The interesting point they make is that the error is non-random, and a significt part of why complex models are better than simple models is they're better at g

The interesting point they make is that the error is non-random, and a significt part of why complex models are better than simple models is they're better at guessing the wrong labels.