The Measurement of Intellectual Influence: the Views of a Sceptic

In an extremely interesting paper, Palacios-Huerta and Volij (2004) [PV] introduce the axiomatic method to the problem of how to rank academic journals on the basis of their mutual citations. They characterize the invariant method as the only one satisfying a list of five appealing properties. In this note, I show an impossibility result, by identifying a sixth property that is violated by the invariant method. Further, I question the appeal of the PV axioms, when applied over larger domains of problems that take into account making distinctions among types of citations.