Chu Shogi

"Would it really be a problem if there was always complete freedom in moving pieces that just got attacked?"

I was initially very hesitant to do a classification of pieces, as someone who is mostly against assigning piece values. However, the rest of the rule as it's written wouldn't work without it. Admittedly, there are times where it's perfectly reasonable to attack (with intent to capture) pieces normally thought of as much weaker, purely for strategical purposes, with no fewer than a dozen such instances in my own games.

Complex tactical reasons and strategical reasons for such attacks still evade the scope of my rule. I am aware of this. For those reasons, I supposed that whoever set up such an attack would deserve to profit from it.

To allow freedom to repeat for all attacked pieces, important or not, would greatly provide players a way to absurdly "earn" the right to repeat, even when clearly the opponent does not have attacking intent. Thus, I felt it more prudent to allow players who truly feel they must defend to judge for themselves, providing a means by which to actively remove pieces from attack where the rule does not explicitly cover it, which encompasses the vast majority of concerns. This is one of the main reasons behind my declaration mechanic.

I could add another attacking criterion...

• "For the third and all subsequent moves in a series of consecutive moves by the same player, that newly attack the same enemy piece, assess one attacking point"

I still want to dwell a bit on this "absudly earn the right to repeat". How bad is it really that people can repeat, if it is not for the purpose of attacking? Note that I did not propose that a player who was attacked can do any move he wants; he can only make a repeat with the piece that was attacked, or perhaps interpose another piece on the line of attack. These will almost always be purely defensive moves. So yes, you will often have the right to repeat without a real reason, but you cannot do much with the granted moves in terms of attacking. Besides, having the right to repeat is one thing, having the opportunity to repeat quite another. When there are futile attacks as side effects, these will in general be ignored, and 'evading' them would never lead to a repeat, as they were not evaded before.

And what we aim to prevent is that one player can force repeats one-sidedly, through attacks that can only be reasonably evaded in one way. If both players aren't interested in making progress, a repetition ban cannot cure that. Even with a few pieces there are so many ways in which you can pointlessly shuffle them without repeating that you can last a lifetime.

Can you think of an actual example of a perpetual chase that you would have liked to be forbidden, but that would be allowed because the attacker derives the right from futile side-effect attacks of the defender's move?

BTW, I have to extend my proposed rule such that you are always allowed to move your King, even when it was not in check. Otherwise dead draws like K vs K could be won by forcing stalemate through repeats. Which I think would be undesirable.

As to the case of mutual perpetual attacks, how about this rule:

* A 'forced evasion' is a move with a piece that suffered a new (possibly futile) attack, or a move onto the ray over which such an atack passes.

* It is only allowed to repeat a position through a forced evasion, except:

* It is forbidden to cause a repeat when all the moves leading from the previous occurrence of that position to the current one are forced evasions.

This puts the burden of deviating from a mutual chase on the one who started it. Which seems reasonable, as he was not forced to start it.

Where an evasion creates a repetition in a position where the "evading" player makes an important attack, it's possible for both players to make a series of moves that is repetitive and undesirable.

Suppose Player A has a free king and moves it such that it incidentally attacks an unimportant pawn. Player B takes advantage of the rule you propose, interposing a rook, thus attacking the free king. Player A moves his free king to another file, incidentally attacking an unimportant silver. Player B once again interposes the rook to "defend" the silver, attacking the free king. 

Using my rule, the side with the rook would clearly be at fault. It would have no means to repeat past the first attack, for which it could potentially declare, for a second declaration would be illegal due to the prohibition against declaring against a defending move.

Suppose Player A defends a gold with his dragon, incidentally attacking a side mover. Player B moves his side mover, revealing a discovered attack on the king by a free king. Player A blocks the "check" with his dragon. Then Player B blocks the dragon with his side mover. Repeat.

Using my rule, Player B would clearly be at fault, and Player A would be completely absolved.

Kv K would terminate long before repetition is possible with any sufficient draw rule. I play with the 199-move rule, which is of reasonable length.

