Great fun watching this with a different engine analysing in parallel - a really tricky position - the engine immediately refutes many of the moves you consider and it suggests moves you never consider. Fascinating!!
The table-base ending was super - just so tricky! The engine i used rated the position +5.3 at depth 24. Well done for getting the king into the corner!
I tried SF v SF and it was able to win. But the moves were incomprehensible and white played far less solid with pieces all disjoint and scattered and the king appeared wide open but somehow it all held together.
@Utsav Joshi It wasn’t using tablebase the way I did it. I was just watching an analysis board top 5 lines and making moves based on that. Evals were changing. If it was tablebase it would eval to mate in X but eval started at +1.9 and changed over time.
@Utsav Joshi Then it doesn’t have a full 7 piece tablebase, otherwise it’s eval would never change it would simply look up the initial position and it’s eval would be mate in X with no searching. Tablebase values are either mate in X or draw.
That's why it's hard to win against engines, it kind of played the move backwards from the desirable outcomes, instead of playing for The outcomes... If you understand what I mean... That's why the moves seem incomprehensible completely!
Something interesting at 14:58, the bishop was tactically defended by Nd6 winning the queen, but doesn't work because the king can go to e5 and take the other bishop and you can't force mate with just two knights. I think that was a cool example of demonstrating how easy missing the smallest detail could result in a loss.
Bd3 is winninig Unfortunately, the 7-man TB Lomonosov server is unavailable for a long time, probably due to the geopolitical situation, but adjudicateing by DTZ 62, it may be a mate in 31 moves
In a perfect world maybe, though chess would never be as beautiful and never get to these beautiful positions if there weren't humans playing and doing "mistakes". So a labeled drawn position might not be draw to to the human eye, on both sides! If both players can't look deep enough to abuse the drawn position to be actually a draw, it's not really a draw. The game is always open based on your skill level... On the other hand I think most of the so called puzzles are pretty nice to look at, were there is only very distinct moves winning you the position/mate pattern, difference is they are somewhat solvable with effort. But a nice puzzle, not some inhuman facts about what is drawn if the other one played only perfect moves *30++turns... !!! wtf??* So I got no clue why some feel obliged to talk about M31, unless it's somehow helpful to us, which it isn't UNLESS you can remember the 31 moves (including variations and all kinds of bad moves of the enemy, too!) for the future and other similar positions and games... (which would be pretty high IQ) - including the fact that some of these endgame positions are so rare it's much more effort to studying opening theory - then sorry it just sounds like some Dunning Kruger wanna sound smart! Anyways I sure hope you have fun with chess in the future the way you fancy.
You can actually find all 12 bad moves pretty easily without engine or tablebase assistance. All of them involve carelessly dropping one of the minor pieces to where the queen can capture them, or moving the king to hang the knight.
I’ve heard Fabi say that he thinks the biggest thing engines have taught us is just how many defensive resources there are in most positions and playing against stockfish in “winning” positions makes that so clear.
@@bocasuja22 Yes those normally solid approaches are what led to complications in this situation. Knowing when to set aside 'best practices' is tough...bordering on impossible. We shy from doing it firstly because it skyrockets our uncertainty as we move into unfamiliar territory. And we can't help but remember plenty of failures when we rolled the dice believing it was time to get creative...only to find out we needed more patience plowing forward with best practices. The game simply becomes too complicated for humans at a level well below the ultimate drawn nature of the game. Timed games only serve to push the bar lower. So probably best to spend our time becoming super competent in common positions. At one time I knew how to mate with B,N in under 50 moves. That is for all practical purposes (human play) worthless info. Much like this problem. It's fine to mess around with occasionally. But not worth getting worked up about.
Brilliant move at the start would likely be Knight-f5 Checks the king, either the king moves away (removing his ability to assist checkmate) or Queen takes f5, leading into Bishop takes f5
18:00 instead of trying to save the knight from the queen, move the other knight to d3 to check the king and force him into the corner. From that position you should be able to keep checking the king until checkmate.
Great video idea, I liked this a lot. Would be interesting to see the other way around too. Have Nelson play the black pieces and see how strong of an engine he can force a draw against.
