Correction: Chess is solved at 7 pieces or less. 8 piece tablebase is still a work in progress. I have even said this in previous videos. Sometimes my mouth moves faster than my brain. This is my apology. Now get out of here.
My favorite part about this video is the low-depth stockfish eval occasionally goes “WAIT A MINUTE” and jumps the eval like 10 seconds after the move was made
Last I heard we were still a long way from 8man tablebases. The bots can use 7-piece tablebases in TCEC and a game is adjudicated at 6 pieces. That's another reason Torch wanted to lose the hpawn. I've seen bots pitch a rook when the remaining position is in tablebase, because maybe there was a way to win faster but they couldn't see it and knew how to win the 7-piece endgame
I still dont understand. So essentially you guys are saying that the bots are gambling on whether or not a 7-piece endgame is a win or not by sacrificing a piece just to get to a 7-piece end game? Then it gets there and sees the future of the game. But with SF not taking the piece, ur saying it didn't want to reduce the board to 7-pieces to stop Torch from potentially seeing a winning scenario because it thought it was probably going to lose and wanted to play without tablebase as it had more chance of winning or drawing??? Obv this is algorithmic, I'm not asserting they are consciously making these abstract decisions.@@Fishy_17
honestly, I think that the ‘piece tension’ - where opposing computers’ pieces see each other but don’t take - is very akin to ‘pawn tension’ for humans. We avoid piece tension (at least compared to the absurd level demonstrated by computers) due to the mind boggingly complex variations. But less experienced players often resolve the pawn tensions very quickly for the same reasons. It’s fascinating.
Torch makes Stockfish's play look...mechanical. Torch's calculations are still insanely complicated, but they almost seem organic. Holy crap, this is terrifying.
At one point bots are going to solve chess It would take a lot lot years But considering exponential growth in processing power. It will be solved just like tic tactics toe.
I'm sure there was atleast one point during the development process where a tester was confused as to why it was "blundering" all its pieces, not realising its sheer genius
that last game reminded me of a game between stockfish 8 vs houdini 5 that was shown by agadmator. stockfish had a queen vs 2 bishops, which is a forced mate in >70 moves, or rather a draw by the 50-move rule, but houdini resigned. I really like to see those unbalanced endgames.
@@donkarnage6986dude, ig you program a chess engine to resign at some point, for example if the advantage for the opponent is too high, or if there is mate on a board, they can resign
@@Gaminguide1000 A UCI chess engine CANNOT resign, it is simply not possible for the engine to do so as it is not contemplated by the UCI specification. The interface that manages the games can adjudicate a game as a win/loss or draw if it sees that both engines consider the game to be winning or drawn.
Actually the opposite. First of all, Stockfish won that opening with white in 83 moves, and it took Torch 202 moves to win. And when Torch played white (the game in the video), it was bugged and was saying that it was a draw for most of the game, even on move 189, Torch was saying 0.00 and Stockfish was saying Mate in 13.
Absolutely fascinating to see Torch in action! The AI's understanding of chess nuances and its ability to sacrifice material for positional advantages is mind-blowing. This truly showcases how AI has transformed the world of chess.
Torch is fascinating man. It’ll choose moves that prioritize destabilizing the opponent’s positions, regardless of the damage it receives, and because the positions of the enemy are ruined, the losses are made up for with the results. The fact it even boggles Stockfish is crazy.
i really want to see a video where we watch bots play the same opening but you analyze the differences in how the bots take that opening into the midgame. it would be interesting to see the different strategies the different bots use to play the same openings.
It's not so much about relative piece value but rather the concept of creating a local advantage. It's not a new concept either, many great generals throughout history used this concept, from Hannibal to Napoleon. Basically it doesn't matter how large your army is(or how many pieces you have left on the board) but rather how many soldiers(or well, pieces) you can bring to the upcoming battle(or well, attack).