Normally, the first evasion in any series of attacks does not result in repetition, but rather when the pieces return, once a player sees that his attempt was futile. The attack itself is not what I seek to punish, but the intent. Thus, if the repetition is not made immediately, as long as he does not make an attacking move himself, a defender maintains the right to repeat far after this initial attack, as a deterrent against overall attacking moves.

The problem with truly neutral repetition by evasion is that it does not advance the game. Progress is desired, one way or another. Allowing such repetition on its own accord can cause great complications. Repetitions could be made indefinitely and in all sorts of arrangements. It would be impossible to arbitrate into anything but a draw. Only by decisively saying whether a move is attacking, defending, or neutral can there be order and fairness.

Most of the wording for my rule proposal covers scenarios that would occur incredibly rarely. They are covered only for the sake of thoroughness and completeness. For over 98% of moves classed as "attacking" by my rule, the amount of attacking points assessed will not exceed 2. Where it does exceed 2, the defender is probably losing anyways. The rule's length does not make it inherently complicated; it was designed to be relatively intuitive.

Count pieces genuinely attacked, count pieces saved. That is enough for the vast majority of cases. Listed exceptions are meant to cover instances where the attack may be incidental or at no fault of the "attacker". 

I think the rule I gave takes already takes largely care of that example. The Queen and the Rook both came from elsewhere. So the position to be first repeated is those where the Rook interposed to attack the Queen. This would be forbidden because the whole cycle consisted of forced evasions. So the Rook must change its move, and cannot use the futile attacks as cover for the fact that it is really chasing the Queen.

Of course this raises the question what would happen if these were genuine attacks, and the Rook tried to parry them by interposing. But that really is a mutual chase, and it will be almost impossible to define by static rules who is forcing who. That completely depends on how many good alternatives there are for rescuing the pieces attacked by the Queen, or for the Queen to escape from the Rook.

Note that in XQ the Rook would be at fault, because the Queen alternately chases two different pieces, (which is always allowed, even if one of the two is the King), while the Rook chases the same Queen on every move Even if you want to make chasing multiple pieces illegal in Chu Shogi, you could still use it as a tie breaker for the seriousness of the offense. Unintended (futile) attacks as side effect of necessary defensive moves will almost always 'attack' different pieces on every move. That alone could be enough to 'forgive' them, even when we unjustly classified them as attacks, against an opponent that is continuously hammering the same piece. So the rule would become that the right to evade the last attack to a repeated position is revoked when all moves since the previous occurrence have been evasions, and you have been evading attacks on more different pieces than your opponent has.

In XQ the situation is simpler, because pieces that can attack each other must be equal pieces there. So if I interpose a Rook to a Rook threat, it would not be considered as a counter-threat, but as an 'offer to exchange'. In Chu Shogi you cannot do that, because too many pieces of different value all move along the same ray.

I don't understand your second example. I suppose Queen, Side Mover, Dragon and King at one time are all on the same file (in that order). The Dragon now moves away, (exposing its King to a discovered check), to protect a Gold. As a defense against what? And it accidentally attacks the SM in the process. But it was already attacking the SM, so this is not a new attack. It doesn't give the SM the right to repeat the discovered check. It seems to me that players have one unforced move in the cycle. I don't see frome what the DK would derive the right to repeatedly protect the Gold.

Anyway, no rule will be perfect, not even an extremely complex one. The best that can be hoped for is to make rules that satisfactorily solve the large majority of situation encountered in practical games.

I don't see the importance of later repetitions; it seems completely hypothetical.

If a particular piece A attacks, from square B, a particular enemy piece C located on square D, and this happens for the 10th time for those two particular pieces and squares, the player with piece A loses, regardless of the consecutive overall positions in which this happened.

Pieces A and C must be the same individual pieces all along, not another instance of the same type of piece.

This is a pretty short read.

What if piece A is a lion avoiding harassment from an enemy free king, and piece C is a defended pawn doing nothing? Also, 10 times is way too many. Also doesn't solve the problem of neutral repetitions. Also, what if attacking that piece is the only way to defend?

An improvement...

