The first game seemed like a game between two professionals. This second game took Ke Ji to places where he simply couldn't follow. When AlphaGo originally came out it was essentially running on a bank of supercomputers. This time it was running on a single node. Once a few grad students build a version that provides position equities for study this is going to revolutionize the game of Go in a way that Neural networks have done to Backgammon over the last 10 years. Absolutely amazing.