On the Proper Classification of Error