• If a particular piece A, on given square B, attacks a particular piece C, on given square D, neither to evade attack nor defend* another friendly piece nor prevent a forced mate, 5 times in any position (resets after every capture and promotion), or 3 times in the same position, or 2 times in the same position if piece C is a king/prince, the player with piece A shall forfeit the game.
• *define "defending" as any move that addresses an "attacking point" as defined in my proposal

Still far from perfect, and I like my rule better, but pretty good. Still doesn't cover mutual repetitions.

Version 2.0

• Clarification of terms "attacked" and "defended" as applies to the rules
• Subdivision of type D pieces
• More defender-friendly criteria
• New measure, "attack count", to better evaluate game flow

Well, I haven't read it, but no matter what it says, the sheer size of it already says this is awful. Games shouldn't have rules as complex as this. The Xiangqi perpetual rules are already awful, but this seems worse. It would hurt a game much less to occasionally have a ruling that doesn't make much sense than to have such complex rules. Perfection is the death of playability.

After all, a player that knows the rules would see it coming when he will be disadvantaged. If for some exceptional reason the rule would specify that I would not be allowed to indefinitely evade an attack on a piece, even though from a realistic point of view I would be the chased party, I should be able to see that trap, and just avoid it. That isn't really much different from saying "hey, that is unfair. He attacks my Queen with a Rook, and normally I would have the right to move it away, but now that my King is standing behind it I cannot, and I will lose it!".

I also have the feeling that this is addressing a problem that for a very large part is purely hypothetical. The proper way to go about this is first play a couple of thousand games under the rules where repetition is a draw, and get statistics on how many repetition draws this results in that on closer inspection could be classified as 'undesired'. And then try to capture the large majority of those in a simple rule. Like "creating a repetition by a move that delivers check is forbidden", or "repetitions are forbidden, except to evade check". Wow, 90% of the cases solved! Of course you would have to generate a new set of games with the new rules, to see if you just didn't displace the problem to another situation.

I would be very surprised if the following simple set would not catch the overwhelming majority of the undesired draws:

Moves that create a repetition are forbidden, except:

1) When they evade check on your only royal.

2) When they solve a mate-in-1 threat without delivering check.

3) When they cure a fresh attack on a piece that has been suffering fresh attacks on every move since the previous occurrence of the position, through moving away, interposing or protecting that piece without checking or threatening mate in 1.

4) When they do not create a new attack on any of the opponent's pieces.

Ideally, the rule would say, in its totality:

"Repetitions are generally forbidden. Use common sense to identify which player, if any, has the attacking initiative from any given position. If Player A has attacking initiative, and Player B has played with sufficiently passive defensive intent, then only Player B is allowed to create a repetition."

This, more or less, was the historical rule - simple but complicated. It depends on the common sense of the players, which can differ. This works for the vast majority of cases, if both players argue in good faith. However, it is helpful to have a clearer guide when disagreements can not be resolved in a friendly manner.

My rule, conversely, is complex but straightforward. Identify the type of move, search the checklist, check the list of exceptions, done. I designed the lists to be intuitive. It's not as if the list items all conflict with each other. All that's required to use the rule is basic position comprehension, plus the ability to follow instructions.

For me, fairness is paramount. It is worth all the effort to convert something from 99% fair to 99.9% fair. Rules with loopholes, unfair results, and odd interpretations drive me absolutely mad. It puts me at ease to fill 50%+ of the rule with complete rulings for bizarre scenarios that will actually occur in fewer than 0.1% of games, just so there never has to be a live adjudication.

'Fair' is a subjective notion. Most players of orthodox Chess would consider it completely fair if a player that is two Rooks behind salvages a draw by perpetual checking with his Queen. Apparently poor King safety compensated the advantage of two Rooks. And they would consider it completely unfair that when my Lion is taken by a non-Lion, I cannot take the opponent's, even though I pseudo-legally attack it. If the rules say you lose on perpetually checking or counter-striking against a Lion, they would consider that very unfair rules. I remember that at one point they changed the blitz-chess rules such that illegal moves did not have to be taken back, but that when they happened you could win by capturing the King. So we often tricked an opponent under time pressure by stepping with our own King next to his in a lost position, at the point where he could make a capture of a fat piece he just attacked elsewhere; in 9 out of 10 times he would then capture that piece, and we would win by KxK. Most opponents considered that extremely unfair, and got very angry about it. But we were just exploiting the new rules. The rules define the game, and when you play by the rules that is by definition fair.

