The journal "Epidemiology" has a series of great articles on why we need to get rid of Thomson Scientific's Impact Factor. I've reported about this ridiculously influencial number before and how irrational, stupid and detrimental to science it is (1, 2, 3). Here the links to the great Epidemiology articles (via Coturnix):

• Epidemiologists (of All People) Should Question Journal Impact Factors
• Impact Factor: Good Reasons for Concern
• How Come Scientists Uncritically Adopt and Embody Thomson's Bibliographic Impact Factor?
• Rise and Fall of the Thomson Impact Factor
• The Impact Factor Follies

I think there's more than enough evidence that the Thomson Scientific impact factor is a pernicious invention that needs to be eradicated and replaced by a multivariate measure consisting of post-publication reviews, ratings, access statistics, citations, media coverage, link-counts, etc.

To say it with the authors at Epidemiology:

Having a collection of impact measures will only be sensibly feasible in a large, decentralized databank in which all peer-reviewed scientific primary literature is collected, cross-referenced and stored. Of course, the algorithms for any such metric need to be transparent. The data is all there, the technology is around. Now we only need to get the word out and bring the majority of scientists behind us. If scientists are as rational as they claim, they have no choice but to follow their rationality and get on with the program

If this is so, why do we still have journals? One of the authors has the answer: