I always find these puzzles so satisfying, knowing that even the strongest chess engine stockfish cant solve it at first glance. But what would happen if you would run stockfish natively on this puzzle
I'm using SF 15 and at first it seemed like it couldn't figure it out, but on my hardware (which isn't anything crazy fast) it needed 6 minutes and 36 seconds and depth 48 to realize that Bc6 is actually winning (+4.91).
The reason Stockfish only sees the mate for white after black plays bf6 (7:37) is that stockfish think that this "main line" is a less resilient defense for black than sac'ing the bishop on g3 (still losing, but not as fast). (It sees the mate in 15 for bf6).
Yeah also you can press the plus button on Lichess to enable local analysys, if you do that it does give an evaluation of +40.4 on the starting position after a while
It's just slightly outside of its depth range If you give it a few more hours to calculate more depth, than it will get this forced mating line. Its basically so bad at evaluating this because it's just one very long forced mate and everything else is either a draw or a lose
This took me about a half hour to solve, but I'm proud to have solved it! The first move is Bc6. This blocks the pawn on c7 and threatens Bd7#. The only way to prevent the mate is moving the room to either a8 or b7. If the rook moves to a8, then play Na5. This threatens Bd7+, and Nc6#. And it's mate no matter what black does. If the rook instead takes the knight on b7, then play Bd7+, and after Kb8, play Na6+ and after Ka8, play Bc6. This completely locks black position, only being able to move their Bishop and h-pawn. Then the white king comes over. Note that the h-pawn can't go to h3 without being inevitably captured by the King, so it remains stuck on h4, until the white king moves to the 5th rank. But before that happens, we need to make sure that black's bishop is not guarding b8. If Black tries to defend b8 from the b8-h2 diagonal. So that the bishop is on g3 and the pawn is on h4. Then go Kg4. I'm pretty sure black doesn't have enough time to set up the Bishop g3 and Pawn h4 position without white's king being on h3. And whatever move black makes either loses the h-pawn or makes the black bishop not guard b8. I should note that if black loses his h-pawn, then white can simply move his king to c8 and proceed to play Bxb7#, and the dark squared Bishop is completely useless. Now if the King starts going to the 5th rank, then the h-pawn starts moving forward. By the time the pawn reaches promotion, the white king will be on d7. Capture the promoted piece with the bishop. And black plays c6+ discovered check (c5+ leads to the same line) then play Kc8. If black does not give a check, then Bxc6 would threaten an unstoppable mate. If Black plays Rc7+, take the rook with the knight, and if Black is somehow guarding c7 with the Bishop, then King takes Bishop is mate. If black instead plays Rb8+, then again take the rook with the knight. And black must move the a-pawn to avoid Bxc6#. There's also the idea of Nxc6, imprisoning the king, and moving the Bishop over to b7 in 4 moves. Black will have enough time to make a queen, but not enough time to guard b7 and stop the mating idea. That's all the cases. This was an insane puzzle, and I took about half and hour typing this comment, because I realized there were a few cases I missed, and I had to think a bit more.
I actually missed the stalemate idea and the fact that the Bishop could not guard c7 either. I originally had the Bishop not guard c7 with the same Zugzwang idea, but I though it was unnecessary because I missed Rb6.
I also feel like this video missed one idea where black tried to guard the c7 square in from the b8-h2 diagonal. But I'm not 100% sure that's an idea by black.
At 5:33 if you press the blue and black cross next to CLOUD on Lichess analysis and wait a few minutes, around depth 45-46 Stockfish starts to recognize the position as a win for white.
so mainline (12:56) forces black bishop into a bad move, but what is stopping them from taking the move right back the next turn (or at any point after the fact, since the king moves slowly)? Back into the defensive position, black pawn can now push at will and the knight is no longer safe to check the king. Once the push has started, the king cant just turn around and stop the promotion, though c6 still guards it. Or is there some sort of ruling i dont know?
If Black moves the bishop back to d8, that slows the Q-promotion and the white king gets to c8 even faster: 12.Kf5 Bd8 13.Ke6 h3 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1. Now that the K is on c8 and not d7, 16...c6 is not a discovered check, so White can play 17.Bxc6 to pin the rook again, then 17...B-any 18.Bxb7#
Really fascinating puzzle, thank you for showcasing it! I was really doubtful that Stockfish actually can't solve it because I expected it to be able to calculate far enough to see the ideas. I'm using SF 15 and at first it seemed like it couldn't figure it out, but on my hardware (which isn't anything crazy fast) it needed 6 minutes and 36 seconds at depth 48 to realize that Bc6 is actually winning (+4.91). So Stockfish can solve it, but it needs relatively long.
Do you know what "depth" actually means? I always assumed it meant that Stockfish has fully searched the game tree that many moves ahead. Not sure though if it's actually the full tree or still a pruned tree. Anyhow, this puzzle wasn't actually 48 moves deep, or was it? So why would Stockfish suddenly at depth 48 see a win if it already existed earlier?
