Material bug

I like best the official rule set of the world association. The most consistent one one can have. It is practically 1:1 transferred to here except for the case which is technically not feasible (fortresses).

So good job FIDE, playchess, chess24 and last but not least, lichess. Thanks @lovlas and friends!

So, good Job to chess.com and ICC for representing the other point of view.

This discussion arises again and again, and Sargon treats them arrogantly again and again. Seems to be his nature, just cope with it.
But the rule Lichess applies is the only sensible one in my (and Sargon's) opinion. If you argue with something like "99 percent it would be draw" you always could ask: Is it 99? Or is it really only 98.5? And why is the limit 99 and not 98? 96? Where is the limit of blundering your advantage away?
It is a very clear rule to say: Time out - you lost if there is any chance of getting checkmated. Play faster or play with increment ...

@Klartext

Don't get me wrong, I agree that playing just for flagging is not cool, but then again I don't play bullet where it seems to be more of a legit thing (or rather: inevatible).
Also, there's the lovely feature of increment. In my point of view, if you play without increment, you implicitly agree to a flagging race and should not be salty when it does occur. Then again, I don't play without increment either. ^^

PS: Despite the "arrogance", they do have a point: Lichess' rule IS closer to FIDE's. That's a pretty solid argument, so it's easy to be adamant about it. *shrug*

With ten second increment and 14 seconds left on my clock, I had 0 seconds to actually make the move because of my connection yesterday.

Cool, learned a new word today: "adamant" sounds reasonable. Thanks @ProfDrHack :D

Calling an example a "bug" which follows the official rules accordingly... Well, it's up to you how to name such a behaviour.

Everyone is entitled to his personal opinion, playground, whatever:

My summary:

-the simplest rule is the best (least exceptions, no arbiter needed, no appeals committee, no tiresome analysis of every timed-out game)
-it is the closest to the official one world-wide (FIDE)

Both points apply: it is not a bug, it's a feature.

@Sybotes but my point is that the remaining 1% can be detected by checking against the endgame database or by running an analysis. So we can detect with nearly 100% safety if it is a forced mate. Constructing a study to prove that it is theoretically possible to not detect it in under depth 20 has been done by sarg0n:

lichess.org/editor/N7/2p4p/5p2/7p/5p2/5p1p/5p2/5K1k

Tablebases wont work here and Stockfish 10 needs depth 26 to detect a mate, but that depth is reached in under one second. If you let him ponder ten seconds he reaches depth 61, 30 moves. ASMFish reaches depth 64. That is a very strong safety net, considering the fact that the study is constructed and will very probably never appear in a practical game.

The blitz/bullet argument is ok. Regarding the FIDE argument, in long FIDE games people can claim, that is not the case in online chess, so the FIDE rules dont fit.

Regarding 'most simple rule', rules are not made to be as simple as possible but to ensure a fair game.

Im anyway fine if it stays like it is. Again, my point is people acting as if it was the logical choice. especially ICC and my above post shows that it is as well doable the other way around.

#16 @Klartext I'm working on a helpmate solver, just I don't think people would be willing to wait a year for games to be decided after the flag falls so probably I need to improve performance of the detector before deploying it:
github.com/ddugovic/Stockfish/tree/helpmate