AlphaZero's bishop pair proves too difficult for Stockfish

When AlphaZero, playing black in this game, encounters the Ruy Lopez, it certainly seems to be in favor of the Berlin Defense. Stockfish injects a minor piece imbalance on move 5. How would AlphaZero play with the bishop pair? Stockfish, as early as move 13, opts for a sacrifice that not only nets two pawns, but also fractures AlphaZero's queenside structure. An important strategy to be mindful of when you have a material plus is that you can return it in exchange for some positional element.
PGN:
1. e4 e5 2. Nf3 Nc6 3. Bb5 Nf6 4. d3 Bc5 5. Bxc6 dxc6 6. 0-0 Nd7 7. Nbd2 0-0 8. Qe1 f6 9. Nc4 Rf7 10. a4 Bf8 11. Kh1 Nc5 12. a5 Ne6 13. Ncxe5 fxe5 14. Nxe5 Rf6 15. Ng4 Rf7 16. Ne5 Re7 17. a6 c5 18. f4 Qe8 19. axb7 Bxb7 20. Qa5 Nd4 21. Qc3 Re6 22. Be3 Rb6 23. Nc4 Rb4 24. b3 a5 25. Rxa5 Rxa5 26. Nxa5 Ba6 27. Bxd4 Rxd4 28. Nc4 Rd8 29. g3 h6 30. Qa5 Bc8 31. Qxc7 Bh3 32. Rg1 Rd7 33. Qe5 Qxe5 34. Nxe5 Ra7 35. Nc4 g5 36. Rc1 Bg7 37. Ne5 Ra8 38. Nf3 Bb2 39. Rb1 Bc3 40. Ng1 Bd7 41. Ne2 Bd2 42. Rd1 Be3 43. Kg2 Bg4 44. Re1 Bd2 45. Rf1 Ra2 46. h3 Bxe2 47. Rf2 Bxf4 48. Rxe2 Be5 49. Rf2 Kg7 50. g4 Bd4 51. Re2 Kf6 52. e5+ Bxe5 53. Kf3 Ra1 54. Rf2 Re1 55. Kg2+ Bf4 56. c3 Rc1 57. d4 Rxc3 58. dxc5 Rxc5 59. b4 Rc3 60. h4 Ke5 61. hxg5 hxg5 62. Re2+ Kf6 63. Kf2 Be5 64. Ra2 Rc4 65. Ra6+ Ke7 66. Ra5 Ke6 67. Ra6+ Bd6
Internet Chess Club (ICC)
Software: Blitzin
bit.ly/179O93N
Discount Code: CHESSNETWORK
I'm a self-taught National Master in chess out of Pennsylvania, USA who was introduced to the game by my father in 1988 at the age of 8. The purpose of this channel is to share my knowledge of chess to help others improve their game. I enjoy continuing to improve my understanding of this great game, albeit slowly. Consider subscribing here on KZbin for frequent content, and/or connecting via any or all of the below social medias. Your support is greatly appreciated. Take care, bye. :D
★ LIVESTREAM / chessnetwork
★ FACEBOOK / chessnetwork
★ TWITTER / chessnetwork
★ GOOGLE+ google.com/+Che...
★ PATREON / chessnetwork
★ DONATE www.paypal.com...

[@runhomie1013]

Stockfish disliked this video

[@dolorconsumer9473]

Alpha Zero：" Again , too weak too slow . "

[@kellyrants7673]

and he created 51 bots to do so.

[@IAMDEATHM4]

I bet so

[@smogblazer4509]

Kelly Rants stockfishes

[@changbohu]

xD

[@shmoe127]

These AlphaZero v Stockfish videos are amazing. Keep them coming Jerry!

[@ChessNetwork]

Thank you Robert.

[@awesomereece123]

Incredible! Thanks for the analysis, Jerry, have a smashing day mate.

[@anonymous-ui7il]

6:46 It was to give the black rook more mobility- more important that the active rook idea you talked about

[@edmundasjauniskis9347]

Time 0:22 1.0-0. I suggest ( without chess engine) to white play 1. Be3 and change bishops now. or after 1.0-0 Nd7 play 2. Be3

[@fudorsprite7370]

Jerry, I have been watching your videos for years now and I still feel that you are one of, if not THE best instructor in regards to taking complex seemingly-incomprehensible moves and positions, and explaining the logic behind them, making them understandable to amateurs. You're seriously like the 3Blue1Brown of Chess. I'm loving this series in particular and am always proud to sport my ChessNetwork shirt ☺️

[@ChessNetwork]

Ha! Thank you for that Fudor. That shirt must be fading by now. :)

[@ayapotato7429]

Hey to a fellow 3Blue1Brown fan!

