Chess324

[lkaufman]

There has been some discussion about how to improve chess960 (Fischerandom Chess) to address the fact that when top engines play against each other on good hardware at Rapid or slower time controls almost all the games end in draws, just as in normal chess (without forced unbalanced openings). Scrapping the symmetry requirement leads to some positions where one side is quite clearly winning.
I believe I have found a solution that is aesthetically pleasing, doesn't require special castling rules, and will dramatically lower draw percentages without any clearly won positions. I call it "Chess324". All rules are the same as in normal chess, including castling, only the start position is modified. The kings and rooks are placed on their normal positions. All the other pieces for White and Black are placed randomly, with no symmetry requirement, with the only restriction being that for each side the bishops must be on opposite colored squares. Unless I have miscalculated, there are 18 permutations for each side, making 324 total possible positions (including 18 symmetrical ones that are legal in chess960 of which 1 is the normal start position of chess).
In order to determine whether these positions are playable, I checked out the most promising-looking ones for White by checking whether White's advantage ever exceeds Black's advantage in normal chess after the Grob (1g4?) is played. There has been much discussion in the past over whether the Grob is losing or not, and I doubt that anyone really knows the answer; the Hiarcs database has Black winning 49% of the games, Lc0 gives Black 54% winning chance, and Stockfish and Dragon give evals suggesting that it is more likely to be a win than a draw but is very near the line. I checked all the promising positions I could think of with recent versions of Stockfish, Dragon, and Lc0, and in no case did I find one that produced an advantage larger than Black gets with the Grob (one position was tied per Lc0 but less per SF and Dragon). Of course the evals are all over the place, sometimes even Black is better, sometimes it's about even but not "balanced", sometimes one side is much better, but never clearly winning (at least not as clearly winning as the Grob as far as I was able to tell). Since many evals clearly favor one side, chess324 should be played in pairs of games, each side having White from the same position once. With humans, that's not essential, just recommended; with engines it would be necessary.
This version has huge advantages over chess960. First, no special castling rules, any engine or GUI or human can play with no instruction after seeing the initial position. Second, since all but 18 of the 324 positions are asymmetrical, opening play should be much more interesting and complex. Third, the normal positioning of the rooks and kings and normal castling makes the game feel closer to normal chess. Fourth, matches of up to 648 games can be played with no repeat positions, generally enough for most purposes. Most important, no matter how many cores or how much time the engines get, there should be plenty of decisive games for the foreseeable future since many positions are at least not too far from the win/draw line. The stronger engine will score 1.5 out of 2 in many of these positions for many years to come, unless chess is truly solved some day.
It is quite possible that a few of the initial positions may ultimately be judged to be won for White, but I am confident that even if they are "won", they will be near enough to the draw line to be playable with any current hardware or engines.

[Chessqueen]

