Its just dilemma that we feel. The term context for us is stupidity for high level AI. Their thoughts if they could explain are far far far more complex than ours. There's no relation with these human context termed explanation to these higher minds' context. It for once might seem obvious once you see the trick these youtubers explain, but a far more higher dimensional context is the correct explanation for these games, which we lack normally. And scientifically, almost all logics they explain behind a move like from upgraded stockfish or alphazero are wrong. And no human(idk if deep meditation minds can compete with them) can explain the logics behind their moves.
@@adhoax3521 the moves seem logical but most GM's won't play them simply because most GM's can't calculate millions of variations in a fraction of a second. strong wacky-looking moves are backed up heavily with like 20 -move variaitons with brilliant tactics that somehow end up winning, no GM realistically could calculate that for every single move he makes. like there were so many points in the video where alpha zero just seems like they're hanging pawns or even pieces but there's brilliant variations at every turn to counter literally every single idea that stockfish has ;-;
The Looinrims It doesn’t matter since it’s the same engine (his picture for AlphaZero looks different from his picture for Leela Chess Zero, for instance).
AG totally changed how we play Go, too... ofc in go (it being that much more complex) the moves AG plays are even more mind-meltingly beautiful ^^ the world no1 (Kie Jie) said after playing a match with it: 'it was like meeting god' - the look in his eyes was something to behold.
How times change! Remember when computers were helplessly materialistic? In this game, white chooses a gambit and doesn't go for the d7 pawn in several ocassions, and black restrains the will to take a2, also on strategical grounds, and then the exchange sacrifice! Wow!
yeah exactly how I saw this game, and every other a0 game for that matter ... they all have the same pattern, squeezing the opponents for space and rendering his pieces useless ... that queen in the corner was so sad at the end .. it could have been just a pawn, or a bishop at best and nothing would have changed ...
re8 is probably a big mistake, too slow to fight against the pawn on e5. shouldve broken free with d5, also there are games with qb8 putting pressure on e5
This series of commentaries on the AlphaZero games took my chess to a new level. I'm a low rated player trying to improve and your commentary on these games cemented into my mind the importance of controlling and utilizing strong and weak squares. Thank you very much awesome videos!
First, you are a great chess analyst. It's really helpful, purelly technical and very insightful. BRAVO. Second, AlphaZero is a monster and humans can learn a lot from its games. I entirely encourage you to keep making as many chess studies as much on youTube video for the sake of the game. THANK YOU. This one of my holidays gifts.
Please keep in mind, Stockfish 8 is the version that struggles to find the solution to this problem. Stockfish 9 and the latest Stockfish Dev build sees the threat instantly.
@@djr5024how is that even possible? It calculates hundreds of millions positions per second and can't spot an Eval drop of 5 in 2? What's going on here?
First of all I would like to thank you Jerry. For analyzing & commenting in a perfect way these amazing videos. The way you explain and break down everything, to the best of your knowledge, in these oh so complex matches is remarkable and much much appreciated. I love it when you say: AlphaZero says No! LoL! All the videos are mind blowing on so many levels... Something that jumps out to me is that Stockfish does not seem to take into consideration in it’s programming the ‘sacrifice’ of a piece for some greater good in the long run AlphaZero’s comments at the are a killer treat!!! Keep up the good work sir and yes! Let’s see the other 90!
It's killing me that Google won't release the rest of the games against Stockfish! it's been enthralling watching the various analysis videos of the ten games and I can't wait for the next instalments!
Why not contribute to lczero.org/ and have an infinite amount of examples. If you don't have a GPU, contribute this way: github.com/LeelaChessZero/lc0/wiki/Run-Leela-Chess-Zero-client-on-a-Tesla-K80-GPU-for-free-%28Google-Colaboratory%29
Très bien expliqué : tout commence par la case f5 mais la suite est difficile à imaginer, l’étouffement des Noirs et la réorganisation avant la conclusion. Il reste étonnant que les Noirs n’aient pas pu se relever. Bien à vous
it's so interesting that Stockfish evaluates the position to be fully equal until the move 50! It cannot comprehend how cramped its position is. It's only after move 51, it assesses the position differently and realizes that black is completely winning.