The only objective measure is how close you can approach the alledged historic intent to penalize the attacking player. How well a given set of rules approaches this ideal can perhaps be expressed as a percentage. But I don't believe that you could get anywhere near 99.9%, not even with rules 10 times more complex than what you already have. No matter how precisely you describe the tactics involved in the attacks and protections, there can always be side effects elsewhere: important squares that must be kept protected to avoid an important loss due to a skewer, a 'mate threat' that is not really a forced mate (and thus flies under the radar of your rules) because the mate can be averted at the expense of a large sacrifice.

So the effect of complicating the rules will not reduce the unsatisfactory cases from 1% to 0.1% and then 0.01% each time you inflate the complexity another order of magnitude, but more like from 11% to 10.1% to 10.01%, where you never get below 10% even with a rule book the size of the universe. And I very much doubt whether anyone but you would care if it is at 10% or at 11%; if you can live with 10% dissatisfaction, the difference with 11% would probably go completely unnoticed.

Fairness. By that, I meant a defender always being allowed to defend, and an attacker not being allowed to repeat an attack. No clear cases to the contrary.

Important squares. One of my defensive criteria allows a player free reign, so long as the action is within their own territory (4 ranks). This seems like a reasonable amount to allow. If one of your "important squares" is outside of your territory, then the fact you have extended that far shows not pure defensive intent.

Mate threat. You can defend as much as necessary. If somehow the mate threat lies partially outside your territory, then you can still declare the move to be defensive. The declaration mechanic is meant to catch all the leftover scenarios not explicitly covered, partially to reward the attacker if they set up an elegant enough attack to evade the bulk of the rule, but still allowing the defender up to 3 consecutive chances to maintain right of repetition, since the attacker is not meant to be able to forcibly repeat to win, even in subtle scenarios.

I am confident my rule as it is currently written adjudicates as desired for over 99% of repetition scenarios.

I would have to see that before I believe it. And even then, the rule is already unacceptably complex.

And as far as I am concerned it does not even address the gravest problem in Chu Shogi, which IMO is a suitable generalization of the 50-move rule, and excessive (but non-repetitive checking). I think Chu Shogi as a game would far more benefit from having simple rules to solve that than from nitpicking about which repetitions are allowed and which not.

I would prefer a pretty low 'budget' of 'quiet' moves that is replenished of every capture or promotion (while the game is drawn as soon as it runs out). To prevent games that make progress to be declared draw too early, I would supplement this with a rule that a player that objects to a thus claimed draw the right to designate a piece he aims to promote, and as long as this piece is closer to the zone than it was N moves before, the draw cannot yet be claimed. (And he cannot change his choice until the next promotion or capture). To allow acceptably low N, I would also award each player with a separate 'check budget' (of, say, 5 checks), which also is only replenished on promotion or capture, where a player would lose after a non-checking move when his budget is negative.

I like the 199-move draw rule. It serves its purpose without any fuss. A top quality opening series can easily see 80+ moves made without capture/promotion. Quiet moves are great! They are the backbone of elite chu shogi. To demand a particular pace shows ignorance of the game.

There is no problem here. The draw rule only becomes a problem when there is an abundance of draws. A potential draw is much less likely than a repetition dispute.

The JCSA has held more online meetings recently. I found some youtube videos from one of the members, under the youtube account "札幌中将棋" (Sapporo chu shogi). If you can understand Japanese, the commentary is not so high level, but he does a good job of letting the viewers know his thoughts. And there are thorough post-game reviews, which are important!! 

Anyways, the next online meet-up is scheduled for 15 August, 10:00-17:00 Tokyo time (02:00-09:00 central european) on 81dojo.

I started a club if anyone interested would care to join. It's foremost a place to learn and get stronger at chu shogi. I'll occasionally post articles on all sorts of topics. Of course, there's no obligation to play, but I would very much like to start 1 or 2 leagues for regular games (round-robin format), so anyone willing to play is greatly appreciated!

https://www.chess.com/club/chu-shogi-league/join