Extended chess (XChess)

[JuLieN]

I just came yesterday with an idea I'll have no time to test before months. But maybe someone will have time to do it? It's an extended version of chess, hence called XChess.

The idea is VERY simple: playing normal chess on a 14x14 board (a-n columns x 1-14 ranks), with doubling the rooks, knights and bishops. For instance, the board would look like this in the beginning :



For their first move, the white pawns could move up to the 7th rank, and the black pawn up to the 8th rank, although they could chose any square between this one and their starting point. Each pawn moving that way (that is, more than one square), could be captured en passant.

For castling, white could chose :
- kingside: the king could chose any square between k-m, with the rook sitting just after it (for instance: king in L1, and rook in K1)
- queenside: the king could chose any square between c-e, with the rook jumping right after it, likewise.

(Equivalent moves for the black king).

Such a chess game would have a much higher branch factor and the game would take longer to finish, being, I suspect, a good compromise between the very tactical modern chess and the much too slow go-moku. I think that would be a come back to more strategy and less tactics.

What do you guys think?

[JuLieN]

EDIT: during the castling, the TWO rooks jump over the king

[Matthias Gemuh]

Your XChess has pathetically weak knights and pawns.

[JuLieN]

Yes, the Knights will be much slower. For the pawns, I'm not sure: their situation won't change much compared with the classic game. Although maybe an idea would be to give each side a double rank of pawns?

Still I'm curious about how playable and interesting such a chess variation would be. If nobody does it before me, I might port my engine to play it, in a few months...

[hgm]

The good news is that the latest WinBoard-Alien binary supports boards upto 20x19. The latest normal WinBoard development version would also support double-digit ranks, but the internal boards are only declared as 14x10 in the version I compile for the binary distribution.

Beware, though, that the convention for boards with 10 or more ranks is that rank counting starts from zero.

Castlings with the corner Rook to any square would be supported when legality testing is off. Double-Rook castling is an interesting novelty. It is currently only supported by playing this as 'variant multi', so that you can make use of the general multi-move capability of WinBoard-Alien. (Which seems like over-kill for just a funny castling move.) Or you could have the engine 'coach' the GUI a little, by letting it move the second Rook after the automatic castling with the corner Rook, using the newly introduced click command.

Are you aware of the existence of Big Chess?



This game is actually being played on-line. So if you make an engine for big variants, you might as well include this one. It would hardly be any extra work. It would also be very worthwile to not be shy with the number of piece types you support. (E.g. upto 8 in stead of the usual 6, so you could include the B+N and R+N compound, and have the engine also play Grand Chess (10x10) and Capablanca Chess (10x8) with virtually no effort.)

[JuLieN]

Thx HG No, I wasn't aware of Big Chess As far as I can see it's the same idea, but with an additional pair of knights, and the R/N/B are interleaved instead of being aligned. it might be a bit too big, especially when your pieces have to cross this huge board.

Is the game interesting?

[hgm]

I never played it, and upto recently, WinBoard could not support it because the need for double-digit ranks. So there also are no engines.

My guess is it will be a slow game....

[Laskos]

I don't think XChess would be particularly hard for computers. EBF would be ~4 instead of ~2, therefore ~8-10 plies deep even today's little modified engines would reach easily, making already hard to beat bots for humans, more so because no human will be the level of IM in the near future. On the other hand, the opening theory would be a completely different animal, but generally, the non-locality of the game for B, R, Q makes it again more of a tactical venture. Arimaa's relative locality, besides the way of making 4 consecutive moves, makes Arimaa probably more of a strategical game.

Kai

[Sven]

I never played it, and upto recently, WinBoard could not support it because the need for double-digit ranks.

Why no hex digits 0..f for boards up to 16x16? Move strings could look like "fcde" but why not?

Sven

[lkaufman]

I think that your idea is overkill, the board is far larger than is reasonable. There is a good case for a larger board and more pieces than in normal chess, but 10x8 or 10x10 boards with bishop+knight and rook+knight extra pieces add quite enough complexity to the game. Last year Don Dailey and I made a Komodo version that plays S-chess (aka Seirawan chess) which has the extra pieces but still uses the 8x8 board. Our program was strong enough to win about ten informal games in a row (after a draw uncovered a huge bug) from Seirawan himself. Probably we could adapt it to the 8x10 or 10x10 board if we were hired to do so. Anyway I think there is no reason to consider boards of more than 100 squares, it just makes for unnecessarily long games.