[fryguy]

This approach is known as Monte Carlo Tree Search, which usually follows a weighted search by how good the move is until it finds a leaf in the explored tree. Then it makes a new set of leaves for its moves and scores the path by randomly making moves until the game is finished and propagating the result up the path to the root. The wikipedia article has a better explanation: http://en.wikipedia.org/wiki/Monte_Carlo_tree_search