Timeout rules

I can't agree more on that last point, especially the "too much work" part :)

Yeah, I think it's fine. It's such a rare occurrence that a checkmate can be delivered with king+knight (i.e. the opponent basically has to blunder themselves into a smothered mating position) that it isn't really worth the effort to correct the "error."

The only real option would be to add an adjudication feature, in which a game could be sent to a mod along with the proposed mating line (no matter how suicidal for the opponent) that is theoretically possible to prove the win and it could be adjudicated as such. However, this too might almost be too much work, as how often would a game truly need adjudicated vs. how often people would send unnecessary adjudications to mods? However...said adjudication feature could be used against suspected cheaters as well, rather than spamming the forums with the weekly witch hunts that hurt the site, imo. Just a thought.

to follow the rules exactly, you'd have to do a search in all future positions that can be reached from the timout position within a maximum depth. if checkmate can be reached, there's no draw.
the maximum depth should be findable in some chess book.
this search wouldn't take that much CPU like the normal AI stuff, because both player aim to get checkmate against the timeout'ed player.

alternatively, you could allow the non-timeout'ed player the hint the computer how a checkmate would be possible, if they know how to, then they win, otherwise a draw.

or, just clearly state that we use simplified rules.
...the official rules aren't computer fiendly anyway :-(

legend, the reason why endgame tablebases wouldn't work is because they only tell you if the game itself is "supposed" to be a draw, not if it's impossible totally for either side to win (an automatic draw).

thibault, if what you said was true about scoring games as draws if the other player times out, that rule of "Insufficient Material" would have been an instantaneous draw, immediately ending the game, and not waiting for a time out.

Time out by the opposing side is always a loss and a win for the enemy. Any draw rules must be applied BEFORE timeout.

The problem is indeed, detecting positions where it is 100% impossible for EITHER side to checkmate.
Which only includes King+Knight if absolutely no other pieces besides the enemy King are in the game.

if your time runs out, you lose only if your opponent could checkmate you, assuming your worst play - otherwise it's a draw. that's the chess rules.

but these rules can only be obeyed if the computer finds some way to detect whether a checkmate against you would be possibly (assuming your worst play), so it has to do a BFS.

"The chess rules" is that the game is IMMEDIATELY ended when checkmate cannot possibly be reached.
http://www.e4ec.org/immr.html

So, no. Insufficient material automatically ends the game immediately before you have a chance to use time-out as an excuse.

And the only case I can think of, where one side is potentially able to checkmate the other, but not the other, (in which case the above rule would not apply, and you could be right about time-out drawing it if the one timing out is the only one who could checkmate theoretically) is if one side has only a King left and nothing else.

All it takes is 1 King + 1 Knight, or 1 Pawn that might never even promote, or 1 Rook etc., versus 1 King + any number of forces capable of checkmating you.
So you cannot draw by timeout unless the other side has nothing but a King, because otherwise the game would have instantly been declared a draw before that by other rules.

>>"The chess rules" is that the game is IMMEDIATELY ended when checkmate cannot possibly be reached.<<
that's right, but not the point;

assume this:

White material: rooks at b1 and c2, king at c1
Black material: king at a8

assume it's white's turn and assume that white times out before doing rook c2->a2,

->then it is a draw, because black couldn't win anyway.

So the implementation for fixing this is very simple.

on timeout,
if (other side has nothing but a King) game == draw;
else { game == loss } // for who timed out

The only real challenge is automatically drawing games immediately by detecting cases where *neither* side can checkmate.

#17

Yes, that falls under what I said about one side having nothing but 1 King left.

I realize now that there may be only that legitimate cause for draw via time-out, only if exactly one side has nothing left but the King.

Otherwise, show me a counterexample.

I guess, something like an inflexible, one-square-color bishop.

- - - - - - - -
- - - - - - - -
- - - - - - - -
- - - - - - - -
b - - - - - - -
k - - - - - - -
- R R - - - - -
- - - K - - - -

If White played suicidally, could he allow Black to checkmate him?

not sure, maybe not

But who gives a fuck about that timeout rule anyway.
It's enough to keep white thinking to try to win so I would rather just say timeout = loss every time kthxbai :) except for enemy has only a King.