Game drawn when it should be a win!

Without making such code it is all just assumption. now let us please try to fix the code and make this program more towards chess rather than what one sole country's chess federation thinks chess should be. btw the link you show litterally has Erik saying the program should be more emcompassing and not that everything should remain as is. Perhaps it is better to not put words in Erik's mouth and let him speak for himself rather than having him portrayed to match your own opinions.

I didn't put words into erik's mouth. I linked to what erik posted and you can read them yourself. He even posted essentially the same thoughts in this topic (but didn't go into the whole backstory which you can read in that topic). I have also read his posts in other topics about the same thing, so while I can't say he won't ever make changes, for certain, the basis for that idea is certainly supported by previous statements.

But as I said you are free to come up with ideas on how to implement something that would work. There are a lot of basic combinations where the FIDE implementation is super easy to check, though, it would take a little thought experiment (or setting up positions) to make sure there aren't positions where those rules fail.

The check only comes about with K+B, K+N and K+2N. If the side with no time has one pawn, or at least one pawn, then mate would be possible in the vast majority of cases. Even with additional material, and at least one pawn, the mate would still be possible, but there might be some instances where it isn't.

Same material and the side has only a queen, mate is possible only with 2N (edited 8/25)

Same material and only a knight, mate is possible.

Same material and only a rook, mate would be possible with the knight and two knights but not the bishop.

Same material and only a bishop,  mate would be possible with a knight (or 2 knights) or a bishop of opposite color.

However, there are likely some configurations of even those combinations, where mate would not be possible and the simple material check like the site has now would not be sufficient to find.

 Daily Chess is essentially correspondence chess, and as far as I can tell, FIDE, unlike the USCF has no specific rules for correspondence chess (does FIDE sanction rated correspondence chess games?).  The USCF rule is simple, if you overstep the time limit, you forfeit the game* (at least that was the 5th edition rule, I don't know about the 6th edition).  Perhaps Chess.com could adopt the USCF rule for Daily Chess.  Of course, that would still mean that Chess.com is not following FIDE rules.

*They do complicate it a bit:  depending on the specific tournament and the round, the player may receive a penalty for the first offense.  The second offense results in a forfeit.

Unfortunately, we cannot add in rules that require "calculation" or "analysis". The volume and engineering effort is far too large to justify given the minuscule number of games this affects. Frankly, we would be better off just manually adjusting the outcome of these games when reported. Would FIDE rules be better? Maybe. And I can see why they are that way when human arbiters are around. But for the sake of practicality, we are using another set of guidelines. It's not about USCF, or FIDE, or which are better. It's about what is reasonably achievable in an online setting. In an ideal world with unlimited resources would it be better to have a computer that would sit and calculate the positions? Possibly. Unfortunately we do not live in that world, and so for us, it cannot be a priority. My apologies to anyone who sees this differently. 

 This topic has been kicked around for the eight years that I have been a Chess.com member.  Typically, the people raising the issue accept the inevitable and life goes on.

I think that Chess.com have come up with a reasonable compromise.  After all, it's just a game of chess, the outcome doesn't have a significant impact on my life, and it's just a way of passing time.

Nobody asks for "calculation" or änalysis". A simple list of pieces left when time runs out would suffice. Positions with a minor piece and the opponent have a passed pawn would lead to possible mates. These positions can be found. No analysis or calculation needed. All we need is find all the positions that will lead to possible mate. Can the community help the developers with that?

I would have under FIDE rules. 1st he timed out. FIDE rules state that he would have lost at the time he timed out. So I would have won. You may argue that I wouldn't have won if he didn't time out. I agree. But let's be honest than I was player a cheater (that's the reason his account was locked. He was caught so even in that case i would have won the match due to a DQ. The only reason he was so highly ranked was he was cheating in the first place. What this topic is about is not about cheaters or the most likely result in which situation. The reason I started this topic is to better the game program that determines a winner. So with that in mind @VicountVonJames what end position would you think would win in the event of a timeout?

but here is the thing. No governing body has the rule that insufficient material should be the decider. FIDE clearly states that the position must be such that no mate is possible. So not only material on the "non timed out" player should be considered but the rules expressively state position should be considered. due to unknown constraints this has never been implemented into the chess.com program that decides the winner. For this reason I started this topic to discuss the positions that should lead to a possible mate when timed out. what position do you think lead to a possible mate when timed out?

 @Hacklover

I don't think that FIDE keeps a list of positions where it's possible to checkmate by a series of legal moves.  It's up to the player to demonstrate that checkmate is possible (the arbiter isn't going to do it for you).  If Chess.com adopts the FIDE rule, how do you prove to the computer that checkmate is possible?

