Bug: lichess does not recognize positions where it is impossible to lose when you time out • page 1/4 • Lichess Feedback • lichess.org

https://lichess.org/NL1DOhZY/white#1

This was a game I played recently, and it should have been a draw, because the only legal move I could make was Kxg7, where it's impossible for me to lose even if I time out.

I suggest that Lichess makes some sort of button after the game that you can press to claim that the ending position in which you flagged was one that was impossible to lose. Then, Lichess automatically evaluates ALL of the legal moves, and if all of them, BY FORCE, lead to a position with insufficient mating material, then it changes the result to a draw.

Just for clarification, the button will only show if you lost by timeout, and simply evaluating whether all the legal moves lead to a dead position should not take much computational power.

https://lichess.org/NL1DOhZY/white#1 This was a game I played recently, and it should have been a draw, because the only legal move I could make was Kxg7, where it's impossible for me to lose even if I time out. I suggest that Lichess makes some sort of button after the game that you can press to claim that the ending position in which you flagged was one that was impossible to lose. Then, Lichess automatically evaluates ALL of the legal moves, and if all of them, BY FORCE, lead to a position with insufficient mating material, then it changes the result to a draw. Just for clarification, the button will only show if you lost by timeout, and simply evaluating whether all the legal moves lead to a dead position should not take much computational power.

nnuunn

#2

On one hand, I do appreciate the simplicity of just saying "if you're in check when the time runs out, you're in check when the time runs out, simple as", but I do think it should be pretty easy for Stockfish to figure out that there is no way to win given you would take the only legal move, and so I don't see why they don't give you the option to appeal in such situations. I think this might get at the purpose of the timer, is it just to keep the game within time constraints, meaning a finish in which there is no real possibility for one side to win even if they happen to have the other in check should go to the former side, or is it intended to be another element of the game that you can use as a weapon against your opponent, so if you can get him in check when the time runs out, you're just playing the game the way it's intended to be played?

NM AustinL0926

edited

#3

One more thing:

Law 6.9 of Fide Rules: "If a player does not complete the prescribed number of moves in the allotted time, the game is lost by the player. However, the game is drawn if the position is such that the opponent CANNOT CHECKMATE the player's king by any POSSIBLE SERIES OF MOVES, even with the most unskilled counterplay." (emphasis added)

So if Lichess wants to be consistent with the FIDE rules, maybe they should do something like what I suggested.

@nnuunn

So from that rule, it's clear that flagging your opponent in a position where it is impossible to win is not the way chess was intended to be played.

One more thing: Law 6.9 of Fide Rules: "If a player does not complete the prescribed number of moves in the allotted time, the game is lost by the player. However, the game is drawn if the position is such that the opponent CANNOT CHECKMATE the player's king by any POSSIBLE SERIES OF MOVES, even with the most unskilled counterplay." (emphasis added) So if Lichess wants to be consistent with the FIDE rules, maybe they should do something like what I suggested. @nnuunn So from that rule, it's clear that flagging your opponent in a position where it is impossible to win is not the way chess was intended to be played.

thomassowell123

#4

the MOST UNSKILLED COUNTERPLAY would include illegal moves like Kh8.

from that position, with black to move, white certainly could be checkmated.

therefore, by a strict literal interpretation of the fide rule, lichess has come to the correct decision.

the MOST UNSKILLED COUNTERPLAY would include illegal moves like Kh8. from that position, with black to move, white certainly could be checkmated. therefore, by a strict literal interpretation of the fide rule, lichess has come to the correct decision.

Magirly edited

#5

@AustinL0926 as far as i know, lichess goes by the USCF rules here, for simplicity reasons. They go by insufficient material, not by the fact whether or not one site can still checkmate.

@thomassowell123

nope, that is factually wrong. Austin misquoted the fide laws. It specifically says:
"However, the game is drawn, if the position is such that the opponent cannot checkmate the player’s king by any possible series of LEGAL moves"

You don't need to downvote this. I'm stating facts here. I'm a chess arbiter, so I kind of know what I'm talking about ;)

@AustinL0926 as far as i know, lichess goes by the USCF rules here, for simplicity reasons. They go by insufficient material, not by the fact whether or not one site can still checkmate. @thomassowell123 nope, that is factually wrong. Austin misquoted the fide laws. It specifically says: "However, the game is drawn, if the position is such that the opponent cannot checkmate the player’s king by any possible series of LEGAL moves" You don't need to downvote this. I'm stating facts here. I'm a chess arbiter, so I kind of know what I'm talking about ;)

Toadofsky

edited

#6

#1 Thanks, this unsolvable issue is recorded (I did develop a helpmate solver, but have no way to test it):
https://github.com/ornicar/lila/issues/6804

#1 Thanks, this unsolvable issue is recorded (I did develop a helpmate solver, but have no way to test it): https://github.com/ornicar/lila/issues/6804

tpr

#7

#1
This is an instance of imperfect lichess implementation of the FIDE Laws of Chess: it should have been a draw, but lichess has not coded it right and it is very hard to code it right.
A better solution would be to play with increment.
All FIDE competitions are with increment: 3+2, 5+3, 15+10, 90+30...
Only on lichess we still seem stuck in the century of analog clocks.
Increment would not only resolve #1, but also KN vs. KN, KR vs. KR, KQRBNP vs. KP

#1 This is an instance of imperfect lichess implementation of the FIDE Laws of Chess: it should have been a draw, but lichess has not coded it right and it is very hard to code it right. A better solution would be to play with increment. All FIDE competitions are with increment: 3+2, 5+3, 15+10, 90+30... Only on lichess we still seem stuck in the century of analog clocks. Increment would not only resolve #1, but also KN vs. KN, KR vs. KR, KQRBNP vs. KP

Cedur216

#8

Well that won't be enough reason for Lichess to suddenly add increment to all the official 1+0, 3+0, 5+0 and 10+0

Magirly edited

#9

@tpr

No, it is perfectly implemented. Lichess follows the USCF laws of chess, which say:

"A game is won by the player who correctly points out that the opponent’s flag has fallen first, at any time before the game is otherwise ended, provided the player has mating material. Mating Material consists of (at a minimum) two minor
pieces, a pawn, a rook, or a queen"

There are some exceptions given but this is not one. Stop spreading misinformation.

@tpr No, it is perfectly implemented. Lichess follows the USCF laws of chess, which say: "A game is won by the player who correctly points out that the opponent’s flag has fallen first, at any time before the game is otherwise ended, provided the player has mating material. Mating Material consists of (at a minimum) two minor pieces, a pawn, a rook, or a queen" There are some exceptions given but this is not one. Stop spreading misinformation.

Toadofsky

#10

#9 I have commented about the USCF chess rules on many occasions, including a 2-hour video:
https://www.youtube.com/watch?v=-sVV1cHmJKU

#9 I have commented about the USCF chess rules on many occasions, including a 2-hour video: https://www.youtube.com/watch?v=-sVV1cHmJKU