I must say, this was the first video of yours that I watched. I enjoyed your narrative and breakdown of the game. I look forward to your future analysis of games. Sir you just gained a new subscriber to your channel. 🤓👍🏾
if only there was some random billionaire interested in chess and willing to finance some TPUs for leela chess zero, we might create another chess god able to demolish SF10+ in a couple days... elon musk, chess needs you
It's seams like when Stockfish calculate branches, Alphazero can understand general long term abstract plans. So yes, I think Alphazero doesn't just caculate chess. it can realy think.
Ori shem-ur i believe with its recognitional powers, it has nicely situated itself superior in chess, because of its methodological-like control over the board. It pushes these positions like the end is from the opening if the right control isnt persued, no matter how minute it may seem
I wonder if AlphaZero gets extremely good at fortress positions if its dueling stockfish all the time since stockfish literally does not see them coming
If I remember correctly, I think it plays itself over and over while it's learning, rather than Stockfish. Each time it wins, that version becomes the new benchmark.
I think it's more along the lines that its probabilities get very slightly adjusted to the new "best" version, as opposed to overwriting previous trials.
I don't think AlphaZero factors the games vs Stockfish into its training. Even if it did, it only played a mere 100 games vs Stockfish, which is basically nothing (and it didn't even lose any, so probably not programmed to change much from draw/victory).
You have a different approach to chess than most of the other voutubers, but i think, you understand this game very well. Without your explanations i woulndn't have understood nothing in this super-complicated game. Thanx for that, so sorry i cannot use it in my games, i have to watch out for not loosing my queen in a 2-move-tactic! But i understand and it is fun!
@@overmeme953 propably not. Leela proofs brute force is not going to take a major spot in the long run anymore. Even stockfish goes into ai territory with stockfish nnue now...
yes . but i've heard alpha zero can only see few position per second like 80,000/s and stockfish is 70,000,000/s so what if we can create an AI that can see 70,000,000 positions/s ... that'd be OP
At 27.41, I came up with Bxf7 instead. Black must take it (the only legal move). Then Qb8 wins. Okay, so it's mate in 4 instead of 2, but after finding a forced check-mate, why look for anything else?
My observation is that, before the position at 27:01, Stockfish never evaluate it being the downside. However, 27:01 is yet again a zugzwang (classical AlphaZero). This shows AlphaZero's evaluation is far better, both in the positional type, and the restrict-the-opponent's-choice type situation. Anyhow, either space advantage or square dominance are just about the same idea: choke the other side of choices. AlphaZero has not only exceled the good old Monte-Carlo based engine, but surely made the game chess cunning once more.
Edit: maybe not necessarily. After ...Kf8 Qf4 Qg8 Qc7 c4 Qxc4 Rd8, Stockfish as of November 2018 gives white a +1.2 advantage, which [may] be winning, but is unclear. Add g4 Qe8 g5, and white has his kingside structure set up but black is trying to slip out. In any case, the situation is much more unclear than the move ...Rf8 played in the game which is evaluated at +7 or higher.
On 27:48 you can also check with bishop to F7, then enemy rook has to F7, then queen checks at b8 and then mate at queen at f8. Right? I'm kinda trying to improve my chess game and this is what I saw, I wonder if I miss something...
at 27:36 when I paused I didn't see the line you played, but instead I saw bxf7+ and only legal move is Rfxf7, leading to Qb8+ and only two choices are Rdd8 which the queen would just take with check, or Rff8 which the Queen would take with checkmate. A slower mate than you posted but I'm happy I found a different line :)
This game is the most astonishing one to me between those 2 AIs. So many counter-intuitive moves, so many poisoned pieces, and the ending game with the whole repositioning from both sides is incredible. How long would it take for human masters to evaluate they need to change gears at that point ? How much time to figure out the right sequence ? Great analysis !
I think Deepmind AI software uses a machine learning approach first pioneered by Arthur Lee Samuel 1n 1949. This could be regarded as trial-and-error but if you think it's intelligent, fine. Certainly very good at chess if not conscious at all ;)
I saw a quicker idea with Bxf7+ followed by Qb8+. The other Rxf7 still wins. I checked with an engine afterwards, and was happy to see my idea still works. It shows both moves secure victory! If my engine sees it very quickly, you can bet that Stockfish 8 saw the dark clouds coming slightly earlier...