This was really instructive, thank you. I think you needed to use the fact that you could attack their queen a bit more; just in the same way that they used attacking your minor pieces as a resource when checks were not available like here 17:24 . You didn't turn to this nearly as much
My man broke stockfish. I gotta say your content has been on point I've been way more interested in your endgame studies and little challenges than I have been other places like Gotham that used to be my goto chess content but his recaps of draw after draw wear on me
@@monke7566 A good game of chess is a good game of chess. Someone slipping up and dropping the half point doesn't enrich the game, a lot of the draws in the last Candidates tournament were more exciting than many of the wins.
Nelson, interesting position. Perhaps you take a wrong road in modelling a systematic procedure the same as two lone bishops mating net (neither side has anything else, except white has 2 bishops). Remember that a "front" can be created in front of the BK with a single knight, using one bishop to fill in the holes in the knight's coverage. I wondered about you trying at move 1 WN d4. If 1 .... Qa2. Interesting is 2 Ng3. Threat here is 3Nc4+... and 4N/g3-a5; where the BQ is trapped firmly as if on flypaper. But not clear to me how white can force a win; and hence 2Ng3 may not in the end be the way to go. The approach of a forward bishop restricting the BK, with the bishop being protected from behind by a knight seemed to me to be rather passive; as both the bishop and knight are somewhat immobilized thereby. Still, I can't say any other clear win is obvious; the position appears to me to be too complicated for exact calculation (by me anyway). I would encourage you to experiment with other approaches; besides simply keeping the two bishops side by side, as is done in a mate with only two bishops.
well its a draw with bishop on b2{a1} and Knight on d4 and king on a2/b1 or any corner where the bishop is the same colour as the corner square. So one would hope you could win with an extra 2 pieces.
Mr. Nelson, I'm so sorry for my comment, which you've deleted. I like your videos very much, ideed. All I wanted to say was that even a strong Stockfish can't win the given position by playing against itself unless using 7-men Syszygy TB. Thus, no human player can win that position against a strong Stockfish either. A strong Stockfish program uses the complete 6-men Syzygy TB, 16 GB Hash and 8 minutes per half-move. I wish you all the best.
Hey bud, sometimes RUclips automatically deletes comments if they think it's spam. Maybe that's what happened cause I don't remember reading this comment before. :-)
I tried defending as black, giving white a position where all their pieces are in the corner and black's pieces are in the middle. I lost in 14 moves by checkmate. Somehow the knights and bishops were able to easily cover all the checks I have and force my king to the corner
I'm going to be real --this one was super boring to me. So I looked it up and this is a tablebase "win", but it takes about 65 moves with no captures, which means it's a draw. Practicing against stockfish is pointless because it will always hold the draw.
well, thats only if everything is centralized when you start and assuming perfect play from your opponent. If your opponent slips up and plays less optimally, you can do it faster than 65 moves, and if your pieces are better placed, or if the black pieces are not as centralized, you can also do it much faster than 65 moves. So in theory, in MOST situations where you could get this kinda endgame, this is going to be a theoretical win. The difficulty just comes with actually practically doing it over the board.
@@maxscherzer9521 Well yea, stockfish wont, But i was mostly talking about in a real game against a human opponent. Like, winning this kinda situation vs a human player is much more manageable than it is against stockfish. While its still difficult, the 4 minor pieces vs queen should still be a win against most humans even if you have to play that ending out. Most good players should be able to checkmate with that against a human. Its only when you get to stockfish with perfect play where the task becomes quite difficult especially with the move limit as not every position is a viable way to win within 50 moves. (although most still should be possible, this was just a pretty optimal position for black to start out with).
Maybe the idea is, you divide pieces, one N+B and K mates the king and the other two handle checks from the queen -- the natural desire to employ all pieces against the king and also to keep them in a tight group is probably inefficient?
lesson of the story: 4 pieces VS Queen strategy is "harass the queen into a shitty position than mate becomes possible" queen can't trade because even with trade it's still a loss probably to knight and bishop or 2 bishops(as losing the two bishops makes no sense as your knight should be guarding them)
I don't even need to open the video to understand that the Qween alone is USELESS even against 3 pieces (not necessarily 4) That formation is absolutely impenetrable. The best queen can do is to pull out a draw game
I actually have 0 ability but about halfway through your video I did start wondering well why instead of constantly running from the queen make the queen run from you. Which in a wide open board and king in the middle probably still extremely difficult
Tips to black player: Keep king and queen distance more than four squares to avoid 🐴 horse 2in1 trap. And also keep King and Queen to different coloured ⬛⬜ squares to avoid Bishop's trap. 👍👍👍
![YoungBoy Never Broke Again - Missing Everything [Official Video]](http://i.ytimg.com/vi/dTYeMZGzOzw/mqdefault.jpg)
Nelson: "We have you outnumbered."