25:30 the thing here is that if you trade the pawns you are left with 7 pieces on the board, so maybe it's using tablebases and prefers to lose the pawn and kerp 8 pieces on the board, which it doesn't know for a fact is losong but it's probably losing regardless.
I love how levy uses Torch for free clout. But forgot the fact that Stockfish is still way ahead above every chess engines existed. The reason i don't validate Torch as a great chess engine is because it has yet to pit itself at the TCEC(World Championship of Chess Engines). And for the matter of fact,Torch struggles to calculate at longer time controls wherein Leela and Stockfish are heads and shoulders above everyone else. Stockfish vs Torch (CCC Blitz Final 3+2): 79-11 Stockfish Stockfish vs Leela (CCC Blitz Alternate Final 3+2): 87-4 Stockfish Stockfish vs Leela (CCC Rapid Finals 10+3): 37-9 Stockfish Stockfish vs Torch (CCC Rapid Alternate Final): 50-2 Stockfish These games that Torch won are rare cases of pair winnings. As much as it thrives at shorter time controls. It still needs a massive improvement at longer time controls hence i don't validate it yet as a great chess engine.
Not playing at TCEC doesn't make Torch less of a great engine, it is a clear top 2-3 engine. And the problem is not that Torch struggles at longer time controls, is that Leela struggles at shorter ones and thus it makes it look as if Torch is weaker at longer ones such as rapid because Leela places higher. You can easily check this by just looking at the previous bullet events, Leela ended up 4th in the CCC 20 Bullet and 5th in the CCC 21 Bullet, it can't even compete with Dragon and Berserk in those time controls. The current ongoing event is rapid again, soon we'll see if Torch also surpassed Leela in this format too.
Rapid? Oh hell no. Torch is good at faster time controls. And yet it got bullied by Stockfish really hard. Torch actually got 2 pairs against Stockfish at Rapid 10+3 which is fortunate already. Rapid is pretty much Stockfish and Leela again.
at the end, "the only way to win this is to lose the pawn" - actually, stockfish finds at least 7 forced checkmates here and they all end with the e pawn promoting to a queen.
I've watched so many of your videos and I'm still no better at understanding what you're saying, but i love the enthusiasm. Honestly if golf commentators had this level of energy, i might watch that too.
Only watched your channel for a short while, but Rock Lee dropping the weights is such a great reference I had to comment. Thank you for your great chess content.
Was I the only one who noticed how Levy started acting all weird after messing up saying "I don't even need it" thinking Torch was gonna checkmate and then realized it was in check
I actually expect these bots to change who gets to be a top player in the future. Because what sets these bots apart from us is that they can calculate 30-40 moves ahead very quickly, memorise it all and then decide on the best move. I think human players with the highest ability to do that will become more and more prominent.
I think it'd be funny to have a game continuously switch between the best chess bot in the world which makes mega brain moves nobody can understand and a 0 elo player who makes mega brain moves that nobody can understand, and levy has to guess the precise moments when the AI takes over
Hey, Gotham, I think you should make a video playing the video game Shotgun King, I want to see how being good at chess affects your playing of Shotgun King.
I like that you say when the AI's are out of book. I was watching a couple other videos today, and they didn't distinguish very well between when they were talking about the ideas behind the book opening and when they started talking about the ideas of the AI's.
Correction: Chess is solved at 7 pieces or less. 8 piece tablebase is still a work in progress. I have even said this in previous videos. Sometimes my mouth moves faster than my brain. This is my apology. Now get out of here.
Pin of shame
Pin of Shame
😂😂
Self shin of pame
Pin of shame
feeling lost? light a torch
Who are you ?
😂😂😂
Stop stalling, release chess 2 already
When does Chess 2: Electric Boogaloo plan on coming out?
Working at the market? Remember to stock the fish
My favorite part about this video is the low-depth stockfish eval occasionally goes “WAIT A MINUTE” and jumps the eval like 10 seconds after the move was made
XD
25:32 WHYYYYYY?
