MatPlus.Net

 Website founded by
Milan Velimirović
in 2006

20:12 UTC
ISC 2024
 
  Forum*
 
 
 
 

Username:

Password:

Remember me

 
Forgot your
password?
Click here!
SIGN IN
to create your account if you don't already have one.
CHESS
SOLVING

Tournaments
Rating lists
1-Jan-2024

B P C F





 
 
MatPlus.Net Forum General Law 9.2: Draw by Repetition
 
You can only view this page!
Page: [Previous] [Next] 1 2 3
(21) Posted by Joost de Heer [Friday, May 3, 2013 11:52]

With my 'right' text I was referring to the list of -possible- moves in a position (which is in a lot of cases different from the list of -legal- moves). The 3-fold repetition rule is about possible moves, not legal moves.
 
   
(Read Only)pid=10368
(22) Posted by Ian Shanahan [Saturday, May 4, 2013 03:09]

Surely a move is possible if and only if it is legal (i.e. legal and possible moves are equivalent)?
 
   
(Read Only)pid=10370
(23) Posted by Andrew Buchanan [Saturday, May 4, 2013 04:02]

I can see what Joost is trying to argue, but I am afraid I agree with Ian. The word "possible" is not used in any specialized sense anywhere in the rules, and I think is lightly intended as a synonym for "legal" here. And the rule actually goes on to say explicitly:

"Positions are not the same if a pawn that could have been captured en passant can no longer be captured in this manner."

This really seems definitive to me.
 
   
(Read Only)pid=10371
(24) Posted by Geir Sune Tallaksen Østmoe [Saturday, May 4, 2013 07:14]

I agree with Ian and Andrew. And in the new wording it looks even clearer:
"Thus positions are not the same if:
a. at the start of the sequence a pawn could have been captured en passant."

This wording seems to state that what matters is whether capturing en passant was legal. If that is not the intended meaning, then the wording looks plain wrong to me.
 
 
(Read Only)pid=10372
(25) Posted by Joost de Heer [Saturday, May 4, 2013 12:35]; edited by Joost de Heer [13-05-04]

 QUOTE 

Surely a move is possible if and only if it is legal (i.e. legal and possible moves are equivalent)?

No, legality can only be determined after the possible move is done, not before the move itself (e.g. 'is my king in check after I've done this move?'). Legal moves are a subset of the possible moves.

I know it can't be used as proof, but take a look at Popeye's output:

begin
pie whi ke6
bla ke8 bh8
stip ~1
opt move noboard
end


1 (Ke6-f6 Time = 0.010 s)
2 (Ke6-e7 + Time = 0.010 s)
3 (Ke6-d6 Time = 0.010 s)
1.Ke6-d6 !

4 (Ke6-e5 Time = 0.010 s)
5 (Ke6-d7 + Time = 0.010 s)
6 (Ke6-f7 + Time = 0.010 s)
7 (Ke6-f5 Time = 0.010 s)
1.Ke6-f5 !

8 (Ke6-d5 Time = 0.010 s)
1.Ke6-d5 !

Popeye tests all possible moves, but only finds 3 legal moves.
 
   
(Read Only)pid=10373
(26) Posted by Andrew Buchanan [Saturday, May 4, 2013 17:46]

An algorithm may indeed take a phased approach to determining legal moves. So I wondered this morning if maybe the rules did too. This is why I'd checked this morning for uses of the word "possible" in the rules, in case it did have a specific definition in the context of move determination (Article 3). But it does not. The rules just specify the general capabilities of the pieces, and in 3.9 say that also you can't self-check. As far as the movement rules are concerned, moves are either legal or illegal. Any halfway house that an algorithm happens to use is entirely invisible to the rules - and quite right too!

It might just conceivably have been that the unusual use of the term "possible" in the rule for Draw By Repetition, indicated that something different is going on there. But surely there would be more than this tiniest of hints if that was the case, and fortunately, we have the specific Letter of the Law, both in the old version of the rules and the new changed version. It's quite clear that FIDE have thought carefully about what they want to have happen here, and that:

"Positions are not the same if a pawn that could have been captured en passant can no longer be captured in this manner."

I hope that finally I have managed to explain this in a way that allows Joost to say: "Oh! Aha!" :D

I would recommend that in the *next* version of the Laws, the word "possible" be replaced by "legal" in 9.2.

Thanks,
Andrew.
 
   
(Read Only)pid=10377
(27) Posted by Joost de Heer [Saturday, May 4, 2013 18:44]

We have a different opinion on what 'could have' means. For me it means 'possible', for you it means 'legal'. So no 'aha' yet.
 
