Congratulations to Sir Demis Hassabis, who last week was awarded a Nobel prize for his work on AlphaFold, which uses artificial intelligence to predict the structure of proteins. Developed by DeepMind, AlphaFold belongs to the same family of work as AlphaZero, which revolutionised computer chess when it was released in 2017, and before that AlphaGo, which in 2015 was the first program to defeat a professional Go player.
I had the honour of partnering with Hassabis for the Pro-Biz Cup at the London Chess Classic in 2021. Now CEO of Google Deepmind, at 13 he ranked second in his age group behind Judit Polgar. There is no doubt that his youthful fascination with strategy games sowed the seeds of his scientific achievement. Between 1998 and 2003, Hassabis was a five-time winner of the ‘Pentamind’ event (playing five different classic strategy games) at the Mind Sports Olympiad. After graduating from Cambridge with a double first in computer science, he founded Elixir Studios, where he applied his early interest in AI to computer game development, before returning to academia to study neuroscience.
In a 2015 interview with the Financial Times, Hassabis reflected: ‘If you take a kid who is fairly thoughtful and introspective, then train them professionally from the age of four to 13, playing adults and grandmasters at chess, they can’t help but think hard about what their brain is doing when coming up with these moves.’
That facility with metacognition is fundamental to developing as a chess player, and intertwines with anticipating the intentions of an opponent. In that respect, the work of the WW2 codebreakers at Bletchley Park provides a clear precedent for chess expertise having a practical application. Alan Turing was not a strong chess player, but many of his colleagues were. Stuart Milner-Barry, who represented England many times, once wrote about their work on breaking the Enigma code: ‘For both Hugh [Alexander] and myself it was rather like playing a tournament game (sometimes several games) every day for five and a half years.’ (Hugh Alexander, twice British champion, wrote The Spectator’s chess column as ‘Philidor’.)
Plenty of Nobel prize winners have taken a casual interest in chess, though in most cases there is no record of whether they were any good. Einstein was said to have been friends with Dr Emanuel Lasker, the second world champion and himself a mathematician, and contributed a preface to Lasker’s biography. A game purported to be between Einstein and Oppenheimer can be easily found online, though its veracity is doubtful.
Before Hassabis, one of the most skilled Nobel winners was John Cornforth (Chemistry, 1975). In 1937 he broke an Australian record by playing a dozen games of blindfold chess simultaneously. One of those is shown below.
John Warcup Cornforth–F.F. Kelly
Blindfold Simultaneous Display, Perth, 1937 (see diagram)
1 e4 e5 2 Nf3 f6? This, the ‘Damiano Defence’, is known to be very bad. 3 Nxe5! Qe7 Better than 3…fxe5 which loses by force. For example: 4 Qh5+ Ke7 5 Qxe5+ Kf7 6 Bc4+ d5 7 Bxd5+ Kg6 8 h4 h6 9 Bxb7! (deflection) Bd6 10 Qa5 Bxb7 11 Qf5# 4 Nf3 4 Qh5+? g6 5 Nxg6 Qxe4+ turns the tables d5 5 d3 Bf5 6 Be2 dxe4 7 dxe4 Qxe4 8 O-O Bd6 9 Bb5+ Kf8 10 Re1 Qg4 11 h3 Qh5 12 Re8+ Qxe8 13 Bxe8 Kxe8 14 Nc3 a6 15 Qd5 Bc8 16 Bd2 Ne7 17 Re1 Nd7 (see diagram) 18 Ne4! A stylish way to win, particularly blindfold. If 18…Nxd5 19 Nxd6+ and then 19…Kd8 20 Nf7# or 19….Kf8 20 Re8#. Rf8 19 Nxd6+ cxd6 20 Qxd6 Black resigns
Comments