Bug: lichess does not recognize positions where it is impossible to lose when you time out

"and simply evaluating whether all the legal moves lead to a dead position should not take much computational power"

Oh dear. Little do you know.

@thomassowell123
I'm sorry, that's completely ridiculous. By that logic, if I have a King and Queen against a King and timeout, then I lose because theoretically I can move my King next to his King and lose. I do, however, apologize for accidentally misquoting the FIDE Rules.

The correct one is: "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 LEGAL moves, even with the most unskilled counterplay."

I accidentally deleted the "legal" because I wanted to capitalize the "possible series of legal moves", so I deleted it and retyped it in caps but forgot to add the "legal" back.

Proof if you don't believe me: official FIDE handbook here: https://www.fide.com/FIDE/handbook/Standards_of_Chess_Equipment_and_tournament_venue.pdf

Go to Law 6.9 on page 10.

@Magirly

Lichess actually uses FIDE rules, not USCF rules.

Proof: https://lichess.org/faq#timeout
Lichess clearly says it uses the FIDE rules when dealing with timeout.

I again apologize for accidentally misquoting the rules, as mentioned earlier.

@Toadofsky Thanks a lot.

@tpr

Yeah, I sort of understand that by playing blitz without an increment, I'm more or less asking to get in a flagging battle. However, that doesn't change the fact that this is a real issue.

@Cedur216

agreed.

@Magirly (#9)

Wrong: as mentioned earlier, Lichess uses FIDE rules. Stop accusing me of spreading disinformation.

@ProfDrHack

I don't think it should take much computing power to just evaluate whether each legal move leads to a dead position.
After all, each second, Lichess evaluates many games where one side timeouts and sees if the other side has sufficient mating material.

*Just to clarify, I'm just asking Lichess to run through every legal move at DEPTH ONE. This will just check if any of the immediate moves lead to a draw. Although there are some more complicated several move sequences and it doesn't solve the problem of locked positions, it's at least a good start and will probably cover at least 50% of these types of timeout vs insufficient material bugged positions.

However, I do admit I'm not completely sure, and maybe there is some hidden catch. Are you experienced in programming?

I feel the best way to really find out how difficult my proposal would be is to ask an actual Lichess dev.

@Toadofsky, it would be greatly appreciated if you could comment on the relative difficulty of adding my proposal.

I agree @AustinL0926 that if you get flagged but in an impossibly losing position you should be able to claim at least a draw. A win might count as well but we don't know what's going to happen. Therefore, Lichess should at least award a draw.

#12 I understand however my official comment (about something different) is at https://github.com/ornicar/lila/issues/6804#issuecomment-724002709

I'm not commenting on that particular suggestion because the next question is "What about depth 2?" etc.
http://xyproblem.info/

It's possible to lose if you resign!

not true the opponent never made that move a saw the game

@AnonymousForumPoster

I said it's impossible to lose if you time out.
And anyway, who would be dumb enough to resign there?

@UWVolsky

i don't think you get my point. My ONLY LEGAL MOVE was Kxg7, after which it would have been impossible for him to checkmate me. My point is under the FIDE rules that Lichess uses, it should have been a draw, because there was no "possible series of legal moves" (FIDE Law 6.9), that would make it possible for him to checkmate me.

@Toadofsky

Yeah, I sort of get your point about "What about depth 2?" and all that.

I do, however want to point out several things. It seems in the project you linked, you've mainly been focusing on locked positions where it is impossible for either side to make progress. Just skimming through the comments, I don't see much discussion about forced lines that lead to insufficient material situations.

Also, about your comment on November 9th: you said, "absent a novel approach, it's too computationally expensive to run this on every timeout". That's why my proposal is that there is a button that you can press after the game to try to evaluate this. I'm not sure how often these situations happen: 1 in 100, 1 in 1000, 1 in 10000 games? Evaluating just the games where the button is claimed will take much less computational power than evaluating every game that ends in timeout.

Now of course, there will be people who just press the button and attempt to make a false claim, and this will waste resources. However, I don't think this will happen too often. Why? Because it's obviously futile. If you press the button on a game where it doesn't apply, nothing's going to change. There would simply be no point for someone to press it on a game that they know doesn't apply.

Disclaimer: I'm not a programmer, so feel free to correct me.

P.S.

By the way, I want to make it clear that I didn't just make this forum topic because I was salty and wanted the game result to be reversed or something. Instead, I made this forum topic with the goal of helping to improve Lichess, so constructive criticism would be highly appreciated.

I dont know. I've resigned before in positions like that :(

@AnonymousForumPoster

Can you show an example?

I do apologize if I accidentally insulted you, but I simply can't believe that you would resign in a position where you're a move or two away from at least a draw.

And anyway, that isn't relevant to the topic.

actually it was not a draw actually if you took and your time did not run out you would win