David Ormerod’s account of the fourth Go game between Lee Sedol and the AlphaGo computer program has sent me off thinking about two aspects of the analysis of performance:
How do we account for and share a story of performance?
What is the place of tactical understanding and personal initiative in an age of analytics?
Accounting for Performance
David’s analysis of game four appears on the GoGameGuru website. In it, he observes:
Having lost the match already (0v3), the heavy burden of expectation had been lifted from Lee’s shoulders and he was now able to play more freely. Once again, Lee reviewed the previous game well into the night with other top professional players, looking for a chink in AlphaGo’s considerable armor.
David discusses Lee’s use of an ‘amashi strategy’ that differed from his approach in game 3 of the match. He describes key moves in this new strategy and David offers an account of what the computer program response could have been.
Move 78 was the key to Lee winning the game. This was described as “a brilliant tesuji ” and “the only move that would keep him in the contention”.
Tesuji are the close range tactics of Go. If you imagine Go as a mental martial art, tesuji are the techniques of hand to hand combat. They are clever moves which contribute to Go’s beauty as an artform.
Strong players can usually spot a tesuji in the blink of an eye, through years of training and experience. So, I imagine, can AlphaGo’s policy network, otherwise it wouldn’t be able to play so well.
But not all tesujis are equal. Some can be found by most players. Others are so rare, so exquisite, that even the majority of professionals don’t see them.
The move Lee played was the latter kind.
David’s article adds layers of detail to the analysis of this fourth game. This layering, including other people’s analysis, really encouraged me to think how we might use a similar approach in sport. It requires immense insight and the ability to share the story. Each move in the game is notated and shared too as an SGF file.
One of the commentaries David Ormerod includes in his account of game 4 is An Younggil’s. This commentary asserts:
This game was a masterpiece for Lee Sedol and will almost certainly become a famous game in the history of Go. After his brilliant move at 78, Lee’s play was perfect.
Another commentator thought the move 78 was “the hand of god”. A move so special that the computer program could not see it coming.
This victory in game 4 raised for me thoughts about preparing for competition in an age of analytics. It coincided with reading Xinyu Wei and his colleagues’ paper on predicting shot outcomes in tennis using style and context priors.
It coincided to with finding Sporticus’s recent post on constraints, teaching and personal learning … and was enmeshed in the fascinating conversations going on at myfastestmile.
I wondered what Lee’s move 78 might look like in dynamical systems approaches to learning and how we create space for personal initiative in individual, co-active and team games.
Limited by our imaginations?
David Ormerod’s analysis of game 4 in the Lee Sedol v AlphaGo match set me of thinking about narratives and learning environments.
As I was concluding this post I read this wonderful exchange shared by Luke on Twitter.
I thought Joanna’s reply was an absolute delight:
I love the idea that ‘secret smiling’ might be the key to extending our imaginations and challenging the overdetermination of playing possibilities in an age of analytics.
Hearing other voices might just help us.
Al Smith shared a link to ‘s post about the match https://googleblog.blogspot.com.au/2016/03/what-we-learned-in-seoul-with-alphago.html
Demis makes this observation:
Commentators noted that AlphaGo played many unprecedented, creative, and even “beautiful” moves. Based on our data, AlphaGo’s bold move 37 in Game 2 had a 1 in 10,000 chance of being played by a human. Lee countered with innovative moves of his own, such as his move 78 against AlphaGo in Game 4—again, a 1 in 10,000 chance of being played—which ultimately resulted in a win.
The Conversation published two posts about Go on 21 March.
Silvia Lozeva on the playfulness of Go.
Jonathan Tapson on artificial intelligence.