Enhanced Chess

For a long time, Chess was considered the “ultimate test of cerebral fitness,” as the 1980’s song “One Night in Bangkok” proclaimed. Its popularity probably peaked in 1972, during the Fischer-Spassky world championship match which seemed to capture the interest of the entire world. Lately, however, its popularity has seriously declined, because of a number of factors:

1. Humans are no longer the best Chess players in the world. It’s somewhat depressing to study and play a game when you know that a program you can buy for $50 can out-play you, or any other human.

2. Chess opening theory has become extraordinarily detailed, and onerous to study. Very often, grandmaster games begin with 20 or more known theoretical moves. Moreover, nowadays most new theoretical novelties are suggested and analyzed by computer programs used by top players before their games. If a player does not keep up with opening theory, he will be at a disadvantage, but many people find the prospect of studying openings in this way increasingly dehumanizing.

3. Most games between top-flight Chess grandmasters end in a draw, simply because Chess is almost certainly a draw with best play. To take one example, Kramnik won his match against Kasparov in 2000 with 2 wins, no losses, and 14 draws.

I want to propose here a Chess variant, that I developed with my son Adam, that addresses all these problems. We call this variant “Enhanced Chess.” In Enhanced Chess, there are no draws, there is no opening theory (or endgame theory) to study, there is no advantage for the first player, and humans should out-class computers for a long time to come. The game is extremely skill-intensive; the more imaginative and “cerebrally fit” competitor will definitely win the game.

The game combines a couple existing Chess variants. The first of these variants is “Bughouse Chess.” Ordinary Bughouse Chess is played by two teams of two players each. On each team, one player plays White and the other plays Black. When you capture one of your opponent’s pieces, you hand it to your partner, who can drop it on his own board rather than making an ordinary move. Pawns may not be dropped on the first or eighth rank, but otherwise there are no restrictions on drops; you can even checkmate with a dropped piece. When a pawn reaches the eighth rank, it is promoted normally, but if the promoted piece is captured, it turns into a pawn again. The game ends when one of the kings is checkmated.

In ordinary Bughouse you may communicate freely with your partner. It is normally played with fast time controls (e.g. 5 minutes for each player for the game) as a very amusing social game with lots of trash-talk. Here’s a video of a typical Bughouse game, where the only unusual thing is that one of the players (at the back left) is Levon Aronian, one of the strongest Chess grandmasters in the world, and a participant in the 2007 World Chess Championship:

In Enhanced Chess on the other hand, a single player will control both the boards on his side, so that the game is again a two-player game. Moreover, the moves are made in a strict turn-based order. First Player 1 makes a move with White, then Player 2 makes a Black move followed by a White move, and then the players continue to alternate, always making a Black move followed by a White move. Notice that because Player 1 only makes one move on his first turn, and then Player 2 makes two moves, there is no obvious advantage to going first or second.

Time controls can be as slow as you like; Enhanced Chess is meant as a serious game.

Turn-based Bughouse is a good game, but if it became standard, it would quickly suffer from the problem of excessive opening theory. But that problem can be remedied by using the idea, proposed and popularized by Bobby Fischer for ordinary Chess, of randomizing the initial position. In Enhanced Chess, the initial random position must be identical on the two boards, as in the following diagram of an initial position:

enhancedchess001.jpg

Chess with randomized initial positions (sometimes called “Fischer Random Chess” or “Chess960” because of the 960 legal starting positions) is already a somewhat popular variant. Levon Aronian is the current “world champion” of this variant, having beaten Vishy Anand (the current world chess champion in ordinary Chess) in a short match in 2007.

One difficult point in Chess960 is how to handle castling. In Enhanced Chess, we propose to handle this problem by simply forbidding castling. Castling is usually of dubious utility in any Bughouse Chess variant in any case, and there’s no reason to carry all the complicated rules baggage. There would also be more than 960 legal starting positions, because while Chess960 tries to ensure that the kings start between the rooks, in Enhanced Chess that is not necessary. (We still required that the two bishops on each side start on opposite color squares).

