We introduce a novel approach to measuring howhumans learn based on techniques from information theory andapply it to the oriental game of Go. We show that the total amountof information observable in human strategies, called the strategicinformation, remains constant for populations of players of differingskill levels for well studied patterns of play. This is despite the verylarge amount of knowledge required to progress from the recreationalplayers at one end of our spectrum to the very best and mostexperienced players in the world at the other and is in contrast tothe idea that having more knowledge might imply more 'certainty'in what move to play next. We show this is true for very localup to medium sized board patterns, across a variety of differentmoves using 80,000 game records. Consequences for theoretical andpractical AI are outlined.
|Number of pages||6|
|Journal||Proceedings of the World Academy of Science, Engineering and Technology|
|Publication status||Published - May 2010|