Stockfish: "True, but we have you outmatched."
Great fun watching this with a different engine analysing in parallel - a really tricky position - the engine immediately refutes many of the moves you consider and it suggests moves you never consider. Fascinating!!
The table-base ending was super - just so tricky! The engine i used rated the position +5.3 at depth 24. Well done for getting the king into the corner!
No
@@arneshpal7702 yes.
Very educational
I tried SF v SF and it was able to win. But the moves were incomprehensible and white played far less solid with pieces all disjoint and scattered and the king appeared wide open but somehow it all held together.
@Utsav Joshi It wasn’t using tablebase the way I did it. I was just watching an analysis board top 5 lines and making moves based on that. Evals were changing. If it was tablebase it would eval to mate in X but eval started at +1.9 and changed over time.
@Utsav Joshi Then it doesn’t have a full 7 piece tablebase, otherwise it’s eval would never change it would simply look up the initial position and it’s eval would be mate in X with no searching. Tablebase values are either mate in X or draw.
@Utsav Joshi there are different versions of stockfish with different plugins not all use tablebase
That's why it's hard to win against engines, it kind of played the move backwards from the desirable outcomes, instead of playing for The outcomes... If you understand what I mean... That's why the moves seem incomprehensible completely!
I tried this for a couple hours last night, too. It was incredibly hard. Best I got was landing in a stalemate 4 times.
Same. I could get the king towards the edge of the board, but couldn't get the queen away at the same time.
Where I can play against stockfish?
Something interesting at 14:58, the bishop was tactically defended by Nd6 winning the queen, but doesn't work because the king can go to e5 and take the other bishop and you can't force mate with just two knights. I think that was a cool example of demonstrating how easy missing the smallest detail could result in a loss.
i read the first part of ur comment and was about to comment that the king could take the other bishop and its a draw lmao
Funny how stockfish just moves the king and it instantly shatters your confidence and breaks your soul.
You made stockfish resign by breaking it. That is a win :)
Fun fact: there are 34 possible moves in the first move, 22 are winning, 12 are drawing
Bd3 is winninig
Unfortunately, the 7-man TB Lomonosov server is unavailable for a long time, probably due to the geopolitical situation, but adjudicateing by DTZ 62, it may be a mate in 31 moves
And zero are losing in that move xd
@@ac0unt_un04 I mean, if you cannot win the game in any way, try to draw the game.
In a perfect world maybe, though chess would never be as beautiful and never get to these beautiful positions if there weren't humans playing and doing "mistakes". So a labeled drawn position might not be draw to to the human eye, on both sides! If both players can't look deep enough to abuse the drawn position to be actually a draw, it's not really a draw. The game is always open based on your skill level... On the other hand I think most of the so called puzzles are pretty nice to look at, were there is only very distinct moves winning you the position/mate pattern, difference is they are somewhat solvable with effort. But a nice puzzle, not some inhuman facts about what is drawn if the other one played only perfect moves *30++turns... !!! wtf??* So I got no clue why some feel obliged to talk about M31, unless it's somehow helpful to us, which it isn't UNLESS you can remember the 31 moves (including variations and all kinds of bad moves of the enemy, too!) for the future and other similar positions and games... (which would be pretty high IQ) - including the fact that some of these endgame positions are so rare it's much more effort to studying opening theory - then sorry it just sounds like some Dunning Kruger wanna sound smart! Anyways I sure hope you have fun with chess in the future the way you fancy.
You can actually find all 12 bad moves pretty easily without engine or tablebase assistance. All of them involve carelessly dropping one of the minor pieces to where the queen can capture them, or moving the king to hang the knight.
30 minutes of stockfish playing for a win, Nelson great job, defending the position! :)
😂
@@ChessVibesOfficial i will be lost on move 4 already))
Anyone can defend if they takeback 😂😂
@@sergiob8501 same
The situation is winnable from either end. I've also done king and 4 bishops vs king and 4 knights.
