[wwarner]

It's great to read someone challenging the hype, but this analysis is a bit too negative. I'd point out that AlphaGo did something a little bit like the author's toddler learning to squeeze through the back of her chair. AG learned its new moves by playing against itself. I still wonder how they avoided over-fitting here. Imagine two twins who loved to play and learn about go, and played constantly against one another, and in the process discover new plays they both believe to be unbeatable. You'd expect some of their new moves to be weak when actually tested against other players. But AG's new moves really were strong.

Admittedly, games with simple scores are the only scenarios where this really kicks in. But then again, the stock market could fit this model.