[@economicsmadeeasy5266]

3 blue 1 brown is great as well

[@Lasselucidora]

I saw Stockfish in the corner of a bar. Empty bottle of Jack Daniels in front of him. So I sold him some pills.

[@johnadams2063]

U got any pills left,?

[@toilaconhaisam3037]

*stuck fish

[@DigitalHaze65536]

Stockfish 12 is the current champ, having have beaten AlphaZero's protégé, Leela Chess Zero very decisively. I think he owns the bar now : )

[@xyon9090]

@@DigitalHaze65536 we need a rematch between Alpha and Stockfish

[@bryanwan6169]

I really love these alphazero "commentaries" at the end of these videos.

[@MrDylanHole]

i really don't

[@lukapodvrsnik9547]

i think it is stupid

[@stephenyarrington5036]

wow you guys are quite the wet blanket

[@SedoKai]

It presents a really clichéd, false characterization of AlphaZero and demonizes it in an 80's Skynet kind of way. He's making AZ out to say "haha, you fool! you play so foolishly!" But you have to remember that AlphaZero is the hero in the minds of lots of people, so it makes sense that we're irritated by it or find it distasteful. That aside, Jerry is the best chess youtuber by far, imo, and it's not a particularly intrusive element, so I can just stop the video before that part starts and continue to enjoy Jerry's content. :D

[@denisdemonte15]

lol me too

[@umchoyka]

I've been loving this series. I wish they released more of the games for analysts like yourself to present!

[@ChessNetwork]

I'm definitely interested in seeing the other 90 games.

[@mrkhoi3]

ChessNetwork there also other 1200 games played with 12 differents opening, 100 each. Hope thay release it in the next few months after the reviewed paper come out.

[@yvesm.8855]

ChessNetwork I'm really much more interested even in seeing A0 self-played games. They are gonna be a true revelation!

[@Mikey-gs1dx]

the preprint is just to get questions so they can be thorough and revise the paper if needed. I'm sure if there is enough pressure, they'll release the rest.

[@crimsonitacilunarnebula]

@@mrkhoi3 are they released yet.:p.

[@DiMono]

It might be coincidence, but ever since I started watching your videos on AlphaZero my chess has apparently improved significantly. It's like you know what you're talking about or something.

[@ChessNetwork]

Think so? :)

[@blugaledoh2669]

TheDiMono Your subconscious are learning and analysing.

[@BeerdyBruceLeeCentral]

Thanks Jerry for these DeepMind games. I insta-click every time I see AlphaZero playing.

[@dominiktristian3139]

i guess it is quite off topic but does anyone know a good website to stream newly released movies online?

[@edwinzaid810]

@Dominik Tristian i would suggest flixzone. You can find it by googling :)

[@judsonethan2648]

@Edwin Zaid yup, I have been using instaflixxer for years myself :D

[@eshneto]

Best analysis of this game I've seen so far.

[@TWPO]

Playing for a draw as white is almost as bad of a crime as not liking a Chess Network upload! You're tremendous, Jerry!

[@drammar1991]

:D :D

[@shadowslave4948]

Well he's analyzing historical games. Keyword analyzing. So the outcome has already been determined. We don't care whether that person won or lost mostly. Why did that person win or lose? That's something you look at when analyzing a game. It's not strictly an entertainment video. Even the description is categorized as Education.

[@petergreen5337]

HAHAHA brilliant.

[@JohnBr0]

Thank you for bringing these games to us in such an understandable, fun and well constructed format.

[@ChessNetwork]

Thank you John.

[@josetedesco440]

Awesome. I was like "It would be nice to watch a new video from Jerry before going to sleep" and here it is. I can't wait to see the fireworks.