“As you can see, pretty straight forward. Anybody can do it.” 😂
I’ve heard Fabi say that he thinks the biggest thing engines have taught us is just how many defensive resources there are in most positions and playing against stockfish in “winning” positions makes that so clear.
I liked how you kept reminding yourself of the two primary objectives:
Avoid checks.
Corral the black king.
at the hind he finds out that he should be treatening the queen.
@@bocasuja22 Yes those normally solid approaches are what led to complications in this situation. Knowing when to set aside 'best practices' is tough...bordering on impossible. We shy from doing it firstly because it skyrockets our uncertainty as we move into unfamiliar territory. And we can't help but remember plenty of failures when we rolled the dice believing it was time to get creative...only to find out we needed more patience plowing forward with best practices. The game simply becomes too complicated for humans at a level well below the ultimate drawn nature of the game. Timed games only serve to push the bar lower. So probably best to spend our time becoming super competent in common positions. At one time I knew how to mate with B,N in under 50 moves. That is for all practical purposes (human play) worthless info. Much like this problem. It's fine to mess around with occasionally. But not worth getting worked up about.
I have never had fun studying endgame in my life but you completely turned the table. Thank you very much.
Brilliant move at the start would likely be Knight-f5
Checks the king, either the king moves away (removing his ability to assist checkmate) or Queen takes f5, leading into Bishop takes f5
Yep i saw it too
After plugging this into stockfish on max depth, it managed to find a mate in 46 moves. So close to 50 move limit.
18:00 instead of trying to save the knight from the queen, move the other knight to d3 to check the king and force him into the corner. From that position you should be able to keep checking the king until checkmate.
kling e7
@@UrMom-xk3od kling
Bert meant d6, as neither of the Knights can move to d3; however, Ke7 then threatens the dark-squared Bishop.
However, after Ke7, the move Nd5 can occur, checking the King again.
Stockfish is so exhausting to fight even when you have the upper hand such as this situation.
Great video idea, I liked this a lot. Would be interesting to see the other way around too. Have Nelson play the black pieces and see how strong of an engine he can force a draw against.
For the position that you set, black pieces in the middle of the board, did you try by choosing black and seeing how the program went about attacking?
Would be very interesting to see!!^
9:45 for those wondering, Ne2 is a blunder because Qd1+ Kc3 Qxd2 Nf4+ Kxe5 Nxe2 Kxe4 and white has insufficient mating material
How does white's blunder lead to black having insufficient material?
@@TJTrickster oops, meant to say white, technically both sides have insufficient material Lol, good catch though
I oftentimes have trouble against stockfish with the bishop and Knight let alone with giving him a queen for an extra two pieces.
18:35 tfw you're just screaming "knight e3 please knight e3"
This was really instructive, thank you. I think you needed to use the fact that you could attack their queen a bit more; just in the same way that they used attacking your minor pieces as a resource when checks were not available like here 17:24 . You didn't turn to this nearly as much
My man broke stockfish. I gotta say your content has been on point I've been way more interested in your endgame studies and little challenges than I have been other places like Gotham that used to be my goto chess content but his recaps of draw after draw wear on me
I appreciate that!
This is by far the best channel for learning and improving.
Draws aren't boring
@@roberthansen5727 draws that happen insanely aren't
@@monke7566 A good game of chess is a good game of chess. Someone slipping up and dropping the half point doesn't enrich the game, a lot of the draws in the last Candidates tournament were more exciting than many of the wins.
Position at 8:37 is a cool 4-piece puzzle-knot where all the white minor pieces are defending each other without the king having to.
Great video! I highly appreciate all of your hard work for us!!
Stockfish: Ay yo what kind of bs you want me to play today?
27:36 "Pretty straightforward. I think anybody should be able to do it."
Yeaaaaaah, suuuuure! :D
8:43 move knight to get in position to fork after bishop sac to fork king and queen.
And if black chooses to trade queen for bishop and knight you will have a bishop and knight mate XD
2:12 bishop isn't hanging, because knight fork f4
Attacking the strongest piece instead of beating up the weaker piece sounds like a sun tzu tactic
Knight F5 check. Take the queen with the Bishop
Theorical win are such a fantasy sometimes.
pure maneuvering. i like this sort of gameplay.
