﻿﻿ MatPlus.Net

Website founded by
Milan Velimirović
in 2006

11:52 UTC
 ISC 2022

Remember me

 CHESS SOLVINGTournamentsRating lists1-Jul-2022
 B P C F

MatPlus.Net Forum General Castling convention

Page: [Previous] [Next] 1 2 3

### Castling convention

Happy new year to MatPlus readers!

Everybody knows the castling convention, article 16 (1) of the Codex: "Castling is permitted unless it can be proved that it is not permissible."

My question is whether this convention applies only to the moving side, or to both sides. To illustrate this point, consider the following scheme:

(= 2+3 )

#1 Disparate

If the castling convention applies only to the white side, there are two solutions 1.Qd8# and 1.Rd8#.

If the castling convention applies to the black side too, then castling is permitted and thus its last move is a Queen one. As a Disparate consequence, there is only one solution 1.Rd8#.

I think this is another example of Partial analysis. If black's last move was with rook or queen, there are different unique solutions (1.Qd8# or 1.Rd8#), if last move of black was with the King, then two solutions!

PRA applies only in case of dependency of castling/ep capture rights (article 16 (3)). In my scheme, considered as a retro problem (a retro content should explicitly be mentionned in case of a fairy problem, if I remember), I just would like to know if the following reasoning is correct or not:

- Nothing can prove that black lost its castling right.
- The castling convention then implies that black may castle.
- It implies at its turn that the last black move was a Queen one.
- Due to the fairy condition Disparate, the stipulation #1 is therefore correct with (unique) solution 1.Rd8#

Nicolas, I think there is a flaw in the above reasoning. You mentioned "Black can castle", but this is NOT mandatory and must not be ascertained.
So actually Black may castle in the above diagram, but this assumption must not be proved, unless the author explicitly authorizes that.

Since no partial retro analysis is involved, I think both moves are mate. We must not care at all about last black move, because there is no history at all!
Just read carefully codex footnote 18: "[Retroanalysis] does not apply to fairy compositions unless it is essential to the content of the composition."

The correct question in this case is: does the author intentionally look for partial retro analysis?
If so, then he must state under the diagram the three magic letters: pRA.

First Nicolas' question was -"..whether this convention applies only to the moving side, or to both sides."-
The convention applies to the position, neither sides has lost the castling-right unless the opposite can be proved. So, who is on the move is generally irrelevant.

In post 3, Nicolas wrote: "In my scheme, considered as a retro problem (a retro content should explicitly be mentioned in case of a fairy problem, if I remember)"

So, CONSIDERED AS A RETRO PROBLEM, his reasoning looks correct, in principle. But the specific features of a particular fairy condition might require the specific retro-logic and a proper reinterpretation of orthodox retro-conventions.
The question is WHY the convention of dependency of castling/ep capture rights is established?
When this is answered, we can see how to apply the orthodox retro conventions to a "Disparate" fairy condition. Hypothetically, the interdependence of "mobility of wQ/black castling rights" might be treated as orthodox "e.p./0-0". But it doesn't seem to be the case here. Nicolas' intention is not pRA, not even much of retro-analysis, only the direct simple consequences of a supposedly legal position.

(Some hypothetical condition "AntiDisparate wQ" might mean that wQ may play only if the last previous black move was played by bQ. In the same diagram with such a condition, wQ could play only if Black will castle a posteriori.)

So Nikola, if I correctly understood your message, the reasoning claiming that my retro fairy scheme admits a unique solution sounds ok, as far as it doesn’t exists any “fairy PRA”. But under such a “fairy PRA”, a Queen move would be the analogue of an orthodox two-step Pawn move allowing en-passant capture and, as a consequence, my scheme would admit two partial solutions.

Is it what you mean?

> "Castling is permitted unless it can be proved that it is not permissible."

I think it's fair to say that your interpretation is incorrect (though I certainly appreciate your efforts to clarify the issue).
This rule implicitly states: "[The] Castling [MOVE] is permitted..." -- therefore, this assumption clearly applies only to the player on the move.