😂😂 me too
Guys he got stuck in 666likes
@@Hitsnotbad it still is haha
Torch: sacrifices rook and is a genius
Me: hangs rook and is reminded that my elo is 450
oh how I wish this was not me every game
Just get better
@@nikbomb3dude this comment screams 200 elo 😭😭
450 elo is still really good!
@@EMTReviewWe all like to sacrifice the game(aka. resign)
Torch's playing style is like:
"Is that a move a degenerate would do?"
"It is!" "I'll play it"
Insane stuff
Last I heard we were still a long way from 8man tablebases. The bots can use 7-piece tablebases in TCEC and a game is adjudicated at 6 pieces. That's another reason Torch wanted to lose the hpawn. I've seen bots pitch a rook when the remaining position is in tablebase, because maybe there was a way to win faster but they couldn't see it and knew how to win the 7-piece endgame
Well that's interesting. Thank you for that information.
And there is still a long way to 8man tbs, Levy made that up. Using engines you can prove that a 8-man position is winning tho.
This is also the reason why SF didn't take on e4 at 25:38
Torch does not play at TCEC. It is highly private and only allowed to play on Chesscom’s own engine tournament.
I still dont understand. So essentially you guys are saying that the bots are gambling on whether or not a 7-piece endgame is a win or not by sacrificing a piece just to get to a 7-piece end game? Then it gets there and sees the future of the game. But with SF not taking the piece, ur saying it didn't want to reduce the board to 7-pieces to stop Torch from potentially seeing a winning scenario because it thought it was probably going to lose and wanted to play without tablebase as it had more chance of winning or drawing??? Obv this is algorithmic, I'm not asserting they are consciously making these abstract decisions.@@Fishy_17
honestly, I think that the ‘piece tension’ - where opposing computers’ pieces see each other but don’t take - is very akin to ‘pawn tension’ for humans. We avoid piece tension (at least compared to the absurd level demonstrated by computers) due to the mind boggingly complex variations. But less experienced players often resolve the pawn tensions very quickly for the same reasons. It’s fascinating.
Torch can literally becomes levy's new magnus for this year.
Torch looks like Tal, crazy sacrifices, relentless attack and powerful engame
@@dman6261 AI is way way more advanced than any human , including great Tal. Comparing human to an AI is like ....
@@vinayjv He is comparing the styles
You don't want an unhinged torch. That frequently leads to fires.
Underrated comment
Bring this to the top of the comments
That pun lit up my day.
These games are fire, so it makes sense
Levy should join FIDE board of directors. Time to modernize the governance of our game.
I think you need to be a Russian oligarch to hold such a position.
@@Jarand91 I mean, he does speak russian, that's halfway there already 😂
You can be highly successful as an influencer but still suck at management
"If I was ever on the FIDE board, something went terribly wrong in my life." Magnus Carlsen
Torch makes Stockfish's play look...mechanical. Torch's calculations are still insanely complicated, but they almost seem organic. Holy crap, this is terrifying.
Stockfish also won with white on that opening but in 83 moves instead of 200+ by the way
As fascinating as the last game was, the attack on that first game was insane.
It's games like that that make me realize I will never understand chess at the level these bots do.
@@webbowser8834don’t know why you would need games like that it’s kind of common sense
No cuz for real, eval of +2 while down 3 pts of material is insane
At one point bots are going to solve chess
It would take a lot lot years
But considering exponential growth in processing power.
It will be solved just like tic tactics toe.
“The continuous threat of something about to happen” sounds like a quote from a series of unfortunate events
I literally thought of the same thing...
"Waiting for something to happen?"
idk about you but personally this is one of my biggest fears.
Is that the one with Count Olaf?
@@WumpusAlpaca Yup
Levy is truly the most unhinged AI of all time.
