Journals with a high "immediacy" tend to get high SCI ratings, to the
detriment of journals which publish papers which have a long term impact.
It would be nice if the ISI would divide the citations to a journal into
say citations to work of the last 10 years and citations to work
published before the last ten years. The ratio of these two might give an
appreciation of the emphemal nature of some journals' publications.
In particular I am thinking here of journals such as the Journal of
Theoretical Biology, which in 1964 published work leading to the modern
"selfish gene" paradigm, an idea apparently ignored at the time.