@@sebastiandierks7919 It's not moves, but half-moves, one move is white moves and then black moves while only one player moving is one ply=+1 depth. Stockfish has a very well optimized and sophisticated algorithm which determines which variations it calculates more deeply, it never goes through the entire tree (the exponential growth makes it to many possibilites). 48 depth mean that all variation SF evaluates as "possibly relevant" are calculated that deeply. It may be that one key move later in the puzzle was just very "unintuitive" to SF so it gets calculated to a necessary depth later than variations which SF deemed more promising. I'm no expert so this is the understanding off a relatively educated layman, but for a very rough understanding this should help.
Actually it's not. The 0 doesn't mean it's 100% a draw, the 0 means it isn't sure about the outcome yet. Thats why it's so annoying to evaluate openings with stockfish. So in this example unlike the ones where it gives advantage to one player it is actually slightly more accurate.
@@anarkonihilizmkaoscunihili9944 Thats how the engine evaluates. Its why it doesnt evaluates attacks as good. Because the engine isn't as scared and gets counterplay in at the right times, making the position even
If you press the plus button stockfish will analyse to a deeper level. I ran it for over an hour, to a depth of 69, and it gave the position a +53.5 point advantage for white, meaning: there must be a mate somewhere
I'm so proud that I've solved this. These types of puzzles show how endgames can be complex and I could miss some cool positional ideas even I'm analyzing with engines. Of course you can get a solution if you run it long enough but you can't spend much time on every position. Btw whoever composed this study is freaking genius
At 6:00, after Bd6 and Kg2, what is stopping black to play h4 then on the next move play Bg3? The pawn and bishop is protecting each other and the white king cannot take anything
7:17 Can’t black draw this, because of the pawns which are not immediately able to be captured? Seems even more difficult for white to win than with just a knight and a bishop.
Guys, I'll explain to you the concept of "depth" which is the Engines' number of moves that they can foretell and evaluate. The puzzle is just too long that stockfish can't evaluate it because it's already past the depth.
@@RapBopzShorts That doesn’t matter really, since stockfish can usually tell a winning position apart from a losing one even if there is no forced checkmate in its search. The issue is probably that when searching moves, stockfish often prunes branches in the search so that it can search less, and therefore be faster, and while doing so, it orders the moves so that it’s likely to prune more branches with a line that is bad for you. However, there’s a high chance that it just got confused and pruned a branch that actually was part of the solution, maybe because it didn’t think that some king move was worth considering.
If black is playing h4 before the white king (min.4.53) - I think is draw !...check..black has enough time to make the queen before the king is reaching line 7. min.4:53 ...h4 Kf3 , Be7 Ke4 , h3 if Ke5 , h2 and is a draw if Kf3 , still h2 can be moved and after Kg2 black can play Bd6...
I tried using Stockfish with as the highest settings possible, and it gives the starting point an evaluation of +27 (it suggests sacrificing the bishop to prolong the game)
im but a lowly 400 rated whos three months late but cant the bishop move back to a position where it protects that pawn at 13:08? please tell me why my thought is wrong
I was thinking the same. Tho, after black Bg3 white simply move kg4 and forces black to move. The basic idea is that black bishop need to guard both c7 and the h file pawn to make a draw. After white kg4, black needs to move the bishop. Black would either lose guard on c7 or lose the pawn in h file.
@@charliezong2938 Black doesn't care about the c7 pawn. It's actually bad for white if he takes the pawn too early because White needs his king to cross the seventh rank. If Black tries ...h5-h4 at the mentioned spot White will just play Kh3, Kg4 and walk his king up to d7. Black has to promote his pawn, White will capture the new queen and play proceeds just as in the main line.
12:19 another thing i don’t understand is why not taking even the pawn with the king ? cuz i mean bro if u take even if he tries to check u with his rook … it’s checkmate cuz u have to move ur king and then discover check with bishop and even if he defends with rook it checkmate that would be amazing checkmate but also yea keep playing like u want guys
after Kxc6, instead of playing Rb6+ black has ...Rb1 attacking the bishop preventing white from moving the king for discovered check, after white moves the bishop black can keep moving the rook on the first rank to attack the bishop preventing white from ever delivering the discovered check, so i dont think this works.
Technology changes are very fast and now Leela + BT4 find it immediately with RTX 2060 or better and SF 17 dev 4 threads and 8GB hash find the solution in a few minutes. When I say find the solution I mean give the right move with > 2.5 eval. The right first move is found instantly with a bad eval first but it remains always in the main line for both engines.
from the position at 12:44, after the black bishop moves out of the way, why doesn't black interrupt its pawn march for a move to play Bd8?
ปีที่แล้ว
Why black does not move pawn at H5 at 6:16 instead and than bishop to E7? In my eyes that creates a draw. We cannot proceed with king to check mate because we let black's pawn to go for a queen. We cannot take the pawn because it is guarded by bishop and only thing black has to do is to move bishop back and forth between E7 and F6. We cannot move anything else because we unstuck other black pawns and king. So isn't that a draw? What am I missing?