[@ChessNetwork]

😎

[@patricks.8506]

What suprises me the most about AlphaZero is what I read about its learning approach from the published paper. As deterministic as chess may seem (each move cause an immediate change in position), AlphaZero is using moveprobabiltiies (!) and analyses the outcome through Monte-Carlo-Simulations. Those simulations evaluate a large number of "random experiments" (!) to find the true value (here the best move). In contrast to Stockfisch's alpha-beta-search ( basically a deterministic "if-then" search algorithm to find the best move under the assumption that both players to "play optimal"), AlphaZeros approach is suprisingly much more accurate by following the law of large numbers and demonstrating that a little bit of randomness in a game like chess leads to superior performance. Simply Stunning. Thank you Jerry, for sharing and analysing those games. The effort you put into the analysis and the passion for the game is clearly noticable in every video. Keep it up!

[@porovaara]

this is true for the stock market as well.

[@Jarretman]

Jerry I'm loving the ending commentary from the robot in these videos! Keep it up man; great work!

[@gamemeister27]

It'd be hard for the Berlin to get any more popular for black.

[@gregzytka]

Anybody knows if AlphaZero is accepting students?

[@derendohoda3891]

Nice to see commentators exhibit patience with stockfish to get to a depth that indicates why the bot played the move it did. You'd think this would end charges of fraud but... anyway, great video Jerry. This game was actually much more subtle than the other games where AlphaZero squeezed stockfish into little balls of useless chess pieces, which shows the power of what the bot learned in self-play more than just a zugzwang style (if there can be such a style). Probably my favorite I've seen analyzed so far.

[@blugaledoh2669]

Deren Dohoda If given the choice, will you versus A0

[@blugaledoh2669]

10 game?

[@lucky45bullets]

AlphaZero for President!

[@slightlokii3191]

Alphazero would Analyse all the shitty things previous presidents have done and learn from their mistakes. That’s better than any other president there’s been so far xD

[@lucky45bullets]

Exactly!

[@klieu90210]

If we had A0 as president, we would be winning so much our heads would spin. I can't take that much winning 😣

[@SuperMitch1995]

Even a human could self learn to be a president in 4 hours and be better than the current one

[@Jay0neDE]

eh, they'd probably just program it to maximize profits in the economy instead of making life better for all humans.

[@MrYoYoBoBoChaddyInsaneZhangII]

Alphazero actually uses bobby Fischer’s brain

[@jeffgreen3376]

Yeah. I think they dug up Fisher's grave and cloned his DNA. Then they created a human-computer hybrid cyborg named Alpha Zero.

[@IsaacT22]

No , Bobby Fischer was a dynamic , tactical , punch throwing player , alpha has a very positionally crushing type of play.

[@andyisyoda]

Excellent analysis. Thanks.

[@cyndaquilpikachu6603]

These games are like a rainbow with the four hours of learning and the pot of gold at the end is all the games I'm just waiting for that one last piece at the bottom which is the hippopotamus defense :p

[@kristiann4346]

The dialouge at the end lol. Imagine robots in the future with chess engines like this built in then they school you like this after the game xD

[@bruceli9094]

All hail AlphaZero, the wise King. The all knowing philosopher

[@burt591]

Dude you should add the Terminator Genesys Main Theme at some point in the video, it would sound awesome. Also make the robot at the end say "You are terminated" :)

[@bruceli9094]

I have a question: Can pure brute Force still beat AlphaZero? Lets say a nation state gave Stockfish 100,000x the CPU power and challenged AlphaZero in 2020. Who wins??

[@jackwuchannel]

maximum brute force is technically impossible to be beaten. it depends how close AlphaZero is to maimum gameplay capability; it just might end up as being a draw every time

[@metalhulk105]

Theoretically, if we could achieve pure brute force we may be able to prove that Chess is, in the end, a kind of Tic Tac Toe game with a draw being the ultimate result always (or maybe the computer when playing as black does a full brute force and decides to resign before the start of the game!). Right now it's not possible for us to conclude that chess can always end in a draw. Brute force means to look at a 10^120 possible combinations at least. I don't know if 100,000x CPU would calculate the first move of the game within the lifetime of a human being. Maybe in the future there will be CPUs which think differently

[@calebgerhard7313]

Big brain

[@kgames7343]

Love theses videos man. You have a way of imparting your knowledge like no other chess channel. Your commentary delivery is spot on.

