St. Louis chess960/FRC

[Modern Times]

[
"Comfortable" may not be the right word here. The players accept the rule, but clearly they sometimes forget that castling is still legal in some positions, and sometimes the game is essentially decided because of that. I haven't heard any of them calling for a change or for playing "shuffle chess" which is basically FRC without castling (and no requirement for king to be between the rooks). They just haven't played enough FRC to be truly "comfortable" with the castling rules. In one game White castled long resulting in his rook on d1 attacking the enemy queen on d8, which could never happen in normal chess.

Yes, better to say that they understand the rules for castling and are "across them". In the cases above, you might say that the player blundered or made a mistake, as sometimes happens in normal chess too. In normal chess it is not unusual for games to be lost because the player didn't see something or made a mistake.

[Time]

Chess960 is brilliant. When the players cannot rely on standard openings and their intensive study and preparation of them, it makes for very interesting and less drawsish games. They have to spend time thinking right from the first move. Only downside to me is the odd castling rules, but even those can make things interesting, Unlike standard chess, move 1 can be a castling move.

Of course the odd castling rules can be circumvented by "chess18" (which is just the 18 positions of 960 with the kings and rooks on their normal squares) or "chess324" (same as chess18 but no symmetry requirement). The problem with chess 18 is that 18 positions (including the standard one) is too small a number; top players could memorize significant theory on 18 positions. The problem with chess 324 is that some of the positions are too favorable, even winning, for one side, which is okay for engines if they play pairs of games, but humans won't like that. For human play, one could prune the 324 positions down to perhaps 200 or so, pruning the most lopsided ones. For engine vs human competition, I think the best solution is to stick with chess 18 with (white) knight odds, so 36 positions, or just 18 if we define some simple rule to choose the knight to remove. This way the humans won't be disadvantaged by unfamiliarity with the castling rules of 960. This is what we did vs. Alex Lenderman a couple years ago, but I selected the positions; for future matches they should be chosen at random from the list or even all 18 (or 17 without the normal one) could be used in a match with a short time limit. This would of course favor the humans somewhat as compared to what we did in the Lenderman match. Perhaps there are other good ideas for improving on the basic concept of chess960.

I actally think in opposite terms. The possibility of establishing a body of theory is a feauture as long as it is sufficiently rich in the number of relevant lines. Would you as a GM mind, say, playing a pair of games with the initial 324 position selected randomly from a list of positions where you and your opponent disagree about the objective outcome (as sketched in one of my posts in the Chess324 thread) or simply from a list of positions evaluated within something like [1.25, 1.85] by SF15 (somewhat higher for SF dev.) on a given high depth? In both cases, there may be a number of "critical" starting positions not that far from 18, and having different books "memorized" can be possible since theory will be more narrow due to white's huge (maybe winnining, maybe not) advantage.

The same idea holds for NBC until move n chess (it seems white is winning in standard NBC after 1. d4), with, say n = 16. (Or selected randomly from disagreement set as sketched in the Karmageddon thread.) Here, actual existing lines and lines closely resembling them will be very relevant, and the memorization environment will be much more familiar than in Chess324.

[lkaufman]

Chess960 is brilliant. When the players cannot rely on standard openings and their intensive study and preparation of them, it makes for very interesting and less drawsish games. They have to spend time thinking right from the first move. Only downside to me is the odd castling rules, but even those can make things interesting, Unlike standard chess, move 1 can be a castling move.

Of course the odd castling rules can be circumvented by "chess18" (which is just the 18 positions of 960 with the kings and rooks on their normal squares) or "chess324" (same as chess18 but no symmetry requirement). The problem with chess 18 is that 18 positions (including the standard one) is too small a number; top players could memorize significant theory on 18 positions. The problem with chess 324 is that some of the positions are too favorable, even winning, for one side, which is okay for engines if they play pairs of games, but humans won't like that. For human play, one could prune the 324 positions down to perhaps 200 or so, pruning the most lopsided ones. For engine vs human competition, I think the best solution is to stick with chess 18 with (white) knight odds, so 36 positions, or just 18 if we define some simple rule to choose the knight to remove. This way the humans won't be disadvantaged by unfamiliarity with the castling rules of 960. This is what we did vs. Alex Lenderman a couple years ago, but I selected the positions; for future matches they should be chosen at random from the list or even all 18 (or 17 without the normal one) could be used in a match with a short time limit. This would of course favor the humans somewhat as compared to what we did in the Lenderman match. Perhaps there are other good ideas for improving on the basic concept of chess960.

I actally think in opposite terms. The possibility of establishing a body of theory is a feauture as long as it is sufficiently rich in the number of relevant lines. Would you as a GM mind, say, playing a pair of games with the initial 324 position selected randomly from a list of positions where you and your opponent disagree about the objective outcome (as sketched in one of my posts in the Chess324 thread) or simply from a list of positions evaluated within something like [1.25, 1.85] by SF15 (somewhat higher for SF dev.) on a given high depth? In both cases, there may be a number of "critical" starting positions not that far from 18, and having different books "memorized" can be possible since theory will be more narrow due to white's huge (maybe winnining, maybe not) advantage.