I went through it with Stockfish 9, several variations, on my laptop and it seems that it gives alphazero a growing edge for quite some time. Although Stockfish 8 in the video is just 0.0 the whole time and v9 actually is aware of this growing advantage, it seems like it's already too far gone and Stockfish 9 can't find a way through before it's already lost. I have not tried Stockfish 10 dev but it seem that v9, while more aware of the slipping position, is unable to do anything about it.
Remember this is a gutted version of SF 8. Also, the first time that I dislike a video from the chessnetwork for not being upfront about the differences in SF. I still clicked liked though as I'm a huge fan of the channel.
Bishop captures. If d pawn retakes, almost nothing changes but white is up material, if f pawn, queen to f4 and then what do you do? if rook recaps secure remaining rook with pawn and queen threatens back rank mate.
7:07 is an interesting instructive moment, with alphazero playing Q-g4. As you said, the move Q-h6 might be expected. In this position, I would've played to h6, even though I saw the obvious two move attack with stockfish's bishop ending up on f8. My gameplan would've been Q-h6, R-e8, then Kn-g5 , threatening mate in one if B-f8, or mate in two with Q-h7 , K-f8, Q-f7. This would basically force black to respond to Kn-g5 with Bxg5, where I'd then go with Bxg5, throwing a punch at Black's queen. I suppose if Alpha zero didn't do that, it wasn't optimal, and being a really poor chess player (1500 or something last I checked), I'm not sure why. As I said, instructional video. 👍 Edit: I'm still paused at 7:07, after rewinding back to it right as Q-g4 is played, and my only guess is that my gameplan kinda traps my own queen for a couple moves, and its cumbersome to then try and protect the e5 pawn. Any other insight? I'm a total chess noob as I said : j
@ChessNetwork Dear Jerry, I fired up the analysis for the move 49. in LiChess and it is evaluated to 0 (deapth 49), equal positions. Instead of Rf8, Stockfish want to move the King to f8 followed by the Queen g8 next move. It seems, for this match Stockfish 8 may have been buggy or did not have the necessary hardware power for an adequate analysis.
It's incredible how obvious the moves are when context is explained.
Its just dilemma that we feel. The term context for us is stupidity for high level AI. Their thoughts if they could explain are far far far more complex than ours. There's no relation with these human context termed explanation to these higher minds' context. It for once might seem obvious once you see the trick these youtubers explain, but a far more higher dimensional context is the correct explanation for these games, which we lack normally. And scientifically, almost all logics they explain behind a move like from upgraded stockfish or alphazero are wrong. And no human(idk if deep meditation minds can compete with them) can explain the logics behind their moves.
@@adhoax3521 the moves seem logical but most GM's won't play them simply because most GM's can't calculate millions of variations in a fraction of a second. strong wacky-looking moves are backed up heavily with like 20 -move variaitons with brilliant tactics that somehow end up winning, no GM realistically could calculate that for every single move he makes. like there were so many points in the video where alpha zero just seems like they're hanging pawns or even pieces but there's brilliant variations at every turn to counter literally every single idea that stockfish has ;-;
@@EM-oe2lz I think it is easier to learn from these AI than from engines though, even though the gap between our abilities is greater.
The only trick part is calculating the whole variation
"What would you do in this position?"
Lose. I would lose, no question.
I could usually make it to about move 12 against LC0 before totally blowing my position. LC0 was based on AlphaZero, but hasn't had the same CPU time.
@@Zenthex i lose in like 5 moves :D
the exceptional thing is, that i would lose with both colors, no question =D
Sorry Jerry you’re gonna run out of colors to spray on that robot placeholder in your thumbnails eventually
The Looinrims It doesn’t matter since it’s the same engine (his picture for AlphaZero looks different from his picture for Leela Chess Zero, for instance).
Wow this is like chess from another planet.
AG totally changed how we play Go, too... ofc in go (it being that much more complex) the moves AG plays are even more mind-meltingly beautiful ^^
the world no1 (Kie Jie) said after playing a match with it: 'it was like meeting god' - the look in his eyes was something to behold.
How times change! Remember when computers were helplessly materialistic? In this game, white chooses a gambit and doesn't go for the d7 pawn in several ocassions, and black restrains the will to take a2, also on strategical grounds, and then the exchange sacrifice! Wow!
He makes it sound like a national geographic documentary. “The knight needs to find a new home”
This is some high level chess. It’s hard to grasp but spectacular to watch. Makes it look so simple.
