I find it interesting how the author's first approach was to use a black box neural network instead of the evidently simpler beam search. As far as I'm aware, beam search was widely considered to be the simple method for game optimization just a decade ago.
Sure, new methods will always replace old methods, just like CNNs replaced SIFT for image processing. However, I feel that beam search is one of those elementary methods that you'd always want to check first, similar to A* and quicksort. Even though there are fancy neural networks for game optimization, pathfinding and sorting, it's easier to get started with elementary methods because they're simple and tractable.
Sure, new methods will always replace old methods, just like CNNs replaced SIFT for image processing. However, I feel that beam search is one of those elementary methods that you'd always want to check first, similar to A* and quicksort. Even though there are fancy neural networks for game optimization, pathfinding and sorting, it's easier to get started with elementary methods because they're simple and tractable.