Aman plays a move that Stockfish can't find
HTML-код
- Опубликовано: 16 окт 2024
- FOLLOW THE CHANNEL
▶ WATCH LIVE / chessbrah
▶ SUBSCRIBE / chessbrah
▶ Play Chess go.chess.com/c...
▶ / chessbrahtv
▶ / chessbrahtv
FOLLOW THE BRAHS:
/ amanhambleton
/ eric.hansen
/ hansenchess
/ amanhambleton
Check out our clothing store! ▶ chessbrahstore...
Chess Books ▶www.amazon.com...
Chess Memberships▶ go.chess.com/c...
Our Music Producers ► MARTIN K4RMA and Jack Ryan (JR)
fanlink.to/MAR...
/ martink4rma
/ official.jr.music
/ official.jr.music
So I never had a brilliant move cause the engine just doesn’t understand them. Thanks for boosting my self esteem aman
I've had the engine call obvious moves brilliant and discredit game winning moves that I knew would work kinda sensing what my opponent would most likely do.
It is a cool tool but not something to live by.
The brilliant isn't amazing. One time I captured my opponents only remaining pawn with my king to get into a winning endgame, when all I had was two pawns and a knight versus their knight. The engine said it was brilliant
@@jackrichards1908 it discredits your winning moves just because there must be a better one or a counter response from your opponent. Its very very rare that you make a move that is better and invincible to the engine.
@KSG the same position happens either way
If you've ever promoted a pawn to queen the engine usually calls it Brilliant
I often play moves that stockfish can't find. The fact that I'm then losing a few moves later is neither here nor there...
Same bro - stockfish just doesn’t understand the brilliancy of me gambiting all my pieces
@@yelloe_birb Ikr, They Should Really Improve it, Because In The
Once I was drawing a game because of repetition and stockfish couldn’t find the move that drew the game
@Utsav Joshi It's not a 'bug'. Engines aren't perfect. Even if your using a normal pc in your home, even there the engines are calculating hundreds of thousands of moves every second. Yes. Even in my not so premium phone I get upto 1000 knodes/sec. (Which means 1 million positions calculated per second). They literally calculate every single legal move in a position. But, because that is very inefficient after the initial calculating they will remove moves that do not achieve anything constructive. So, it's not a bug. It's a feature that backfires sometimes. This can be solved once our AIs get stronger at considering candidate moves in a position instead of calculating every legal moves.
He was playing brilliancies when Stockfish was still PiPing in its pampers.
Bro do you sleep
What the f*** are you talking about. Stockfish is the biggest looser.
@@grantdillon3420 Yes,stockfish is cheater
@Garrett Long hes everywhere meaning hes on youtube too much
@Garrett Long Levent posts on every chess video ever. He posted on a video I saw as I went to sleep last night and I wake up to see he commented on a video from an hour before.
Stockfish doesn't even understand the first rule of chess: if there is a mate in 1, look for better.
Aman has AlphaZero in his veins get that fish outta here.
I like pie!
I like lizards more.... The big carnivorous lizards.
that fish clapped alpha
People don't understand what Alpha zero actually is. It is a computer, not an engine. And while it initially beat stockfish (after learning the game in 4 hours, the next version of the fish already beat it realibly. This version of stockfish beats any current engine and it would beat alpha zero.
@@Jartran72 Thank you for the enlightenment. I know better now~
Engine when someone plays the only legal move: Wow, brilliant!
Engine when someone plays a move so good it can't even see it: Inaccuracy, take it or leave it.
I mean if it can’t see it
It's not even the engine that assigns those ratings for moves, they're just arbitrarily chosen by chess.com
The engine understands his moves more than himself
Tbh, that was scary, he calculated so ahead and so quickly, sac material and a queen, in order to get a wining position and fixed the opponent queen. And for what is worth, the guy acted like it was nothing. Man that was frightening, what a brain.
Yeah during the game I was thinking there's no way he can sack the queen, that just looks like suicide.
Frightening? Did you have trouble sleeping over a chess game?
Maybe an invitation to the Alpha Zero headquarters soon.
69
They are going to upload Aman's brain to the computer and make him even more powerful
@@Ninjalectual you can add makeup to a pig but adding a pig to makeup is senseless. The alpha zero team will most likely buy a big computer case & just have Aman inside. Be rid of pigs.
Imprisoning the Queen like that was Wow. I've never seen that before.
