Is this position draw?

Not so hard, really: en.lichess.org/j7wEemcPGrEX

It's true that I spent a little time looking for the fastest wins (I think I did miss the fastest mate, though; mine was maybe 3 or 4 moves slower than best, I'm guessing), and some of the tactics used were a bit funny.

Still, more straightforward moves would also have won, just not quite as quickly.

Cheers!

@OneOfTheQ Very nice playing! Then i tink this endgame is just hard for beginners (me) but that's how people learning :) . I am sure there are much harder endgames than this. That's why i love this community because people help each other.

Also i am wondering why stockfish level 8 (2500) make 3 errors and 1 blunder? Is not that the hardest stockfish level?

@Diszpecser

It is the hardest level that plays on the site, but even that level has its search depth explicitly limited to a fairly low level.

The other part of it is that the position is losing quite badly to begin with, but it takes a while for the engine to realize just how bad the position is.

Taking them move by move:

The first "mistake" it made, playing Rab1 instead of Rad1, isn't actually a mistake at all. It thinks Rad1 is only -4 or so, and sees that Rab1 is around -5, but if you let SF think for a little bit, it realizes both moves are around -6. So the first mistake isn't a mistake at all.

The next one, playing Rbc1 instead of dxc4, also isn't really a mistake. For that one, if you let SF think for a while, both moves end up around +8 (and climb as you let it think longer).

The next one, where the analysis engine suggests Rxc6 is a blunder and Rcd1 should be played, is also not quite right.

After Black plays Rc6, it's mate in 22 no matter what White does.

Rxc6 does speed that up, since after Rxc6 Bxc6, it's mate in 12 no matter what White does.

So, Rxc6 is some sort of mistake. It's just not the sort of mistake analysis suggests, since its proposed move is also a forced mate (in 21 moves).

The next "blunder" is maybe the funniest. The analysis suggests that f6? is #19, while thinking Rc1 would have been around +9.

The truth is that after Bxc6, black is getting mated in 12 moves no matter what, and that evaluation is the same for f6 and Rc1, so there is no difference in those two moves.

Finally, where it plays Rc5 and the analysis suggests Rf1, Rf1 is mate in 10, and Rc5 is mate in 9, so Rc5 hardly deserves any criticism.

In all of these cases, the problem is just that neither the playing nor the analyzing SF at the site saw deeply enough.

For the one playing, that means it really did make one mistake (Rxc6 cut the checkmate time in half), because it didn't find the mate.

For all the rest of the "mistakes", the analyzing SF didn't see deeply enough to correctly evaluate the moves, so it saw differences where none existed.

Cheers!

Nice topic, learned a bit from it

@OneOfTheQ

WOW! The most awesome answer i have ever read! Thanks for the deep analysis, and move by move explanation. I learned from it too, and it is awesome how you figured out the outcome of the match. with explanation. You are a awesome guy!

Cheers!