Mates that are difficult for engines

Draw with Black to move. He plays into the drawing zone.

(Drawing zone for the pieces on the h file as shown is illustrated here.)

It is not the case that the knights have no wins if the pawn is beyond the Troitzky line, nor that the pawn has no draws or wins if the pawn is on or behind the Troitzky line.

I think +M42 at 19:10 followed by +10.74 at 22:07 has him bang to rights.

You can never tell whether you have a "final outcome", but here's one with switched mates (+M39 at 4:13 then +M40 at 29:52 for example).

Maximum is 36 moves from any king bishop knight position that's not an immediate fork. That's the position I used to set up against stockfish as well lol!

All wrong.

It's maximum 33 moves from any winning position, not 36.

Nothing at all from these, for example, which are not immediate forks.

That it's your position is not an accident. I tried the White pieces against SF15.1 when you posted it. It's a mate in 31 and I managed mate in 23 against SF15.1+NNUE and mate in 27 against SF15.1-NNUE.

SF's performance without tablebase in this endgame has disastrously collapsed since SF8 in terms of accuracy. I used to outperform SF8 with this material by only about 1 move in two or three games. ( Edit: No it hasn't - I must have been thinking of a version of Rybka with an "e" on the end of the version number. I tried it against SF8 and it's just as bad.) Here is an objective comparison with the SFs as White against Rybka+Nalimov.

SF15.1-NNUE appears to going for a draw under the 50 move rule.

I saw all those already, I meant those too. Inevitable stalemate or loss of the piece.

Then it would be a good idea to say what you mean.

Jeez, you people are being really nit picky lately, btw you forgot one:

Which side to move?

But seriously you should think before posting "facts" about endgames.

There's so much crap posted and written about this endgame already. When you post that the maximum mate depth is 36, you can expect about 50 follow on posts over the next few years saying the same thing because people believed you.

It's easy enough to check before you post. There are no pawns or winning conversions, so you can just look it up on the Syzygy site.

Oh jeez big deal, so people would think it takes 36 instead of 33 or 34 who cares? What's important is that people know how to do it, not random factoids about tablebases (assuming they are even accurate).

White to move can't avoid stalemate or loss of the knight in that position.

That one is excellent my friend, thanks for creating and posting it!

Ke8 avoids both, but the black king is forking White's pieces. It may be relevant for people learning to play the endgame that they're not necessarily winning when that is not the case.

For example they may be inclined to play BxP here if they believe your original post.

As for the 33 v 36 maximum depth, it's not particularly important, as you say, but what's the point of spreading misinformation?

I checked the tablebase, its 15 moves to win the knight, and it involved bishop to c1 and to g5 at one point. I may add this one to my underpromotion compilation thread cause of the interesting tactics and traps here!

Bc1 instead of b4 stops the black king from protecting the knight, try it from there.

Full analysis here, maybe I'll go back and delete all my other accidental 'misinformation' posts (although ironically stockfish saw the win immediately when I plugged it in, so maybe this entire puzzle is "misinformation to this thread 😉)

https://www.chess.com/play/computer/Jimmy-Bot

Stockfish doesn't solve this at first. Note white is still winning without this sequence with a +8 advantage, but this line is the fastest win.

The latest Development version of Stockfish from 02/11/2024 (today) finds mate in 12 in 36 seconds on my computer:

Since 12 moves is almost the same as 10 moves, I guess its not a unique solution so is invalid.

LOL, you're right. Let me try again...I didn't look at the solution.