Comment by funstuff007

> because it's often true that you can deliver better by focusing on a smaller number of things.

This is true / dogma in linear / non-linear regression world, but of no real import in deep learning or Bayesian methods.