Strange draw awarded

@Peter-developer01 said in #9:
> Or if the side with queens decides to give them to a bishop or a king freely (so that the side with bishop can take the queens away), but that's very rare variant.

Nope. The bishops by themselves cannot mate if they are all on the same colour squares.

@Brian-E said in #11:
> Nope. The bishops by themselves cannot mate if they are all on the same colour squares.

ah, exactly. My bad.

the fide rules of checkmate by any sequence of legal moves means it's possible in a fide tournament for it to be a draw even with sufficient material.

with online chess, it's only about sufficient material.

lichess.org/editor/8/8/5k2/1p2p2p/1P2P2P/3K4/8/8_w_HAha_-_0_1?color=white

This would be a draw in a fide tournament, with online chess, if black's time ran out, white would win.

@h2b2 said in #13:
> the fide rules of checkmate by any sequence of legal moves means it's possible in a fide tournament for it to be a draw even with sufficient material.
>
> with online chess, it's only about sufficient material.
>
> lichess.org/editor/8/8/5k2/1p2p2p/1P2P2P/3K4/8/8_w_HAha_-_0_1?color=white
>
> This would be a draw in a fide tournament, with online chess, if black's time ran out, white would win.

This sort of fortress position is where it gets interesting. If I understand things right, that position would only be a win for the opponent of the player running out of time because it's very difficult for software to detect such positions where neither side can make any progress. But isn't there at least some attempt to do so in the software? Are some fortress positions successfully detected?

as far as I know, at least on lichess, it's just about the material and the position is ignored.

software could do it for sure but I don't know if it's difficult or easy to implement or if something exists already, like maybe stockfish could do it? no idea. but regardless, it's extra computer time/expense, and you'd have to check every game that ends in a timeout for something that happens very rarely.

So far the response has been it's not worth the effort for something that happens so rarely.

and whatever the rules are, if it's the same for both sides, it's fair.

@h2b2 said in #15:
> [...] software could do it for sure but I don't know if it's difficult or easy to implement or if something exists already, like maybe stockfish could do it? no idea. but regardless, it's extra computer time/expense, and you'd have to check every game that ends in a timeout for something that happens very rarely.[...]

I'm not so sure about this.

Okay, in a different universe with no limit on resources you could create a 32-piece tablebase storing all legal positions and detect which of them never lead to mate.

But failing that (and it obviously does fail), I think detecting all fortress positions which can never lead to mate may be an example of a non-computable problem. For example, the task of writing a program which can detect if any input program terminates is impossible. It is non-computable. Detecting whether any input position can lead to mate with any legal (unbounded in number of moves) sequence of legal moves seems to me to be a similar problem.

EDIT: for clarity, I was talking about the so-called "halting problem". See en.wikipedia.org/wiki/Halting_problem . I think the problem of determining all positions which can lead to mate may be similar.