The same idea holds for NBC until move n chess (it seems white is winning in standard NBC after 1. d4), with, say n = 16. (Or selected randomly from disagreement set as sketched in the Karmageddon thread.) Here, actual existing lines and lines closely resembling them will be very relevant, and the memorization environment will be much more familiar than in Chess324.

It seems that part of the appeal of chess960 for human (unassisted) play is that both sides can realistically play for a win from the start even though White is always technically "better" in the beginning. If the games start from positions near the win/draw line (as some of the chess 324 positions are), that is better for engine competition, but I think that most strong human players just won't enjoy defending the inferior side. Also as you say, such positions may have long forced lines that can be memorized, whereas most of the nearly equal positions of chess 960 (or a subset of the more equal ones in chess324) will have enough plausible moves right at the start that no one can memorize much theory on hundreds of them. But certainly one could memorize lots of theory on 18 positions, for example I know a huge amount of theory on more than 18 different openings in normal chess. It seems really difficult to settle on one variant that is ideal for both unassisted human play and engine-assisted human play.

[Time]

Correction: No black short castling (NBSC) was what I meant. (That white is probably winning under NBC is not new information.)

Personally, I would find it easier to play "restricted" classical chess with the black pieces, e.g. with black short castling not allowed for the first 15 moves, than a symmetrical Chess960 position, even if I would have an enormous defensive task in the former case.

A possible example line with the above restriction (based on what Chessify-SF suggests for pure NBSC for the initial moves, assuming for the sake of argument and example that the line is still relevant) could be [pgn]1. d4 d5 2. c4 e6 3. Nc3 a6 4. cxd5 exd5 5. g3 Nf6 6. Nf3 c6 7. Bg2 Be7 8. O-O
Nbd7 9. Re1 Ne4 10. Nd2 f5 11. Ndxe4 fxe4 12. f3 Nf6 13. fxe4 Nxe4 14. Nxe4
dxe4 15. Bxe4 Bf6 16. d5 O-O 17. dxc6 Qb6+ 18. Kg2 bxc6 19. Qd6 Rd8 20. Qxc6
Qxc6 21. Bxc6 Rb8 22. Bf4 Rb6 23. Bf3 Rxb2 24. Reb1 Bf5 25. Rxb2 Bxb2 26. Rf1
Ba3 27. Bg5 Rd4 28. Be3 Rd8 29. Bg5 Rd4 30. Be3 Rd8 31. Bb6 Rd6 32. Ba5 Rd4 33.
Bc3 Rd8 34. h4 *[/pgn] and it is still uncertain to me if white has a winning advantage in the endgame. Here, one could also imagine what Fischer described as a drawback but what I actually would find attractive as a scientific aspect of chess: the possibility of very late novelties. If these novelties actually matter, i.e. it might be the case that they improve over their predecessors in absolute terms rather than just practical ones, then I cannot see why this form of chess should be any less appealing than, say, FRC. I would rather think that people with a classical intuition should be able to enjoy it a lot more.

[lkaufman]

Correction: No black short castling (NBSC) was what I meant. (That white is probably winning under NBC is not new information.)

Personally, I would find it easier to play "restricted" classical chess with the black pieces, e.g. with black short castling not allowed for the first 15 moves, than a symmetrical Chess960 position, even if I would have an enormous defensive task in the former case.

A possible example line with the above restriction (based on what Chessify-SF suggests for pure NBSC for the initial moves, assuming for the sake of argument and example that the line is still relevant) could be [pgn]1. d4 d5 2. c4 e6 3. Nc3 a6 4. cxd5 exd5 5. g3 Nf6 6. Nf3 c6 7. Bg2 Be7 8. O-O
Nbd7 9. Re1 Ne4 10. Nd2 f5 11. Ndxe4 fxe4 12. f3 Nf6 13. fxe4 Nxe4 14. Nxe4
dxe4 15. Bxe4 Bf6 16. d5 O-O 17. dxc6 Qb6+ 18. Kg2 bxc6 19. Qd6 Rd8 20. Qxc6
Qxc6 21. Bxc6 Rb8 22. Bf4 Rb6 23. Bf3 Rxb2 24. Reb1 Bf5 25. Rxb2 Bxb2 26. Rf1
Ba3 27. Bg5 Rd4 28. Be3 Rd8 29. Bg5 Rd4 30. Be3 Rd8 31. Bb6 Rd6 32. Ba5 Rd4 33.
Bc3 Rd8 34. h4 *[/pgn] and it is still uncertain to me if white has a winning advantage in the endgame. Here, one could also imagine what Fischer described as a drawback but what I actually would find attractive as a scientific aspect of chess: the possibility of very late novelties. If these novelties actually matter, i.e. it might be the case that they improve over their predecessors in absolute terms rather than just practical ones, then I cannot see why this form of chess should be any less appealing than, say, FRC. I would rather think that people with a classical intuition should be able to enjoy it a lot more.

