Mates that are difficult for engines II

Dear friends,

For some reason unbeknownst to me, chess.com felt the need to remove my thread entitled "Mates that are difficult for engines" without any warning to me, or any reason as to why, so I am going to speculate as to why. I am guessing it is because of all of the screenshot images that were in the thread (possibly using a lot of bandwidth from their servers). Now I will post a continuation of the thread in this new thread, hope you enjoy.

A mate in 82 to start off with (somewhat easy for humans, and difficult for engines):

Accidents happen on chess.com. A few years ago there were lots of issues with corrupted and disappearing posts. Since then things have quieted down on the chesstern front but I still see the occasional mishaps. Looks a lot like failed roll-backs after corruption. Also I have seen chess.com remove individual posts but not complete topics on purpose. At least not besides spammers.

Thanks for the explanation.

Here is a mate in 77:

Here is a mate in 22 that most engines seem to struggle with:

Here is a mate in 17 that seems difficult for most engines, but I did find some ancient engines from 2006 and 2008 that can solve it, but no modern ones:

Here's a Mate in 72:

Here's a mate in 69:

Here's a mate in 20 that engines struggle with:

I speak for the mates, for the mates have no tongue!

These are excellent puzzles but in the last puzzle is two mates but no problem!

Could you show the other mate please, I would appreciate it, but keep in mind that this is a mate in 20, so if the other mate is longer it doesn't matter.

Yes i agree

This one can be solved by engines, but it might take your engine a while. Stockfish on my computer using all 20 cores and 16GB hashtable took 3 minutes 55 seconds to solve this mate in 12:

Enlighten me, why is the rook on h8? Not only does it make the position look ugly, it also appears to play no role in the solution. Everything is the same with or without it!

It is there to see if you would notice and complain about it! . Looks like it served it's purpose , but seriously though the more pieces the engine has to think about the better to make a mate more difficult for them, without the Rook on h8 Stockfish only needs 17 seconds on my computer to find mate in 12.

That's weird! Considering that any move played by Rh7 or Nf7 leads to an obvious #1 we can see that Rh8 will never play a move in any sort of properly pruned search strategy. Could SF be more stupid than even I imagined?

Well, if you noticed the bean counting by StockFish with and without the Rook you'll see that the SF eval shows that white is "down" by 6+ points with the Rook and without the Rook on h8 it thinks the position is pretty even, so perhaps it is dumber than you imagined, but it gets "smarter" if you let it think about the position long enough.

Here is a difficult mate in 23 for engines: