American Go E-Journal


Saturday August 7, 2010

Peter Shotwell, author of Go! More Than a Game, stopped by the U.S. Go Congress Friday evening to talk about some of the updates coming in a new revision of the book due in the next few months.  In the seven years since the publication of the first edition in 2003, Shotwell has done additional research in several areas that will appear in the new edition, including a re-examination of the attitude of Confucians toward go, advances in computer go, the combinatorics of go, and the possible spiritualization of Tibetan go.   It was long presumed that the Confucians did not think very much of the game of go, said Shotwell, identifying it with gambling and laziness, but a better dating of some of the source documents has led him to see an evolution in their thinking — including seeing some value in the game by the time of the last mention.  The Confucian writings were from the third century B.C. but the game was not explained at all, which implies that it was very well known and thus quite old at that time.  The earliest go board that has been found was from 141 B.C.E. in the guardhouse of a Han emperor’s tomb and go was alternately praised and damned in the writings of the Han period.  By the Three Kingdoms period in the third century, go was played by many and by 600 it was getting high praise in poetry.   An archeological find along with a game that Shotwell played in Tibet got him thinking about the connection between spirituality and go in Tibet.  He played with a government official while visiting Tibet and found out that there were some very different rules including only being allowed to move up to one space or a knight’s move from an existing stone, being awarded five points for taking the center, and losing twenty points if you lose all of the corners.  The starting position placed five stones of each color in a pattern on the fourth line around the whole board, which made for a kind of “race to the center.”  Shotwell thinks that these rules may have added a spiritual air to the game, which may have been done to “convince the early Buddhists that this new game from China was OK.” Two stone boards from the seventh century have been found in Tibet since the first edition and interestingly, one of the boards had two depressions on each side, which may indicate that Japanese scoring with prisoners may have been used.

There have been huge advances in computer go since 2003 as well. At that time “any kid could beat the computer programs.”  But, since the advent of Monte Carlo simulations and advances in the tree pruning algorithms in 2006 and 2007, “the top programs are at an amateur 1D level.”  Once in a while, those programs can beat professionals when getting seven stones on the 19×19 board.  On the 9×9 board, some programs are at a mid-level professional strength.  Those programs use a lot of computer power, though, with up to 112 cores allowing them to do 100,000 simulations per second.  Shotwell also pointed out some advances in go combinatorics, which is a branch of probability that studies the number of possible go games or positions.  Two of the more interesting results from studies by Dr. John Tromp compared possible chess positions with that of go: for go it is a 171-digit number whereas for chess it is only 46 digits. Even more astonishing, perhaps, is that just the number of digits in the total number of possible go games is larger than the total number of possible chess games.  Much of that information will be reflected in the new edition of his book, but Shotwell has also contributed numerous essays to AGA’s Bob High Memorial Library.
– Report/photo by Jake Edge