AlphaZero aka Boa Constrictor. That was beautiful
yeah exactly how I saw this game, and every other a0 game for that matter ... they all have the same pattern, squeezing the opponents for space and rendering his pieces useless ... that queen in the corner was so sad at the end .. it could have been just a pawn, or a bishop at best and nothing would have changed ...
If a0's mO is always the same why does a0 win so often?
crazy that Stockfish still thinks evaluation is zero up to move 50 and then...bang! thanks for the great commentary (as usual :-), Jerry!
Pasquale R I agree, I'm fascinated by Stockfish's interpretation of its position right up to the death knell.
Pasquale R no it's not lol check on lichess stockfish 9 says that the evaluation is +12 for white
Muhammad Amir this is stockfish 8, not 9
Yesto I meant stockfish 8 not 9 sorry about that anyways stockfish 8 says it's +12 for white so I don't get what pasquale is talking about
re8 is probably a big mistake, too slow to fight against the pawn on e5. shouldve broken free with d5, also there are games with qb8 putting pressure on e5
This series of commentaries on the AlphaZero games took my chess to a new level. I'm a low rated player trying to improve and your commentary on these games cemented into my mind the importance of controlling and utilizing strong and weak squares. Thank you very much awesome videos!
👍🏼
Maurice Ashley: "So Stockfish, how do you think Alpha Zero was able to dominate so well on the dark squares?" Stockfish: "What do you want me to do?"
Lmao! tbt Carlsen
First, you are a great chess analyst. It's really helpful, purelly technical and very insightful. BRAVO. Second, AlphaZero is a monster and humans can learn a lot from its games. I entirely encourage you to keep making as many chess studies as much on youTube video for the sake of the game. THANK YOU. This one of my holidays gifts.
9:05 It's not just a home, it's an actual house.
a tiny house
a castle...fit for a king
Soon it will be: "Alpha Zero dominates any square!"
eule franz Alpha Zero dominating the human race!
Soon it will be: "Alpha Zero dominates the world!"
soon it will be "... " forgot what I was gonna say but trust me oh man was it clever
has john connor actually been born? does anybody know?
“The Queen can’t just go into Pacman mode” 😆
@ Kristina Morrow: Do you play chess with friends?
27:04 Stockfish has a score of 0.00 but then two moves later it's +6 for white ... That's incredible
"What would you do in this position?"
Ehh, not blunder a piece, hopefully?
Interesting how long Stockfish thinks it's an equal position.
Please keep in mind, Stockfish 8 is the version that struggles to find the solution to this problem. Stockfish 9 and the latest Stockfish Dev build sees the threat instantly.
Yea notice that it was a pawn move when the evaluation shot up.
The way alphazero plays with the queen is absolutely masterful
I saw this same game on agadmator’s chess channel. I love seeing two perspectives to get more potential ideas behind moves on both sides. Love it! :)
27:02
Watch the engine evaluation jump from 0.00 to +4.80 in just two moves...!
Bro that was crazy
@@djr5024how is that even possible? It calculates hundreds of millions positions per second and can't spot an Eval drop of 5 in 2? What's going on here?
First of all I would like to thank you Jerry. For analyzing & commenting in a perfect way these amazing videos. The way you explain and break down everything, to the best of your knowledge, in these oh so complex matches is remarkable and much much appreciated.
I love it when you say: AlphaZero says No! LoL!
All the videos are mind blowing on so many levels...
Something that jumps out to me is that Stockfish does not seem to take into consideration in it’s programming the ‘sacrifice’ of a piece for some greater good in the long run
AlphaZero’s comments at the are a killer treat!!!
Keep up the good work sir
and yes! Let’s see the other 90!
at the end*
Really wonderful game !
Funny how stockfish's evaluation drops at some point when it realizes its queen is trapped :D
It's killing me that Google won't release the rest of the games against Stockfish! it's been enthralling watching the various analysis videos of the ten games and I can't wait for the next instalments!
Why not contribute to lczero.org/ and have an infinite amount of examples. If you don't have a GPU, contribute this way:
github.com/LeelaChessZero/lc0/wiki/Run-Leela-Chess-Zero-client-on-a-Tesla-K80-GPU-for-free-%28Google-Colaboratory%29
Stockfish owns Alphazero on equal hardware and classical timeformat, hence why google choose not to play on equal hardware and classical timeformat
@@1345-v2e that's a lie
@@1345-v2e isn't that the whole point of Alphazero? ... that she has better hardware?
