I would start with implementing the Ikeda Area rules III ( http://gobase.org/studying/rules/ikeda/?sec=e_rules and more precise http://gobase.org/studying/rules/ikeda/?sec=e4010000 and following ) they approach to the complete Japanese rules (where Robert Jasiek is an expert) but i do think they are easier to implement in a program. (just area score but check for the first pass, the rules make the first pass valuable, so you do need to implement a routine that passes can be valued and investigated.
the only change i would make to them is where Ikeda has: Rule 7 Rule of scoring: A player's score is the number of that player's played stones plus the number of grid points in that player's territory. If the first pass was made by White, however, then 1/2 point is subtracted from Black's score and added to White's score. The winner is determined by comparing the players' scores. i would use If the first pass was made by White, however, then 1point is subtracted from Black's score(or 1 point is added to white's score. The winner is determined by comparing the players' scores. I am not sure how far these rules differ from the japanese rules, (Robert did once compare the japanese rules with the Ikeda Territory rules, but this is no territory rule, hope he will compare them sometime) On Fri, Feb 22, 2013 at 6:41 AM, Robert Jasiek <[email protected]> wrote: > Here are conceptual rules ideas: > http://home.snafu.de/jasiek/rules.html#ruletextsJapaneseStyleRules > http://home.snafu.de/jasiek/rules.html#RuleTextCommentaries > > In particular, the Simplified Japanese Rules can be a pragmatic start: > http://home.snafu.de/jasiek/sj.html > > Obviously, you would need to replace superko, exclude territory from sekis > and add Japanese style game end phases. What you get your program should be > able to handle: detecting independently alive by the possibility of creating > the player's two-eye-formation. > > Needless to say, pathological, rare cases can lead to differences from real > Japanese rules. Otherwise, Simplified Japanese Rules with the mentioned > modifications should be good enough for a program's internal move > prediction, if only it can construct representative playout sequences from > typical territories to two-eye-formations. > > -- > robert jasiek > > > _______________________________________________ > Computer-go mailing list > [email protected] > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go _______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