Wish I wasn't a copywriter for an AI company 💀
Not the most unhygienic , he's second most unhygienic after GM Osama Bin Biden
underrated comment
Hey there, TypicalTard comment guy.... I didn't know you were a chess guy too!
@@SoulStar222_ I miss tardy
I'm sure there was atleast one point during the development process where a tester was confused as to why it was "blundering" all its pieces, not realising its sheer genius
Stockfish: how dare you challenge my power
that is so hilarious im laughing so much rn bro
edit: im being sarcastic if you somehow didnt realise
Torch is "a Stockfish Jr."
@@Evex6stop yapping
Let’s make this the top comment so none of those “I’ll do one push-up for each like 🤓” or “Gotham never fails to mention Magnus in his videos! 🤓 “
@@WaChessOfficialStockfish is a Rybka Jr.
that last game reminded me of a game between stockfish 8 vs houdini 5 that was shown by agadmator. stockfish had a queen vs 2 bishops, which is a forced mate in >70 moves, or rather a draw by the 50-move rule, but houdini resigned. I really like to see those unbalanced endgames.
you tell us that a chess engine resigned! yes yes...
Are you sure that is true?
@@donkarnage6986dude, ig you program a chess engine to resign at some point, for example if the advantage for the opponent is too high, or if there is mate on a board, they can resign
@@Gaminguide1000 A UCI chess engine CANNOT resign, it is simply not possible for the engine to do so as it is not contemplated by the UCI specification.
The interface that manages the games can adjudicate a game as a win/loss or draw if it sees that both engines consider the game to be winning or drawn.
In the last game Stockfish experienced what we experience when we play against Stockfish
I wonder how would Magnus Carlson play it
Actually the opposite.
First of all, Stockfish won that opening with white in 83 moves, and it took Torch 202 moves to win.
And when Torch played white (the game in the video), it was bugged and was saying that it was a draw for most of the game, even on move 189, Torch was saying 0.00 and Stockfish was saying Mate in 13.
He doesn't. He believes playing AI is beneath the spirit of the game @@DearMan-we2md
26:24 is one of my favorite Levy noises that I've heard in a while.
I'm watching this high asf and I died lmfoAoaoao I'm so sure he has autism
Absolutely fascinating to see Torch in action! The AI's understanding of chess nuances and its ability to sacrifice material for positional advantages is mind-blowing. This truly showcases how AI has transformed the world of chess.
I'm still waiting for a Hannibal Barca chess bot to release (that will be stronger than torch)
Stockfish:why you sacrificed the pawns ?
Torch: well its mate in 150 ...
I don't know how that guy makes me watch a bot chess game without even editing
Torch is fascinating man. It’ll choose moves that prioritize destabilizing the opponent’s positions, regardless of the damage it receives, and because the positions of the enemy are ruined, the losses are made up for with the results. The fact it even boggles Stockfish is crazy.
Sf won with white in that same opening against torch but in 83 moves instead of 200+
Levy never fails to be scared of bots
Man its difficult to watch human chess after watching computer chess
Torch should be called campfire because it is nonstop cooking.
Chess bots never fail to include Levy in their video
0:54 levy and I will never.
Levy never fails to make me watch an uncut 30 minute video everyday
It's funny how many pieces they hang, it's like they're so good they've wraped back to 100.
i really want to see a video where we watch bots play the same opening but you analyze the differences in how the bots take that opening into the midgame. it would be interesting to see the different strategies the different bots use to play the same openings.
imagine if one of the candidate uses torch to prepare instead of standard stockfish.....
There was a free bishop against ethereal that didnt get taken
it's crazy to me how well torch utilises tempo and baiting the opponent into losing tempo in order to get such an insanely crushing advantage. so cool
Would love to see a Torch win with Black!
Levy not shouting the rook had me very sad
Stockfish: I'm stronger, I'm smarter, I'm better... I AM BETTER!!!!!