You'll be glad to know that they have released more games now, if you haven't seen them already
i love that the move that made stockfish go "OH SHIT" was that innocent little pawn move g4
Years later....Alpha zero's world domination.
Such a beautiful game...
I can't wait for more games of AlphaZero with your wonderful commentary.
AlphaZero for president!!!
Très bien expliqué : tout commence par la case f5 mais la suite est difficile à imaginer, l’étouffement des Noirs et la réorganisation avant la conclusion. Il reste étonnant que les Noirs n’aient pas pu se relever. Bien à vous
it's so interesting that Stockfish evaluates the position to be fully equal until the move 50! It cannot comprehend how cramped its position is. It's only after move 51, it assesses the position differently and realizes that black is completely winning.
Alphazero is the king of chess!
so lets promote it to king of earth, too
@@Pintkonan more importantly - it's the god of Go... so yeah, qualifications for ruling earth seems fine to me ^^
@@sjs9698 has john connor actually been born? lets hope so =b
Yay! Another AlphaZero video!
Stockfish created 16 accounts to dislike the video.
newp gaston it is now 84
Nice smooth talk,can follow really well
Thanks Jerry, extremely useful mentoring.
I must say, this was the first video of yours that I watched. I enjoyed your narrative and breakdown of the game. I look forward to your future analysis of games. Sir you just gained a new subscriber to your channel. 🤓👍🏾
Wow I didn’t even realise it was a 31 minute video! Great commentary and analysis Jerry. :)
It's interesting that alphazero neglected development that long
watch the game where he moves the rook multiple times in the opening ...
if only there was some random billionaire interested in chess and willing to finance some TPUs for leela chess zero, we might create another chess god able to demolish SF10+ in a couple days...
elon musk, chess needs you
I'm from the future and i have good news
@@matiasgarciacasas558 Dammit, one month too late :D
I don't get the point since we already have alpha zero
@fazega He wants something that can waste A0 like A0 is capable of destroying SF.
Daily uploads man! God I love it when you post videos every other day so I have something to watch while eating lunch
Thanks from Brazil! Very good explanations!
Thank you and hello 🇧🇷😊
The value of activity over material is a reoccurring theme in the Alpha One vs Stockfish games
It's seams like when Stockfish calculate branches, Alphazero can understand general long term abstract plans.
So yes, I think Alphazero doesn't just caculate chess. it can realy think.
Ori shem-ur i believe with its recognitional powers, it has nicely situated itself superior in chess, because of its methodological-like control over the board. It pushes these positions like the end is from the opening if the right control isnt persued, no matter how minute it may seem
I wonder if AlphaZero gets extremely good at fortress positions if its dueling stockfish all the time since stockfish literally does not see them coming
If I remember correctly, I think it plays itself over and over while it's learning, rather than Stockfish. Each time it wins, that version becomes the new benchmark.
I think it's more along the lines that its probabilities get very slightly adjusted to the new "best" version, as opposed to overwriting previous trials.
I don't think AlphaZero factors the games vs Stockfish into its training.
Even if it did, it only played a mere 100 games vs Stockfish, which is basically nothing (and it didn't even lose any, so probably not programmed to change much from draw/victory).
AlphaZero learns only from games that it plays with itself
AlphaZero only learns via self-play.
Love your analyses Jerry. Thoughtful and helpful.
Jeez, it's as if AlphaZero zugzwanged all of Stockfish's pieces. Each move made it that much harder for it to progress. Brilliant!
17:06 Chop, chop, chop chop, chop... a lot of chops! hahahahaah!!! Best moment of the game Jerry, thank you for such beautiful moments :D
Clear and instructive as always. Thank you!
instructive video. keep up the good work!
@17:17 Jerry meant "team white"
TY Jerry, I learned a ton from this one!
what a game... wow....
Thanks Jerry for another amazing video!
Excellent videos as usual, Jerry. Love these analyses.
You have a different approach to chess than most of the other voutubers, but i think, you understand this game very well. Without your explanations i woulndn't have understood nothing in this super-complicated game. Thanx for that, so sorry i cannot use it in my games, i have to watch out for not loosing my queen in a 2-move-tactic! But i understand and it is fun!
