Why are researchers citing fewer papers than ever?

Via this article about James Evans's work in the Economist, referring to Electronic Publication and the Narrowing of Science and Scholarship, published yesterday in Science, I came across this (poorly produced!) National Science Foundation short video of Evans discussing why researchers cite fewer research papers despite having access to more. Blurb:

"Thanks to the Internet, scientists now have access to an astonishing number of research papers, scholarly journals and other papers. But according to new research conducted by James Evans, a professor sociology at the University of Chicago, researchers are actually citing fewer papers than ever, and they tend to cite newer papers that are also cited by many of their peers. In this interview, James discusses what got him interested in the topic, how he conducted his research and what he believes are some of the implications of this trend."

The key cause of the change - which Evans alludes to towards the end of the interview - is surely "search", and in particular the ranking technologies that search engines employ: once a paper on the Web has a lot of citations from articles also on the Web, then that paper's search ranking rises; if its search ranking is high it is more likely to be cited.  You wonder what account is being taken of this by proponents of metrics-based assessments of research excellence.