There has been some discussion about how to improve chess960 (Fischerandom Chess) to address the fact that when top engines play against each other on good hardware at Rapid or slower time controls almost all the games end in draws, just as in normal chess (without forced unbalanced openings). Scrapping the symmetry requirement leads to some positions where one side is quite clearly winning.
I believe I have found a solution that is aesthetically pleasing, doesn't require special castling rules, and will dramatically lower draw percentages without any clearly won positions. I call it "Chess324". All rules are the same as in normal chess, including castling, only the start position is modified. The kings and rooks are placed on their normal positions. All the other pieces for White and Black are placed randomly, with no symmetry requirement, with the only restriction being that for each side the bishops must be on opposite colored squares. Unless I have miscalculated, there are 18 permutations for each side, making 324 total possible positions (including 18 symmetrical ones that are legal in chess960 of which 1 is the normal start position of chess).
In order to determine whether these positions are playable, I checked out the most promising-looking ones for White by checking whether White's advantage ever exceeds Black's advantage in normal chess after the Grob (1g4?) is played. There has been much discussion in the past over whether the Grob is losing or not, and I doubt that anyone really knows the answer; the Hiarcs database has Black winning 49% of the games, Lc0 gives Black 54% winning chance, and Stockfish and Dragon give evals suggesting that it is more likely to be a win than a draw but is very near the line. I checked all the promising positions I could think of with recent versions of Stockfish, Dragon, and Lc0, and in no case did I find one that produced an advantage larger than Black gets with the Grob (one position was tied per Lc0 but less per SF and Dragon). Of course the evals are all over the place, sometimes even Black is better, sometimes it's about even but not "balanced", sometimes one side is much better, but never clearly winning (at least not as clearly winning as the Grob as far as I was able to tell). Since many evals clearly favor one side, chess324 should be played in pairs of games, each side having White from the same position once. With humans, that's not essential, just recommended; with engines it would be necessary.
This version has huge advantages over chess960. First, no special castling rules, any engine or GUI or human can play with no instruction after seeing the initial position. Second, since all but 18 of the 324 positions are asymmetrical, opening play should be much more interesting and complex. Third, the normal positioning of the rooks and kings and normal castling makes the game feel closer to normal chess. Fourth, matches of up to 648 games can be played with no repeat positions, generally enough for most purposes. Most important, no matter how many cores or how much time the engines get, there should be plenty of decisive games for the foreseeable future since many positions are at least not too far from the win/draw line. The stronger engine will score 1.5 out of 2 in many of these positions for many years to come, unless chess is truly solved some day.
It is quite possible that a few of the initial positions may ultimately be judged to be won for White, but I am confident that even if they are "won", they will be near enough to the draw line to be playable with any current hardware or engines.

Instead of Chess960 (Fischer-Random Chess it should be called (Kaufman-Random Chess) Chess324 your name will be remembered forever. I already have collected the 324 positions and replaced them with the 960 position and will be the first to start a Kaufman Random chess match between Stockfish and Komodo Dragon once I get your Dragon 3.1 which I will purchase at the end of this weekend

[Frank Quisinsky]

Hi Larry,

we had the discussion with Reinhold Scharnagl in Arena Support Forum for many years (around the year 2004 / 2005). I remember that Chess960 / FRC should be not changed with building a new chess variant. Furthermore, if we give Chess960 / FRC again more attention is this for the long-term of classical chess in my opinion not the best way.

I haven't the opinion that more as 75% are draws if strongest chess programs are playing a match, if the style and the techniques are different and most complicated openings are inside a test-set. Most of the complicated opening-lines comes from King-Indian and Dutch systems. It make much more sense to search in most complicated opening systems for new lines / ideas with strongest chess programs. Here I can see a better order for chess engines in the next years.

It make no different to use classical chess or different chess960 / FRC positions for reduced the draws.

I think we are in the near of 90% draws if Stockfish or Komodo are 100-150, maybe 200 Elo stronger.

Stockfish and Komodo are to equal in strengths and this is the main reason for the higher draw quote we can see today. Or better, the differents between Stockfish and Dragon by Komodo are all the years in reality very small.

Furthermore, the draw quote is not the main problem !!!

The main problem is the move-average with nonsensical contempt programmings.
Fore people like to looking in the games is that more and more a much bigger problem.
Differents are speaking from "To fall asleep boring engines".

More human-like is to give a game faster draw if the game is draw !!
There is no reason to wait hundreds of moves for an opponent to make a mistake.

People lost interest with such a playing style.
Unfortunately, different programmers going the wrong way!

Much more interesting is to give users during the games the following information:
Probability of a draw = x% after each move during a game is still running. And if engines find out that a game is to 95% draw game should be ended with draw (maybe UCI protocol can changed here with such new ideas).

Means, the higher the Elo for TOP-Engines the more programmers should be self-confident in questions of daring forecasts. That will give the others ... the normal chess players ... the information ... oh, chess programming ideas goes in a new generation.

I think with more and more Elo, programmers fall into a deep sleep and are not able to develops new interesting things. So different like to work on bad contempt settings for 3 Elo more. That is not the right way !!!