At 3:52 please someone explain why taking the rook with the knight doesn’t work? He didn’t give it as a possible move in the video, but it looks like the best move for white at this moment
But at 6:06, couldnt black just push the pawn 1 square so its defended by the bishop, and at that pointit would be impossible for white to make any progress
This piece of Genrich Kasparjan is very nice. I tried Stockfish 15 here to crunch the numbers down to depth 52, the machine was not able to find it and always shows -0.16. At depth 51 it even goes to -0.18 The main theme is that the w♚ has to cross the 7th rank somehow, so it's quite obvious that White has to block the a/c-pawns. And there is this nice dance of the w♚ with the b♝! The winning king move is quite hard to find...I cannot believe that this puzzle only got a 4th Honorable Mention!
9:37 a nicer cleaner option is to have the night go to b 4, causing the king to move to a 7 then moving the bishop to b 7 cutting off the kings escape. Then moving the night back to c6
6:14 Why not play pawn h4 for later playing bishop g3 and blocking you from threatening the pawn.. that way the pawn would be one square further down the line.
this is effectively the same thing as what was played a few moves later. once black moves the pawn to h5, white gets the king to g5, then once blacks bishop moves away from g6 it is no longer guarding f7 and white can march the king to c8 etc as was shown. its the exact same position except blacks bishop is on e2 or f1 rather than d7 which doesnt change anything
I think, Stockfish is actually right You just have to h4 earlier than you did at 6:11 and defend with bishop on g3 if necessary instead of trying to cut of the king. This means black can queen on h1 before white king get to c8 and gets the draw.
If Black plays ...Bd8, that slows the Q-promotion and the white king gets to c8 even faster: 13....Bd8 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1. Now that the K is on c8 and not d7, 16...c6 is not a discovered check, so White can play 17.Bxc6 to pin the rook again, then 17...B-any 18.Bxb7#
On my laptop, it takes SF16 about 8.5 minutes (1.8 G nodes, depth 86) to see that Bc6 is winning. After 16 minutes and more than 4G nodes, it reaches +200.
There are some people on here that say local Stockfish (meaning Stockfish installed on your computer and not the internet version) solves this, but local Stockfish does NOT solve it on my computer using 20 cores even after more than 8 hours with a depth of 104 and over a trillion positions searched: Stockfish_22121919_x64_bmi2: NNUE evaluation using nn-ad9b42354671.nnue enabled 104/82 8:17:49 1,314,424,433k 44,006k 0.00 1.Bf3xh5 c7-c6 2.Nb7-d6+ Kc8-c7 3.Nd6-e8+ Kc7-b6 4.Nc5-d7+ Kb6-b7 5.Nd7xb8 Kb7xb8 6.Bh5-f3 Bd8-g5 7.Bf3-g4 a7-a5 8.Ne8-d6 a5-a4 9.Nd6-c4 c6-c5 10.Ke2-d3 a4-a3 11.Nc4xa3 Kb8-c7 12.Bg4-f3 Bg5-e7 13.Na3-c4 Kc7-d8 14.Bf3-d1 Be7-h4 15.Kd3-e4 Bh4-e1 16.Ke4-d5 Kd8-c7 17.Bd1-e2 Be1-f2 18.Be2-d3 Kc7-b7 19.Kd5-e4 Kb7-c6 20.Bd3-e2 Bf2-h4 21.Be2-h5 Kc6-c7 22.Bh5-g4 Bh4-d8 23.Bg4-d1 Kc7-b8 24.Ke4-d5 Kb8-c8 25.Bd1-g4+ Kc8-c7 26.Bg4-h5 Bd8-h4 27.Kd5-e5 it is still stuck on 1.Bxh5
Given the importance you placed on the bishop staying in contact with c7, I think you also need to show the line where black chooses a set-up based on Bg3 and h4. It may only be a side-line in the computer analysis but it seems to me to be relevant to a position given as a chess problem.
I do not understand. At 13.07, after Kf5, the Black does not have to move the pawn to h3, it can just play Bd8 again covering c7. What am I missing? BTW, Stockfish at depth 75 cannot see the victory for White. Incredible.
5:23 I don't get it... I go with the Stockfishmove h4 for black and leave the bishop on d8. As soon as the King comes up, you go towards the first rank with the pawn eventually forcing the white bishop to take the promoted pawn. And black waits with the promotion until the black King hits the 7ths rank. Why? cause you play c6 with check. If the white King doesn't go to the 7th rank, you play waiting moves with the bishop.
After kg2, h4 allows other lines of defense for black, including the sacrifices that lead to stale mate. But it doesn't free the king because the pan is closer to promotion.
If Black plays ...Bd8, yes, that continues to protect the f7 square... but the White King gets to c8 a move earlier: 13....Bd8 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1. Now that the White King is on c8 and not d7, 16...c6 is not a discovered check, so after 16... c6 White can play 17.Bxc6 and continue to pin the rook. And then 18.Bxb7 is mate.