Not only has this been the commonly accepted interpretation of the rule's wording, there exists a long established convention which nullifies the contrary interpretation (which you have advanced, here)...

For example, consider how entangled castling situations would be impacted by your interpretation of this rule...
If either white or black may castle, but not both, your interpretation (which considers the opponent's castling rights, prior to their having the move) would effectively nullify castling for both sides (and, we all know this is incorrect)!
The long established precedent holds that assumed castling rights are exclusively granted to the player on the move.

You have exposed an implied ambiguity (and raised a very interesting question).
And, we should all welcome problems (even joke problems which are not entirely correct) which help to clarify the wording of rules (something you have clearly done here).

Perhaps this rule should be restated:
The player on the move may assume to have some castling right, unless there exists no retrograde possibility which allows for the legality of said castling.

Your attempt to consider, before the fact, the assumption of castling rights, can be viewed as something akin to "A Posteriori" reasoning ...
Except, you never actually castle in the course of your solution -- which is necessary under A.P. convention (to establish the validity of any assumption drawn before the fact).
As a result, the reasoning behind your assumption is vacated.

If you want to consider castling assumptions before the fact, I am afraid you must resort to some fairy convention.
I sincerely hope that the Codex will take seriously your exposure of ambiguities implied by a poorly worded standard convention, and seek an explicit improvement!

I'm sorry Nicholas, I have overlooked your question.
According to the Codex there are two mates because the fairy retro-legality is irrelevant if this is not clearly marked in the stipulation.
If you mark PRA, then it is as Seetharaman wrote in post 2.

But PRA is a specific stipulation. If I understand the logic right, it is like having two or more twins. The same diagram represents several positions with different histories and therefore different possibilities of future play. Partial solutions might be considered as twins.

You did not mark PRA, but since there is a fairy condition, you have mentioned in text (post 3) that retro-legality is included in the stipulation. You mentioned it, and that makes the problem correct with unique mate 1.Rd8#.
Trouble is how to mark shortly that the author wants to include retro-legality in a fairy problem, but not PRA (orthodox problems require retro-legality by default, without mentioning PRA). The essential difference is that without a mark "PRA", a diagram represents only one position which is determined by the castling and e.p. conventions (possibly something else in fairies).

I don't think that "a Queen move would be the analogue of an orthodox two-step Pawn move allowing en-passant capture" in this case, at least for "a posteriori" reasoning. But I'm not sure what you exactly mean. When you declare that stipulation includes retro-legality, the castling is legal by the convention, this means that last move was by bQ, so wQ is immobile for a moment. The existence of castling-right proves that a move by wQ is ILLEGAL, without PRA or AP. In case of e.p., the existence of castling-right DOES NOT prove that e.p. is LEGAL. Castling must be executed AP to prove that e.p. was legal or a mark "PRA" might allow e.p. partially.

Only one proof-game is enough to legalize castling, but each one of all possible proof-games must end with P's double step to legalize e.p..
A specific case is when all proof-games which legalize castling also legalize e.p., but the two equally strong conventions do not allow us to choose a history and determine the position precisely. The castling-right exists but the present position (its history) is not determined untill the castling is executed. e.p. capture in the diagram (and the whole play afterwards) is theoretically illegal but in the moment when castling is executed, the history is determined and everything is legalized a posteriori.

Tomorow I'll read what I've just written, that might be funny :-)

@ Nicolas : It took me awhile until I understand what you mean !!
Just tell me if I understood well :
You mean :

1- there is a convention that 0-0 (0-0-0) is possible unless it is proved not to be.
2- In the above diagram I cannot prove the 0-0 to be impossible, it means that it is possible.
3- this means that the last move was made by the queen.

Well... If so, I would say : what a good joke !

For me it is a matter of obviousness that 0-0(0-0-0) possible does not mean that is mandatory possible. It means only that you cannot forbid it if played.
So in your diagram, for me, 0-0 is possible and it does not mean anything about the last move ! Because black can say that 0-0 is broken.

