given that drilling puzzles will lead to overall improvement. I don’t think I’ve ever heard an explicit justification for why puzzles are supposed to be the best way to improve your chess, but let’s try to supply one here.

  • In a position with exactly one winning move, it’s extremely important that you find that move. In this sense puzzles can be justified as training for high leverage opportunities.

  • In positions with many equivalently good moves, it doesn’t really matter which you play, so long as you play one of them. In other words, not only are puzzle-type positions especially important, the reverse is true as well, non-puzzle positions are less important.

  • Puzzles allow you to work on chess without the stress of playing a full game. Your ego is not on the line as much as when you’re competing against another person.

  • Puzzles are very time-flexible, you can do them for as long or as short as you want.

  • Puzzles have a clear answer, which means you can get immediate and unambiguous feedback on your performance. This is often understood to be a key element of deliberate practice.

I believe these justifications for puzzles make sense, as far as they go, but they don’t rule out the central problem: a method where you only train on a small, artificially selected subset of positions cannot prepare you to handle every position you’ll encounter in a game.

For these reasons puzzles deserve a place as part of your training program, but they shouldn’t be the whole program. You need to learn how to handle non-puzzle positions, which, after all, are most positions.

A few caveats. First, when I talk about puzzles, I’m thinking about the sort of relatively quick, straightforward puzzles you’ll encounter in Puzzle Rush or lichess puzzles. If we expand the definition of puzzles to include more challenging positions, positions without a win, or positions with multiple solutions, at some point we’re just talking about analyzing, which is a much bigger part of chess.

Second, the advice to deemphasize puzzles applies most to intermediate players. For beginning players, doing a lot of puzzles early on can be an effective way to get up to speed. In contrast, advanced players usually understand that puzzles alone are unlikely to propel them to further improvement. It’s at the intermediate level where you’re most likely to find players who are much better at puzzles than they are at chess.

The downside of spending too much time on puzzles is you end up trying to apply puzzle-solving techniques to every position. When all you’ve got is a hammer everything looks like a nail. Many players suffer from puzzlitis, which presents with

  • Looking for forcing moves in every position.

  • Becoming confused or discouraged when you can’t find a clear win.

  • Going for aggressive moves even when your analysis indicates they won’t work, because you don’t know what else to do.

The cure for puzzlitis is to play the whole game. As a road to chess improvement, playing whole games of chess is regarded with suspicion by adults, possibly because it’s too fun and possibly because it’s too obvious. But if you want to improve at chess, you need to play a lot of chess. Another good route is to review whole master games. As with playing, you’ll encounter all the types of positions that occur in a real game.

Puzzles can work as part of a balanced training program, but their position as the default way to work on chess for large numbers of players has more to do with convenience than effectiveness. If you want to beat tough opponents, you need to be ready to handle all the positions you’ll encounter in a game, not just the kind of positions that go in puzzles.