Here's the FEN/PGN for the starting position; 1rkb4/pNp5/8/2N4p/8/5B2/4K3/8 w - - 0 1 You can run through it yourself with whatever tools your preferred chess site offers. I really wish he would put the code in the video description or something, I like to get hands-on with these puzzles and play against 3200ELO stockfish
why king cant step on black squares? cant he just move away if he gets checked by the bishop? 8:36 what would happen if white king steps on a black square?
This is wrong, White will just play Kg2-h3-g4-f5-e6-d7. You are forced to promote your pawn to a queen, I will take it and play will proceed just like in the main line.
If Black plays ...Bd8, that slows the Q-promotion and the white king gets to c8 even faster: 12.Kf5 Bd8 13.Ke6 h3 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1. Now that the K is on c8 and not d7, 16...c6 is not a discovered check, so White can play 17.Bxc6 to pin the rook again, then 17...B-any 18.Bxb7#
Guys, a Chess Vibes bot is coming soon to Hercules Chess: ⏩ herculeschess.com/chessvibes ⏪
what, video is minutes ago and this comment is 13 days?
Bishop could have just keep the pawn alive and move back and forth that’s why stock fish was right with 0.0
This app is sooo laggy tho. Took me 2 min to get 1 puzzle
@@GivingOthersAMentalBreakdown 😊😊😊
at which moment does the engine realise that white wins ?
I finally have something in common with Stockfish. Now.. I’m not saying I’m stockfish, but no one has ever seen me and stockfish in the same room
And i also have something common with " Stockfish " , i can not solve this puzzle . 😅
I always find these puzzles so satisfying, knowing that even the strongest chess engine stockfish cant solve it at first glance. But what would happen if you would run stockfish natively on this puzzle
Stockfish wasn't running at a high level, at the level stockfish was running a very strong player ie Magnus could beat it.
@@maimonguy123 Pretty sure a cloud Stockfish 14 with depth 36 and NNUE enabled is a high enough level
@@maimonguy123 Stockfish, even at regular depth 20 could beat pretty much everyone on planet earth in any time format
@@maimonguy123 No, a depth 36 browser Stockfish is plenty enough to wipe the floor with Magnus and every human.
I'm using SF 15 and at first it seemed like it couldn't figure it out, but on my hardware (which isn't anything crazy fast) it needed 6 minutes and 36 seconds and depth 48 to realize that Bc6 is actually winning (+4.91).
The reason Stockfish only sees the mate for white after black plays bf6 (7:37) is that stockfish think that this "main line" is a less resilient defense for black than sac'ing the bishop on g3 (still losing, but not as fast). (It sees the mate in 15 for bf6).
Yeah also you can press the plus button on Lichess to enable local analysys, if you do that it does give an evaluation of +40.4 on the starting position after a while
After the first two moves are given, Stockfish 15.1 with Syzygy gets it almost instantly. For some reason though, Stockfish can’t see it before this.
It's just slightly outside of its depth range
If you give it a few more hours to calculate more depth, than it will get this forced mating line.
Its basically so bad at evaluating this because it's just one very long forced mate and everything else is either a draw or a lose
This took me about a half hour to solve, but I'm proud to have solved it!
The first move is Bc6. This blocks the pawn on c7 and threatens Bd7#. The only way to prevent the mate is moving the room to either a8 or b7.
If the rook moves to a8, then play Na5. This threatens Bd7+, and Nc6#. And it's mate no matter what black does.
If the rook instead takes the knight on b7, then play Bd7+, and after Kb8, play Na6+ and after Ka8, play Bc6. This completely locks black position, only being able to move their Bishop and h-pawn.
Then the white king comes over. Note that the h-pawn can't go to h3 without being inevitably captured by the King, so it remains stuck on h4, until the white king moves to the 5th rank. But before that happens, we need to make sure that black's bishop is not guarding b8. If Black tries to defend b8 from the b8-h2 diagonal. So that the bishop is on g3 and the pawn is on h4. Then go Kg4. I'm pretty sure black doesn't have enough time to set up the Bishop g3 and Pawn h4 position without white's king being on h3. And whatever move black makes either loses the h-pawn or makes the black bishop not guard b8. I should note that if black loses his h-pawn, then white can simply move his king to c8 and proceed to play Bxb7#, and the dark squared Bishop is completely useless.
Now if the King starts going to the 5th rank, then the h-pawn starts moving forward. By the time the pawn reaches promotion, the white king will be on d7. Capture the promoted piece with the bishop. And black plays c6+ discovered check (c5+ leads to the same line) then play Kc8. If black does not give a check, then Bxc6 would threaten an unstoppable mate. If Black plays Rc7+, take the rook with the knight, and if Black is somehow guarding c7 with the Bishop, then King takes Bishop is mate. If black instead plays Rb8+, then again take the rook with the knight. And black must move the a-pawn to avoid Bxc6#. There's also the idea of Nxc6, imprisoning the king, and moving the Bishop over to b7 in 4 moves. Black will have enough time to make a queen, but not enough time to guard b7 and stop the mating idea.
That's all the cases. This was an insane puzzle, and I took about half and hour typing this comment, because I realized there were a few cases I missed, and I had to think a bit more.
