Does electronic publishing diminish the breadth of the scientist's scholarly attention span?

In this article in Science Magazine (sorry, subscription required), Dr. Evans argues that it does. By reviewing citation data covering over 50 million articles going back to 1945, he presents evidence that the more a journal makes it's backfiles available electronically, the less the author of a manuscript is likely to cite papers that are further in the past and/or less closely relevant to her research. There are several dissenting voices with regard to this analysis, but let's ponder what it might mean. Dr. Evans suggests one cause may be the relatively poor pre-electronic indexing which compelled readers of print articles to necessarily cast a broader net while browsing the literature. A less charitable hypothesis would be that in the face of an overwhelming number of relevant articles, a prospective author will find enough relevant articles to cite within the most recent published segment of the bibliome and/or the most high profile journals. When enough authors follow this trend (perhaps by following the the citation styles within the most cited articles), the result is a general narrowing of scope of attention of the scientific community within each discipline, both in subject matter and in time. A more optimistic view is that we have all become more agile in our use of electronic search and that in the face of a mountain of less relevant publications, we have become much better at winnowing out the chaff. Or it could be that the ordering of search results by year could be biasing investigators (sorry, subscription required) in ways that they are not aware of. Regardless of the cause, in an era in which we have all called for "evidence based medicine," it should give us pause if our view of evidence is overly myopic due to its electronic immediacy and organization.