AlphaZero returns to play more chess | No Castling Chess

AlphaZero returns to play more chess against himself with modified rules. No castling is allowed.

AlphaZero vs AlphaZero
Queen’s Pawn Game: Chigorin Variation (ECO D01)
Jobava London System
No-Castling Chess – 2019 article:

👉Visit the store:

If you like my videos and you would like to support my channel, please consider donating via PayPal at one of the addresses below. Thank you!

Contact me on:
GMail: [email protected]

My profile:

Official site of the Computer Chess Championship (CCC):

Official site of the Top Computer Chess Championship (TCEC):

Official site of the Leela Chess community:

Official site of the Stockfish community:

If you want to get better at chess fast, i recommend you to study simple and complex endgames:
Endgame Fundamentals:
Endgame Practice:
Pawn Structures:
This is the microphone i use for recording:

DISCLAIMER: This description contains affiliate links (microphone link, links), which means that if you click on one of the product links, I’ll receive a small commission.
AlphaZero image by
Game moves:

1. d4 d5 2. Nc3 Nf6 3. Bf4 a6 4. Nf3 e6
5. e3 c5 6. Bd3 Nc6 7. Kf1 h5 8. dxc5 Bxc5
9. Qe1 h4 10. h3 Rh5 11. e4 Bb4 12. Rd1 Bd7
13. e5 d4 14. exf6 Qxf6 15. Bg5 Rxg5 16. Ne4 Qxf3
17. gxf3 Bxe1 18. Nxg5 Bb4 19. a3 Be7 20. f4 Bd6
21. Rg1 Bxf4 22. Rg4 e5 23. Rxh4 Ke7 24. Ne4 b6
25. Re1 Be6 26. Ng3 b5 27. Be4 Kd6 28. Ne2 Bc4
29. b3 Bxe2+ 30. Kxe2 Rc8 31. Rg1 Bh6 32. Rh5 Ne7
33. f4 Bxf4 34. Rxg7 Rc3 35. Kf2 Re3 36. Bb7 Rc3
37. Be4 Re3 38. Bb7 Rc3 39. Be4
#AlphaZero #ElzChess


  1. I agree with Kraminik. If we change everything just for sake of entertainment, life can became totally non-sense after sometime. If people change to much in somethings, it will lose its meaning!

  2. I 100% agree with your comments opposing Kramnik's comments.. Fischer Random does what it's supposed to do, which is provide unfamiliar positions so players can use their talent and creativeness, rather than play lines through wrote memory! It's also not any less appealing.. mainstream and the elites of chess were against Fischer, as you probably know, and they don't want to promote it, that's why I think it hasn't taken off as much as it should have

  3. Non-castling is OK for friendly games during coffee or something, but I´d never make it something "official". Replacing chess? /laughs/

  4. there is nothing wrong with chess its a great game, draws are part of chess and always will be.

  5. Agreed, 960 should be the future – and the memory gurus such as the likes of Kramnik are just afraid that the real Bobby Fischers out there like Wesley So will eat them for lunch in such matches.

  6. However interesting the idea is, it would be ages before anybody changes the official rules of chess. Maybe in 20 years from now. Not yet. Time has not come. The Fabiano vs Magnus 12 games draw was nothing. Nobody would change rules because of that. Until the draws become so commonplace, to be in almost every top game.
    PS Wanna see more AlphaZero chess games, preferably against other engines than itself!

  7. Get rid of the AI..they it are replacing humans in the workplace..makes it boring….Peter Parker deserve the truth.

  8. Even when humans adjust to 'no castling' chess and opening theory is developed it will not be as routine as 'castling' chess.
    1. Castling chess makes it easier to get your king safe, (if castling is a good idea in a particular game).
    2. In castling chess you have two options castle or not which makes it more likely that a player can be safer.

  9. I agree with you that any changing of the rules will not change anything about the excitement on chess, once people or machines are accustomed to the changes. That goes even for Fischerandom: give AlphaZero a year and it will have analyzed all the 960 variations exhaustively. Let's face it: at the absolute top of AI-Zero, chess will produce only draws in the long run, if you let the players choose their own openings. That's the fate of the game, but it doesn't have to affect the joy human players can take from it. More than 99,99% of the human players will stay at sub-grandmaster level, winning and losing often against their opponents. And grandmasters will learn from top-AI, the top grandmasters will hardly ever lose because of this. If you like to win every game, or lose some of them, at absolute top level, you have to take up another game, like go.

