Comment by zardo

It would be interesting to see how strength and "human like play" scales with the depth of the search.

There's also a really interesting possibility in training policy networks with different attributes by using games from players with certain styles of play.