That first game was the most insane game ive ever seen in my life
today's stare was a classic. lasted around a sec and was really absorbent. I'll rate it a 8.23 out of 10
"This move, mind blowing, the next one bananas". Like when your 1st move at Wal Mart is pulling down your pants and the next move is to wet the floor
lmao🤣🤣
It's not so much about relative piece value but rather the concept of creating a local advantage. It's not a new concept either, many great generals throughout history used this concept, from Hannibal to Napoleon. Basically it doesn't matter how large your army is(or how many pieces you have left on the board) but rather how many soldiers(or well, pieces) you can bring to the upcoming battle(or well, attack).
one might say that torch burns bright enough to be scary
Levy is getting symptoms of withdrawl due to a lack of Magnus videos.
The first game was one of the best chess games ever played
This is absolutely a 10 thousand elo. WTF!!
And choosing the title is expertly done, well done Levy
25:30 the thing here is that if you trade the pawns you are left with 7 pieces on the board, so maybe it's using tablebases and prefers to lose the pawn and kerp 8 pieces on the board, which it doesn't know for a fact is losong but it's probably losing regardless.
That last game is like the decider between the carnivorous and mankind, with the latter's imminent demise on wager.
Respect for the internet chess teacher
even if he is not a GM
25:36 the only reason I can think of is because if the trade happens, then the bots might get access to the tablebase.
Levy never fails to correct the computer as a GOAT.😂
If Leela is Magnus and Stockfish is Kasparov, this one is Tal.
I love how levy uses Torch for free clout. But forgot the fact that Stockfish is still way ahead above every chess engines existed.
The reason i don't validate Torch as a great chess engine is because it has yet to pit itself at the TCEC(World Championship of Chess Engines). And for the matter of fact,Torch struggles to calculate at longer time controls wherein Leela and Stockfish are heads and shoulders above everyone else.
Stockfish vs Torch (CCC Blitz Final 3+2):
79-11 Stockfish
Stockfish vs Leela (CCC Blitz Alternate Final 3+2):
87-4 Stockfish
Stockfish vs Leela (CCC Rapid Finals 10+3):
37-9 Stockfish
Stockfish vs Torch (CCC Rapid Alternate Final):
50-2 Stockfish
These games that Torch won are rare cases of pair winnings. As much as it thrives at shorter time controls. It still needs a massive improvement at longer time controls hence i don't validate it yet as a great chess engine.
Not playing at TCEC doesn't make Torch less of a great engine, it is a clear top 2-3 engine.
And the problem is not that Torch struggles at longer time controls, is that Leela struggles at shorter ones and thus it makes it look as if Torch is weaker at longer ones such as rapid because Leela places higher.
You can easily check this by just looking at the previous bullet events, Leela ended up 4th in the CCC 20 Bullet and 5th in the CCC 21 Bullet, it can't even compete with Dragon and Berserk in those time controls.
The current ongoing event is rapid again, soon we'll see if Torch also surpassed Leela in this format too.
Rapid? Oh hell no.
Torch is good at faster time controls. And yet it got bullied by Stockfish really hard. Torch actually got 2 pairs against Stockfish at Rapid 10+3 which is fortunate already.
Rapid is pretty much Stockfish and Leela again.
You switched torch and stockfish names in the last game
levy never fails to contradict the eval bar
“Like rock Lee dropping the weights” great analogy
The scariest chess bot on Earth is Levy
at the end, "the only way to win this is to lose the pawn" - actually, stockfish finds at least 7 forced checkmates here and they all end with the e pawn promoting to a queen.
If Levi should be a GM in anything, it should be in thumbnail pictures!
He doesn’t make those
NGL, I weeped like a little child on chrismas getting his first xbox, when I saw the first game. Absolutely beautiful!
chess engine running on amd epycs while stockfish eval bar runs on levy's browser: "Stockfish doesn't even understand"
meanwhile stockfish: +1.3
The first game was one of the coolest I’ve ever seen. The discovered checkmate was pretty classy too…
I know I already made a comment but when do you use torch in the evaluation bar instead of stockfish or in game review?