[@Jobmaofficialstar]

Jerry, I have been watching your videos for years now and I still feel that you are one of, if not THE best instructor in regards to taking complex seemingly-incomprehensible moves and positions, and explaining the logic behind them, making them understandable to amateurs. You're seriously like the 3Blue1Brown of Chess. I'm loving this series in particular and am always proud to sport my ChessNetwork shirt ☺️ Read more

[@ChessNetwork]

+Jobma Perangin Thanks a lot for the Jobma. I appreciate the compliment.😊

[@nr1771]

Jerry, have you seen the analysis of the Alpha Zero games by Jose Camacho Collados? He argues that Alpha Zero's victories over Stockfish may not be as impressive as they appear, for many reasons including the way the experiment was set up: "However, the experimental setting does not seem fair. The version of Stockfish used was not the last one but, more importantly, it was run in its released version run on a normal PC, while AlphaZero was ran using considerable higher processing power. For example, in the TCEC competition engines play against each other using the same processor. Additionally, the selection of the time seems odd. Each engine was given one minute per move. However, in the vast majority of human and engine competitions each player is given a fixed amount of time for the whole game, and then this time is administered individually. As Tord Romstad, one of the original developers of Stockfish, declared, this was another questionable decision in detriment of Stockfish, as “lot of effort has been put into making Stockfish identify critical points in the game and decide when to spend some extra time on a move” " His analysis is interesting and worth a look.

[@dannygjk]

+nr1771 “lot of effort has been put into making Stockfish identify critical points in the game and decide when to spend some extra time on a move”. Try it yourself. See how long it will take Stockfish to realize that AlphaZero's sacs etc. are excellent moves.

[@pandabearguy1]

hard to agrgue over processing power in stockfish's favor when stockfish is calculating 1000x more positions per unit time. The hardware alpha zero used during the game was not the insanity it used during it's learning phase

[@medajig]

erlend myhre If that is how you see it then you are wrong.. stockfish and other top engine like komodo and houdini have different algorithms.. and its impressive the way Alphazero look at the position like humans, it doesnt look position tree where by experience its bad.. the point is it should be nice if the stockfish version is the latest.. and should have its own opening book. Stockfish 8 will be annihilated by the latest version of stockfish, komodo 11.2.2 and the current tcec champion houdini 6.03.

[@grantcivyt]

+nr1771 I agree the AlphaZero results are questionable. If you're the challenger, you have to win on the winner's turf and on an otherwise level playing field. The games are certainly impressive, and I've no doubt that AlphaZero is innovative and formidable. I'm reserving final judgment until we can get a clearer picture of the experimental controls.

[@pandabearguy1]

im just saying that giving stockfish ultra hardware doesnt make much of a difference in its performance because of its inherent design, and alpha zero had greater gain in performance with increased time per move than stockfish aswell. alpha zero didnt use insane computing power during the game. in any case the fairness of the match is only relevant to the actual match. The whole point of the experiment is FAR beyond it's score against stockfish, it is simply a test to see the effectiveness of self learning AI. If google and deepmind wanted to make the strongest possible chess playing entity they would have given it far more time and computing power.

[@modolief]

Ah, poor Stockfish. We shed a tear for ya, mate.

[@lochvids108]

modolief 1 minute moves, plus stockfish running on a pc vs googles supercomputers of deep mind. Idiotic test

[@dandiaz19934]

wait is that really how it happened? if so, that's kind of terrible!! :( where can I read more about this?

[@lochvids108]

Dan-The-Man Funnily enough, google!

[@modolief]

LochVids: Here's my criteria for fairness regarding CPU power: How much power does each system draw? Stockfish and AlphaZero were probably both running on systems that drew about 300 Watts. The talkchess forum has some discussion on this if you care to dig it up. But I can criticize the following: Stockfish didn't have enough RAM for hash; Stockfish didn't have an opening book where AlphaZero effectively did; the time control was weird--should have been something more like game in 60 minutes plus 30 second increment; Stockfish 8 is a year old, the development version is much stronger.

[@lochvids108]

modolief yeah im not sure. I just read an article on the register saying it was manipulated googles way

[@maksudulhaque3227]

Alpha jerry evaluating xD

[@ChessNetwork]

😎

[@maksudulhaque3227]

Are human games coming back or alpha zero will go on? 😁😁😁

[@luckygozer]

You know you are really making me start to dislike AlphaZero with your video titles. He's like ohohoh i am 4 hours old let me show you how to play this game of chess. What about poor Stockfish? Has he been doing nothing all this time? Was all his effort in vain!? Wonderful video as always. These games are very intresting to watch.

[@Yzeyr]

he's like that 12 year old in video game that has 10x the aim you have and you sit there thinking wtf have i been doing all my life.