The appeal of FRC is that you can be pretty confident that after the first couple of moves, your opponent has to think for himself, he is unlikely to be confident that the computer has recommended f4 for example in this precise position. In any variant where the initial position is specified, then you are essentially playing against an engine (and the memory ability of the opponent) for many moves. It pretty much comes down to whether you want chess to be a test of memory or a test of the ability to find the best moves yourself. For engine-assisted (correspondence) play, then the considerations are different, and some "restricted" classical chess makes more sense since randomization doesn't really help in this case.

[Albert Silver]

Chess960 is brilliant. When the players cannot rely on standard openings and their intensive study and preparation of them, it makes for very interesting and less drawsish games. They have to spend time thinking right from the first move. Only downside to me is the odd castling rules, but even those can make things interesting, Unlike standard chess, move 1 can be a castling move.

Perhaps there are other good ideas for improving on the basic concept of chess960.

https://en.chessbase.com/post/double-sh ... itz-online

Have you read about Chess324 which Mr. Kaufman created?
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.

Yes, this is the fear, not the reality. I have not encountered one to date after several hundred games, but no doubt it is possible. However the purpose is to reduce the drawishness, not just remove opening theory, and symmetry is the number one cause of drawishness.

[lkaufman]

Chess960 is brilliant. When the players cannot rely on standard openings and their intensive study and preparation of them, it makes for very interesting and less drawsish games. They have to spend time thinking right from the first move. Only downside to me is the odd castling rules, but even those can make things interesting, Unlike standard chess, move 1 can be a castling move.

Perhaps there are other good ideas for improving on the basic concept of chess960.

https://en.chessbase.com/post/double-sh ... itz-online

Have you read about Chess324 which Mr. Kaufman created?
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.

Yes, this is the fear, not the reality. I have not encountered one to date after several hundred games, but no doubt it is possible. However the purpose is to reduce the drawishness, not just remove opening theory, and symmetry is the number one cause of drawishness.

Yes, I quite agree, but I've seen that some of the double FRC positions are evaluated as way beyond the probable win/draw margin, whereas only perhaps 5 of the chess324 subset are beyond it, and not by very much. I think that the advantages of keeping the evals to within striking distance of drawing plus the advantage of being able to play using normal chess rules (no FRC castling rules needed for 324) exceed the benefit of having an enormous number of starting positions. Once you get beyond a number that people can memorize much theory on, and reach a number that will allow any broadcast event (like tcec or ccc) to hold matches (in this case 648 games), the benefit of even larger numbers of positions is mostly confined to engine developers (ok, that includes me!).

[M ANSARI]

I never really paid attention to these Chess 960 tournaments but this time I did and have to say it was very impressive and very interesting! One think I did not like though is that they gave the position to the players so they could study and discuss it before the game. I think better is to just enter the room and go to the board and the board would have the position chosen there already. No preparation ... just start.

Interesting that Fiourzja did so well in this tournament and really shows how much he has matured as a player.

As for engine play .... I mean it will probably give insights that we have no clue about. I can easily see a situation where every single position would have a custom neural network built. Engines are strong enough where they can play millions of games against each other and create a custom network for each of the 960 positions. Bigger and larger computational tasks are never an issue with computers. Maybe it will turn out that some positions are a forced win from the opening and those openings would be removed. Or maybe they could keep such positions but then give each player a chance to play the same opening reverse colors.

[lkaufman]

I never really paid attention to these Chess 960 tournaments but this time I did and have to say it was very impressive and very interesting! One think I did not like though is that they gave the position to the players so they could study and discuss it before the game. I think better is to just enter the room and go to the board and the board would have the position chosen there already. No preparation ... just start.

Interesting that Fiourzja did so well in this tournament and really shows how much he has matured as a player.

As for engine play .... I mean it will probably give insights that we have no clue about. I can easily see a situation where every single position would have a custom neural network built. Engines are strong enough where they can play millions of games against each other and create a custom network for each of the 960 positions. Bigger and larger computational tasks are never an issue with computers. Maybe it will turn out that some positions are a forced win from the opening and those openings would be removed. Or maybe they could keep such positions but then give each player a chance to play the same opening reverse colors.

Regarding the first point, I think it is reasonable for the players to be able to think about the initial position for 15 minutes before the game starts, without access to engines. Whether they should be allowed to discuss the positions with other players is a minor point, they are all close enough in strength for that to be insignificant in terms of affecting the results.

The engines are already trained on the 960 positions. They don't need separate networks for each position, any more than they need separate positions for the Italian Opening or the King's Indian. They already play FRC at an astronomically high level, and draw 98% of games (SF vs Dragon) in Rapid on big hardware. It is almost certain, both based on this statistic and on actual evals, that none of the 960 positions are winning for White, or even close to it.

[M ANSARI]

[/quote]It is almost certain, both based on this statistic and on actual evals, that none of the 960 positions are winning for White, or even close to it.
[/quote]

OK now that I didn't know and is an interesting data point. Somehow some positions look completely lost ... that just shows how rich chess is! I would have expected at least 100 of the positions as forced wins for one side ... but that engines don't see that then for sure it means that this is not the case. I do think that customized neural nets for each starting positions would be useful as the castling is completely different.