It would in fact probably be sufficient if at the beginning of the game, the players alternated on deciding the positions of one piece at a time. For example, my first move would decide that the Kings go on b1 and b8. You then decide that one of the Knights should be placed on f1 and f8, and so on (one needs to forbid placements that place the bishops on the same color square or leave no option but for that to happen in the future). This version would remove any element of luck from the game, and I doubt that studying openings would pay with even this amount of “pre-processing.”

Making a pair of moves which repeats a position (on the combined boards) is forbidden in Enhanced Chess, and stalemate is a loss. These rules remove any theoretical possibility of a draw.

It is interesting to speculate on what the “correct” result would be in Enhanced Chess if God played against God. There is no obvious end to the game until positions start repeating. It’s clearly just idle speculation, but how many moves do you think would need to occur in the perfect game tree before one side won?

In practice, I don’t think that Enhanced Chess games would typically last any longer than ordinary Chess games, although a big difference is that players would definitely need to rely much more on intuition and much less on knowledge and calculation. One other big difference, which would definitely be a drawback for some players, is that reduced-material end-games would cease to exist. Enhanced Chess would be much like Shogi (Japanese Chess) on steroids, but maintaining all the familiar pieces (and removing the advantage of moving first).

Finally, I should address why I think computer programs will play Enhanced Chess poorly, at least in comparison to ordinary Chess. The main point is that programs basically use a search algorithm, that would be hobbled by the much larger number of possible moves in each position (remember that for each turn you get one move on each board, and all the drops mean many more possibilities as well, so for each turn there might be 10,000 possibilities for the computer to consider, instead of 30 or so). Evaluating a position will also much more difficult for a computer in Enhanced Chess compared to ordinary Chess, because material and mobility are less important. I would anticipate based on these points that Enhanced Chess programs would perform significantly worse than Shogi (Japanese chess) programs (where there are drops, but only one move per turn), and even in Shogi, computer programs still fall somewhat short of the top level.

In the 15th century, a series of major rules modifications were accepted into Chess (the modern Queen, Bishop, and Pawn moves, Castling, and en passant) that succeeded in reviving what had become a rather dull game. It seems to me that something similar will need to occur again before Chess can reverse its decline and regain its status as the King of Games.

 

Tags: , , , , , , , ,

4 Responses to “Enhanced Chess”

  1. Anonymous Says:

    This variant seems like fun – I’ve played Bughouse (or “team”) chess and it’s definitely lively. But if the goal is to produce something which resists computerization, why not just take an existing, already-popular game with a game-tree complexity a couple of hundred orders of magnitude greater than chess – such as Go? I don’t mean to raise the “whose game is better” argument – but game complexity means that Go should resist computerization (at least by brute-force methods) for many years to come.

  2. Jonathan Yedidia Says:

    Go is a great game (I’ve posted about some starting points for beginners already), but you almost answered your own question. For many people, myself included, Chess is more fun then Go, which is an important quality for a game. Enhanced Chess is even more fun. Fun is obviously a matter of taste, but even in Japan and China, more people play Shogi and Chinese Chess than Go. I respect Go a great deal, but I enjoy Chess a lot more.

  3. drj11 Says:

    Well, it’s fun to invent new games based on equipment that you happen to have lying around for one thing.

  4. David MacKay Says:

    Hey Jonathan,
    (1) I’d love to hear what you think of Ultima – another fun game that can be played with chess hardware.
    There are various rule-sets on the internet, but I learnt from an intensely logical mathematician in Cambridge so obviously the rules I learned
    http://www.inference.phy.cam.ac.uk/mackay/ultima/ultima.html
    are the best. 🙂
    Perhaps you could play it with your family for a few days and let us know?

    (2)
    If you ever need a bughouse player, Sanjoy Mahajan (at MIT) is keen on it.

Leave a comment