U kill my brain
@@johnappleseed6926 Sorry, my bad.
69
It's like Max Warmerdam vs Annmarie Mütsch
@@JoeeyTheeKangaroo exactly like that.
Engine simply had its ego hurt and didn't want to admit your move was indeed brilliant
Brilliant move. This deserves an Agadmator review.
agadmator can make a draw a amazing video, imagine what he can do with this
yes
I would like Gotham's review on this
@@ettaprimm4448 Gothamchess for recaps everyday :)
@@ettaprimm4448 congratulations, you are an excellent recommender of chess video channels
Stockfish 13 sees the queen sac at depth 27.
Does it really?
Mine found it at depth 30 (Using the dev version)
But does it see the knight move two moves before?
@@redhadoken3172 Wdym?
So basically what's happening here is probably Aman just figured "I'm gonna win the endgame because I have an extra knight" on which he was correct because he's playing a 2600. But to stockfish, the endgame is way more tricky and requires higher depth to figure out. If he played that position against stockfish, stockfish would probably find a way to maneuver those pawns and the queen to actually win it.
I've been saying for a min; this is the beauty of Aman's game. He has some Nezhmitdinov in his spirit. His game is artful, and he plays it intuitively. I'd rather watch Aman lose an interesting game, than watch a "stronger" GM win an essentially predetermined one.
well said! so entertaining
You need to give the person who made that intro a shit ton of money.
The reason why alphazero won the tournament against stockfish was because aman IS alphazero.. it all makes sense now
that was 2017 now stonkfish has asserted his dominance
No it's because aman wasn't in the tournament
@@junwei8480 just you wait until the alpha zero team buys a computer case... & puts Aman in there.
Legend has it that the engine uses Aman to cheat
holy shit, I was really impressed a few weeks ago when Aman sacrifaced two rooks to checkmate the opponent, but one could say that it still was within a spirit of the london system middlegame plan. but this queen sac right here is simply another level, no way I could see it in a milion years, absolutely amazing.
Agree, a sacrifice that creates a positional advantage is much harder to find than a sacrofice that allows for a mate
This is mind blowing. How quickly Aman saw that idea and then calculated all the way through it is unbelievable. So freaking cool.
"Alright, I'm going for the fun"
Aaand he went for perfection again KEKW holy smokes
Holy Jesus.
Congratulations, man.
That's a feat, seriously. What an amazingly titled article.
Indeed. It is unbelievable that the article is real...like, literally unbelievable. :)
@@brettjones327 On youtube, where people believe in flat earth, you genuinely can't tell who is being serious and who is joking lol
@@radscorpion8 This has a name! It's called "Poe's law."
@@Zack_Taylor poe's law, thanks for this gem, gonna use it
@@Zack_Taylor thanks dude
0:35 "when asked to comment Mr Hambleton just said ' don't forget to subscribe to our channel ' "
Aman is way too modest, this was such a beautiful idea haha
The engine always says I made mistakes and never gives me any "brilliant" moves neither.
What if you put some depth? I noticed that Aman chose depth 18, usually after depth 30 +, engine will understand what is going on
Edit: after checking with lichess stockfish 13 nnue, it didnt found D3 after 10m, until you play it and then it realized it's also a good move. Yeah i think it's a briliant Move engine couldnt find, engine prefer g4 for immediate check mate threat and white needs to give back materials and then black will win endgame with pawns advantage, but human wise, d3 is much better to calculate and to convert
SF13 likes both options but prefers Neg4.
21. ..Neg4 +6.57 @ 58 depth
21. ..Nd3 +5.98 @ 59 depth
@@Corteum Neg4 is 58 depth and the other one is 59 depth tho
@@junwei8480 It doesn't make a difference. Neg4 is SF13's preferred move all the way through, even if you go to 60+ depth. Try it.
@@junwei8480 Nd3 @ 58 depth is +5.64
@@Corteum how did you even go to 60 depth
Literally broke the engine and maintains the most chill Canadian Chessbrah vibe. Aman is literally POG
After many many weeks of teaching everyone good habits (and for free at that) as well as starting a new educational series, he's now surpassed the engine. It's like these educational series as unlocked something in Aman to show us where we all can be by following his series. And, again, all his educational stuff is free, unlike a lot of other chess stuff. A true hero of chess.
Can Aman defeat Eric now?
