We could play in hill-climbing fashion by making the move with the highest static value; but this produces weak play by failing to take the future into account.

A better way is to search to a depth bound, evaluate positions statically at the bound, and ``back up'' the values to the top using the minimax algorithm:

• The value of an OR node (Max's move) is the maximum of its successors' values. (Max can choose any move, hence will choose the best one.)
• The value of an AND node (Min's move) is the minimum of its successors' values. (Min will choose the move that is best for Min (worst for Max).)

Contents    Page-10    Prev    Next    Page+10    Index