@Jacques,

As I understood the question -- perhaps I have misread it -- the crux of the issue here pertains to whether the castling assumption can be applied to a player, prior to their having the move.
In the diagram, white is on the move, and is attempting to reason based upon the assumption that castling legality applies to an opponent not in possession of the move.

And, if this is indeed a correct interpretation of the question being posed here, as I have noted in my previous post, such a convention would violate a well established precedent (e.g., there are no entangled castling rights, because the castling assumption implicitly applies to a move, which can only be considered by the player on the move).

I can certainly appreciate the humor in attempting to apply the assumption before the fact (though I can not consider it correct).
But, after reading your post (which does not mention this essential point), I am left wondering whether I have missed the punch-line.

Kevin, you can castle only if you're on the move, but the castling-right is a feature of the posiotion, whoever is on the move. That's the basis of the triple-repetition rule. In this very forum, that rule was even interpreted as that castling-right might exist even when it's clear that the castling itself can never be executed (in future).

After reading my own reasoning in post 5, I must admit that my presumption was wrong. Now it seems that a hypothetical convention of "temporary Disparate paralysis" may be equally strong as the convention of "temporary e.p. legality". So, 1.Qd8# might be legal, since the temporary paralysis of wQ could be proven only by a posteriori castling. Black can't castle after 1.Qd8 is played, but may claim that it was an illegal move, because his castling-right was never lost. In case of helpmate in 0.5 moves (1 halfmove), Black wouldn't protest and the mate is legal -:)

Hello and happy new year!
I'm new to this forum, but the question posed by Nicolas occured in a discussion with me.

I think the problem proposed by Nicolas mixes two different (and interesting) issues:
1) does the castling convention only apply to the moving side,
2) does the castling convention permit an retrograde analysis TO DETERMINE FORWARD PLAY.

Concerning the second issue, I think that the following are commonly accepted (but I have no doubt that someone will contradict me...) :
– castling and pep conventions are used to forward play,
– they can't be used for retroanalysis.
But with disparate chess, a retroanalysis seems necessary to forward play.

For not being disturbed by who is to move, I suggested to Nicolas the following problem:
(= 4+12 )

#1 Disparate
White is to move and can actualy castle according to convention.
Does it mean that Rb6 moved last and only 1.cxb8=Q or R (sorry for the dual!) mates?
I don't think so and have a schizophrenic approach (if PRA is not stipulated): white can castle AND last move was not necessarily played by Rb6.

Concerning the first issue, I think exactly the opposite of what Kevin said :-) :
QUOTE
If either white or black may castle, but not both, your interpretation (which considers the opponent's castling rights, prior to their having the move) would effectively nullify castling for both sides (and, we all know this is incorrect)!

I would say, if either white or black may castle, but not both, my interpretation (which considers the opponent's castling rights, prior to their having the move) would effectively ALLOW castling for both sides (and, we all know this is correct, and this is precisely why PRA or RS is necessary)!

QUOTE
The long established precedent holds that assumed castling rights are exclusively granted to the player on the move.

I really think that the long established precedent is in favour of my interpretation! :-)

If the legality of diagram is included in the stipulation, I agree that "white can castle AND last move was not necessarily played by Rb6".
1.0-0 would prove that wRc8 was paralysed on the diagram, but it would be irrelevant. 1.Rxb8# is legal but for instance, 1.Rxd8 would prove that the last move prior to the diagram was h7-h6 and that castling-right must have been lost before that. The position is not exactly determined by the diagram, but when one of the potentially legal possibilities is executed, the position becomes determined.
Without such a concept of undetermined position, I can't understand a posteriori reasoning.

@Francois,

Thanks for sharing alternative viewpoint...
I certainly do appreciate this attempt to reinterpret the castling assumption, using PRA; but, you should be aware that the CODEX already expressly forbids this end-around attempt (and you haven't succeeded in getting around the CODEX).