   
(Read Only)pid=10378
(28) Posted by Andrew Buchanan [Saturday, May 4, 2013 19:27]; edited by Andrew Buchanan [13-05-04]

Hi Joost,

"Positions are not the same if a pawn that could have been captured en passant can no longer be captured in this manner."

I don't think that the interpretation of 'could have' is as wide open as you claim. Your position is untenable.

A pawn that 'possibly could have been captured' in your sense would always be possibly capturable in later versions of that position. A transition from capturable to non-capturable is only achievable when we are talking about legality, not 'possibility' in your sense.

So basically your argument implies that FIDE have deliberately inserted and now retained a rule which they know can never be triggered. It cannot be!

I don't think there's anything further I can add. I am just going to claim "checkmate"! :)

Thanks for the game!
Andrew.
 
   
(Read Only)pid=10379
(29) Posted by Joost de Heer [Saturday, May 4, 2013 19:47]; edited by Joost de Heer [13-05-04]

 QUOTE 

A pawn that 'possibly could have been captured' in your sense would always be possibly capturable in later versions of that position. A transition from capturable to non-capturable is only achievable when we are talking about legality, not 'possibility' in your sense.

Can I say 'huh?'. In what way did I ever imply this? The e.p. capture is only possible if the previous move was the necessary double-step. The whole point of 'possible moves' is that the e.p. move actually disappears from the list of possible moves, thereby creating a different position. And the whole point of the 'illegal e.p. capture' is to show that there's actually a difference between possible and legal.
 
   
(Read Only)pid=10380
(30) Posted by Andrew Buchanan [Saturday, May 4, 2013 20:21]; edited by Andrew Buchanan [13-05-04]

I'm tired here it's late - I'd forgotten which of the criteria you are choosing to apply before you say "OK, that's great work so far - now time for a cup of tea. Let's call the candidate shortlist the 'possibilities'" and then you will come back when fortified and apply the other criteria later.

Let's back up.

If e.p. was illegal for whatever reason, then it is reasonable to say that "the pawn could not have been captured". If you say that with an illegal e.p. the pawn "could have" been captured, then I think this requires a specialized explanation of the semantics here, and of other parts of the rule. Such an explanation is not to be found in the rules, suggesting that this strange interpretation was not the intended. Moreover, if your interpretation was the intended one, there are other more direct and unambiguous ways that the rule could have been expressed.

There is also the evidence of the state transition model in my letter to Guert, published in Chess Cafe, quoted earlier. He was satisfied that actual capturability was the right characteristic that determined position.

Enough of this nonsense - time for bed.
 
   
(Read Only)pid=10381
(31) Posted by Joost de Heer [Saturday, May 4, 2013 21:34]

I think we agree to disagree.

But let's go back to Gyorgy's idea, a bit simplified:
(= 2+3 )

White to move

I think we can all agree that after 1. Sc7+ Ke7 2. Sd5+ Ke8 3. Sc7+ Ke7 4. Sd5+ black can't claim a draw after 4. ... Ke8. But can white claim a draw after 5. Sc7+? Technically, the first and third repetition of this position are different (in the first position, after 1. Sc7+, black still has the right to castle), but there's no possible legal continuation in which black could castle. I say that these positions are indeed different, and I think the rules agree with me (based on the line 'castling rights are lost only after the actual move'). But is this really different from the 'illegal ep' case, where there is no legal possibility to capture ep?

And I think we can both agree that black has a correct claim after 5. ... Ke7 :).
 
   
(Read Only)pid=10382
(32) Posted by Andrew Buchanan [Sunday, May 5, 2013 03:42]; edited by Andrew Buchanan [13-05-05]

Please don't think me unfriendly Joost if I say that I don't think there is space for unresolved disagreement about interpreting the basic rules. We might agree to disagree about the conventions. We might agree to disagree about what constitutes a good or artistic composition. But if there is a hole in the rules then we should get back to the arbiters and determine what is the correct interpretation. Actually, I think this has already been done, and I reported the results earlier in this thread. You haven't answered this, or the other point made in my last post. :)

I am glad we are in agreement on castling. (Phew!) But yes it is fundamentally different from e.p. and your plea for analogy between the two is a rather weak argument. Castling rights are persistent through the game, while en passant is dynamic.

http://chessprogramming.wikispaces.com/En+passant also addresses this issue:

"The legality test should be best applied in making of the double pawn push, also considering updating Zobist keys to avoid dissimilarity of otherwise repeated positions if the first occurrence happened after a double pawn push with no en passant capture actually possible."

If you care, there is a long forum thread you can then click on at [2] which looks at hashing implications. But all the participants are in agreement what is the right rule for en passant repetition.

All the best!
Andrew.
 
