Computer Go

    Keywords: Software

Table of contents


Introduction

While many strategic games, such as chess and backgammon, have in the last few decades been conquered by machines, there is much yet to be done in the field of computer Go. In 1995, professional player Janice Kim defeated the top program of the day despite of a whopping 25 stone handicap. Since then, rapid progress has been made. By the early 2000's, computer programs were at the footsteps of amateur 1 dan - quite a large improvement, but still lightyears away from the realm of top human professionals. However, around 2008 a new approach to Go AI began sweeping the Go scene: Monte Carlo. This method basically relies on playing out millions of "random" games, and using statistics from these simulations as a sort of positional judgment. As of 2011, several programs playing of KGS have reached 4 or 5 dan. These ranks are a bit deceptive, however, because with sufficient practice playing against computers, a human can often adapt to the weaknesses of the programs and subsequently perform much better against them. Also, it is a common belief that on the go server people tend to take games against computer opponents less seriously; this coupled with the computer's tendency to make very strange moves makes human-computer matches likely to be Maynerds. Today, the top programs play competitively against professionals with around a 7 stone handicap.

Thus computer Go is a very exciting part of Artificial Intelligence (AI), and many new ideas and techniques are yet to be discovered. It's interesting to note that compared to the approach which proved so successful with Chess, brute force, appears to be not nearly as useful for solving Go problems because of the huge branching factor of the game tree.

In fact, Go endgames have been proven to be PSpace-hard, let alone other parts of the game. Also, many other aspects of Go, including life and death, are also known to be NP-hard. This means that it is very unlikely to be able to find a reasonably fast algorithm for playing perfect Go. So it looks like it's all about heuristics (surprise surprise).

Patrick Taylor: I'm not sure heuristics are all that bad. Humans don't know the perfect move sequence any more than computers do. Therefore, we basically play with heuristics as well. Proverbs are essentially heuristics used by human players to approximate good play.

In July 2011, KGS records accurately show computer program Zen19s achieving 4 dan. In June and July 2011, Zen19s played 700 games on the KGS Go server. Zen19s plays at 20 minutes main time and then 30 seconds per move. A player can download the games of Zen19s from the KGS server. The player can study the games to find the program's weakness and try to exploit it. After 700 games and almost 2 months, Zen19s is 4 dan on KGS. This shows that finding the weakness of computer players is just as hard as finding the weakness of human players. Computers have truly reached the 4 dan level. Successful chess programs do not use brute force, but are selective in finding a move. Chess programs that use brute force, play at a low level because of the huge branching factor. Chess programmers do not use fast algorithm for playing perfect Chess. Mark Schreiber, July 22, 2011


References

See the [ext] Computer Go Bibliography maintained by the [ext] Computer Go Group at the [ext] University of Alberta.


Philosophy

Before delving into computer Go, it would behoove the novice to consider some philosophical questions about computers and Go.


List of Existing Go Playing Programs

Please see Go Playing Programs for a discussion about the best programs currently available.


List of Existing Problem Solving Programs

Robots

Some online Go servers, such as KGS, provide software opponents or robots to clients. Robot is derived from the Czech word robota meaning drudgery or slave labour which certainly describes the work needed to pummel double digit kyu players like me. PatG


Anti-Computer Strategy

It might be entertaining and educational to consider strategies for beating computers at Go.


Competitions

Below are competitions where Go playing programs can be tested.

International Competitions:

Regional Competitions:

  • European Championship (1987-2005,2008)
  • US Championship (1988-2000,2008)
  • USENIX (1984-1988)

Small Board Competitions:


History of Go Programs vs Human Professional Matches

Program vs Professional


History of Go Programs

  • [1989]: GNU Go 1.1 was posted to comp.sources.games March 13 1989
  • [1968]: Albert L. Zobrist wrote the first ever program which played complete Go games.
  • Ishi - old file format

Articles in Magazines


Links

(Smaller Go Bibliography, with comments)

  • [ext] http://gobase.org/go-7.html : Jan van der Steen's computer Go page has numerous links to famous programs and programmers as well as programming resources and articles.

Maillist(-Archives)

  • gnugo-list

Authors: Gounter, Chestnut


Notes:

[1] [ext] Will a Go Program Defeat the Meijin by Aratani Kazunari.

Bill: Note on the article. Typically, when computer go tournaments are held in Asia, some child will play a program an even or low handicap game, and the program does reasonably well. Claims about the strength of programs are made, based upon such results. Compare that with the claims about the strength of the tsumego solver, Go Tools, which has been thoroughly tested over the years.

Hyperpape: Note that Bill made this comment in 2006, before some recent advances.

Mschribr: I find this to be a problem with Sensei's Library. It is difficult to see when something was written. Something written 10 years ago would be correct. However, the same thing written yesterday would be wrong. Everything should be automatically date stamped.


This is a copy of the living page "Computer Go" at Sensei's Library.
(OC) 2012 the Authors, published under the OpenContent License V1.0.
[Welcome to Sensei's Library!]
StartingPoints
ReferenceSection
About