A study of millions of journal articles finds that the availability of info online has paradoxically led to fewer studies being cited. Christopher Intagliata reports