Improving Chess

I have recently been thinking about how to improve the game of chess. There are two main problems that I have identified.

Problem 1: Role of Memorization

Problem: Memorization gives a player huge advantages. For me, the ability to strategize well in novel situations is much more valuable than memorizing the best moves in many scenarios. This is why I view it as a problem that memorization gives a player huge advantages in chess. One can easily beat a sizeable percentage of chess players just by reading a book on chess openings. This is a natural implication of the fact that early mistakes are hard to recover from.

Solution suggestion(s):

1. Chess960: Because there are 960 initial piece configurations (as opposed to 1 in standard chess), it is a much more difficult task for a human to memorize opening lines. Double Fischer Random Chess (like Chess960, but with different positions for White and Black) would also be interesting.

2. Allowing players to choose where to start their pieces. In this scenario, it would be a good idea to prevent each player from seeing their opponent’s starting configuration before they commit to their own starting configuration, as that might give a player an advantage that I see as unnecessary and meaningless. Some ideas for rules on where the pieces can be placed: 1) pieces can only be on the first two rows (as in standard chess), 2) the second row must consist entirely of pawns, 3) or just require each configuration to be one of the 960 combinations allowed in Chess960.

3. Dark chess: This is the most interesting one among the three suggestions in this section. Because there are squares on the board that you can’t see, a book on opening theory cannot help you much here. There is just not much to memorize. Notice how closer Dark Chess is to a real-life battle than is standard chess: the fog of war is a central theme in virtually all real-life warfare, so Dark chess is a much better approximation of real life warfare than standard chess. I like saying that standard chess is a match between two machines, and Dark chess is a match between two humans.

Problem 2: Aesthetics

Problem: Castling, the ability to move a pawn for two squares, and en passant do not look aesthetic and/or natural. It feels more natural for a piece to have only one way to move. According to Wikipedia, the two-square advance was introduced “between the 13th and 16th centuries, to speed up games.” It looks like I prefer aesthetics over shorter game durations.

Solution suggestion(s):

1. Let’s try playing chess without castling or two-square advances for pawns (so, also no en passant). In a variant like Dark chess, the two-square advance for pawns might not even make the game faster, so there might be even fewer reasons to keep this rule in some variants.

Conclusion

Making Dark chess and/or Chess960 as popular as standard chess would significantly increase the number of people who enjoy playing chess.

On a separate note, DeepMind has done some interesting research on alternative rule sets in chess. A highlight of the paper is Table 6 on page 16, which lists the “estimated piece values from AlphaZero self-play games for each variant.”