  10. If you want pure creativity, randomize the back rank pieces and only reveal the chess board 15 minutes before the game starts. Each player would have the same back rank setup, thus forcing them to think through each move as if it was their first time playing.

  11. Original (old Indian) rules for chess had no castling and also no 2-hop pawn moves.

  12. Re Chess960 "Fischer random" or double-random chess(960^2) yielding possibly-biased initial setups, that is no problem for computer chess tourneys since they play color-balanced game-pairs. For human double-960 chess without bias, one could have ploys like "black sets up the white men as he wishes subject to Fischer restrictions; white sets up the black men; now black can optionally choose to become white; now commence play." Point of those ploys is, if white evilly set up a biased white-favoring position, black could exploit that by becoming white. Therefore white tries to keep it even. There are many possible such ploys to eliminate bias, resembling the "brother cuts the cake, sister chooses" ploy to inspire children to cut cakes fairly. This actually would be MORE fair than ordinary chess. So Kramnik's bias criticism then is completely destroyed.

  13. No castling is still static unlike fischer random which is dynamic.

  14. To Kramnik's credit, if you're looking from the perspective of entertainment, it helps to appeal to a wider range of player base by showing familiar setups. Yes, people will adjust to that so it will become pointless. In my opinion, all those crazy variants of chess just don't interest me. I feel the better way to go is to do what we do to chess engines. Force randomized openings and have both players play both sides to see who can hold better. It doesn't even have to be familiar lines. They can be purposely imbalanced or even dubious. If both sides get to play it, it is fair enough. This is after all the reasoning behind why we do this to engines….entertainment.

  15. Comparing chess 960 to real classical chess & saying " it's more exciting " is like comparing indoor soccer to real clsassical soccer and claiming it to be more exciting 😆

  16. Cool. Totally did not expect them to bring A0 back.

  17. Hey Elz could you show us some deceives games that engines are free to choose the openings they are going to play? Thx…

  18. Another advantage of Fischer Random is that it encourages players with different levels of opening theory prep to start playing again with each other. I stopped playing classical chess with some friends of mine, who enjoy the game, but are amateurs with a low level of opening preparation, because it's not fun at all to constantly win (or lose) a game gaining decisive advantage in the first 10 moves. Fischer Random gives less prepared players the opportunity to reduce the opening theory gap and start having fun again together. Thumbs up for this format.

  19. Moreover Fischer Random encourages players to think tactically and strategically since the first move, which is a great benefit for youg players who approach the game and for amateurs who don't have the time to study opening theory of classical chess.

  20. Who want to play the game which takes 5 to 10 minutes to play first move correctly? The future is no castling chess not 960

  21. I think classical chess is the most flexible variation of all these post modern new rules chess because you always have that opportunity to choose if you are going to play by memorised games or you are going to create your own masterpiece during the game. In classical chess that both opportunities are possible, open and can be mixed (you can play some part of the game from memory and only then try to create or create right from the beginning but that must be very good creativity othervise you'll be punished or maybe not).

  22. Fisher random is nice but will never replace standard chess. In contrast, no castling is an earlier and purer form of the game, which is still played casually in many parts of the world. This makes it a better candidate for the new FIDE standard. It’s not clear whether it would in fact lead to more interesting chess, but if king safety is harder to achieve it could increase the advantage of the first move enough to lead to more decisive games.

  23. This is nothing new, lots of people have been playing no castling chess known as shatranj for years during the Islamic golden age in Spain.

    These were the rules noted
    ● The board was not yet checkered.

    ● Stalemating the opposing King resulted in a win for the player deliver-

    ing stalemate. (In modern chess, stalemate results in a draw.)

    ● Capturing all of the opponent’s pieces except the King also counted as

    a win, provided that one’s own King could not be left alone on the

    very next move.

    ● There was no castling option (wherein the King essentially changes

    places with one of his Rooks

  24. no other computer will have any chans if they traning alpha zero i bit more.
    I have 2000ELO in chess so i am not a newbe
    I have also see the alpha go that is faar more impresiv then chess.
    it have so much more possibility combinations then chess, how in hell they computer figure out that game to cruch the woulds best player on that game in insane.
    this is how many moves there is in chess.

    this is the count for alpha GO

Leave a Reply

Your email address will not be published. Required fields are marked *