Global context influences local decisions

This paper studies the development of human expertise in the game of Go. Although superficially a simple game, Go is the most difficult of all established games for artificial intelligence, no computer program yet achieving top international level on a full 19x19 board. On smaller boards, such as 9x9 computers \emph{are} competitive, implying that the understanding of the complex global interactions is the key to human superiority. By mining thousands of positions online, we show that at some player levels the sequence of plays leading up to a \emph{local} position is a stronger determinant of the next move than the position alone. This suggests that the sequence of plays is an indicator of global strategic factors and thus provides a context for the next move in addition to the local position itself.