Best
Frank

[Lazy_Frank]

Sounds very promising.
Thanks, Larry, definitive this chess variant need to be given a chance.

Speaking of double Fisher random chess (DFRC).

It is very rich in openings count, its also do not request book to differentiate engines.
Main problem is many DFRC openings exceeds the draw range.
I am currently try to clear DFRC from "busted" openings.

[Frank Quisinsky]

Forgotten ...

Contempt is interesting only for avoid draw in the early mid-games or better with x-pieces on board.
The longer the games, the fewer the pieces on board contempt ideas make no longer sense in times of neural network.

The fewer the pieces on board, games should be faster ended with draw if the game is draw.
I think in most of programms made the quanity of pieces for contempt ideas no different.

Can not otherwise explain the high move average for different engines.
Again, this is the much bigger problem ... not the draw-quote.

The draw-quote never changed with more as 2% (for ~ +100 Elo) if two diffferent engines in style playing a match and Elo is equal.
Engines, different in playing style, like to have 3000, 3100, 3200 or 3300 Elo produced ~68%, ~70%, ~72% and ~74% draws.

Why it should be a big different to engines like to have 3400 Elo or 3500 Elo?
Maybe 2-3% more draws, not more!

We can't see that if the engines we have with more as 3400 Elo are equal in his playing-style, or better ... have the same strengths.

Sorry, can't build an other opinion after so many stats I made or engines-styles I saw.

[jkominek]

I find this a very attractive proposal. I like it a lot.

If we are on the topic of chess variants that are only a small perturbation of standard chess, I'll mention four variations of the variant that come to mind. It is based on the observation that, from a certain viewpoint, chess324 does in fact have a single starting position: it is with the queens, bishops, and knight standing off-board waiting to be placed on-board. Or, if you prefer, during "move 0" the empty squares of the back ranks transform into pieces according to a special start-of-play rule.

Variant A. During move 0 the blank squares are randomly assigned. This is your proposal.
Variant B. During move zero the White player then Black player choose the arrangement on-board of their off-board pieced. The Black player moves in response to White's opening arrangement.
Variant C. The White and Black players alternate placement of pieces until all ten are on-board.
Variant D. (Somewhat like Bughouse.) The player to move has the option of either placing a piece on-board on first rank, respecting the bishop color constraint. or moving a piece already on board.

The variants with a gradual deployment phase result in a more complex, intricate game, and that probably works against chances at adoption. Human players will want to develop pre-opening theory before feeling comfortable. Engine programmers would have to write code to add the necessary support, and it might not be elegant code. There is no shortage of chess variants (see for example the reference encyclopedia by Pritchard), but not many ever "make it."

Besides the merits of chess324 that you outlined, familiarity is I think part of the appeal. It is easy to wrap your head around the idea without feeling far away from home.

[bnst]

Why reduce the number of possibilities making it easier to prepare openings.
I would say that the old proposition of Chess2880 where there is no castling and the only rule for the pieces is that the bishops must be on different colours is much better.
More possibilities and randomness, no special rules and any engine that can take a FEN input can play it.

I have for many years wondered why people are so hung up on castling that they even had to implement it in Chess960.

Best regards
anst

[Chessqueen]

The standard position of Chess has many many openings which are hard for human unless they are top GMs to remember them all. What make you think that we can remember 100th of thousands of openings from 324 positions if we are currently having a hard time remembering all the openings and variations from one single position, which is the standard Chess position.?

Note: If Fischer had created the 324 GREAT positions, nobody would had questioned him, since he became such a celebrity.

[Modern Times]

Larry, do you have a pgn of the 324 positions ?

[Chessqueen]

Larry, do you have a pgn of the 324 positions ?

I already have them, it took me over 3 hours to select them from the already Fischer Random Chess 960. I will purchase Dragon 3.1 tomorrow to test it against Stockfish 15. I do NOT want to send them to anybody because I want to be the first to test them with the best two engines. It is very simple, simply look at every 960 chess position and select those that meet the criteria.