[@ashharkausar413]

Pretty sure alpha made stockfish rethink his life choices those 100 games....

[@ChessNetwork]

You scared me for a moment. Thank you 😎

[@squintyasians]

Ashhar Kausar the

[@RuLeZ1988]

Stockfish won't RIP by that, since it is opensource and needs only few ressources to be used on conventional PC's. So the replacement by AlphaZero will have to wait for some time, mainly because of the high ressource requirements and the huge investment into this running project. Other than that... this Chess Match is just a side project by the Team behind this AI.

[@zvit]

At 7:01 why didn't white take the pawn with the knight and not lose a rook?

[@klosnj11]

I have watched a few of these, but this time my wife and kids were sleeping still and i have no headphones available. Watching this muted makes me fill in both what the AI are thinking anf what you are likely explaining. It is an interesting experience.

[@YTcrazy]

Recently just started watching your vids, great in depth analysis! Love how you also go into hypothetical situations, and even entertain the thought so amateurs an easily see the outcome of it. Question though, at 6:17, you mentioned the rook to b6 ties the queen down to b2, but I'm a bit confused how. If white bishop at e3 takes knight on d4, then pawn at c5 takes bishop on d4, then queens on c3 takes pawn on d4. At this point, if black rook takes pawn at b2, the white queen gets a free rook...so unsure of how the rook ties down the queen, if the exchange went through. Thanks!

[@anthoras]

Rb4. You either drop the b pawn, or Qc3 and have to first defend b2 with Ra2 before you free the queen, but at this point white is completely on the defensive

[@ETBrooD]

I don't understand the tactic at 5:50 Qxc5 Rxe5 Qxe5 What then for black?

[@petergreen5337]

Beautiful lesson and demonstration.Thank you very much.

[@Krmpfpks]

Thank you. I kind of am frightened by the beauty of AlphaZeros play style. When Stockfish plays, you can see it’s a computer playing making it somewhat acceptable that it’s better than any human. AlphaZero plays like a well prepared human, just infinitely better... I for one welcome our AI lords.

[@jobe4279]

AI is taking over.....

[@sebastiancastro7382]

19:23 When AlphaZero gets the cool voice of Starcraft 1 Goliath

[@BattleFieldGalaxy]

btw jerry could one of your videos for beginner to chess masters series be about how to analyze a chess game? I find myself just pressing the arrow key without taking much learning points from the game. Thanks!

[@user-uq3iw2nf6x]

so what's the difference between stockfish8 and the stockfish that played alphazero? because the one in the analysis doesn't 100% approve with the moves played by stockfish.. what was the time control by the way?

[@grev7794]

the stockfish in the analysis may not given 64 threads to analyze.the one that played alpha zero is much stronger in hardware. the time control is 1 minute per move.

[@wasp89898989]

Jerry, are you gonna upload the zugzwang/queen-trapped-in-the-corner gamer?

[@Pipiopy]

is it ok if we give it the laws of physics instead of the rules of chess?

[@Baekstrom]

The Master Control Program in the movie Tron started out as a chess program. I always thought that was bullshit, because a program doesn't change it's purpose like that. However now I am not so sure about that any more. Maybe that movie was more prophetic than anyone realized.

[@bruceli9094]

A self learning program changes everything

[@Pintkonan]

hey if you're into prophecies, you should read 1984 by george orwell =b the compliances are insane =b

[@chessanalysis64]

Unique and deep analysis by a talented chess master , thanks Jerry !!!!

[@ChessNetwork]

Thank you

[@varishnakov]

I like these videos, but I'm scared of the robot on the end. I just end it early though so it's fine.

[@ChessNetwork]

Your neural network figured out a simple solution. :)

[@timmy1729]

He will be back

[@zogzog1063]

A also stop there. It is not a pleasant sound. Still hit the thumbs up!

[@michaelmikhael5203]

The only crime committed here is not giving stockfish a post interview like alphazero

[@jamesrav]

would Alpha Zero be infallible if it was 'forced' to start from a certain opening position, say 4 or 5 moves in? For example, the Fischer - Fine 'Evans Gambit' game you analyze ... could Alpha Zero 'blunder' in the same way Fine did , without benefit of an opening book? Is it basically incapable of making fatal errors like that ?

