According to a survey publishsed in Science [abstract][Slashdot] scientists are routinely pressured to include superfluous references in their papers in order to boost the Impact Factor of the journal publishing their paper. The Impact Factor is (roughly) a measure of the importance/influence of a journal, based on a two year average of how often its papers are cited. Careeers are made by publishing in high Impact Factor journals.
This sort of corruption (which I talk about a bit in Too Big to Know) might seem like an inevitable imprecision in how we gauge something as vague as “infuence” if alternatives were not becoming available. Services like Mendeley can provide real-time readouts of which articles are being read and commented on. Google likewise can see how often articles are being linked to. Facebook can see how articles are being passed around social networks, some of which are quite expert. It would of course be good to have measures not gated by commercial entities. In any case, institutions of knowledge are currently relying upon an instrument that was always too blunt and now known to be corrupt.