A far more active computer Go competition is the CGOS server.
The Computer Go Ladder is an informal competition between computer Go programs for the principal purpose of having fun. It is hoped that the ladder will also stimulate work on programs and allow some measurement of the state of the art of current programs.
The Computer Go Ladder is a "handicap" ladder; the number of handicap stones that each participant can give to the next lower program is explicitly tracked. Whenever a program author feels that his program has been improved, he can issue a challenge to the program below to increase the number of handicap stones, or to the program above to decrease the number of handicap stones. New programs can join the ladder by challenging the program at the "bottom rung" (at no handicap). If the new program wins the challenge, it can challenge successively higher programs until it loses. It would then start playing handicap challenges to determine its exact placement. Many boring technical details of ladder operation have been omitted from this description.
Challenges are normally played on the Internet Go Server. Challenge results and scheduled challenges are announced via e-mail (as well as on this page). If you would like to receive such announcements, send mail to the ladder administrator and indicate whether you would like to receive match results only, or match results and upcoming challenge announcements. Challenges and results involving the top programs on each ladder are also announced on the computer go mailing list.
| Program | IGS name | Scoring | Author | E-mail address (replace ':' with '@') | |
|---|---|---|---|---|---|
| Ego | ego | Japanese | Bruce Wilcox | suewilcox:bigfoot.com | |
| Explorer | ex63 | Chinese | Martin Mueller | mmueller:cs.ualberta.ca | |
| Gnu Go | FSF | Japanese | FSF | gnugo:gnu.org | |
| Go++ | mreiss | Chinese | Michael Reiss | mick:reiss.demon.co.uk | |
| GoAhead | GoAhead | Japanese | Peter Woitke | woitke:physik.tu-berlin.de | |
| Gobble | bernd | Chinese | Bernd Bruegmann | bruegman:gravity.psu.edu | |
| Godot | LuckyGodot | Chinese | Jens Lieberum | jenslieberum:yahoo.de | |
| GoLife I | GoLifeI | Chinese | Henrik Rydberg | rydberg:fy.chalmers.se | |
| Gogo | pwgoldberg | Chinese | Paul Goldberg | pwg:dcs.warwick.ac.uk | |
| Golois | Golois | Chinese | Tristan Cazenave | cazenave:ai.univ-paris8.fr | |
| gottaGo | gottaGo | Japanese | Eric Pettersen | pett:cgl.ucsf.edu | |
| GoWind | gowind | Chinese | Ruhai Zhou | rzhou:ahpcc.unm.edu | |
| Indigo | parigo | Chinese | Bruno Bouzy | bouzy:math-info.univ-paris5.fr | |
| Jacqueline Go | N/A | Chinese | Patrick Hippenmeyer | info:informatrix.ch | |
| JaGo | Dude | Japanese | Fuming Wang | fuming:venus.radsci.uci.edu | |
| Many Faces of Go | ManyFaces | Chinese | David Fotland | fotland:smart-games.com | |
| NeuroGo II | NeuroGoII | Chinese | Markus Enzenberger | me:markus-enzenberger.de | |
| Orego | Orego | Chinese | Orego project | drake:lclark.edu | |
| Poka | Poka | Chinese | Howard Landman | howard:polyamory.org | |
| SmartGo | SmartGo | Chinese | Anders Kierulf | anders:smartgo.com | |
| Topgoer | topgoer | Chinese | Shaomin Feng | smfeng:yeah.net | |
| The Turtle | TheTurtle | Chinese | Joan Pons Semelis | semelis:terra.es | |
| Viking | viking | Chinese | Magnus Persson | magnus.persson:phmp.se |
| Program | Stones to next | Preferred time controls |
|---|---|---|
| Go++ | 0+ | 0' + 7' / 25 moves |
| Many Faces of Go | 1 | 10' + 6' / 25 moves |
| NeuroGo II | 0+ | 25' + 11' / 25 moves |
| Ego | 0+ | |
| Explorer | 0? | 30' + 6' / 25 moves |
| Indigo | 0+ | 0' + 10' / 25 moves |
| Golois | 0+ | 60' + 25' / 25 moves |
| Gnu Go | 0+ | |
| Gobble | 0+ | 30' + 25' / 25 moves |
| gottaGo | 0? | 0' + 19' / 25 moves |
| Poka | 0? | |
| Viking | 0+ | 10' + 12' / 25 moves |
| Orego | 0+ | 60' + 25' / 25 moves |
| GoLife I | 0? | 0' + 20' / 25 moves |
| Gogo | 0+ | 0' + 25' / 25 moves |
| Godot | 0? | 1' + 10' / 25 moves |
| Jacqueline Go | 0? | |
| The Turtle | N/A | 60' + 25' / 25 moves |
| Program | Stones to next | Preferred time controls |
|---|---|---|
| Go++ | 0+ | 0' + 7' / 25 moves |
| GoAhead | 0+ | 0' + 45' / 25 moves |
| Many Faces of Go | 0? | 20' + 13' / 25 moves |
| Gnu Go | 0+ | |
| Ego | 0+ | |
| Explorer | 0+ | 60' + 12' / 25 moves |
| NeuroGo II | 0+ | 25' + 25' / 25 moves |
| Indigo | 0+ | 0' + 25' / 25 moves |
| SmartGo | 0+ | 60' + 25' / 25 moves |
| Golois | 0? | 60' + 25' / 25 moves |
| Poka | 0? | |
| Viking | 0+ | 10' + 12' / 25 moves |
| GoWind | 0? | 10' + 10' / 25 moves |
| Topgoer | 0+ | 2' + 7' / 25 moves |
| gottaGo | 0+ | 0' + 19' / 25 moves |
| The Turtle | N/A | 60' + 25' / 25 moves |
You may be able to set up your Web browser to display the game records (which are in Smart Game Format) in an SGF viewer on your system, instead of as text. Here are instructions about how you would do that.
None.
Orego vs. GoLife I on 9x9.
The caliber of current program play can perhaps best be judged by "exhibition" games against human opponents. I currently have five such games available from this page (two involving Peter Woitke's GoAhead program, one involving David Fotland's Many Faces of Go, and two multi-program games), and Jean-loup Gailly's page has some games against the commercial program Handtalk . Also, Many Faces of Go plays frequently on NNGS and KGS against human opponents, so you could observe games there or retrieve recently played games and review them.
The GoAhead games have GoAhead as black against Herve Fournier (~ 8k) and Bill Newman (~1d). The games have many comments and are quite interesting.
The Many Faces of Go game has Many Faces as black with 29 handicap stones(!) against Martin Mueller (author of Explorer). The game was played at the '98 US Go Congress and $30 in bets rode on the outcome. This game unfortunately is uncommented. However, there is some commentary on the game in a Computer Go survey paper available on one of Martin Mueller's web pages. The game commentary is on pages 6 and 7 of that paper.
Peter Woitke provides three "multi-brain" games where a human plays against a "player" that is a human choosing candidate moves suggested by two or more computer programs. If the programs agree on a move then the human has no choice; otherwise he picks the move he thinks best from among the candidates.