@UWVolsky
That's true, but that's not relevant to the topic. If I had more time, the position would be winning. However, the point of my post was to point out that when I timed out, then it was impossible for him to checkmate me, so it should have been at least a draw (even if objectively it was winning for me).
#17 I understand; indeed it was for these very reasons that it took a very long time for me to reach a consensus with other developers (through my own difficulty of understanding) when designing Lichess' timeout rules as they are today.
The problems with adding an "arbitrate" button is that every feature adds complexity (e.g. having to add code for overturning results) and every feature can be prone to abuse.
(Personally, the "fair" way I'd like to see it implemented is: if your opponent's clock runs out, you are free to move both players' pieces and need to earn the win by demonstrating a checkmate. But that's more complex and confusing, and less versatile, than the button approach.)
@Toadofsky
The way you described is probably the "fairest" way, but is of course nowhere near practical; what if you're in a tournament, and have no time to demonstrate a helpmate, even if one obviously exists?
Revisiting your "Why not depth 2?" argument in post #14; I do understand it's a slippery slope, but I think from a practical point of view, depth 2 sequences or higher will be far less common.
Although I understand that a depth 1 run-through of all the legal moves would not catch all of these situations; but it would be a good start, and I don't think it would take much computing power.This proposal won't even use any engines, and will only evaluate all the legal moves to see if they immediately lead to insufficient material for the side opposing the one that timed out. If the claim is legitimate, then Lichess will probably have to evaluate no more than 3 or 4 moves, since these kind of forcing sequences, are well, forcing by nature, so there will be just a few responses.
Anyway, if you do believe it takes too much computational power, I'll be happy to accept that, although I would be curious where majority of the computational power would be spent.
TL;DR: It's very hard to completely fix the problem, but it's definitely possible to get a good start. I don't think it should take much computing power, but feel free to object.
#23 "what if you're in a tournament, and have no time to demonstrate a helpmate, even if one obviously exists?"
So... your opponent ran out of time, then you ran out of time without demonstrating a mate. What's the problem again?
"I think from a practical point of view, depth 2 sequences or higher will be far less common."
I don't disagree.
"Anyway, if you do believe it takes too much computational power, I'll be happy to accept that"
That's what I've been saying... https://github.com/ornicar/lila/issues/6804#issuecomment-724002709
@Toadofsky
Okay, got it. Thanks a lot for your insight.
@AustinL0926 if you had taken the queen it would have been drawn.
@Vertonghen
Yes, I know, but that's irrelevant. My point is that even though I timed out and his queen was still on the board, under FIDE rules it should have been a draw because there was no possible series of legal moves through which he could checkmate me. (see post #3)
Lichess detects from the material on the board, not the position, because there are no arbiters on Lichess.
@AustinL0926
@AustinL0926 Thanks and likewise it's been a good conversation. Among the devs I suspect I'm the one who criticizes the most & begs the most for difficult changes... here I spent months writing and improving a helpmate solver which I have no way to test, which would need to work and perform well.
@Vertonghen
That's the reason why I submitted a bug report; because according to FIDE rules, games that end by time out should technically be arbitrated by the position, not the material. Of course programming that is a different matter entirely.
See post #3 again for how FIDE governs these kind of situations.
This topic has been archived and can no longer be replied to.