They follow a simplified version of USCF rules. Basically, the only look at the material the side with time has. If they have only a bishop, knight or two knights, it is a draw. There are some edge cases where Chess.com's implementation doest follow USCF rules either; i.e. some of the material depends on what material the side with no time has too, such as a pawn in the case of two knights, and the site doesn't look at that at all.
If chess.com does not incorporate the USCF forced win then it should follow the FIDE rules. It is simply funny that a leading chess site invents its own rules. By these rules, for instance, many endgame studies would be unsolvable were they to occur in a game. The endgame of 2N vs P would be a draw in its entirety though it is often winnable.
Some odd bugs in Chess.com application of draw rules. I think this position (White to move) would be ruled a win for Black if white runs out of time for e.g.
https://www.chess.com/dynboard?fen=r1b1k3/pR6/Kpp3P1/6bB/6P1/3R2P1/4q3/8%20w%20-%20-%200%201&board=marbleblue&piece=classic&size=3
If chess.com does not incorporate the USCF forced win then it should follow the FIDE rules. It is simply funny that a leading chess site invents its own rules. By these rules, for instance, many endgame studies would be unsolvable were they to occur in a game. The endgame of 2N vs P would be a draw in its entirety though it is often winnable.
Determining if mate is possible, for FIDE rules, by algorithm, is harder than comparing material. Also, I would guess less than a tenth of a percent of games end in an incorrect draw under USCF rules and wouldn't be worth implementing engine checks at the end of games during time loss in insufficient material situations.
Some odd bugs in Chess.com application of draw rules. I think this position (White to move) would be ruled a win for Black if white runs out of time for e.g.
https://www.chess.com/dynboard?fen=r1b1k3/pR6/Kpp3P1/6bB/6P1/3R2P1/4q3/8%20w%20-%20-%200%201&board=marbleblue&piece=classic&size=3
@texaspete:
Are you suggesting that the position you posted should not be a win for black if white runs out of time? Why should the position you posted be a draw?
BMK
@texaspete:
Are you suggesting that the position you posted should not be a win for black if white runs out of time? Why should the position you posted be a draw?
BMK
Because under either FIDE or USCF rules it would be a draw. White cannot win as he has no time left. Black will win as long as a mate (FIDE) or a forced mate (USCF) for black is possible on the board. If play had continued, white's only possible move leads to immediate checkmate. So black cannot win either. So the result is a draw.
On this website it would be a win for black because it only looks at piece count.
Solution. Just do nothing and wait for your time to run out.
Or "flag" if that is possible..
@texaspete:
Are you suggesting that the position you posted should not be a win for black if white runs out of time? Why should the position you posted be a draw?
BMK
Because under either FIDE or USCF rules it would be a draw. White cannot win as he has no time left. Black will win as long as a mate (FIDE) or a forced mate (USCF) for black is possible on the board. If play had continued, white's only possible move leads to immediate checkmate. So black cannot win either. So the result is a draw.
On this website it would be a win for black because it only looks at piece count.
Uhm, NO! You just contradicted yourself.
If White's flag falls, this is a win in FIDE because a sequence of legal moves, logical or illogical, is possible to lead to a mate in favor of Black.
If White's flag falls, this is also a win in USCF. In USCF, lone K, K+B, K+N, or K+N+N with the opponent have zero pawns (he may still have pieces), are considered insufficient mating material. However, in this case, it's a 100% forced. White's only legal move is 1.a7 and Black mates with 1...Nc7. Because of that, this would also be a win for Black in USCF.
On Chess.com, with how stupidly written the algorithm is, it would reward White a draw for timing out! So you had the whole thing backwards!
Uhm, NO! You just contradicted yourself.
No I did not. Re-read the post i quoted. This is the board position https://www.chess.com/dynboard?fen=r1b1k3/pR6/Kpp3P1/6bB/6P1/3R2P1/4q3/8%20w%20-%20-%200%201&board=marbleblue&piece=classic&size=3
So you had the whole thing backwards!
No I did not. re-read the post i quoted
I mentioned this problem like a month ago - thrillerfan absolutely correct.
All I could do was accept that the chess.com algorithm allowed for some fringe cases like this..
Requests for adjudication could be suggested, but that would likely quickly overflow with ridiculous complaints.
Determining if mate is possible, for FIDE rules, by algorithm, is harder than comparing material. Also, I would guess less than a tenth of a percent of games end in an incorrect draw under USCF rules and wouldn't be worth implementing engine checks at the end of games during time loss in insufficient material situations.
Actually, there is a 'safe' solution to all this:
(a) only automatically draw endings that we know are certain draws: K+N/B vs K or K+B vs K+B (bishops of same square colour).
(b) always permit the game to continue in other cases thru 50M
(c) this is the beautiful part: in endings chess.com knows are very likely draws, increment the clock time for both players to at least 1 minute and indicate this event to the players with a flashy light signal. A minute is enough to reach the 50M limit for routinely drawn endgames.
So why is the beautiful part the beautiful part? It is because it doesn't matter which or how many endings you include in that category! One would obviously include things like K+N vs K+B but one might also include K+R vs K+B or even K+R+B vs K+R though the latter is no easy draw. There is no harm to the time incrementals since it will generally discourage the superior side from continuing aimlessly. This approach will also protect the superior side from losing on time for instance in the engame of K+R vd K+B. Both sides have at least 1 minute to get to 50M!
Do live games on chess.com follow the FIDE rules?
BMK