Mates that are difficult for engines

Here is a nice mate in 16 that engines struggle with:

Here is a nice Mate in 27 that engines struggle with:

perfect play has two options:

promote a pawn to a second queen 

dont, and checkmate with the one

it chooses the first, if the pawn is in the 6th rank

thats weird, ig at some skill level 2 queen mate becomes better than 1 queen mate

also en passant (!) and three queen promotions 

Here's a Mate in 74 that engines struggle with:

nice promotions lol

I thought someone might enjoy that.

Here is a mate in 172 that engines won't solve, it's another triangulation/Zugzwang mate in x:

Wow 172 moves?? And that endgame even looks somewhat realistic..chess is amazing.

Here's a mate in 549 just for you, enjoy .

It would be even better starting it at 1..kg2 and saying mate in 541 moves. The fact that the best starting move in a position where you have a queen and knight is a backward king move away from the center when he's not even in check, and this is the only move that leads to mate 540 moves later is hilarious!

This is simply the longest 7-piece tablebase win. Supposedly it is quite easy for a general purpose chess engine which can simply look it up in the tablebase. No need to waste precious energy on cooling our processors for this task. Note that the winning line is not unique and it therefore fails by directmate composition standards!

In general diagram checks are acceptable for compositions. Even starting with a checking move is OK for long moremovers as they are closer in character to endgame studies where about 30% of all compositions start with a checking move!

Comment #246 the game is too laggy

But it's good

what did you expect

storing 547 moves isnt an easy task

Yet another trianulation/Zugzwang mate, this time it's mate in 121. As EndgameEnthusiast2357's favorite chess YouTuber Suren would say "Easy for humans Impossible for Chess Engines".

Doesn't work with chess.com chess.

i have a serious doubt

why is sometimes Ka2 Ka1 and sometimes Ka1 Ka2

if you have an irrelevant choice of triangulations many times in a puzzle, why not make it easier for the solver by making it always the same?

Basically it's an error of the puzzle interface which is incapable of accepting more than one correct answer per move turn. There are simple ways of handling this, e.g. by forcing one of the king moves at that point automatically or say something like "please try another move" as in the chess.com lessons. Understand that chess.com's puzzle interface is absolutely barbaric (read this again in 25 years) and does not allow you to do these things. Technically, different move orders are a composers error but these are so hard to avoid in triangles that they are tolerated.

Though the burden should not be on the puzzle poster he could use an algorithm that works as well for the first triangle as for the repeats without distinction between black and white:

• first play the "diagonal" move if possible
• if not possible play the "level" move (same rank)

If this does not decide all choices then it is not a triangle!

if there are two people who can fix a problem, one of them should do it but doesnt, should the other have to fix it?

thats the question here

but yeah thats what i am saying

if the poster wants to fix the problem then the thing to do would be to be consistent between all the instances of triangulation

but i mean, i dont use any thought when solving these i just click the hint button lol, i just want to see how its done