   
(Read Only)pid=10383
(33) Posted by Joost de Heer [Sunday, May 5, 2013 09:20]

I really find it strange to separate right and legality in one case (ep) and not in the second (castling). Your claim that ep is a temporary right and castling is a persistent right is IMO weak too: castling is a temporary right too (you have the right until king or rook moves), the window of opportunity is just a lot larger.

Just look at these three cases:
- If castling is illegal in a position and the position repeats itself a third time with avoidable loss of castling rights then the position isn't the same
- If castling is illegal in a position and the position repeats itself a third time with unavoidable loss of castling rights then the position isn't the same
- If ep is illegal in a position and the position repeats itself a third time with (inherently unavoidable) loss of ep rights, then the position is the same

I really cannot see why these three cases are different. For me, all these cases should lead to 'position isn't the same'. If you want to make a differentiation, then, to me, it would be more logical to make the second case (unavoidable castle right loss) equivalent to the ep case, and not to the avoidable castle loss case, as is now the case.
 
 
(Read Only)pid=10384
(34) Posted by Sarah Hornecker [Sunday, May 5, 2013 17:27]

 QUOTE 
And why would an arbiter give the draw in any case if they are sane?

Because this is an unique case in chess. So this is a case that must be decided by a judge. It is open to appeal any way, but the appeal is more likely in case of a draw.

If I follow the words, then analogous to en passant the castling is listed below. So if we agree that en passant does not count if it is permanently impossible then castling also can not count if it is permanently impossible, regardless of whether rook or king have moved yet.

So in my opinion the only correct decision is to make a draw, and hope the players will appeal.
 
   
(Read Only)pid=10387
(35) Posted by Andrew Buchanan [Sunday, May 5, 2013 19:25]

I really don't care which way the rules are. What I insist on is that the answer be clear. "Agree to disagree" is the one outcome which is unacceptable. These are the Rules of Chess, for gosh's sake! :) They should be definitive (but see my footnote). The Story of Solomon's Judgment has just floated up in my mind for some reason... :)

A few years ago, I thought that the rules were as Joost suggests. When I realized I was wrong, that cooked irretrievably a problem that I'd composed with Guus Rol. But what can I do? I think the rules are pointing in a clear direction, so that's what I'm arguing for. If I thought they were pointing in the direction that Joost claims, then I would argue for that. And then I would be able to rehabilitate that old composition.

But that's not what I honestly think. Castling and e.p. are treated a bit differently in the rules. If Joost wants to campaign to change them to make them more "consistent", then God Bless Him go for it. But that's a future exercise. With respect, I think he is confusing the rules as he would like them to be with how they are today. I just want a common understanding of the rules today.

Footnote: Chess Tournament Guidelines are looser than the Chess Rules and cannot hope to be definitive. One of the problems with the Laws is that the two documents (Rules & Guidelines) are mixed together - but that's another story. To my mind, the determination of repeated position is definitely a Rule - the mechanism for claiming a draw under these circumstances is a Guideline. I think we ought to tease these apart.

All the best,
Andrew.
 
   
(Read Only)pid=10388
(36) Posted by Dupont Nicolas [Monday, May 6, 2013 06:10]

It is not very important for the discussion here, but let me recall that “possible moves” is the disjoint union of “legal moves” and “auto-checks”. See http://christian.poisson.free.fr/problemesis/glossaire.html

To my eyes this is clear that the sentence “if a pawn that could have been captured en passant” means more precisely “if a pawn that could have been captured en passant via a legal move” (and not “via a possible move”).

The castling framework addresses new questions. For example, what are the “castling rights” in a position where the moving-side has no other option than destroying this right?

There is some similarity here with dead reckoning, dear Andy! And also with the pending question of the goal in a self-problem: is it already achieved at the pre-terminal move, i.e. when white has forced black to reach the goal, or do we have to wait for black’s answers to claim that the goal is reached?

For me there are no logical answers to those questions, this is a matter of convention. But conventions have to be consistent one respectively to others, i.e. to share the same “philosophy”. As far as the dead reckoning rule asserts more or less “if you have no other option than dying, you are already dead”, it seems consistent to also claim “if you have no other option than destroying your castling rights, they are already destroyed”.
 
   
(Read Only)pid=10393
(37) Posted by Andrew Buchanan [Monday, May 6, 2013 20:03]; edited by Andrew Buchanan [13-05-06]

Dear Nicolas,

 QUOTE 
It is not very important for the discussion here, but let me recall that “possible moves” is the disjoint union of “legal moves” and “auto-checks”. See http://christian.poisson.free.fr/problemesis/glossaire.html

I didn't know any problemist had attempted to formalize the word "possible". It's naughty to try to formalize a generic word, because it bequeaths confusion to all future generations. A much better term for this concept is "psuedo-legal" in the excellent site http://chessprogramming.wikispaces.com/Pseudo-Legal+Move.