I actually missed the stalemate idea and the fact that the Bishop could not guard c7 either. I originally had the Bishop not guard c7 with the same Zugzwang idea, but I though it was unnecessary because I missed Rb6.
I also feel like this video missed one idea where black tried to guard the c7 square in from the b8-h2 diagonal. But I'm not 100% sure that's an idea by black.
As a 700-800 rated player, I tend to get some moves on these puzzles correct without knowing why they're correct
A stopped clock tells the right time twice a day.
An airplane flies higher than a light bulb.
Sometimes we forgot we are breathing
Every 60 seconds in Africa a minute passes.
As a 300-400 rated player, I tend to get some of these moves wrong without knowing why they're wrong.
At 5:33 if you press the blue and black cross next to CLOUD on Lichess analysis and wait a few minutes, around depth 45-46 Stockfish starts to recognize the position as a win for white.
When the game review tells you you had a missed win.
The missed win:
so mainline (12:56) forces black bishop into a bad move, but what is stopping them from taking the move right back the next turn (or at any point after the fact, since the king moves slowly)? Back into the defensive position, black pawn can now push at will and the knight is no longer safe to check the king. Once the push has started, the king cant just turn around and stop the promotion, though c6 still guards it. Or is there some sort of ruling i dont know?
If Black moves the bishop back to d8, that slows the Q-promotion and the white king gets to c8 even faster: 12.Kf5 Bd8 13.Ke6 h3 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1. Now that the K is on c8 and not d7, 16...c6 is not a discovered check, so White can play 17.Bxc6 to pin the rook again, then 17...B-any 18.Bxb7#
Really fascinating puzzle, thank you for showcasing it!
I was really doubtful that Stockfish actually can't solve it because I expected it to be able to calculate far enough to see the ideas. I'm using SF 15 and at first it seemed like it couldn't figure it out, but on my hardware (which isn't anything crazy fast) it needed 6 minutes and 36 seconds at depth 48 to realize that Bc6 is actually winning (+4.91).
So Stockfish can solve it, but it needs relatively long.
Do you know what "depth" actually means? I always assumed it meant that Stockfish has fully searched the game tree that many moves ahead. Not sure though if it's actually the full tree or still a pruned tree. Anyhow, this puzzle wasn't actually 48 moves deep, or was it? So why would Stockfish suddenly at depth 48 see a win if it already existed earlier?
@@sebastiandierks7919 It's not moves, but half-moves, one move is white moves and then black moves while only one player moving is one ply=+1 depth.
Stockfish has a very well optimized and sophisticated algorithm which determines which variations it calculates more deeply, it never goes through the entire tree (the exponential growth makes it to many possibilites). 48 depth mean that all variation SF evaluates as "possibly relevant" are calculated that deeply.
It may be that one key move later in the puzzle was just very "unintuitive" to SF so it gets calculated to a necessary depth later than variations which SF deemed more promising.
I'm no expert so this is the understanding off a relatively educated layman, but for a very rough understanding this should help.
@@ElKartoffelbreii thanks, if it's 48 plys rather than full moves it's much more realistic and correlates better with the actual length of the puzzle.
😅 I didn't understand anything
@@BabitaKumari-hk9ci I didn't understand the Russian either😅
This is like the reverse of some of the puzzles, in which stockfish shows a drawn position as an advantage for one player.
Actually it's not.
The 0 doesn't mean it's 100% a draw, the 0 means it isn't sure about the outcome yet.
Thats why it's so annoying to evaluate openings with stockfish.
So in this example unlike the ones where it gives advantage to one player it is actually slightly more accurate.
@@mauer1 Proof? Also still it's good puzzle due to engine didn'r see win.
@@anarkonihilizmkaoscunihili9944
Thats how the engine evaluates.
Its why it doesnt evaluates attacks as good. Because the engine isn't as scared and gets counterplay in at the right times, making the position even
@@mauer1 Do you actually watch chess engine programmers videos or just read codes of engine?
@@anarkonihilizmkaoscunihili9944 that's how engines work, unless it's too obvious or a forced line they don't have the enough depth to get it
If you press the plus button stockfish will analyse to a deeper level. I ran it for over an hour, to a depth of 69, and it gave the position a +53.5 point advantage for white, meaning: there must be a mate somewhere
Nah bro asking me to solve a puzzle that a computer cannot solve 💀 💀
I'm so proud that I've solved this. These types of puzzles show how endgames can be complex and I could miss some cool positional ideas even I'm analyzing with engines. Of course you can get a solution if you run it long enough but you can't spend much time on every position.
Btw whoever composed this study is freaking genius
At 6:00, after Bd6 and Kg2, what is stopping black to play h4 then on the next move play Bg3? The pawn and bishop is protecting each other and the white king cannot take anything
7:17 Can’t black draw this, because of the pawns which are not immediately able to be captured? Seems even more difficult for white to win than with just a knight and a bishop.
Wow thanks for Hercules Chess very helpful!
Stockfish is just able to calculate billion times faster than humans, but it doesn't understand the game.
