The last science paradigm has marked the beginning of the e-Science, or Science 2.0: we are immersed in an enormous amount of data and are equipped with the computational resources and infrastructure needed to make sense of these data. However, the process of scholarly communication and especially the one of research evaluation need to still shift the focus from the traditional research outputs (aka, the paper) to data. In this talk, I will make the case that the 21st century academic production can no longer be PDF-centric, but needs to look at data as first-class citizens of science, recognizing that the publishing system, as well as the assessment criteria, need to move towards dataset publication, citation, evaluation.