The article doesn't exactly make a case against it. It even explicitly says it's useful! It just says making a newtype isn't as strong a guarantee as a proper t

The article doesn't exactly make a case against it. It even explicitly says it's useful! It just says making a newtype isn't as strong a guarantee as a proper type. Newtypes do work to assert certain checks have already been done and need not be repeated.