But does it really not understand?
Maybe stockfish does know but he tricking us to think it doesn’t?
Guys, I'll explain to you the concept of "depth" which is the Engines' number of moves that they can foretell and evaluate. The puzzle is just too long that stockfish can't evaluate it because it's already past the depth.
@@RapBopzShorts That doesn’t matter really, since stockfish can usually tell a winning position apart from a losing one even if there is no forced checkmate in its search. The issue is probably that when searching moves, stockfish often prunes branches in the search so that it can search less, and therefore be faster, and while doing so, it orders the moves so that it’s likely to prune more branches with a line that is bad for you. However, there’s a high chance that it just got confused and pruned a branch that actually was part of the solution, maybe because it didn’t think that some king move was worth considering.
This problem hasn’t anything to do with understanding, it is all calculating. But at first to deep for Stockfish.
Stockfish right its a draw..with perfect play...but he makes wrong move for black that gives white win
If black is playing h4 before the white king (min.4.53) - I think is draw !...check..black has enough time to make the queen before the king is reaching line 7.
min.4:53
...h4
Kf3 , Be7
Ke4 , h3
if Ke5 , h2 and is a draw
if Kf3 , still h2 can be moved and after Kg2 black can play Bd6...
I really enjoy the puzzle videos. Thank you for sharing your knowledge with us.
I tried using Stockfish with as the highest settings possible, and it gives the starting point an evaluation of +27 (it suggests sacrificing the bishop to prolong the game)
12:19 king takes pawn rook b6 check king d7 rook b7 check and defends king king c8 and its still checkmate after bishop takes rook
6:19 Can t black play pawn h4 then bishop g3 ?
That's so cool. I checked out Hercules chess a while ago and it was pretty fun so its cool that they are adding more stuff
Genrikh Kasparyan was a genius.
im but a lowly 400 rated whos three months late but cant the bishop move back to a position where it protects that pawn at 13:08? please tell me why my thought is wrong
What
6:09 what if black plays Ph4? and if Kh3 black can Bg3 and keep the bishop on the e diagonale? Would that help to keep the game a draw?
I was thinking the same. Tho, after black Bg3 white simply move kg4 and forces black to move. The basic idea is that black bishop need to guard both c7 and the h file pawn to make a draw. After white kg4, black needs to move the bishop. Black would either lose guard on c7 or lose the pawn in h file.
h4, not Ph4
@@charliezong2938 Black doesn't care about the c7 pawn. It's actually bad for white if he takes the pawn too early because White needs his king to cross the seventh rank. If Black tries ...h5-h4 at the mentioned spot White will just play Kh3, Kg4 and walk his king up to d7. Black has to promote his pawn, White will capture the new queen and play proceeds just as in the main line.
@@eniszuferi Black care about bishop guarding c7 to prevent the Nc7 checkmate not the pawn.
@@charliezong2938 which checkmate bro? Nxc7+ isn't checkmate unless the white king reaches c8
12:19 another thing i don’t understand is why not taking even the pawn with the king ? cuz i mean bro if u take even if he tries to check u with his rook … it’s checkmate cuz u have to move ur king and then discover check with bishop and even if he defends with rook it checkmate that would be amazing checkmate but also yea keep playing like u want guys
after Kxc6, instead of playing Rb6+ black has ...Rb1 attacking the bishop preventing white from moving the king for discovered check, after white moves the bishop black can keep moving the rook on the first rank to attack the bishop preventing white from ever delivering the discovered check, so i dont think this works.
6:16 bishop to E7 and pawn to H4. Bishop protect pawn and its a draw.
Technology changes are very fast and now Leela + BT4 find it immediately with RTX 2060 or better and SF 17 dev 4 threads and 8GB hash find the solution in a few minutes. When I say find the solution I mean give the right move with > 2.5 eval. The right first move is found instantly with a bad eval first but it remains always in the main line for both engines.
from the position at 12:44, after the black bishop moves out of the way, why doesn't black interrupt its pawn march for a move to play Bd8?
Why black does not move pawn at H5 at 6:16 instead and than bishop to E7? In my eyes that creates a draw. We cannot proceed with king to check mate because we let black's pawn to go for a queen. We cannot take the pawn because it is guarded by bishop and only thing black has to do is to move bishop back and forth between E7 and F6. We cannot move anything else because we unstuck other black pawns and king. So isn't that a draw? What am I missing?
The thumbnail is what I say when stock fish calls my brilliant sac a blunder
I love these puzzles!!
At 3:52 please someone explain why taking the rook with the knight doesn’t work? He didn’t give it as a possible move in the video, but it looks like the best move for white at this moment
Managed to get it right until the moment where the king starts to move, then got completely lost. Still proud for having figured out some of it.
13:13 What if the black doesn't play rook and play knight on d6. That is better move then playing rook on b6.(correct me if I am wrong)
But at 6:06, couldnt black just push the pawn 1 square so its defended by the bishop, and at that pointit would be impossible for white to make any progress
This is probably something like mate in 40 with perfect play
Stockfish no longer uses pruned trees, the AI recognizes lines
It is a draw because at 6:24 black will push a pawn then anywhere the king moves bishop will come to g3
Latest stockfish can solve this. It took 11 mins on my server and result is +M53.
