The catalyzing event for the article is a talk at NIPS drawing a comparison between ML and alchemy (i.e. experimentation by intuition and rules of thumb without an underlying theoretical foundation). The author presents this as a novel bombshell declaration (an "intellectual hand grenade" he calls it), but that is a sentiment that has been raised and broadly agreed with in every conversation along these lines I've ever had with actual practitioners. The other comparison I hear a lot is to bridge-building before formal structural analysis. People were perfectly capable of building functioning bridges, and some through practice were much better than others, but no one could tell you e.g. the maximum load for a given margin of safety and maintenance schedule.
Anyway, point is, the author is not wrong but this is far from some eventful revelation.