good point. this algorithm doesn't take account of the fact that one can make a threat (x's 147) by parrying a threat (o's 357). nevertheless i'm still certain that the components the algorithm is built from are the same components i use myself to play tictactoe, and i'm still certain that my own tictactoe playing is optimal. that's my motivation for doing it this way: i'm trying to formalize the algorithm i use myself. besides, exhaustive minimax search seems like cheating to me, since it doesn't generalize to less-trivial games. anyway, thanks for pointing out the error
peace stm -- https://mail.python.org/mailman/listinfo/python-list