And of course the authors of the Laws don't read Problemesis to get their definitions. We don't read in the Laws: "The first player to execute an Umnov Pronkin must replace each nightrider with a royal princess." :)

 QUOTE 
To my eyes this is clear that the sentence “if a pawn that could have been captured en passant” means more precisely “if a pawn that could have been captured en passant via a legal move” (and not “via a possible move”).

Hurray!

 QUOTE 
The castling framework addresses new questions. For example, what are the “castling rights” in a position where the moving-side has no other option than destroying this right?

The new rules are clear.
Law 3.8.b.1: The right to castle has been lost:
(a) if the king has already moved, or
(b) with a rook that has already moved.
Law 9.2.b The castling rights are lost only after the king or rook is moved. (e.g. Umnov Pronkins won't do it...:)

 QUOTE 
...the pending question of the goal in a self-problem: is it already achieved at the pre-terminal move, i.e. when white has forced black to reach the goal, or do we have to wait for black’s answers to claim that the goal is reached?

It depends on the goal. In most cases we wait.

 QUOTE 
For me there are no logical answers to those questions, this is a matter of convention.

Logic will take us a long way in any specific situation. My personal taste is to be extremely parsimonious about conventions, and only apply them in specific cases where there is inherent ambiguity. The vagueness you sense is just because your question is over-broad.

 QUOTE 
But conventions have to be consistent one respectively to others, i.e. to share the same “philosophy”. As far as the dead reckoning rule asserts more or less “if you have no other option than dying, you are already dead”, it seems consistent to also claim “if you have no other option than destroying your castling rights, they are already destroyed”.

In life, I think there is a real art in knowing how far to impose "philosophical" consistency. But here, dead reckoning and castling rights are both covered by good rules, so no conventions are needed.
 
   
(Read Only)pid=10395
(38) Posted by Dupont Nicolas [Tuesday, May 7, 2013 01:16]

Dear Andy,

Of course I do agree, lesser we have to use "conventions", better is structured the theory. But when some conventions are really needed, it seems important for them to go in the same way. Let me give you a practice example, namely a white pawn standing on the first row. This is possible under numerous fairy conditions but it exist at least four different conventions for such a "delocalized pawn":

- wPa1 is illegal (e.g. Sentinel chess)
- wPa1 is paralyzed (e.g. Annan chess)
- wPa1 can move on a2, capture on b2 and then becomes a normal pawn (e.g. Circe Parrain)
- wPa1 can moreover move to a3 or a4 (e.g. Einstein chess)

I’m far from being convinced that such diversity is good for fairy chess, to my eyes this is much more a source of trouble and an illustration that fairy theory is not well-structured. It would be one of the missions of a fairy Codex to stipulate a unique convention for delocalized pawns (i.e. to enlarge orthodox rules), even if it would bother some inventors and made some problems unsound.
 
 
(Read Only)pid=10396
(39) Posted by Geir Sune Tallaksen Østmoe [Tuesday, May 7, 2013 01:49]

 QUOTE 
Because this is an unique case in chess. So this is a case that must be decided by a judge. It is open to appeal any way, but the appeal is more likely in case of a draw.


Okay, I see what you mean now. I read your first post too quickly, so I missed the point you made of "inviting" an appeal. Still, the wording of the rules looks quite clear to me on this point, so I don't see why we would need a judge to decide.
 
   
(Read Only)pid=10397
(40) Posted by Geir Sune Tallaksen Østmoe [Tuesday, May 7, 2013 02:12]

Regarding the logic of the rules, I think there could simply be practical reasons for treating castling and en passant differently.

Regarding en passant, it is quite straightforward to determine whether there are any legal en passant captures in a position, and I think the average arbitor will find it more natural to consider only legal captures and not "captures which would have been legal if the king hadn't been in check."

When it comes to castling, it is straightforward to determine whether the kings and rooks have moved previously, while it is not necessarily as straightforward to see whether it will ever be possible to castle in the future. Consider the following example:

(= 4+4 )


In this position, Black can never castle.

(= 5+4 )


This is a different story. White can promote to a knight and sacrifice it on b4. Then Black gets a passed pawn which can promote and then capture White's pawn on h7, enabling Black to castle.

These considerations are unneccesary under the current rules: as long as Ke8 and Rh8 haven't moved, Black is considered to have castling rights.
 
 
(Read Only)pid=10398

Read more...
Page: [Previous] [Next] 1 2 3

MatPlus.Net Forum General Law 9.2: Draw by Repetition