😎 Thank you. You are capable of preventing those 2 move tactics.👍🏼
@@ChessNetwork ...on a good day...
The content you put up is very good, I believe your channel will grow to be the biggest on the chess segment if you keep up in this manner. Thank you!
"Black is in a giant pickle." I don't think those words have ever been said about Stockfish. That was unbelievable.
I love this game! I watch it all the time.
Your view on a chessgame is a bit different, but very instructive! I like the way you explain weak squares and diagonals and files - thank you!
Great explanation/ analysis
You analyze alphazero chess games better than anybody 👍🏿☺️
If we need a player to represent Planet Earth at chess, Alpha Zero is probably that player we'd want.
Too bad stockfish would destroy alpha zero if they were on the same classical format
@@overmeme953 propably not. Leela proofs brute force is not going to take a major spot in the long run anymore. Even stockfish goes into ai territory with stockfish nnue now...
@@overmeme953 Stockfish 9 got destroyed by Alphazero even after having 20/1 time.
yes . but i've heard alpha zero can only see few position per second like 80,000/s and stockfish is 70,000,000/s so what if we can create an AI that can see 70,000,000 positions/s ... that'd be OP
I like how the eval went from 0 to straight 5
Alpha Zero is like Tal who doesn’t make mistakes.
This is probably the most beautiful chess game I've ever seen. Black fleeing to their castle, queen and king locked up during the siege by alphazero.
Great vid Jerry! Love your commentary on these games
excellent work there ---Kid. Loved the tutorials.
At 27.41, I came up with Bxf7 instead. Black must take it (the only legal move). Then Qb8 wins. Okay, so it's mate in 4 instead of 2, but after finding a forced check-mate, why look for anything else?
spicy, was waiting for another alpha zero vid. thx man!
My observation is that, before the position at 27:01, Stockfish never evaluate it being the downside. However, 27:01 is yet again a zugzwang (classical AlphaZero).
This shows AlphaZero's evaluation is far better, both in the positional type, and the restrict-the-opponent's-choice type situation. Anyhow, either space advantage or square dominance are just about the same idea: choke the other side of choices.
AlphaZero has not only exceled the good old Monte-Carlo based engine, but surely made the game chess cunning once more.
AlphaZero puts a pawn on e5. Again.
I have watched it at least 2 or 3 times. Every time I see this game, I feel sorry for the suffocation of the black king. Total annihilation by AZ.
wonder what alphazero's message is this time
"GG" - Alphazero
"get rekt nub" -AlphaZero
Based on the games I have watched, AlphaZero is such a control freak!
Incredible repositioning of attackers by alpha zero!
So methodical disassembly of Stockfish here. Wow.
26:51 I disappointed by you. You not pointed that 49. ... Rf8 it was losing move for stockfish (and king f8 was a draw)
No, Kf8 also loses, just more slowly
Edit: maybe not necessarily. After ...Kf8 Qf4 Qg8 Qc7 c4 Qxc4 Rd8, Stockfish as of November 2018 gives white a +1.2 advantage, which [may] be winning, but is unclear. Add g4 Qe8 g5, and white has his kingside structure set up but black is trying to slip out. In any case, the situation is much more unclear than the move ...Rf8 played in the game which is evaluated at +7 or higher.
Very good video and very well explained thanks 👍 👍 👍
Alpha playes like it's real war going on and survives at any cost..
On 27:48 you can also check with bishop to F7, then enemy rook has to F7, then queen checks at b8 and then mate at queen at f8. Right? I'm kinda trying to improve my chess game and this is what I saw, I wonder if I miss something...
Yay more alpha zero !
at 27:36 when I paused I didn't see the line you played, but instead I saw bxf7+ and only legal move is Rfxf7, leading to Qb8+ and only two choices are Rdd8 which the queen would just take with check, or Rff8 which the Queen would take with checkmate. A slower mate than you posted but I'm happy I found a different line :)
this game is really the go-to for playing chess positionally instead of materialistically
WOw what a fantastic video and analysis thanks Jerry
This game is the most astonishing one to me between those 2 AIs. So many counter-intuitive moves, so many poisoned pieces, and the ending game with the whole repositioning from both sides is incredible. How long would it take for human masters to evaluate they need to change gears at that point ? How much time to figure out the right sequence ?