>"I would say, if either white or black may castle, but not both, my interpretation (which considers the opponent's castling rights, prior to their having the move) would effectively ALLOW castling for both sides (and, we all know this is correct, and this is precisely why PRA or RS is necessary)! "

You want to presume that the assumption of castling rights applies even to a player not in possession of the move.
However, your quip (that "BOTH SIDES" might have castling rights) completely misses the point of the entangled castling situation.

Clearly established precedent has maintained that castling rights (in a mutually dependent castling situations) are laid at the feet of the player on the move.
These rights are explicitly NOT settled by either PRA or RS (and everyone should be well aware that standard convention already settles this matter, without needing resort to either PRA or RS).

This long established precedent completely invalidates your claim (and your reinterpretation stands in stark violation of the CODEX).
If you want to establish such a convention, you would have to invoke it explicitly in the stipulation (to get around the standards of the CODEX).

CORRECTION: It seems I failed to notice that the CODEX has already been rewritten, in an excessively careless manner, such that it now attempts to use PRA (or RS), to interpret castling permission.

This rewrite is a travesty... and it certainly does open the door to your interpretation...

Castling is a move, which is explicitly "PERMITTED" by the CODEX, so long as the player intending to take permission meets the criteria described therein.
Having the move is implicitly a part of that criteria, because without the move, the player is CLEARLY not "PERMITTED" to play such a move (or any other).

I stand corrected -- the CODEX needs to return to the prior standard (the rewritten rule -- 16.3 -- only makes matters worse).

I think my 'schizophrenic approach' is probably unsustainable in the long run...

In my problem (12), I said that castling was possible by convention but that it could not be used to prove that last move was played by Rb6.
So I would have allow 1.Rxb8#.

But consider this:
(= 8+8 )

h#1.5 disparate

Retroanalysis: white can castle only if last move was played by a rook.
So if white can castle, Rf7 can't move.

According to my 'schizophrenic approach', white is allowed to castle, but it can't be used to prove that last move was played by Rf6, so Rf7 is allowed to move.
But is the solution 1...Rxf6 2.Qxh2 O-O-O# acceptable? Certainly not.
And is the solution 1...Rxf6 2.Qxh2 Rf1# acceptable only because white does not castle in actual play? I really don't know :-(

Btw...
I assume that everyone should know what is "Disparate"
However I give here a translation of the french definition I could find :
A move cannot be done by a piece of the same nature just played by the opposite side.

Why not : 1...Rxf6 2.Qxh2 Kf1# ?
Or also 1...e4 2.Ra6 Kf1# ?

It is commonly admitted that 0-0-0 is a move of the king.

So 1...e4 2.Ra6 0-0-0# should work too.

Yes Jacques, there seem to be many solutions. But 1...e4 2.Ra6 0-0-0+ 3.Rg~/Ra~ !

Nevertheless, if Kings are hypothetically immune to paralysis, Francois' reasoning looks OK. 1...Rxf6 2.Qxh2 O-O-O# is illegal and 1...Rxf6 2.Qxh2 Rf1# is questionable. I think that it is legal if the basic logic of the castling convention is applied on Disparate convention.
If there is any history which allows castling, castling is a legal possibility.
Similarly, a piece is NOT paralysed if there is any history in which the previous opponent's move was not played by a piece of the same nature.
If the castling-right is not stronger or having a priority, the moving-right of wR is equally possible as the castling.
1...Rxf6 is legal and the castling-right should be considered as having been lost even before the diagram. No castling in the future play.
1... 0-0-0+ is also legal and now wRf7 should be considered as having been temporary paralysed on the diagram, but that is irrelevant for the future play.

"1...e4 2.Ra6 0-0-0+ 3.Rg~/Ra~ !" Nice !

And so, 1...Rxf6 2.Qxh2 O-O-O+ is not only illegal but also refutated by 3.Rg2 plays.
So you have to add some more conventions
a) disparate rex exclusive (as you said it)
b) 0-0-0 is a move of both king and rook (or a move of the rook only)

And I wonder how bR came to g2. This could have happen earlier but I can't quickly find possible last moves by both W&B to legalize 0-0-0.

This does not seem to me so difficult...