BGJ 132 Autumn 2003
Reviewer: Nick Wedd
Previous issues of this Journal have carried reviews of Go Professional, Go Professional II, and Go Professional III. These programs used the Go-playing engine written by Dr. Michael Reiss, which competed in interna- tional Computer Go tournaments under the name Go4++. Dr. Reiss was under contract to Oxford Softworks, which later became part of Purple Software, and received a salary from them and small royalties for each copy of his program that they sold. They embellished it with a decorative user interface and marketed it as Go Professional, and in other ways.
However Purple Software recently went into liquidation and was unable to meet its debts. Dr Reiss was able to retrieve full rights to the Go-playing engine he had written, and is now marketing it himself as Go++.
Go++ is a program for Windows. It plays Go, and plays it as well as any program; it does not do anything else. It is notoriously difficult to assess the strength of Go-playing programs, because humans soon become accustomed to their weaknesses, and learn to take advantage of them. Dr. Reiss claims that Go++ is about 7-kyu, and supports this claim by giving its results against some members of North London Go Club. These were, I assume, people who had never played against it before. I tested it myself by giving it nine handicap stones on a full board, with it set to its maximum playing strength, and I beat it comfortably (I am 2 kyu on the European Rating List). I had never played Go++ before, but I have played its predecessor Go4++ in its various incarnations, and therefore know something of its weaknesses. I also tested it by playing it on KGS, against a KGS 3-kyu. He gave it only two handicap stones, and only narrowly beat it.
It was interesting for me to see how he treated it with respect, in positions where I, knowing it weaknesses, would have bullied it mercilessly.
I will not describe Go++’s weaknesses here, as to do so would reduce its value to purchasers who are looking for a challenging program to play.
It has options of 19 by 19, 13 by 13, and 9 by 9 boards; You can get it to play as Black or as White, but not as both, and you can set the handicap and the komi. You can set it to any of five playing strengths, each about 1 grade apart. At its fastest setting, it takes less than a second a move. At its slowest setting, on my 1 GHz system, it averages under half a minute a move.
It only uses Japanese rules, with their fixed placement of handicap stones. The user interface is clean, simple, and easy to use. It does not use a clock – its human opponent can play as slowly as they like. It can record partly-played games for re-loading later, but it does so in ‘Ishi’ format, rather than the more widely used SGF format.
Go++ (formerly known as Go4++) has entered many international Computer Go Tournaments, and in the last five years has never been placed below third in one. Its main rivals have been Many Faces of Go, and GoeMate. The latest versions of these three programs all have good chances of beating one another, but it is only Go++ whose latest version is available to purchasers. The 2002 version of Many Faces of Go offers many more features than Go++, and costs US $89.95. GoeMate is not available to purchasers, but its ancestor program HandTalk is available at US $60, or US $50 for a DOS version. So Go++ is not only the strongest-playing program that is available to the public, it is the least expensive of the strong programs.
Go++ is available from the web site www.goplusplus.com