Game Changer: AlphaZero’s Groundbreaking Chess Strategies and the Promise of AI by Matthew Sadler & Natasha Regan 2019 New in Chess 416pp
In December of 2017 DeepMind released a paper showing that their self-learning AI, AlphaZero, had defeated the powerful and popular engine Stockfish in a 100 game match by what seemed to be an inconceivable score of 28 wins, 72 draws, and no losses.
However, a deeper look showed that the terms of the match were deeply flawed. The playing field was nowhere near level, and so as many people in the chess world went all agog at the results, I was in the small group of non-believers. Yes, it was impressive *how* AlphaZero played, making speculative sacrifices, etc. but as my friend Hikaru Nakamura said in an interview “I don’t necessarily put a lot of credibility in the results simply because my understanding is that AlphaZero is basically using the Google supercomputer and Stockfish doesn’t run on that hardware; Stockfish was basically running on what would be my laptop. If you wanna have a match that’s comparable you have to have Stockfish running on a supercomputer as well.”
And as far as I was concerned that was that.
Then, a few weeks ago I was listening to Ben Johnson’s excellent podcast Perpetual Chess. The authors of this book were on and one of the things that they mentioned early on was that the DeepMind staff was also receptive to the criticism and as a result they had set up a second match.
As Wikipedia notes about the second match “In the final results, Stockfish ran under the same conditions as in the TCEC superfinal: 44 CPU cores, Syzygy endgame tablebases, and a 32GB hash size. Instead of a fixed time control of one move per minute, both engines were given 3 hours plus 15 seconds per move to finish the game. The version of Stockfish used was version 8. AlphaZero won with a score of 155 wins to 6 losses, with the rest drawn. DeepMind also played a series of games using the TCEC opening positions. AlphaZero won 95 out of the 100 mini-matches from these positions.”
Suddenly I found my interest in AlphaZero piqued. Perhaps there was more to this after all.
Then, like a beam of light shot straight into my soul I come home one day to find the book Game Changer in my mailbox.
The content is laid out in eighteen chapters in five parts.
Part I AlphaZero’s history
Chapter 1 A quick tour of computer chess competition
Chapter 2 ZeroZeroZero
Chapter 3 Demis Hassabis, DeepMind and AI
Part II Inside the box
Chapter 4 How AlphaZero thinks
Chapter 5 AlphaZero’s style – meeting in the middle
Part III Themes in AlphaZero’s play
Chapter 6 Introduction to our selected AlphaZero themes
Chapter 7 Piece mobility: outpost
Chapter 8 Piece mobility: activity
Chapter 9 Attacking the king: the march of the rook’s pawn
Chapter 10 Attacking the king: colour complexes
Chapter 11 Attacking the king: sacrifices for time, space and damage
Chapter 12 Attacking the king: opposite-side castling
Chapter 13 Attacking the king: defense
Part IV AlphaZero’s opening choices
Chapter 14 AlphaZero’s opening repertoire
Chapter 15 The King’s Indian Samisch
Chapter 16 The Carlsbad
Part V Conclusion
Chapter 17 Epilogue
Chapter 18 Technical note
After playing through the games of AlphaZero I really can’t rave about the games enough.
In addition to the book, the authors have created a YouTube channel where they only include games not otherwise included in the book.
The truth about the games of AlphaZero is that they are amazing works of art in so many cases.
Rather than listening to me ramble on, look at this game from the YouTube channel.
Then listen to the authors on Perpetual Chess:
Then go buy this book. You won’t regret it.
Til Next Time,
If you like this blog, please consider becoming a Patreon supporter. Any money I raise will go towards lessons and stronger tournaments.
If you can spare it, please click here and become a supporter. Even $1 a month can help me achieve my dream.