Checkmate In Zero

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Sven
Posts: 4052
Joined: Thu May 15, 2008 9:57 pm
Location: Berlin, Germany
Full name: Sven Schüle

Re: Checkmate In Zero

Post by Sven »

hgm wrote:However, playing with two Kings does not violate any FIDE rule. AFAIK there are no rules for setting up positions.
The FIDE rules clearly define the game of chess as a game where each player always has exactly one king. This follows from
- the definition of exactly one starting position where each side has exactly one king,
- the promotion rules that disallow promoting to a king,
- the rule that disallows capturing a king, and
- the various places where the term "the king", "his king" etc. are used, e.g. in the definition of checkmate, which unambiguously confirms that the rules are explicitly made for board positions with exactly one king of each color.

Whenever you set up an arbitrary chess board position, on a physical board or within a chess program, you are immediately faced with the question whether that position is reachable from the official starting position with a sequence of rule-conforming moves. There are obvious and non-obvious cases. Obvious cases where the answer is "unreachable" include those where the number of existing pieces of a certain type is already "out of range", e.g. more than 10 white knights or more than one black king. A couple of other simple indicators for obviously "unreachable" positions exist, e.g. pawns on ranks 1/8 or "out of range" piece counts when also considering the maximum possible number of promotions (= number of missing pawns) of a color. For the non-obvious cases you often need a retrograde analysis to decide about "reachability".

The question whether a chess program shall allow setting up a position that is definitely unreachable from the official starting position of the game must be answered individually for each engine. Opinions differ a lot, and I think everything between strict refusal of all obviously non-conforming positions (as in Gaviota for instance) and permissive acceptance of almost all positions you can think of (as in Micro-max for instance) can be acceptable. The permissive way can be seen as a kind of "FIDE rules extension" by allowing starting positions that are unreachable from the official one.

What I consider as incorrect, though, is to say that "playing with two Kings does not violate any FIDE rule". Furthermore, allowing to play with a number of kings different from 1 for any color immediately creates the need to redefine the semantics of the "checkmate" rule, which is a topic for which there is no accepted standard to my knowledge. I would even say that playing with two kings should be disallowed since that would be a game different from chess due to a different strategy (e.g. "give checkmate to ALL kings of the opponent").

Sven
User avatar
hgm
Posts: 27796
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Checkmate In Zero

Post by hgm »

I consider 'reachability' another issue than 'legality', and I don't think anything useful can be achieved by erasing the difference between these two concepts. I would not consider initial positions of shuffle Chess as 'illegal' according to FIDE rules.
Sven
Posts: 4052
Joined: Thu May 15, 2008 9:57 pm
Location: Berlin, Germany
Full name: Sven Schüle

Re: Checkmate In Zero

Post by Sven »

hgm wrote:I consider 'reachability' another issue than 'legality', and I don't think anything useful can be achieved by erasing the difference between these two concepts. I would not consider initial positions of shuffle Chess as 'illegal' according to FIDE rules.
1. Shuffle chess is not subject to FIDE rules of standard chess. The rules of shuffle chess extend the FIDE rules by allowing more "legal" initial positions.

2. Which "unreachable" position would you consider as "legal" in standard chess when strictly following the FIDE rules?

For me the question whether "reachability" and "legality" are different comes down to the question to which degree you allow "unreachable" positions and therefore extend the rules. If you are strict then there is no difference, if you are permissive then there is. Just as I already wrote before. So I accept your personal decision, but I don't think it can be applied globally.

Sven
PK
Posts: 893
Joined: Mon Jan 15, 2007 11:23 am
Location: Warsza

Re: Checkmate In Zero

Post by PK »

Fide rules specify that if a game has started form an illegal position and no player issued a complaint during the first three moves, then the game continues. This way one of my friends, a 1900+ player, ended up on a receiving end of a vicious attack carried out with the help of opponent's same coloured bishops. The position was unreachable, but the game has been scored and rated, and therefore, in a sense, legal.
Sven
Posts: 4052
Joined: Thu May 15, 2008 9:57 pm
Location: Berlin, Germany
Full name: Sven Schüle

Re: Checkmate In Zero

Post by Sven »

