Mark Boon wrote:
I think what you describe is a general case of a well-known specific problem. And that is that sometimes the same position can be reached in different ways where one ends in a ko-capture and the other doesn't. This can result in incorrect results because the number of legal subsequent moves is different in the positions that otherwise look the same. The solution you described is probably formally the correct way to solve it, but as you already noticed, not very efficient.
The problem is known in the literature as the "Graph History Interaction" problem. See the 209 hits that Google produces on this keyword... regards, -John _______________________________________________ computer-go mailing list computer-go@xxxxxxxxxxxxxxxxx http://www.computer-go.org/mailman/listinfo/computer-go/