In the context of the feverish pace in which the social sciences are grappling with the implications for a turn toward ‘big data’, I suggest a different starting point: that big data are not necessarily social science data. In this somewhat speculative provocation, I argue that we should lean more on the notion that social media are phenomena and less on the notion that social media are evidence of phenomena. In doing so, I sketch four areas of potential criticality for an emerging big data studies.