On the Kullback information measure as a basis for information theory: Comments on a proposal by Hobson and Chang