> ... This is a very minimal reduction in code complexity. > Instead of the move generator having to consider all legal plays plus > their neighbours, > this rule allows it just to consider the former. Routines for adding > the set of neighbours > to a point set are already quite common in Go engines. > > I think adoption benefits more from having a clean ruleset.
I think, John's proposal is the better one here: allowing also throws on occupied cells. For the Leiden tournament with manual operation it will also be not a problem, because a participating bot who has the "free cell restriction" does not get known which cell was the target of the opponent's last move. It only gets the realized cell. (Of course, extrem optimizers might like to get both informations for their bot: both the intended cell and the realized one of the opponent's move. But in Leiden we should live with the simplified message: only the realized cell.) By the way: Do you, John, intend to participate in the Frisbee Go simulation in Leiden? Ingo. _______________________________________________ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go