gnugo homepage seems quite spectacularly inaccurate. http://www.gnu.org/software/gnugo/gnugo.html says: This is a tough question. Against humans, its results vary a lot depending on how well the human knows its weaknesses, and on other factors (time settings, hardware, etc.), but most robots running GNU Go on one of the go servers achieve a rank of 8k-9k. Compared to other go programs, it may be 1-2 stones weaker than the top commercial go programs.
Wikipedia [http://en.wikipedia.org/wiki/GNU_Go]: At this level of performance, GnuGo was between six and seven stones weaker than the top commercial programs on good hardware as of early 2009 Sensei's Library [http://senseis.xmp.net/?Zen] Zen is a strong Go engine by an individual Japanese programmer Yamato (real name not disclosed yet). A bot running this engine played on KGS in April and May 2009 as Zen19 and was the first to hold a KGS 2d (and 1d) rating for more than 20 rated games in a row. That doesn't seem like anywhere near 1-2 stones weaker than top commercial go programs, more like 8-10 stones weaker. _______________________________________________ gnugo-devel mailing list gnugo-devel@gnu.org http://lists.gnu.org/mailman/listinfo/gnugo-devel
