click for larger view

Analysis by Houdini 4 d 20 dpa done only

1. - / + (-1.02): 20.Be5 f6 21.Bh2 h4 22.Qc6 Kf7 23.Nc1 Rd2 24.Re1 Qc8 25.Qxc8 Rxc8 26.Rb1 e5 27.Nb3 Rd6 28.Red1 g4 29.Rxd6 Nxd6 30.Nd2 Kg6 31.Kf1 Bf8 32.b3 Bh6 33.Rd1 Bxd2 34.Rxd2 Ne4 35.Rd3

click for larger view

Analysis by Houdini 4 d 20 dpa done

<<1. - / + (-0.92): 19...g5> 20.Be5> f6 21.Bh2 h4 22.Rad1 Rxd1 23.Qxd1 Kf7 24.b3 g4 25.Re1 Qxd1 26.Rxd1 e5 27.Kf1 Ke6 28.Re1 Kd6 29.f4 gxf3 30.gxf3 Kc6 31.Kg2 h3+ 32.Kf2 Rd8 33.Rg1 Rd2 34.a3 Bf8

<The Byrne game was quite fabulous, and I cannot call to mind anything to parallel it. After White's eleventh move I should adjudicate his position as slightly superior, and at worse completely safe. To turn this into a mating position in eleven more moves is more witchcraft than chess! Quite honestly, I do not see the man who can stop Bobby at this time...>>

click for larger view

Analysis by Houdini 4 <d 23 dpa done

1. = (-0.17): 19.Ng3> Nh4 20.Ne4 Kg8 21.Be3 f5 22.Nc3 Kh7 23.Rae1 Qc8 24.f3 Bf6 25.Bf2 Re8 26.Qa3 Rd2 27.Qxa7 Bxc3 28.bxc3 Qd8 29.f4 Qf6 30.Bxh4 Qxh4 31.Rd1 Red8 32.Rxd2 Rxd2 33.Qxc5 Rxc2

2. = / + (-0.38): 19.Nc3 h4 20.Rfd1 Nd4 21.Rd3 Rh5 22.Rad1 h3 23.gxh3 Qc8 24.Be3 Rb7 25.Bxd4 cxd4 26.Rxd4 Rxb2 27.Ne4 Qb8 28.R4d3 Qa8 29.Ng3 Rxh3 30.Rd7 Rb8 31.Qa5 g6 32.a3 Rc8 33.Rc7 Qb8 34.Rdd7 Rxc7 35.Qxc7 Qxc7 36.Rxc7 Rh4 37.Kg2 Bd6 38.Rxa7 Rxc4

3. = / + (-0.63): 19.Be5 h4 20.Nf4 Rh6 21.Qc6 g5(T) 22.Nd3 h3 23.gxh3 Rxh3 24.b3 Nh4 25.Rae1 Nf3+ 26.Kg2 Nxe1+ 27.Rxe1 Rh7 28.Re3 Kg8 29.Rg3 Rh4 30.a4 a5 31.Bc3 Rc7 32.Qf3 Rc8 33.Ne5 f5 34.Ng6

4. = / + (-0.63): 19.Qc6 h4 20.Be5 Rh6 21.Nf4 g5(T)

1. After 6 h4 h5 both sides have advanced the h pawn but Black has the bishop pair and it is White who has exposed himself more.

2. 8 exd6 spends two tempi for development.

3. Nfd2 clears the square f3 for the queen but it costs considerable time for development.

All this suggests that White's overall advantage is not as great as Black's loss of castling may suggest and indeed it took only two errors on Giri's part to pass a great advantage to Black.

< Score: -0.57

"...involved exactly 3 of the key elements identified by Sandler and Regan as key to the successful attacking technique of AlphaZero against a castled King--namely, the combination of:

1. controlled diagonal
2. controlled file
3. push of h pawn."

As fully explained with examples dating back to Greco by Vukovic in 'The Art of Attack in Chess.' (1965), he too took a lot of his material from earlier writers on the subject, Tarrasch, Spielmann and Alekhine...

Chapter 7 "Ranks, Files and Diagonals in the attack on the Castled King."

Alpha Zero, and it's fans, (a machine with fans...good grief) some think this tin can has revolutionised chess. It is simply re-inventing the wheel - it's added a few new spokes, I'll give it that, but we cannot allow it to remove the saddle and handle bars. We are still in the seat and steer chess. It's a human v human game.