A Battle between these two more with 10 minute games would be wonderful. And get dylan to commentate
+ Yasser to commentate
It’s kind of an unspoken rule that they never play each other lol
@@tommykimberlin7528 eric is rated far above 3k on chess.com He's a tad weaker than the supergm's, maybe right about where danya is.
@@tommykimberlin7528 Even if they did, I have a feeling it might be a private game.
This is Amans best game on RUclips for me
I remember Hikaru saying something along the lines of, “Oh Aman? Well, Aman isn’t that good at chess,” as he responded to someone asking in his live chat about chessbrahs. I think Aman is showing what he’s really capable of here!
Aman isn’t really that much to Hikaru tho, hikaru is like top 3 bullet and blitz
hikaru is a man child with an ego the size of canada
Hikaru also isn't the most likeable chess player
Stockfish has a hard time finding my moves too. But not because they are too good 😂
that was a beautiful idea. the position after the sacrifice is still quite sharp (e.g. if white plays Qe5, black can only play Rd7 or Rd8 to win). great video and game
My entire chess 'career' can be summed by "It's a bold move Cotton, lets see how it plays out for him"
There is nothing more satisfying than clicking the like button on a cbrah video
I'm confused. Is it a problem with the depth setting of the engine, or does Stockfish genuinely not see the move?
Yes, he has it on a max depth of 18. He doesn't know how to use an engine
@@matthewviramontes3131 Is any human going to see more than 18 moves deep?
Maybe its a practical decision
@@Halibut86 i mean, aman did see all the endgame, he just didn't analize every move but figured that the queen was stuck, while the computer is calculating every line even the ones where it hangs the queen which are nonsense
so is chess brah just a group of gms that stream
By the way Brandon, thank you very much for the release of ROW during lock down. you saved me from a bad case of anxiety.
SF13 says...
21. ..Neg4 +6.57 @ 58 depth
21. ..Nd3 +5.98 @ 59 depth
Amazing how you saw Nd3 in a 3+0 game, though. Killer move.
If Aman goes missing after this, it’s probably fair to say he got kidnapped and his brain is being used to program the next great chess engine.
you have to expand the lines out to see what the differences are, the depth looks 18 or more moves ahead, the lines can be exactly the same up to the 16th or 17th move ahead.
Just checked this game with the engine today and it will say a move is -0.1 until you play it, then the eval jumps to -5.
Further analysis needed!
brilliant strategy Aman. Huge props for this great instructive game
I think the problem is optimization. There's no way the engine wouldn't see that move if it was just brute-forcing all the options, but something that seems like a queen-sacrifice with only very long-term benefit (by keeping the enemy queen out of play) get discarded in the search-tree too early on.
If you increase the depth it would eventually find it. Stockfish uses iterative deepening so any time it looks one more move into the future it reconsiders some of the moves it previously discarded.
straight fire Aman....txs for proving the limitations of chess engines.
I just checked it on Lichess - 3 years later Stockfish still doesn't see the move lol xddd
Ideas transcend depth. Hope Stockfish improves their architecture with these examples, like how Tesla improves the AI in their cars with data from every autopilot override.
The engine doesn't see my moves either
The engine didn't see any brilliant moves because Aman is on a whole different level
Aman, you are the Man! [smile] Congratulations on confounding the A.I & proving imagination rules!
Just found this video today, and checking out that game in your archive, it looks like it does give a brilliant to that queen sac now, and says Nd3 as the best move. Interesting to see how a few years of engine development make a difference.
Aman Anti Cheat. Bots cant see him cooking.
PUT THAT FISH IN THE DIRT AMAN
8:26 Stockfish found the 7 innacuracies, 8 mistakes & 2 blunders thou!
For a class on Tolkien I had in college last year, I composed a dragon sicillian game based on the Hobbit and I found a move for white that the engine just did not see until a move or two after it was played, and then it started going crazy for it. I was really proud of the game, I put a lot of time into making it beautiful and make sense in the context of the book, and the engine didn't catch a move.
Do you still have the game recorded? I'd love to see that game!
Legend has it, Hikaru’s ceiling monitor has now been replaced with Aman taped to the ceiling
This was absolutely wonderful
The engine after queen e1: *this is either madness or brilliance*
Pretty crazy tbh. To outsmart an engine in 2021 is some God tier stuff
Thanks for keeping language clean. Not like some other chess sites.