Now.
Why would you thought, Stockfish is way superior to Torch, most encounters with these 2 end up with Stockfish stomping.
I disagree entirely. My Chessbot is 100x more deadly than Torch. I call him Murdertron 5000. It's really bad at chess but has chainsaws for arms.
AI never fails to Levy.
I've watched so many of your videos and I'm still no better at understanding what you're saying, but i love the enthusiasm. Honestly if golf commentators had this level of energy, i might watch that too.
I'm curious, what is the time control? If even there is one? Like, do these programs make moves in a couple of seconds or it takes some time?
I think this is from either ccc blitz or ccc bullet
3 min + 2 sec
0:28 Scariest part of the video right here when he says society might last another 10-20 years 💀
Only watched your channel for a short while, but Rock Lee dropping the weights is such a great reference I had to comment. Thank you for your great chess content.
You've been using a lot of ableist language for a guy who can't shave
Levy fails to never
Was I the only one who noticed how Levy started acting all weird after messing up saying "I don't even need it" thinking Torch was gonna checkmate and then realized it was in check
GothamChess never fails to show us chess
Torch: Hold, hooold, Hoooold, HOOOOOOOLLLD, HOOOOOOOOOOOOOLLLLLLLDDDDDDFIIIIIIIIIIIIIIIIIIIIRREEEEEEEEEEEE
I was about to sleep but a Gotham vid never fails to disappoint
AI chatbots will never think of stealing pawns. They just don't.
I like the moment when torch said "It's torchin' time" and torched all over the other engines
I actually expect these bots to change who gets to be a top player in the future. Because what sets these bots apart from us is that they can calculate 30-40 moves ahead very quickly, memorise it all and then decide on the best move. I think human players with the highest ability to do that will become more and more prominent.
Levy never fails to change the title
Ohh now they put Tal's brain into a super BOT...
flashy titles and clickbaits. Levy doesnt hesitate in losing self respect for money. Shameless behavior
In order to make this move i cannot play, I guess I'll just play it anyway. Genius.
Levy, any way you could put the evaluation from both engines on screen? I'd love to see the difference in their evaluation of certain positions
Chesscom’s UI for their engine tournaments is incredibly ugly. Maybe he should do it for TCEC games but I do agree with his decision here.
I'd be like "this is too difficult, I'll sacrifice my Queen for a draw."
Let's see how long will it take this time for Levy to change the title of video
UPD: it took 21 minutes lol
I think it'd be funny to have a game continuously switch between the best chess bot in the world which makes mega brain moves nobody can understand and a 0 elo player who makes mega brain moves that nobody can understand, and levy has to guess the precise moments when the AI takes over
Is it unhindered now like what bro huh
I just realized that I play like TORCH, Sacrificing my pieces but I don't know how I always end up losing the game.
Levy never fails to fill me up with his juicy content.😋
Ayo
“Me see rook, me attack rook” Very wise words levy😂
levy never fails to milk AI's for content
Hey, Gotham, I think you should make a video playing the video game Shotgun King, I want to see how being good at chess affects your playing of Shotgun King.
I ❤ AI content Levy, please keep it going please!!!
Torch is playing like a modern day Tal.
Im not a human
I liked your comment
@@cygnustsp thanks human
I commented your like
The last game Torch was just the Terminator. Torch just hunted that king down like it was nothing.
These AI videos getting super realistic
Fly high Levy 🙏🙏🦅🦅
R.I.P
@@monsterexterminator??
I like that you say when the AI's are out of book. I was watching a couple other videos today, and they didn't distinguish very well between when they were talking about the ideas behind the book opening and when they started talking about the ideas of the AI's.
Let's see if my 'Levy never fails to ...' comment goes viral.
virial theorem…
Torch vs Ethereal? Feels like we just witnessed a murder