PK wrote:Fide rules specify that if a game has started form an illegal position and no player issued a complaint during the first three moves, then the game continues. This way one of my friends, a 1900+ player, ended up on a receiving end of a vicious attack carried out with the help of opponent's same coloured bishops. The position was unreachable, but the game has been scored and rated, and therefore, in a sense, legal.
"In a sense", maybe. But the fact that the FIDE rules describe more than just what comprises a legal chess game, e.g. how to deal with deviations from the rules, does not make an illegal game itself legal only because it has been scored and rated. The position from which that game started did not conform to the rules, that can't be discussed away. But there are good reasons for the FIDE rules to define what should happen when rule violations are not discovered. I leave it up to you where you want to place that on the scale between "strict" and "permissive" that I mentioned before.

I have no problem to accept that someone wants his program (or other programs) to allow setting up such an "unreachable" position, as long as we continue to call it what it actually is: "illegal" when strictly following the FIDE rules as far as they describe what a correct game is, and maybe "legal" from a different viewpoint where starting from unreachable positions is allowed under certain conditions.

Sven
User avatar
hgm
Posts: 27796
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Checkmate In Zero

Post by hgm »

Seems better to call it just "unreachable" rather than legal and illegal at the same time...
User avatar
lucasart
Posts: 3232
Joined: Mon May 31, 2010 1:29 pm
Full name: lucasart

Re: Checkmate In Zero

Post by lucasart »

Christopher Conkie wrote:[D]8/8/8/3Kk3/3kK3/8/8/8 w - - 0 1

Or......

[D]8/8/2K2k2/8/8/2k2K2/8/8 w - - 0 1

You may say that these positions are not chess......but we could say the same about some programmers.

:)

Chris
And what's your point ? That these illegal positions crash chess engines ?

If the position is invalid (several kings on the board, for example) you should assume the behavior of a chess engine to be undefined:
- It can reject it
- crash
- spit out any kind of non sense

shit in -> shit out...
Theory and practice sometimes clash. And when that happens, theory loses. Every single time.
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Checkmate In Zero

Post by zullil »

lucasart wrote:
Christopher Conkie wrote:[D]8/8/8/3Kk3/3kK3/8/8/8 w - - 0 1

Or......

[D]8/8/2K2k2/8/8/2k2K2/8/8 w - - 0 1

You may say that these positions are not chess......but we could say the same about some programmers.

:)

Chris
And what's your point ? That these illegal positions crash chess engines ?

If the position is invalid (several kings on the board, for example) you should assume the behavior of a chess engine to be undefined:
- It can reject it
- crash
- spit out any kind of non sense

shit in -> shit out...
I believe Chris is suggesting that, by studying how an engine responds to such "crazy" positions, one may attempt to determine its origin(s).
User avatar
hgm
Posts: 27796
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Checkmate In Zero

Post by hgm »

lucasart wrote:If the position is invalid (several kings on the board, for example) you should assume the behavior of a chess engine to be undefined:
Actually that is only true for UCI engines. WB protocol prescribes that the engine should generate an error message when presented with positions it cannot handle, and reject any input move as 'Illegal move' when such a rejected position is not yet replaced by a new one.
Christopher Conkie
Posts: 6073
Joined: Sat Apr 01, 2006 9:34 pm
Location: Scotland

Re: Checkmate In Zero

Post by Christopher Conkie »

zullil wrote:
lucasart wrote:
Christopher Conkie wrote:[D]8/8/8/3Kk3/3kK3/8/8/8 w - - 0 1

Or......

[D]8/8/2K2k2/8/8/2k2K2/8/8 w - - 0 1

You may say that these positions are not chess......but we could say the same about some programmers.

:)

Chris
And what's your point ? That these illegal positions crash chess engines ?

If the position is invalid (several kings on the board, for example) you should assume the behavior of a chess engine to be undefined:
- It can reject it
- crash
- spit out any kind of non sense

shit in -> shit out...
I believe Chris is suggesting that, by studying how an engine responds to such "crazy" positions, one may attempt to determine its origin(s).
Thank you Louis, that is exactly what I meant and you described it better that anyone here, where I might add we are supposed to have the crème de la crème of chess intelligence.

:)

Chris