If you read back you would have known how i would implement it. We don't need to implement it like FIDE on the board events. We can do it differently. The number of positions where this occurs are limited. Not few but not infinitely many either. We can list the (types) of positions that could end up in a mate. If this list is implemented into the engine than Bob's your uncle. All we need to do now is list the types of endings where this occurs. What are your opinions on which endings would suffice @woton?

You're inconsistent.  First you're upset because Chess.com does things differently than FIDE.  Then you suggest that part of the solution is for Chess.com to do things differently than FIDE.

As far as my opinion on which endings would suffice.  I think that there are numerous endings where it is possible to checkmate by any series of legal moves.  Even simple drawn positions (K+R vs. K+R for example) can result in checkmate if one side blunders.  Then there are the complicated endgames (your game for example) where checkmate by any series of legal moves is not obvious, the position has to be analyzed.

Either you include every possible position (which you will never know),or you do as Chess.com has done, make it simple.

chess.com does loads of things differently than FIDe, we have no arbiters for instance. That i'm ok with. What I find odd is that chess.com uses different rules than FIDE. and with this we as a community can change that.

PS. We are only interested in positions where (self) mates aren't obvious. So that would mean K+N vs. K+X(+X) with an X number of pawns,  K+B vs. K+X(+X) with an X number of pawns and K+N+N vs. K(+X) with an X number of pawns. All positions with K+R and K+Q and K+pawn doesn't give any problems against FIDE rules at the moment. So in short we are concerned with the pieces and pawns of the timed-out player more than of the non-timed-out player.

There are only 3 types of material combinations that would need looked at, for the side with time on the clock. So, K+R vs K+R would be out of scope in such a situation. 

I have enumerated a number of situations where mate is possible in the case of normal insufficient material considerations. The question is, are there exceptions in those, where mate is not possible and are all positions where more material exists, trivially reducible to one of the postions where mate is possible ?

The thing is, there is no way to fully follow FIDE rules for all situations (not just insufficient material, even if the algorithm ends up being simple). For example, there are positions with sufficient material that are draws on timeout under FIDE (and USCF) that get ruled as wins here. Think dead positions or forced mating combinations where neither side has a choice in moves. Rare in the latter case, for sure, and probably a little less rare in the former.

That all said, even if the list is small and manageble from a coding/resource perspective, it is possible that nothing will change.

 You may think it's odd that Chess.com doesn't follow FIDE rules, I don't.  The games on Chess.com are casual games.  Most of the players don't really care as long as we know what the rules are and they are uniformly applied.

Martin

I guess that I chose a bad example.  The point that I was trying to make is that for all positions (and there are many of them) you have to show that checkmate is possible by any series of legal moves and that this requires that all positions, even the simple ones, be evaluated.  Take the OP's position.  Suppose instead of K+ B, he had K+N or even K+R.  Is it obvious that checkmate by White is possible (with K+R I would guess that it is, but it still has to be shown)?  Now add one or two pawns or maybe even another piece to Black's position, is checkmate by White still possible?  It's not obvious.

For insufficient material checks, the only positions that would really need to be checked would be K+B, K+N and K+2N, for the side with time.

K+R is sufficient material to mate, with worst play by the defending side, regardless of material still on the board; or at least I can't think of many positions where that would not be the case. There would be some positions where there would be a forced recapture but as I mentioned in my previous post, the site already doesn't look at such situations, which is completely understandable from a computing and complexity standpoint. Most any other material combinations, other than the ones currently checked, would fall under similar considerations.

In that case, I enumerated some basic guidelines in post 22. The whole question comes down to whether or not those always hold or if positions can be created where mate is not possible with worst play by the defending side. I'm sure some exist and I think both the current method and any other solution miss edge cases. 

I'm not trying to get the site to change the implementation. Just trying to discuss whether or not it is possible to create a simple material check algorithm to implement FIDE rules in those 3 situations.

Martin

The OP wants to abandon the present Chess.com criteria and somehow implement the FIDE criteria.  My comments are aimed at what I think would be required to implement the FIDE criteria.

I think that the Chess.com criteria are more than acceptable.  In a few cases, they will differ from either USCF or FIDE rules, but I don't think that the majority of people care.  Most of the threads that I have seen ask why a game was drawn when the opponent ran out of time.  They didn't know about the requirement to have sufficient mating material.  In most cases, the games were draws under both USCF and FIDE rules.