Great way to wake up to find out I’m just as good as Stockfish.... that is, until Stockfish gets his coffee and get that W!
9:51 "We literally don't even care" + that smirk hints that this is an obvious Hikaru reference
13:01 why can't black bishop just go back to d8?
This piece of Genrich Kasparjan is very nice. I tried Stockfish 15 here to crunch the numbers down to depth 52, the machine was not able to find it and always shows -0.16. At depth 51 it even goes to -0.18
The main theme is that the w♚ has to cross the 7th rank somehow, so it's quite obvious that White has to block the a/c-pawns. And there is this nice dance of the w♚ with the b♝! The winning king move is quite hard to find...I cannot believe that this puzzle only got a 4th Honorable Mention!
13:14 what is balck moves the C6 pawn then thats a win for black correct me if i am wrong
It would've been cool if there was a segment at the end where you beat stockfish in this position
Stockfish is bad now
Not stockfishs he is quite wrong theres other moves to start trying to trade pieces and they could always repeat moves after he locks his king
Btw by defending his pawn in his bishops and King cant go for bishop cause then pawn woud queen
@@boazscheper9743 This guy thinks he really better than stockfish and still believe its a draw
*Proceeds to beat mittens*
9:37 a nicer cleaner option is to have the night go to b 4, causing the king to move to a 7 then moving the bishop to b 7 cutting off the kings escape. Then moving the night back to c6
Why can’t the King move from a7 to b6 and escape?
@@frankieheng4977 touche
Honestly, Hercules Chess is dope.
I find these puzzles absolutely amazing at first glance it’s easily a draw but the way you mate with white is beautiful
Bro its still draw but he blunder with black
@@alsaid9 watch the vid bro
At 2:40 what about the move knight to c7?
Why can horse on b7 go to d6 and force checkmate?
11:53 a brilliant move
6:14 Why not play pawn h4 for later playing bishop g3 and blocking you from threatening the pawn.. that way the pawn would be one square further down the line.
this is effectively the same thing as what was played a few moves later. once black moves the pawn to h5, white gets the king to g5, then once blacks bishop moves away from g6 it is no longer guarding f7 and white can march the king to c8 etc as was shown. its the exact same position except blacks bishop is on e2 or f1 rather than d7 which doesnt change anything
I think, Stockfish is actually right You just have to h4 earlier than you did at 6:11 and defend with bishop on g3 if necessary instead of trying to cut of the king.
This means black can queen on h1 before white king get to c8 and gets the draw.
Wow another great one, thank you!
13:08 what about d8 with bishop?
13:07, why can't the bishop just go back instead of pushing the pawn? still same issue and pawn is free and clear.
If Black plays ...Bd8, that slows the Q-promotion and the white king gets to c8 even faster: 13....Bd8 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1. Now that the K is on c8 and not d7, 16...c6 is not a discovered check, so White can play 17.Bxc6 to pin the rook again, then 17...B-any 18.Bxb7#
Can't wait for 9-men tablebases to see the perfect play for this position.
On my laptop, it takes SF16 about 8.5 minutes (1.8 G nodes, depth 86) to see that Bc6 is winning. After 16 minutes and more than 4G nodes, it reaches +200.
13:07 what about bishop to d8?
I'm all about smartypants Stockfish getting owned. Nevermind that I had no idea what to do, either. :/
Why not Bishop F4 at 9:18?
You mean at 9:21? ...Bf4, Bxc6 pins the rook and the bishop mates next move.
@@Rocky64oh ok
6:17 why did you play be5 instead of just pushing the pawn?
That was fascinating
8:03 I think the only useful move for black here is putting the bishop on f4 and leave it there if the king isnt going to take it.
Indeed!! He is dumb that black would go for the queen while you can clearly see if he has the queen he wil lose it immediatly..
There are some people on here that say local Stockfish (meaning Stockfish installed on your computer and not the internet version) solves this, but local Stockfish does NOT solve it on my computer using 20 cores even after more than 8 hours with a depth of 104 and over a trillion positions searched:
Stockfish_22121919_x64_bmi2:
NNUE evaluation using nn-ad9b42354671.nnue enabled
104/82 8:17:49 1,314,424,433k 44,006k 0.00 1.Bf3xh5 c7-c6 2.Nb7-d6+ Kc8-c7 3.Nd6-e8+ Kc7-b6 4.Nc5-d7+ Kb6-b7 5.Nd7xb8 Kb7xb8 6.Bh5-f3 Bd8-g5 7.Bf3-g4 a7-a5 8.Ne8-d6 a5-a4 9.Nd6-c4 c6-c5 10.Ke2-d3 a4-a3 11.Nc4xa3 Kb8-c7 12.Bg4-f3 Bg5-e7 13.Na3-c4 Kc7-d8 14.Bf3-d1 Be7-h4 15.Kd3-e4 Bh4-e1 16.Ke4-d5 Kd8-c7 17.Bd1-e2 Be1-f2 18.Be2-d3 Kc7-b7 19.Kd5-e4 Kb7-c6 20.Bd3-e2 Bf2-h4 21.Be2-h5 Kc6-c7 22.Bh5-g4 Bh4-d8 23.Bg4-d1 Kc7-b8 24.Ke4-d5 Kb8-c8 25.Bd1-g4+ Kc8-c7 26.Bg4-h5 Bd8-h4 27.Kd5-e5
it is still stuck on 1.Bxh5
6:17 black should push pawn and play bishop g3…
push pawn then king h3, bishop g3 then king g4 - mate in 11
Then they won’t be guarding c7, leading to that mating line where it involves Nxc7#
Great video
Given the importance you placed on the bishop staying in contact with c7, I think you also need to show the line where black chooses a set-up based on Bg3 and h4. It may only be a side-line in the computer analysis but it seems to me to be relevant to a position given as a chess problem.
