The Limits of Citation Counts

I study the measurement of the influence of scientists based on bibliographic data. The main result is an axiomatic characterization of the family of citation-counting indices, a wide class of influence measures which includes the popular h-index. The result highlights several limits of these indices: they cannot be used to compare scientists across fields, nor can they account for indirect influence. I discuss how these limits can be overcome by using richer bibliographic information.