Great analysis !
Head Minerve this game is not between two a.i.'s
What do you mean ? Wikipedia tells me they're indeed AIs.
Artificial Stupidities ;)
Head Minerve stockfish is not an A I. It is a chess engine.
I think Deepmind AI software uses a machine learning approach first pioneered by Arthur Lee Samuel 1n 1949. This could be regarded as trial-and-error but if you think it's intelligent, fine. Certainly very good at chess if not conscious at all ;)
The best game in the series IMHO, worthy of immortal status ?
I agree with you man!
DEED I
I saw a quicker idea with Bxf7+ followed by Qb8+. The other Rxf7 still wins. I checked with an engine afterwards, and was happy to see my idea still works. It shows both moves secure victory! If my engine sees it very quickly, you can bet that Stockfish 8 saw the dark clouds coming slightly earlier...
My mistake. On my board I misplaced white king. Now I've corrected, both lead to mate in 4!
Your a better commentator than agadmator in my opinion, you explore more options and your tone of voice is better
two videos in two days? Fantastic weekend!
"stockfish you underestimate the value of throne pawns in proximity to your king."
What the heck!
That made me laugh.
30:35
26:45 Rook e6 sacrifise, because black have 2 pons more and win with Queen vs Queen ? Please answer..!
Awesome game. Great analysis
Thanks, Jerry
Why don't you use Stockfish 9 or Stockfish 10 Dev? Stockfish sees the threat instantly with these versions.
Because I want to give some idea of what move AlphaZero’s opponent was considering.
Ah, thank you!
Would it be possible to show both? (I don't know if that is difficult to do, but I thought it would be cool!)
I went through it with Stockfish 9, several variations, on my laptop and it seems that it gives alphazero a growing edge for quite some time. Although Stockfish 8 in the video is just 0.0 the whole time and v9 actually is aware of this growing advantage, it seems like it's already too far gone and Stockfish 9 can't find a way through before it's already lost. I have not tried Stockfish 10 dev but it seem that v9, while more aware of the slipping position, is unable to do anything about it.
pringlerjr +1
Nice job,thank you!
Question: at around 26:30 in the video, could stockfish have returned the exchange by playing Re6 while also threatening to trade the rook off?
Remember this is a gutted version of SF 8. Also, the first time that I dislike a video from the chessnetwork for not being upfront about the differences in SF. I still clicked liked though as I'm a huge fan of the channel.
Supernova ... you do know that he's using stockfish 8 because that's who AlphaZero is playing against right?
Bishop captures. If d pawn retakes, almost nothing changes but white is up material, if f pawn, queen to f4 and then what do you do? if rook recaps secure remaining rook with pawn and queen threatens back rank mate.
regarding the d pawn recapture, wouldn't black be up material?
7:07 is an interesting instructive moment, with alphazero playing Q-g4. As you said, the move Q-h6 might be expected. In this position, I would've played to h6, even though I saw the obvious two move attack with stockfish's bishop ending up on f8. My gameplan would've been Q-h6, R-e8, then Kn-g5 , threatening mate in one if B-f8, or mate in two with Q-h7 , K-f8, Q-f7. This would basically force black to respond to Kn-g5 with Bxg5, where I'd then go with Bxg5, throwing a punch at Black's queen. I suppose if Alpha zero didn't do that, it wasn't optimal, and being a really poor chess player (1500 or something last I checked), I'm not sure why.
As I said, instructional video. 👍
Edit: I'm still paused at 7:07, after rewinding back to it right as Q-g4 is played, and my only guess is that my gameplan kinda traps my own queen for a couple moves, and its cumbersome to then try and protect the e5 pawn. Any other insight? I'm a total chess noob as I said : j
@ChessNetwork Dear Jerry,
I fired up the analysis for the move 49. in LiChess and it is evaluated to 0 (deapth 49), equal positions. Instead of Rf8, Stockfish want to move the King to f8 followed by the Queen g8 next move.
It seems, for this match Stockfish 8 may have been buggy or did not have the necessary hardware power for an adequate analysis.
Stockfish 270918 evaluates Kf8 at +1.7 and Rf8 at +7.8 at d28. Kf8 more than likely just loses more slowly.
Can't wait for your analysis of the world camp battle come nov.
27:12 Stockfish realizes the pozition is lost