I do not understand. At 13.07, after Kf5, the Black does not have to move the pawn to h3, it can just play Bd8 again covering c7. What am I missing? BTW, Stockfish at depth 75 cannot see the victory for White. Incredible.
5:23 I don't get it... I go with the Stockfishmove h4 for black and leave the bishop on d8. As soon as the King comes up, you go towards the first rank with the pawn eventually forcing the white bishop to take the promoted pawn. And black waits with the promotion until the black King hits the 7ths rank. Why? cause you play c6 with check. If the white King doesn't go to the 7th rank, you play waiting moves with the bishop.
At 5:00, why isn't h4 played? A covered pawn is a draw.
12:19 why not taking bishop with the king after discover check ? so if he does another move it’s simply checkmate with ur bishop
there are so many wonderful checkmate parterns in this puzzle
13:07 Cannot this black bishop come back to d8?
No he loses a tempo and the king can take
After kg2, h4 allows other lines of defense for black, including the sacrifices that lead to stale mate. But it doesn't free the king because the pan is closer to promotion.
if you ever feel useless, remember the dark squared bishop in this game.
Amazing, thanks so much
seems I'm still bad .. why in 6:15 not move the pawn, then bishop g4 and is a draw...
okay but at 13:06 what if black moves bishop d8 instead of pushing the pawn once
why can't black just move the bishop elsewhere at 11:31
like e7
Brilliant Puzzle!
What if instead of Bishop D1 -> Bishop E2 you play bishop D1 -> bishop G5 check? 12:48
@Chess Vibes I have a question, see this position -> 13:07 "What if the black play Bd8 instead of h2?"
If Black plays ...Bd8, yes, that continues to protect the f7 square... but the White King gets to c8 a move earlier: 13....Bd8 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1.
Now that the White King is on c8 and not d7, 16...c6 is not a discovered check, so after 16... c6 White can play 17.Bxc6 and continue to pin the rook. And then 18.Bxb7 is mate.
@@MrEdwardCollins thanks
I would like to see the same puzzle with stockfish on aswell
Here's the FEN/PGN for the starting position;
1rkb4/pNp5/8/2N4p/8/5B2/4K3/8 w - - 0 1
You can run through it yourself with whatever tools your preferred chess site offers.
I really wish he would put the code in the video description or something, I like to get hands-on with these puzzles and play against 3200ELO stockfish
personally i prefer seeing the mouse, hard to explain why
9:47 the easier way to win is to go knight to a5
Then bishop B7 and knight C6? I was wondering if that was correct
At 5:42 what if black pawn moves forward once
Ok so this is the hardest move to make, we’re gonna send our king to the moon
why king cant step on black squares? cant he just move away if he gets checked by the bishop? 8:36 what would happen if white king steps on a black square?
At 6:09 black can push pawn forward and after Kg4 they can simply play bishop to g3, after that it's infinity loop of king and bishop playing
Yea he clearly ignored pawn to h4
I am pretty sure this is why Stockfish can't solve this
This is wrong, White will just play Kg2-h3-g4-f5-e6-d7. You are forced to promote your pawn to a queen, I will take it and play will proceed just like in the main line.
At around 13:00 wha happens after Kf5 and Bd8?
If Black plays ...Bd8, that slows the Q-promotion and the white king gets to c8 even faster: 12.Kf5 Bd8 13.Ke6 h3 14.Kd7 h2 15.Kc8 h1=Q 16.Bxh1. Now that the K is on c8 and not d7, 16...c6 is not a discovered check, so White can play 17.Bxc6 to pin the rook again, then 17...B-any 18.Bxb7#
Stalemate is from leaving black pawn on blacks square while bishop dances to keep the king away and threatens queen.
Hi, I appriciate the puzzle, but I have a question. Does not King to G8 make checkmate on 9:55 ?
Thank you for the answers :)
You can't attack a king with another king
@@turicagamer9951 thank you, I didn’t know :)
At 14:03 with K - c8 why can't Black play B -f4?
This is the hardest puzzle I have ever seen in my life