3rd repetition, a case where not cause castle rights... but.

[Luis Babboni]

I´m here bother you again!

In the game I adjonted, Arena did not determined a draw cause castle rights was not the same in the 3 "same" positions and my engine crashed so I need something to fix there.

But my question is other.
I always thought about not tie cause differents castle rights in a situation where you in fact have differents options in the 3 "same" repetitions.
That is, if in one of the repited positions you had differents options, is correct that those 3 positions are not really the same so is correct that it not deservs a tie.
But in this case, nevertheless the black king had not the same castle rights in the 3 "same" positions, the option black had was the same in the 3 cases!

May be is not a "question", is just a comment cause the rule seems to be that and not other, but surprised me that "the idea" I understood is behind why differents castle rights do not mean a tie, is not happens here.




https://dl.dropboxusercontent.com/u/905 ... c%20T4.pgn

[Evert]

Positions that differ in castle rights do not count as the same because players do not have the same move options in all otherwise identical child positions.

In other words, if castling rights have changed then the positions count as different (the right to castle is distinct from the ability to castle, the formulation in FIDE rules is a bit confusing).

What is perhaps confusing here is that castling does not reset the 50 move counter: it is considered a "reversible" move.

[Luis Babboni]

Positions that differ in castle rights do not count as the same because players do not have the same move options in all otherwise identical child positions.

Not understood.

In other words, if castling rights have changed then the positions count as different (the right to castle is distinct from the ability to castle, the formulation in FIDE rules is a bit confusing).

Good point, what is the same here is the ability to castle but rule just talka bout rights, I´m right?

What is perhaps confusing here is that castling does not reset the 50 move counter: it is considered a "reversible" move.

Logical, is reversible.

[hgm]

This is sort of a hole in the FIDE rules. Formally you have castling rights, but because all legal moves you have will destroy them, the game state is actually the same as if you did not have castling rights. FIDE rules still consider the positions with and without rights different.

Not that it is of any practical interest. If people want to repeat, they will just repeat one time extra. If they don't want the draw, they should not vene have repeated the first time.

[hgm]

Logical, is reversible.

Well, it is not reverisble in the sense that you can ever get your castling rights back.

The point here is that it is more or less coincidental that captures and Pawn pushes are both irreversible and signs of progress towards a decision. It is really the latter that is reason to reset the ply counter. Castling is not really progress.

This causes a big problem in games like Chu Shogi, where decisive promotions can be made by pieces that move reversibly. It is not really possible to define an equivalent of the 50-move rule, in that case. Imagine that in Chess Pawns could also move one step backward... What should you do now to decide whether a game makes progress? Obviously moving a Pawn backwards would not be progress. But resetting the counter on forward Pawn moves only is also no good. Because you could step your Pawn back and forth to reset the counter without really advancing the Pawn. I guess you should do something as that no Pawn is closer to promotion than it was 50 moves earlier. That still makes it a bit unclear what should happen when one Pawn advances, but another retreats (and then the next 50 moves you would do the reverse). So I guess the rule should be that the total number of steps of all your Pawns to promotion together must decrease within 50 moves.

[kbhearn]

There's very few cases where it's important that a pawn push resets the 50 move counter though (at least in human play - in computer play it gives them plenty of time to find the win that they don't know how to progress). And furthermore there's cursed wins that show the 50 move rule isn't designed to not alter the game. They flirted with the idea of removing/altering the 50 move rule so that theoretical wins could be demonstrated but ultimately kept it so it's just a practical decision to limit how long one may torture the opponent without bringing the game demonstrably closer to the end (even at the cost of sometimes turning wins into draws).

In the case of your oversized shogi game where i assume like regular shogi captures leads to drops and therefore are not bringing the game closer to an end, i'm not sure a 'no progress' rule makes sense at all even. You may have to rely on the honor of the players to declare when there's just no more game to be played.

[hgm]

No,Chu Shogi dates from the 13th century, long before drops were invented. So it is much more Chess-like than modern Shogi. An extra complication there is that not just Pawns promote (these really are irreversible), but just about anything. So doing arithmetic for how close the total of your stepping pieces are to promotion might take longer than the game itself.

Perhaps a two-step rule would be feasible: if nothing is captured or promoted for at least 30 moves, a player could issue a 'progress warning'. If his opponent wants to deny a draw, he must then single out one or two of his (promoting) steppers, pledging to advance these at least once every 20 moves. Meaning that as long as nothing is captured or promoted, the player issuing the progress warning can claim a draw any time the sum of the distances of these two pieces to the promotion zone is not smaller than it was exactly 20 moves before.

The reason I propose to do this for two pieces is to prevent that a defender will throw all his material at obstructing a singled-out piece, letting the other run. You must allow people to convert the advantage of having material on both wings, wich the defender could not both stop.

[kbhearn]

Perhaps just forget about mere 'pawn moves' as progress qualifiers then and set a rule for X moves until promotion or capture. Make X higher if you need to, but keep the rule simple! Take a look at typical endings that 'should' be a win and see what a comfortable value of X that they stay a win with room for inaccuracy should be (equivalent of winning KPvK, KBBvK, KBNvK, KRvK etc - not so much neurotic ones like KNNvKP).

[xmas79]

Mmmh... that reminds me a discussion on this forum...

http://www.talkchess.com/forum/viewtopic.php?t=49362

[syzygy]

Logical, is reversible.

Well, it is not reverisble in the sense that you can ever get your castling rights back.

The point here is that it is more or less coincidental that captures and Pawn pushes are both irreversible and signs of progress towards a decision. It is really the latter that is reason to reset the ply counter. Castling is not really progress.

And neither is a king move or rook move that loses the (or a) right to castle. (Logically such moves should also reset the 50-move counter if castling did.)