[@jamiepoush6732]

At ~14:00, Rc1 is discussed and Jerry points out that a black pawn sacrifice leading to the bishop covering g1 is why Fish didn’t do it... but I couldn’t help thinking, with four to five straight files of pawns unopposed, wouldn’t white be better off encouraging Alpha to go ahead and sacrifice that lone pawn? The black rook can’t force his way into the back ranks and do anything and the white King can just sit and wait while one of his pawns is eventually promoted... the rook and two bishops, could they prevent that? If not, it goes back to Fish’s original decision to sac the knight for two pawns... trying to take advantage of Alpha’s tendency to lean on its higher computing power, toss away its pawns, and win anyway. Being that Alpha’s rating is 100-500 points higher than Fish’s, it’s a no-wonder strategy that any player can always execute against a player several levels weaker... a master teacher guiding a grade school student, for example. But... was Fish wrong to abandon his pawns in favor of apparent over defensiveness when rejecting c1 for the rook there? Could a pawn be put through to promotion in a long unblocked chain with a knight and rook against a rook and two bishops? Or would black not have sacrificed that sole pawn opposing that chain? That was the critical moment in this game... the moment where Fish best had a chance to cash in on its thus far game long pawn gobblings.

[@darthartagnan4737]

Jerry.. do you play chess? I know nothing about you.. other than I've watches 12 of these videos and I love your detailed explanations.. I'd love to see you play sometime.. I know nothing about chess but I can't stop watching these videos.

[@blacktimhoward4322]

At 5:20 you're explaining that black's queenside is not coordinated but stockfish's evaluation was still slightly in favor of black. I'm no supercomputer but I'd rather be white than black at that point in the game, curious to know what stockfish saw

[@DjVortex-w]

Somewhere around the 9-minute mark, at a quick glance it looks like white is in a much better position. Tons of passed pawns, all pieces look active, black looks cramped... Yet, somehow, black is still better and proceeds to win.

[@stevenb4956]

6:21 why doesn't white take d4 knight with bishop and then pawn with queen if pawn kills bishop? It would be knight+pawn for just a bishop yes? White b2 pawn is still protected. Any downside to this or am I missing something?

[@themadmaxxp]

So you are telling me that AlphaZero had this move Re7 but he, on purpose, played Rf6 so that he can laugh at stockfish at the end for going for a perpetual draw with white pieces. It's all fun and game until someone teach him how to hold a gun :D :D

[@josephlombardo1246]

I think the idea at 2:53 of alphaZero "baiting," in the sense of tricking an opponent into suboptimal play, is unlikely. It was only trained to play against itself, I don't believe its evaluations would take into account suboptimal play by another engine. It can "bait" in the sense that alphaZero knows that knight sacrifice is not a threat, and if stockfish doesn't recognize that the sacrifice is not a winning strategy, so much the worse for stockfish.

[@Osiris261]

14:43 no def? e5 to g6 ? proberly followed by black g7 to d4 check. forcing white g1 to h1 now black f8 to f6 white g6 to e7 check. black has to move king then then e7 to d5 protecting that weak pawn

[@dragmio]

The burning desire of humanity to be enslaved is such that they're dreaming up "artificial intelligence" to bow to, not long after they got fed up with bowing down to dreamed up gods. Pathetic.

[@blahlool]

Stockfish just strayed a bit too deep into absolute zero waters. Go back to fish school, go back to shallow waters where the two-leggeds providing easy nourishment will tip the scales in your favor.

[@felipejhony6039]

Jerry, just a remark, alphazero doesnt know any master chess player or tatics or nothing of the sort. It doesnt have any knowledge about external chess games besides those played by itself.

[@maxnullifidian]

It was said on black's move 24 that "white is giving up a pawn," but it is black giving up the pawn. Just thought I'd point that out...

[@daninja98]

Did AlphaZero just punk Stockfish for playing for a draw with white? I think that's what I heard.

[@israelwasaness]

please make stockfish reply to tal's quote

[@modolief]

Yes, surely Stockfish has something to say at the weigh in before the match. No?

[@ryanswanson126]

beep boop beep.

[@danielloeb2044]

At 5:50 I'm confused. If queen takes pawn and rook takes knight, why wouldn't queen take rook?

[@gildeddrake1479]

I've seen 2 of these games and in both, A0 wins with black. Did he perform better with black than with white or is it just the games you decide to show because they're more interesting ?

[@inikous58]

so its like alpha zero knows how stockfish is gonna play and take advantage of it...I mean its like alpha understands whats the best move of stockfish to be played and it has already a plan about it..Poor stockfish

[@CloneHat]

This is an epic series of videos.

