Luca de Alfaro, a colleague of mine, is publishing a paper on how to automatically derive "reliability" information from the reputation of the authors of an article, this reputation being itself obtained from how much these authors are reverted or not.