Insufficient material -- why are unwinnable interlocked pawn positions difficult for computers?

I understand that in a position where no mate is possible for black, if white's time runs out, the game should be counted as a draw. I just tried what happens if there is enough material in principle, but the pawn chains are sufficiently interlocked such that no piece can cross to the opponents side.



a) am I correct that this should be a draw?

b) if so, is there a good explanation for why it is hard for the computer to be sure that there is no possible mate for white in such positions?

Yes , this should be draw (even according to Stockfish). Unfortunatly this happens always (read some months ago in lichess forum about that, appereantly that is not solved yet or too hard to programm )

github.com/ornicar/lila/issues/9249#issue-929530390

There is a proposed solution.

@anodynos said in #1:
> I understand that in a position where no mate is possible for black, if white's time runs out, the game should be counted as a draw. I just tried what happens if there is enough material in principle, but the pawn chains are sufficiently interlocked such that no piece can cross to the opponents side.
>
>
>
>
> a) am I correct that this should be a draw?
>
> b) if so, is there a good explanation for why it is hard for the computer to be sure that there is no possible mate for white in such positions?

b) Draw detection is not one of the objectives of the Stockfish project.

Stockfish does exactly what it's supposed to do when a position is drawn.

My suggestion:

1. Stop wanting it to do something different.

2. Learn what it does.

Engines could never recognize fortresses anyway

If the only valid move is always a king move is easy to calculate...