I hate the threefold repetition rule

FIDE and others constantly mortifying themselves about such things is indescribably anal. No offence to present participants of course.

A good way to look at it is, it's not the 3 fold repetition is taking the easy way out, like escaping with a free draw, but it is the result of being in a position where the only good move is to check the king, and in the next position once again the only good move is to check the king, and so on, each position requiring a check to avoid getting mated for example. This likely results in positions being repeated obviously, but it's not the like the goal is repetition as an end in itself. The goal is to make legal moves for as long as possible without getting mated.

The interesting point is that FIDE rulings related to dead positions are much more relevant to the domain of composition because:

1. The Codex has no DP-rule of its own other than stating that the FIDE-rule only applies to the retro-composition type - which leaves its concepts and interpretations to inheritance from FIDE as with most chess rules. This is the first time where it is made clear that FIDE's DP-rule handling is fundamentally different from the one assumed in the composition environment. Without either side having taken an official stand, having identified and clarified the difference or having justified why that difference exists.
2. This is a major blow to the composition field which already produced a significant body of good compositions on the reasonable expectation that "no way to checkmate" does mean "no way to checkmate".
3. The distinction between FIDE and WFCC-Codex cannot be brushed off as a distinction between the 3R/50M set and the 5R/75M set since the DP-ruling does not attach itself to any particular termination rule (other than checkmate). Its evaluation algorithm for both termination sets should therefore be identical apart from the "numbers". 

I predict a war.

I agree with the arbiters.

It's quite simple for over the board play. The FIDE Laws define DP in terms of legal moves, which themselves are defined very clearly. There is no term for playable or unplayable moves after the action of DP/75M/5Rep rules. Why would there be? From a FIDE perspective the game is over.

Clearly, one cannot refer to "legal and playable"/"legal but unplayable" moves simply as "legal"/"illegal", or the definition has become circular. Of course one can use the terms "legal"/"illegal" casually in this altered sense, but when it comes to working out the exact rules, we must be precise.

Since DP is defined in this way, it has no visibility over 5Rep or 75M.
Now what effect do we want to see? Although it might seem desirable to truncate overlong games by having DP see 75M, I suspect that the amount of hypothetical reasoning and argy-bargy required late at night would take more time than just playing out the moves.

I thought you might say that! Simply because you never understood "legal moves". But fortunately you produced the example that brings your viewpoint down yourself! The SPG 12.5 "game over" is now invalid since the FIDE judges will tell you that your diagram is not "game over" until the third repetition is actually produced - under their interpretation of DP. You actually agree to both sides - whichever is convenient!

Please calm down

I merely spoke of FIDE Laws over the board. "Part Deux" of all this must be compositions, which is a different case. I don't "agree to both sides", rather I want to find the best place for a fence.

I went through your problems in PDB yesterday, and I couldn't find one in which the last move was forced, but maybe I'm being dense. In both P1004354 & P1011937 the final move is optional. Please remind me of the detailed situation, because I don't want to opine on DP/3Rep for problems until I do.

I did see your sublime https://pdb.dieschwalbe.de/P1080442 3Rep/castling/RS which was missing a detailed solution so I added one.

There is no "self-reference problem". Just read "legal move" as it is defined in the Laws and all is good over the board. Algorithm:

1. Do I have a legal move?
2. - If no then the game is over by mate or stalemate depending on whether there's a check.
3. - If yes then is the game over by DP, 75M or 3Rep?

DP, 75M & 3Rep are all completely separate considerations. DP just looks at the tree of legal moves, which has nothing to do with 75M/3Rep termination.

I agree some clarification is needed for the composition world, but over the board it's fine

Won't work. Read especially point 3 of my post above the previous one in which I anticipated this trick as well. This is precisely the point where you cannot make a distinction between FIDE games and Codex. The Codex has no different DP-rule or different official interpretation or different understanding of legal moves - only a different DP application domain (obviously).

You will not find the fence you're looking for except by permanently diverging from FIDE on fundamentals.

I don't know what's in the PDB but I won a 2nd PB prize behind Caillaud with a 3-rep PG41.5 problem with a 12-move repeat cycle. The last move of that cycle is forced. Year should be around 2009 but I don't know exactly.

OK, let me spoon feed you: https://pdb.dieschwalbe.de/P1011937 42. Tg8+= is the final move, but it's not forced. **Please explain your exact issue**, without skimming off into generalities.

Rules for chess problems must always have minor differences from FIDE Laws (1) to automate human decisions, and (2) provide necessary game history. We do have a problem notion of "retro" which is like a mini-reality of its own already, and that's fine. I am looking to understand DP/3Rep/50M in that context.

I am flabbergasted! Can't see what's wrong with the problem any more. Of course it is also DP but that does not change the fact that it looks like a valid drawn composition. Have to sleep on it! Had no clue it was from 2003. Btw, I only visit the the PDB once about every 10 years and then I have to restudy the instructions.

Arguments around "automating human decisions" carry no weight since the 5R and 75M moves have precisely the same characteristics. Actually, that will prove to be their main application. Computer-interfaces can now replace humans in automatically terminating dead positions since absolutely nothing could happen (not even flagging, or resigning or checkmate) to change the outcome after 5R or 75M lines were crossed.

I suddenly remember something about my problem. Something to do with a dual repetition cycle which loses a tempo which comes back by relying on the certainty that the ..Bf8 move need not be played due to DP! Just as in your PG12.5 but with a different history.

Hi @arisktotle,

Thank you. I will enjoy going through this problem, and posting a detailed solution. My own thinking has evolved in the last few years, so please bear with me. I am glad there is a coherent picture for otb chess. That's a good foundation. Now onwards to problem chess.

