Abstract

The spread of fake news online has far reaching implications for the lives of people offline. There is increasing pressure for content sharing platforms to intervene and mitigate the spread of fake news, but intervention spawns accusations of biased censorship. The tension between fair moderation and censorship highlights two related problems that arise in flagging online content as fake or legitimate: firstly, what kind of content counts as a problem such that it should be flagged, and secondly, is it practically and theoretically possible to gather and label instances of such content in an unbiased manner? In this paper, I argue that answering either question involves making value judgements that can generate user distrust toward fact checking efforts.