"there are no checks"
*nelson literally ignoring a good knight move*
Now try to defend the position
14:55, king to G3 is fine, if queen takes f5 bishop, knight to D6 forks the king and queen, just saying 🤷🏻♂️
'pretty straightforward'
I'm a bit late, but I watched the other video yesterday, and I love that you made this followup vid
Chess Vibes is the best!
stonkfish is tough!
I admire your persistence.
Nelson, interesting position. Perhaps you take a wrong road in modelling a systematic procedure the same as two lone bishops mating net (neither side has anything else, except white has 2 bishops). Remember that a "front" can be created in front of the BK with a single knight, using one bishop to fill in the holes in the knight's coverage. I wondered about you trying at move 1 WN d4. If 1 .... Qa2. Interesting is 2 Ng3. Threat here is 3Nc4+... and 4N/g3-a5; where the BQ is trapped firmly as if on flypaper. But not clear to me how white can force a win; and hence 2Ng3 may not in the end be the way to go. The approach of a forward bishop restricting the BK, with the bishop being protected from behind by a knight seemed to me to be rather passive; as both the bishop and knight are somewhat immobilized thereby. Still, I can't say any other clear win is obvious; the position appears to me to be too complicated for exact calculation (by me anyway). I would encourage you to experiment with other approaches; besides simply keeping the two bishops side by side, as is done in a mate with only two bishops.
It makes me feel like im a grandmaster every time i predict their moves i predicted 21 moves this makes me get better
Every chess RUclipsr should do a stockfish content once a week
well its a draw with bishop on b2{a1} and Knight on d4 and king on a2/b1 or any corner where the bishop is the same colour as the corner square. So one would hope you could win with an extra 2 pieces.
At 1:35 you started analyzing the line where you play Kb3, and after Qd1+ you block the check with Nc2... That drops the d3 bishop
14:55 if the queen takes the bishop Nd6 is a fork
try the opposite, and see how fast stockfish can do it with four pieces :)
Mr. Nelson, I'm so sorry for my comment, which you've deleted. I like your videos very much, ideed. All I wanted to say was that even a strong Stockfish can't win the given position by playing against itself unless using 7-men Syszygy TB. Thus, no human player can win that position against a strong Stockfish either. A strong Stockfish program uses the complete 6-men Syzygy TB, 16 GB Hash and 8 minutes per half-move. I wish you all the best.
Hey bud, sometimes RUclips automatically deletes comments if they think it's spam. Maybe that's what happened cause I don't remember reading this comment before. :-)
Wow, that was tough!
0:53 Isn't that every Game of 2-Person Chess ever?
Early on your bishop wouldn't be hanging. 😂
14:54 bishops not hanging because after queen takes, knight can fork king and queen
He missed a opportunity to capture the queen
I tried defending as black, giving white a position where all their pieces are in the corner and black's pieces are in the middle. I lost in 14 moves by checkmate. Somehow the knights and bishops were able to easily cover all the checks I have and force my king to the corner
Pretty straightforward. Just calculate everything a super computer is going to do to make your life a living hell and prevent it.
Isn't this playing the computer where there are so few pieces that it can rely upon a gigantic lookup table of every possible move and game state?
“I broke stockfish” 😂😂😂🤣😂
At 15:30, Ke3. Even as you confine his king, you free his queen behind you.
I'm going to be real --this one was super boring to me. So I looked it up and this is a tablebase "win", but it takes about 65 moves with no captures, which means it's a draw. Practicing against stockfish is pointless because it will always hold the draw.
well, thats only if everything is centralized when you start and assuming perfect play from your opponent.
If your opponent slips up and plays less optimally, you can do it faster than 65 moves, and if your pieces are better placed, or if the black pieces are not as centralized, you can also do it much faster than 65 moves.
So in theory, in MOST situations where you could get this kinda endgame, this is going to be a theoretical win. The difficulty just comes with actually practically doing it over the board.
@@eragon78 Stockfish is not going to slip up.
@@maxscherzer9521 Well yea, stockfish wont, But i was mostly talking about in a real game against a human opponent.