* I don't think problems ever become unsound because of rules/conventions changes/clarifications. This is what the "Golden Age" tag is for, to allow us as proper anthropologists of our hobby to track the context in which problems were conceived.

* OTB, humans choose moves, resign, make claims, offer draws & respond to offers. All of that gets automated in problems (e.g. in directmates, white attempts to mate in n, black defends).

"DP just looks at the tree of legal moves, which has nothing to do with 75M/3Rep termination."

Not quite as the FIDE laws are currently printed. It also looks at whether series in the the tree (which is not the game tree - serialmover mates are also series of legal moves, for example) can be played.

5.2.2 The game is drawn when a position has arisen in which neither player can checkmate the opponent’s king with any series of legal moves. The game is said to end in a ‘dead position’. This immediately ends the game, provided that the move producing the position was in accordance with Article 3 and Articles 4.2 – 4.7.

I know you're up to almost any challenge, but I challenge you to find a series of legal moves with which either player can checkmate the opponent's king under FIDE competition rules from this (partially specified) position.

If you believe it can't be done then you have to accept that the above blue paragraph tells you straightforwardly that the game is terminated by DP under FIDE competition rules. (It may also be terminated by other rules, but that's by the by.)

The self reference comes from the fact that the condition that a player can checkmate with a series of legal moves constrains the series to be playable under the rules, but, without any caveat, "the rules" includes the above blue paragraph.

That's all without any mention of the WFCC Codex.

Note that the 75 move rule says "9.6.2 any series of at least 75 moves have been made by each player without the movement of any pawn and without any capture. If the last move resulted in checkmate, that shall take precedence."

The key words are "have been made", NOT "will be made".

Also note that 9.6 is not part of articles 3 and 4 and thus it is not part of the dead position rule.

From a practical standpoint it is safer to avoid the correct ending of a game to dependent on an arbiter's analysis ability. I've seen arbiters looking at White Kd3, Rf2 vs Black Ke1 thinking that White's fastest mate is playing Ra2 Kf1, Ke3 Kg1, Kf3 Kh1, Kg3, Kg1 Ra1# - totally missing the simple two-mover (such an arbiter might call a draw if the position was on move sequence 72 even though there is a forced mate on move sequence 74). If an arbiter has to carry around an analysis tool and put each position into it looking for a possible forced 75-move draw then that takes a lot of the arbiters time that needs to be spent monitoring games to make sure they comply with the other rules - such analysis only being practical in smaller round robins where the players are likely already strong enough to make the 50-move claims that don't require arbiter intervention.

I sympathize with "Golden Age" classifications as long as their "Golden Age" actually exists(ed). I am convinced (and I predicted) that FIDE always believed that 5R and 75M would be "invisible" to DP and probably believed that stalemate is invisible as well. But they never thought this through and never incorporated it in the rules. And now that they betrayed their inner thoughts to @jetoba, it is still not official in the texts. Which raises the question on when exactly that "Golden Age" exists(ed) for the DP-rule interpretation that agrees with your PG12.5 and when the other "Golden Age" exists(ed) which ratifies my 2003 PG41.5 which requires invisibility of 5R/75M draws to the DP-rule. "Golden Ages" can only be assigned when the rules were clear to everyone.

The peculiar situation is that the introduction of the 5R/75R drawing automats in the FIDE laws - effectively the same as the already existing 3R/50M automats in the Codex with different numbers - brought to light an ambiguity in the FIDE laws on DP. Those DP-rules were not changed and only recently received some "clarfication". The Codex has no DP-rule of its own and is now faced with a FIDE-clarification which has gone opposite to the one adopted in the Retro-community. Both choices were made to fill the same "law void" as this situation existed and required resolution for both communities!

Yes, FIDE has to deal with all those "OTB" (which today includes digital enviroments) issues which sets it apart from the problem-community. But what happened here is that FIDE actually eliminated the option of human choice in 5R and 75R moving more in the direction of the problem-environment and less dependent on human interaction. And BANG! the 2 environments clash, not on human interaction, but on visibility between interacting laws!

Arisktotle, note that I was only consulting experienced IAs about OTB, not about problem-solving.

I know, and all of us know! The point is that the FIDE laws automatically flow to the orthodox composition environment (unless specifically amended like 3R and 50M). Not by necessity but by convenience. The WFCC would hate to define castling moves which they can inherit from the FIDE laws. There is an issue of synchronization points, like which FIDE laws version is invoked for which Codex version. However that is not an issue here since the DP-rule was never amended only clarified.

The DP-rule is only about the future, never about what has taken place. The DP-rule has no interest in any other rule except in the definition of checkmate. All it does is continue the game by the rules in all the ways it could have been continued by the players to find a "checkmate" and rule according to its findings. It could not have found a checkmate beyond the 5th occurrence of the same position and would base its conclusion (among others) on this finding.

Note that all complexity around the DP-rule - like the relationship with 5R and 75M - comes from FIDE, not from the compositions community. It only suggests DP-analysis by "playing on by the rules" and now FIDE says "Ho, ho, but not by the 5R/75R rules!!" MARattigan explained that but apparently it didn't stick.

Many decades ago, when I first read about the dead rule(s), my first thought was "THIS IS GONNA GIVE TROUBLE, especially for compositions". I actually remember thinking that. I have not been disappointed in my predictions!

It happened in FIDE - just as it also happens in the WFCC - because you can't develop a rule system for a reasonably rich abstract game without mathematical scrunity. You always need the User Manual to communicate it to the community and the System Manual to provide formal definitions, design concepts, mathematical soundness checks and intricate algorithms.

The DP-rules go wrong on line 1 not understanding you need slightly advanced mathematical logic (like RAA = Reduction Ad Absurdum) to avoid contradictions, circularity and undecidables for this issue. Of course, that's for the System Manual, the User Manual need not be that precise.