[@ThoseWhoStayUofM]

I'm confused. Since these are AI making decisions, can we not know exactly why they made the decisions they did? I'm just saying, how is it that we humans can create an AI and not have access to the reasoning for why it acts?

[@Banikan]

After seeing what AlphaZero is capable of I would argue that playing for a draw against AlphaZero is not a crime despite playing as white.

[@traplover6357]

Since AlphaZero tries to replicate human thinking instead of the brute force method of other chess engines, are humans technically still smarter in their approach to situations?

[@alanwhite3154]

Many people are saying that the march was not fair hardwarewise and timecontrol wise. Could you comment on that?

[@starplatinumzawarudo8755]

Yay more alphazero

[@JHsillypantsMcGee]

17...Nd4 would be something my feeble human mind would try, to threaten a fork on the queen and rook combined with a discovered attack against the e5 knight.

[@skuchessmania1991]

Stockfish rating 2652 will crush a lphazero

[@rmendeljacobs2832]

At 6:05, why can't Qxc5 and then if Rxe5 then Qxe5???

[@mw2nuke610]

because of rook f7 pinning the pawn and potential check mate.

[@kller04]

what they don't tell you is this game probably happened in less then 5 seconds. computers fast af

[@Michael-iw7ov]

lol well if the AI knew chess from a human input of chess it might have quoted Tal.

[@גידיפלדמן]

Srs now..im 2185 fide rated and even kasparov looks like a patzer compared to this games.

[@WisdomVendor1]

When my dad and I used to play, we had an agreement that if there was a three move repetition, whoever initiated it lost the game. Therefore we never had a three move repetition.

[@puneetkumarsingh1484]

We need more of Stockfish vs AlphaZero hames. Please!!!!

[@richardmartinez9328]

I really want to play with. Alpha zero I beat the shadows out this master program That I have and want to try something. More challengeging. Where can find this alpha zero..

[@abduljalal5727]

Its ironic you used stockfish to analyse the game parallel, that fishy fella has no idea whats going on.

[@rodsitvideos]

I do not look at chess games UNLESS they are alphazero then I am interestedin what you experts think. Sure, I can play chess and am interested in openings/defences but never "learn" them. So please, more alphazero stuff...

[@ChessNetwork]

+Steven Brown I anticipate covering all 10. 5 more to go...

[@ForeverDayGreen]

Why doesnt White go for the capture with the knight instead of the rook? I just cant See the downside to that move

[@cptnoremac]

Why would queen take knight here? Just take the rook. That square's defended.

[@patrickleblanc7751]

i feel like stockfish is just so defensive all the time and thats what makes it weak and Alpha keeps alphaing his brain out

[@lks9988]

The guards were bribed... Soon, the kingdom falls

[@michaelmoran4420]

Who programmed stock fish to think giving up a power piece was a good idea?

[@agusavior_channel]

I want see stockfish playing Crazyhouse

[@timmy1729]

What If stockfish learned from playing itself ?

[@DocsDota]

after 3:00, I think A0's thought process is rook's biggest challenge in mid game is getting pinned and forked by minor pieces, as rooks naturally have a back-line role and usually never gets trapped by pawns. The F7 rook is there since there's no more white bishop, both King and rook can take advantage of these squares and make room for other pieces on the dark squares maybe?

[@omenuk1566]

i keep trying to watch this video but your voice keeps putting me to sleep

[@josephjoestar3631]

I wonder how Bobby Fischer would do vs AlphaZero

[@daffaabhyass3166]

One question , the song in the very end of your every video ,whats that song?

[@hotfever]

I imagine in the future that all we will do is watching robots compete at each other

[@imarmanogaran3674]

At the end of the commentary was stockfish talking?

[@Vendavalez]

Whether Stockfish was given a weaker computer on purpose so that Alpha-0 had an advantage or not is irrelevant to the instructive moments highlighted in the video. Can we at least recognize that?

[@peterdao7346]

Alphazero will crush any chess program ever

[@timotejxxxl]

Stockfish had computer with 64 cpu cores. If that is not as strong as "normal computer" can get then I don't know. It was not run on a laptop if thats what you belived.

[@BusDriverGames]

I don't even play chess but I like your videos bud.

[@artemislogic5252]

i like this videos, but i cant stand that grinding robot voice at the end., it ruins it for me,