[Retros] rights & ocassions /answering Andrew

[Andrew Buchanan]

Dear Roberto,

This issue is not *quite* as academic as angel sex - it does affect the
soundness of certain chess compositions. :o) 

The chess programming position, which seems to pass there without
controversy, is given at:
http://chessprogramming.wikispaces.com/Repetitions

Now, a SPOILER ALERT :o) for those enjoyably busy reading through the
matplus discussion of this issue, located at:
http://matplus.net/start.php?px=1399373473&app=forum&act=posts&tid=1235
The climactic denouement of the thread is when I quoted senior FIDE arbiter
Stewart Reuben in May 2013 [my annotations added in square brackets here]:

"Geurt Gijssen and I differ over the repetition rule and castling. His view
will continue to hold sway for the new version [of the Laws] (probably now
no change [in that draft version] until 1 July 2013 [when it will be
ratified, going live in 1 July 2014]).
Position Qa4 Ka1. ke8, rh8. The laws state that the right is not lost to
castle until the king moves. Thus 1...kf7 2 Qf4 ke8 2 Qa4 is a new position.
I think that is wrong. We can look into the future and know black will never
castle.
But Rh2 Pg2 Kf1. ka2 pf4. 1 g4+ Kh3 2 Rh3+ Ka2 3 Rh2+ is the same position.
Black could never capture en passant. 
To my mind this is illogical. The two possibilities [i.e. castling & en
passant] should be the same.
But it is the law and I have no idea whether it has ever arisen. So I do not
care too much."

So Stewart is unhappy the fact that castling and en passant approach things
differently. I think the position that castling *rights* are what counts
will survive until 2017. However, then those "troublemakers" :o) who want to
change the criterion for castling may have a champion in Stewart Reuben.
Stewart is a very nice guy, but is focused on the game, and I don't think he
feels any particular duty to problemists.

OK, so why should castling rights rather than opportunity be the driver?
(1) It's the status quo. Why change such a tiny thing, which will force
programmers to alter their core code, if they want to continue to capture
the exact rules of chess (as many will)? (And some problemists may be
inconvenienced as well.)
(2) It's simplest. Castling rights are just 4 bits of information, 4
characters in FEN, which change only when a king or rook moves for the first
time. Easy to compute; easy to consult.
(3) According to Guus Rol, there was a FIDE ruling at the time of Fischer.
(It would be nice to see that.)
(4) Guus also points out that dynamic interpretations may lead to
undecidable positions. Although he invokes retro strategy and proof games,
remember that we are talking about games of chess, where none of these retro
concepts are available. So his argument is even stronger.
(5a) An interesting question is: why can't we just use the FEN notation to
represent the state for en passant? Recall: FEN records the en passant
target square in algebraic notation. If there's no en passant target square,
this is "-". If a pawn has just made a two-square move, this is the position
"behind" the pawn. This is recorded regardless of whether there is a pawn in
position to make an en passant capture. (Clearly the rank number, 3 or 6, is
redundant.) But it would be absurd to use the FEN square as an indicator of
en passant capability - there may be no neighbouring pawn to make the
capture. But once we've opened the door to that concern, then why stop
there? One king may be in check, or there may be direct or indirect pins. So
rather than invent some halfway house, where SOME but not ALL necessary
criteria are checked, the approach taken is that whether a position is
different or not depends on whether the pawn can capture legally.
(5b) So to revert to castling: if we feel that there is need for
"consistency" between castling convention and en passant approach, and the
en passant approach is hard to modify, then why not change the castling
approach? Well, *which* castling approach? We could either look at whether
the rooks can castle right now, or whether they could castle some time in
the future? This question doesn't come up in en passant, which is
now-or-never.

So that's 5 good reasons to leave the current rule well alone.

I have just seen a new tomelike email from Guus arrive in my inbox, so I
think I will send this one.

Thanks,
Andrew.

-----Original Message-----
From: retros-bounces at janko.at [mailto:retros-bounces at janko.at] On Behalf Of
raosorio at fibertel.com.ar
Sent: 06 May 2014 06:10
To: retros at janko.at
Subject: [Retros] rights & ocassions /answering Andrew

Hi Andrew,

Long time without these "sex of the angels" discussions.


I hope you find the MatPlus discussion useful. To my mind, it was
conclusive, and filled the key hole in the following strategy
*********************************************************************
I find it useful indeed. I'm afraid I did not read it carefully enough to
realyze that conclusive side.


For example the chess programming community have converged on a 
standard set of rules to judge any engine, and this community agrees with 
the arbiters as to how to work castling and en passant in the context of 50M

& 3Rep.
*********************************************************************
this is really new for me. Could you please indicate where to find this
standard
set of rules? All of the present discussion should start from there.


There is a Golden Principle in this, that conventions *only* come in to 
remove areas of ambiguity where we cannot deduce using information in the 
stipulation and logic what must have happened. Any convention which does not

respect the Golden Principle is worded in too strong a way, and must be 
toned down because it's taking us into Fairy Chess.
*********************************************************************
Simple and common sense. 

I think the codex writers had a better lunch, with more wine, than the 
laws writers.
********************************************************************
Yes, but I suspect it was at the same restaurant.

Best wishes,
Roberto

Hi Roberto, 

I hope you find the MatPlus discussion useful. To my mind, it was 
conclusive, and filled the key hole in the following strategy. 

There may be always fun ways to interpret wordings of rules. But there is 
surely a vanilla set of rules which is how they are intended to be. This 
intention is not so difficult to discern, because the rules of chess are so 
simple. For example the chess programming community have converged on a 
standard set of rules to judge any engine, and this community agrees with 
the arbiters as to how to work castling and en passant in the context of 50M

& 3Rep. I agree with this general interpretation. This allows us to play 
chess unambiguously. This programming interpretation incidentally bounds the

scope of chess: it excludes touch move, the clock, responses to errors etc. 
Basic robust vanilla chess. 

Now, and only now, do problemists come into the discussion. There is a 
minimal set of conventions which is then needed to make the transfer from 
vanilla game to problem. These fall into two areas: 
(1) history. Just looking at a diagram is not sufficient to see what moves 
may be legal, due to who's move, castling, ep, number of times position has 
repeated, number of moves since last . We have now reasonably comprehensive 
understanding of how conventions will work in this area. 
(2) decision-making. what can players do? stipulation type (#, h#, s#...) 
gives a lot of information, but also decisions about whether draws of 
various kinds would be proposed or rejected. 
There is a Golden Principle in this, that conventions *only* come in to 
remove areas of ambiguity where we cannot deduce using information in the 
stipulation and logic what must have happened. Any convention which does not

respect the Golden Principle is worded in too strong a way, and must be 
toned down because it's taking us into Fairy Chess. 

A few random remarks to finish... 
- I remember encountering Sergio Orce in the old days in france-echecs.com. 
I hope he is well, wherever he is. 
- I think the codex writers had a better lunch, with more wine, than the 
laws writers. 
- Actually, it's not a problem with the writing, but the subsequent editing.

Nobel Laureate Ernest Hemingway said: "The first draft of anything is s**t."


Thanks & all the best, 
Andrew.
_______________________________________________
Retros mailing list
Retros at janko.at
http://www.pairlist.net/mailman/listinfo/retros