Like, winning this kinda situation vs a human player is much more manageable than it is against stockfish.
While its still difficult, the 4 minor pieces vs queen should still be a win against most humans even if you have to play that ending out. Most good players should be able to checkmate with that against a human.
Its only when you get to stockfish with perfect play where the task becomes quite difficult especially with the move limit as not every position is a viable way to win within 50 moves. (although most still should be possible, this was just a pretty optimal position for black to start out with).
So like 3 knights vs a bishop, you attack the piece.
Wow, you don't even have to kill the queen to checkmate, I guess it would be harder to capture the queen than just checkmate lol.
11:26 you’vs to play Kd4. If check (only one is possible) then u can block with night and u at the same time also delivered the check to black
I didn't play the game for a year, so glad i can see someone against stockfish
11:45 I was thinking Bishop to d4 to prevent that annoying check...
14:58 Kg3 is actually okay bc if Qxf5 then Nd6 forking the king and queen
Maybe the idea is, you divide pieces, one N+B and K mates the king and the other two handle checks from the queen -- the natural desire to employ all pieces against the king and also to keep them in a tight group is probably inefficient?
Follow up idea: you with K and Q, Stockfish with N,N,B
Your last comment!!!!! LMAO
1:53 can we go for Bd4 instead? It forks both the king and queen and will likely end up with a king+bishop+knight vs king mate?
lesson of the story: 4 pieces VS Queen strategy is "harass the queen into a shitty position than mate becomes possible"
queen can't trade because even with trade it's still a loss probably to knight and bishop or 2 bishops(as losing the two bishops makes no sense as your knight should be guarding them)
18:39 was knight to e3 a move? It cuts off the kings escape square
it hangs the bishop
Wow! Incredibly tricky.
26:03 look closer, then do King to e7
What appraisal does Stockfish assign to the initial position? Did you try it w/you having the Q? Stockfish playing itself?
I love the comment about how it's pretty straightforward everyone should be able to do it. Rofl.
8:40 I think you allowed black to sac the queen for both bishops, forcing a draw
I don't even need to open the video to understand that the Qween alone is USELESS even against 3 pieces (not necessarily 4)
That formation is absolutely impenetrable. The best queen can do is to pull out a draw game
Everytime he makes mistake, he just "well, i wont do this, let s rewind". That was horrible downer, cant win - lose with dignity.
Have to spread out all your pieces, thats the trick
Looking at this with an endgame tablebase you played a lot of the top moves, and the computer played almost all the top moves
I tried to win a +3 endgame (queen vs rook+1pond) against stockfish and he chipped away like nobodys businnes..
in 18:30, could've played Nf2 for same threatening reasons
I was thinking sacking a couple of pieces to get the queen
On 18:30 you missed a fork.
You could have made the fork the outher way, Not K c4., instead you could do K f2
:26.06 if nelson puts his horse e 7 checkmate
How do you do custom games like these?
I actually have 0 ability but about halfway through your video I did start wondering well why instead of constantly running from the queen make the queen run from you. Which in a wide open board and king in the middle probably still extremely difficult
Dude, why are you taking back bro? How is this even considered a challenge at this point.
This so heavily computational, with no first principles to guide you, I don't see how you could beat even a 20yo program
11:32 Ne2 seems playable. Ne2 Qa2+ Kd4 Qxe2 Nf4+ and fork
And 17:06 Ke4 also seems playable. Ke4 Qh4 Kf5 Qh7+ Kg5 Qg8+ Kh4 Qh7+ Nh5 and no more checks + King in action
should let stockfish have the 2 knight and 2 bishop and seeing how it executes vs you
Tips to black player:
Keep king and queen distance more than four squares to avoid 🐴 horse 2in1 trap.
And also keep King and Queen to different coloured ⬛⬜ squares to avoid Bishop's trap. 👍👍👍
You didn't take advantage of your knights indirectly defending pieces through forks
i feel like when no pawns left GMs usually offer draw.
or resign if the lead is big enough.
Magnus once win a game where there was bishop knight and rook vs knight and rook.
Those endgames are just incredibly rare.
LOL... straight forward, mate in 60 moves.
What’s the app he is using pls
You should start with a check and keep with never ending checks. Force the opponent, a check is a forced move, don't give him options.
Nah he took moves back this not accurate.