In the 60s everyone thought that expert systems were the future and if we could just figure out all the right rules we could do anything. In the 90s the pendulu

In the 60s everyone thought that expert systems were the future and if we could just figure out all the right rules we could do anything. In the 90s the pendulum swung all the way towards the kind of statistics/calc based learning that they are complaining about here. It already came back a little bit when people realized that you should always exploit domain specific knowledge as much as possible if you expect to get good results. It seems like a lot of whining to me without any real grounds or ideas about alternatives. Not even any real direction about research, just "we want groundbreaking and prolific results instead of incremental research!"