I'm gonna send this game to agadmator
Stockfish: Bxd4 and we're not too bad after Rxd4, only -0.87!
Aman: Captures bishop
Stockfish: Perfect! -2.21
Aman: Captures bishop
Stockfish: Exactly what we wanted, -4.23
the engine doesn't understand me either
I love it when that happens. There's a line in the traxler where the engine gives something like +3, after one more move it says 0, then after another move it says -7.
It's like it's saying "dude come on, I'll just play this and WAITWHAT"
Stockfish 14 NNUE immediately says that Q d2, c1 or b1 are all the same.
It also says that Nd3 is not inaccurate but the best move along Neg4.
Might wanna use higher depth than 18 for better analysis
the "a" in alphazero stands for aman
What does the other a stand for?
@@Ninjalectual aman
And the g stands for good
More of these explanations please, from a humble 1400. Thank you.
All jokes aside, this was a beautiful game, and beautiful idea by Aman
so engine cant see the idea of taking a piece out of the game without actually capturing it until later in the line
only watched the first minute so far but this is already a great video
The engine saw the mate but was looking for better! (it watched Aman's vids)
engine destroyed chess
Actually the engine understands them more than him
Dang that was def a brilliant move
My arousal is maintained. Props to Aman for putting some Elo-points on the line to create this master piece.
On the bright side at least it proves you're not cheating 🤣
Whenever I play the engine is also confused by my moves.
Damn bro been waiting for this update for a while
Theory time... Interestingly, the value of the queen in the starting position is 9. A rook is 5, a knight is 3 and a pawn is 1. So in theory, a rook, pawn and knight is worth 1 queen. In this case, it should be a draw. However, again, in theory, a pawn advanced to the 7th rank is worth a rook, so once the pawn was advanced to d7, just the rook and pawn is worth 1 point more than a queen. Together with the knight, overall, black has a 5 point advantage and is a material evaluation of a rook up in this position.
Took my Stockfish almost three minutes to find the correct continuation and evaluation after Nd3
Could you send the links for the articles? hihihi
Aman: Qxe1
chess players: dam this is so deep 😤👊🏻
the right move was actually Neg4 his move is only better if the opponent ended up taking the queen with the rook
Where is the link of the match
official jr needs to drop cant beat it on soundcloud
Put the position in rignt before Ne5 d3, engine is saying the best move is Ne5 g5 is the best move from a depth of 27-41, I’ll keep y’all updated as I go up to 99, or attempt to get to 99 depth, then I’ll put in Ne5 d3 and see what is says
That intro was amazing
We haven't seen a player>engine moment like this since PICK UP THE DEPTH
The computer is just on depth 18. Amans idea that white won't be able to use his queen, is more long-term, Qxe1 is a positional sac. Computers are often bad recognizing such ideas, but especially if it's only on such low depth. Maybe the winning line takes around those 18 moves with perfect play, considering that the computer suddenly sees it one move before playing Qxe1. You can put analysis on unlimited depth with a diamond membership, if it still doesn't see it at higher depths I'm really impressed. It was a very brilliant idea and positional understanding of Aman anyway of course, independ of whether the computer sees it or not.
really beautifully done
We aren't smart enough to even comprehend stockfish so clearly WE don't have any idea what's going on
Dude! Dude! Dude! Had to give you a dude hatrick. Love the channel. My chess is improving cause ive been watching. Outstanding!
That intro is GOLD!
Me a 1500 wondering how many best moves the engine has missed in my games
Holy fucking shit aman that’s so freaking Genius.
The BGM in starting of this video, cool af
I was so upset with a puzzle once. I had spent upwards of 5 or so minutes trying to figure it out. Finally made my move and it was wrong. Analyzed the move and it said I had mate in 8. Then I did the correct answer to the puzzle and it said +7 . Why!?
my Stockfish 14 sees ..Qxe1+ at a depth of 33 ply and rates it -4.75 - a black win. Amazing. It suggests Bc1,instead of Bxd4,as does Fritz 18.
@chessbrah Fun fact: You actually need Stockfish 14 to run about 2 min (depth 29), until it recommends Aman's Nd3!
So how do they determine the ELO of the move? Is it just equating the minimum depth of the engine that can find it?
I've seen this before and am still blown away. When the AI takes over, our only chance will be to challenge them to a game of chess and send in Aman. Good luck us!