AlphaZero is a chessplayer. Questions of hardware and configuration aside, it's arguably the strongest chessplayer on earth, certainly far stronger than Carlsen or any human, and it plays beautiful chess. It's never hijacked a national chess federation, or dodged an opponent, or accused the Jews of trying to take over the world. It's also a fascinating application of reinforcement learning. Why shouldn't it have fans?

<As fully explained with examples dating back to Greco by Vukovic in 'The Art of Attack in Chess.' (1965), he too took a lot of his material from earlier writers on the subject, Tarrasch, Spielmann and Alekhine...>

Since there is nothing completely new under the sun, this sort of argument can easily be used to claim that no one has learned from anyone post-Morphy. But the style of the strongest players always tends to influence everyone else. Chessplayers were surprised, in my recollection, when Kasparov came along and showed that Karpov's style was not the only style that could succeed at the very top. When Stockfish and other engines seemed to show that nearly any position could be defended, that had an effect on how humans played the game. Now that AlphaZero has come along and shown that even Stockfish can be overcome with aggressive, dynamic play, that's also having an effect.

I think some over-the-top claims of its influence on Carlsen have been made, but he does seem to be playing rather differently this year, and the change has been welcomed by everyone, except his opponents. Has A0 had an effect on him? Yeah, I think so.

<It's a human v human game.>

Hi K.P.

(here we go again, usually we debate in a civil manner, let's keep it so.)

Not afraid, just bemused at the glorification being bestowed on a computer program which does not even know it's playing a game.

First this bit:

"It's never hijacked a national chess federation, or dodged an opponent...etc."

You are correct there, computers don't hurt, cheat or have any racial discrimination - only people have that particular gift.

"...AlphaZero has come along and shown that even Stockfish can be overcome with aggressive, dynamic play."

I howled like a lost dog in the wilderness for years decrying the faults in Stockfish. Alpha is highlighting these defects.

Thank you Alpha. I'm a human with a ton of baggage, the last three letters 'age'. being my biggest suitcase.

(I'm 68 on June 30th K.P. - I'll be touring in the desolate blissful peaceful wastes of north Scotland - but you can still wish me a Happy Birthday.)

Where was I?....Oh yes thanking Alpha because it can calculate deeper without tiring, without worrying about winning or losing and frying S.F's chips (fish and chips...geddit...).

"... no one has learned from anyone post-Morphy."

We could go back to Greco. Every combo played in one form or another can be found in the manuscripts of Greco.

Greco's analysis is dodgy (all the games are made up) but the idea behind the combo's are sound.

Later players came along used these combo's in different settings and then figured out and taught how to set them up by looking for landmarks in positions.

As decades passed the level of the game rose higher and higher not only opening analysis but in the areas of positional play, endgame theory and dynamic play.

Alpha is showing us nothing new, it's been infront of us all the time. This Alpha tool is simply blowing fresh (and welcome) air.

I cannot credit Carlsen's amazing run in the past six months down to Alpha. (I think we agree there) He is a genius who blows hot and cold.

His cold makes him a standard good GM.and admits when he blows cold he's been lucky. When he is hot he is magnificent.

"....and it plays beautiful chess."

Beauty is in the eye of the beholder, if you like what it does, then so be it. Afraid I cannot get excited about a machine beating another machine (Stockfish) which I know was flawed.

Need to see Alpha playing a 3rd choice move in a lost position going for a swindle. That is human v human chess. (our game) playing v human foibles.

That to me from a computer would be, for want of another word, 'beautiful' because it's not the move but the idea behind it I've always found beautiful and stimulating, Alpha does not have ideas.

Reply soon mate, at noon tomorrow the mobile is switched off and I'm off planet till the end of next week.

Anyway, I've pointed out Stockfish vs AlphaZero, 2018 a few times. That's as close as a computer can come to playing as badly as you want it to.

<Beauty is in the eye of the beholder, if you like what it does, then so be it. Afraid I cannot get excited about a machine beating another machine (Stockfish) which I know was flawed.>

You should be ashamed to write something so head-poundingly dumb.

Never said **** traps, I'm talking about a swindle based on a two move trick when a player realises he going down and switches into rat mode.

You can set traps when you are winning, level or losing and you can have such a thing as coincidental trap.

I want it to realise it is losing and go for a roll of the dice. A swindle, which is a very important part of human - human chess. It even has it's wiki page.

https://en.wikipedia.org/wiki/Swind...

Never seen the game you linked to, Thanks. I'll reply there as it is more relevant. Stockfish vs AlphaZero, 2018.