In his post Matthew Cockerill lists three "practical things that Google could do to improve the communication of scientific research", one of which I'd like to highlight:
Generate alternative citation metrics for the scientific literature
Journal citation metrics from a single company currently dominate the research evaluation process. This is unhealthy in all sorts of ways. When (and whether) a journal gets tracked by Thomson Scientific is something of a lottery, and even once a journal is tracked, it typically takes 3 more years before an official impact factor is made available. Google Scholar could generate citation metrics more rapidly and more comprehensively than this, and in doing so would help level the playing field between established journals and innovative new journals which, while highly cited, do not yet have impact factors..
I think this is timely because I've been thinking about citation in the broader sense (e.g., citation networks for GenBank sequences, museum specimens, and taxonomic names). Part of this has come out of thinking about using citation networks to find good phylogenies, but also the concern raised by some taxonomists that their work is under cited (see Taxonomic inflation: two additional causes). The solution often suggested is to cite the publication of original scientific descriptions more often in the literature, but this seems to me to be a poor solution. It needlessly inflates the list of references cited (biologists are already pretty bad at this), and I think it tries to play a game that taxonomy will fail to win. Simply trying to increase the number of times taxonomic work is cited will do little in then long term unless that work is accessible, i.e., available online, with unique identifiers (such as DOIs).
Furthremore, I think we should extend this notion of "citation" to include GenBank records, specimens, etc. In this way we could compute the "impact factor" of, say, a museum specimen, and develop metrics of its true value. See also the commentary "Publishing perishing? Towards tomorrow's information architecture" (doi:10.1186/1471-2105-8-17).
