This is a farcical idea if the author actually is serious. The polynomial method is deep, dangerous overfitting. It’s the same reason why you don’t make a hand-

This is a farcical idea if the author actually is serious. The polynomial method is deep, dangerous overfitting. It’s the same reason why you don’t make a hand-made decision tree: you don’t know which features of the data are important or not. Presuming you do know, like basing the car’s steering on a polynomial fit of the three directional probes, will be a disaster in non-toy problems. This example shows why you _do_ need machine learning. The toy steering problem is a place for you to work out how the car can learn to drive with as little structure or assumption baked in as possible. Putting less a priori structure on it directly means you need models with higher capacity to figure it out. To put this all more succinctly, “never use machine learning when business logic would do” is a statement about how commonly needed machine learning is, not business logic.