system should set draw

Im restricting it to K(B|N) vs K(B|N) exclusively now, because i just found other positions where a mate in +1 can be enforced *blushing*:

black mates in 4:

lichess.org/editor/8/8/8/8/8/kB1B4/B1BN4/KBbN4_b_-_-_0_5

i would argue for adding K(B|N) vs K(B|N) because while theoretical possible it is just too obvious to miss it. And it is a simple case.

@Toadofsky #20

i would also add KN vs KR for the same reason. Its too obvious.

I just wanted to point out that you came up with really old stuff. It was once called the "helpmate minus one" rule (if one's turn can prevent mate and so on)

It was a couple of years part of the FIDE rules but has been replaced by the consistent set now. Because it has also created other problems.

My suggestions are: play in the US or on a US server like chess.com, play without clock or with increment, as usual. I cannot help further.

@Sarg0n
Which problems has it created then?
Which problems would it create on lichess now?

I see the advantages: Less angry players. Angry players complain all the time in the forums because they lose a position on time which they would otherwise draw against Magnus Carlsen using Stockfish.

I consider getting a full point for winning such a position on time simply bad sportmanship. Why support this?

@Toadofsky i just did some statistics with KR vs KN in my database, from 2113 games the knight wins 15 games, on time, the mate never appears.

To play on to win on time with K+N vs K+B is just as stupid as K+R vs. K+R or K+Q vs. K+Q. The solution is not to tamper with the rules, but to play with increment. The present Laws of Chess are as they are and for good reason.

Yeah, as @Sargon and others have pointed out, losing KBkb or KBkn because of a flag isn't fundamentally different than losing KQQQRRkp with respect to how "ridiculous" it is.

To use your example, @impruuve, Carlsen using SF would lose, not even draw, such a position against basically anyone who wasn't actively trying to avoid a win.

Yet calling that a win for the side with a lone pawn when the attacker flags is somehow less of a problem than calling KBkn a win for the side with time remaining?

Any flagging rule is going to result in losses that wouldn't happen if the flagging side got even a small amount of time per move for the rest of the game; that's just the way it is (and is why as @tpr points out, the real solution for avoiding time losses in trivial positions is to pick a time control that facilitates that).

Tinkering with the rules to address particular piece configurations that some people's intuitions find more ridiculous than others just doesn't have a lot of appeal to me.

Have some simple, consistent rule and stick with it.

The traditional internet rules (used on ICC, lichess until fairly recently, also used by USCF, etc.) where certain fixed material configurations are deemed insufficient is one way.

The FIDE rules, where existence of a possible mating sequence is the determining factor, is another way.

Both award wins in "ridiculous" positions like KQQQRRkp if the side with the material advantage flags, so it's not like either is inherently more reasonable.

At the end of the day, if you agree to play under some time constraint, there has to be a penalty for overstepping the bounds.

Complaints of "Well, if it weren't for the clock I'd draw this" ring a bit hollow when you agreed to a game with a particular time control and then used all your time :)

@a_pleasant_illusion the main reason why i would like to have these two endgames (and maybe KNN vs K) added is that in FIDE tournaments there is a referee and that is why the theoretical rules can be strict. The player can stop the clock and claim and the referee will agree.

There is no referee in online chess. Therefore i would favour a more tolerant approach regarding what is considered draw, aka also adding endgames without pawns, where the player not losing on time can never enforce a mate.

Aka K(R|(B|N))(loses on time) vs K(B|N) and KNN vs K (always draw).

This is simple to implement.

Other types of positions where one has many pieces, and the other for example just has a pawn are a) not trivial to implement and b) we dont know for sure if forced mates in 2+ exist (aka, the player with the material advantage, losing on time, may actually be really lost). For example i erred in KB vs K(Bs and Ns), see #11 and #21. Other examples have already been given, for example the Knight versus a-pawn example in #9. and even the single pawn can enforce a mate against a full piece set: lichess.org/analysis/standard/2k2bnr/K2ppppp/1P1p4/1b1pn3/1q1p4/8/8/1r6_w_KQkq_-_0_1

So the answer to the dudes complaining about this would be: Sorry, it is not trivial to implement.

#24 "I consider getting a full point for winning such a position on time simply bad sportmanship."

In bullet and blitz chess, players are frequently awarded a full point in positions (such as K+P versus K+Q) which, if there were an increment, the opponent should never lose. Blitz and bullet chess do not encourage your concept of sportsmanship (and classical chess is best enjoyed with an increment, even something as small as +2 sec/move).

#27 There is no referee in online chess. If K+B versus K+N were defined as a draw, I can guarantee someone will ask "Why is K+B versus K+N+P a win for the K+B?" Okay, so then you make that endgame a draw, but then you get a position like:
lichess.org/editor/8/8/8/8/7n/8/7p/3B1K1k_b_-_-

and now Black times out and White is denied a half-point. So then you make an exception (to your earlier exception) for forced mate cases, but then you have to rule on a different game:
lichess.org/editor/8/8/8/8/7n/7p/8/3B1K1k_b_-_-

and in time-trouble Black might have blundered into mate. Should White be denied the full point?

Simple, consistent rules are important. As long as checkmate is possible, a player should be able to win on time.

@Toadofsky

Exactly because of the example you gave i _dont_ want endgames with pawns to be included. Sarg0n gave other examples for this type of position in #9.

Once again, i want exclusively these endgames to be added and nothing else:

KR vs KN (if R loses on time).
KB vs KB
KB vs KN
KNN vs K

This is simple to implement, because all that is necessary is to count the pieces. No other exceptions.

Notice also that the 'draw if no mate possible' rule is _not_ simple and consistent. Example:



A mate is impossible. Still black wins (Edit: i linked to the wrong position).

So both the current rule and my extension are imperfect. And people will complain about both rules. But my rule will add some more easily detectable, obvious drawn positions. and thats all what can be done if all we have is an algorithm and no referee: Making it slightly less broken.

Let us just stick to the FIDE rules and try to make increment more popular. That solves everything without tinkering with the FIDE Laws of Chess and introducing weird rules that other people will complain about.