26 July 2019

AlphaZero's Zeros

Earlier this year, in Kindle Best Seller (February 2019), I mentioned a Google Books preview version of Sadler & Regan's book 'Game Changer':-

I noticed a section in the book on engine evaluations of '0.00'. I think it was at the end of chapter 2, titled 'ZeroZeroZero'. [...] Now the preview has disappeared and the page is marked 'No preview'. I've been looking into the topic of engine evaluations '0.00' for some time, but I'm not prepared to write a full post on it. I had hoped to use the half-dozen (or so) 'Game Changer' pages to launch a discussion, but that will have to wait for another day.

Last week, while working on the post Interpreting AlphaZero, I discovered another preview copy of the book, which gives me the chance to continue that 'Best Seller' discussion. Chapter 2 of the book, titled 'ZeroZeroZero', starts,

The end of 2017 saw the release of 10 games from a match between the strongest superhuman-strength chess engine -- Stockfish -- and DeepMind’s AlphaZero. The match result was a comprehensive victory for AlphaZero.

A few pages later, the discussion of 0.00 kicks in with a section titled '0.00'. I assume it was written by GM Sadler:-

A couple of months ago, I was reviewing an excellent middlegame book of extremely complicated positions analysed by a world-class grandmaster. The author mentioned a few times when the crisis in a particular position was at its highest that his engines were assessing the position as 0.00. Judging from his comments, he wasn’t completely sure what to make of such an assessment which didn’t always tally with his intuitive or practical feeling about the position.

Any chess player who has analysed complicated positions seriously with an engine running in the background will recognise this scenario. You summon your inner genius, think up a brilliant doublepawn sacrifice to set the board on fire… and your engine responds with an evaluation of 0.00 and a best line ending in you forcing a bizarre repetition of moves. And it’s not just one engine: they all want you to force a repetition (though the ways to do so often vary). It’s perhaps the most irritating and obstructive thing that these otherwise fantastic engines do during analysis.

The discussion goes to the heart of engine analysis and I'll continue it in another post. As an aside, who is the 'world-class grandmaster' mentioned in the first paragraph of the previous excerpt?

No comments: