Ignorance in Social Networks: Discounting Delays and Shape Matters

Knowledge is the proper basis for action (Williamson, Timothy, Knowledge and its limits. Oxford University Press, Oxford, 2000). But disinformation causes ignorance, whether through error (false belief) or omission (agnosticism). We use philosophical simulations (Mayo-Wilson, Conor and Kevin Zollman, Synthese 199:3647-3673, 2021) to study how ignorance persists in networks of inquiring rational agents. Following (Zollman, Kevin J. S., Philosophy of Science 74:574-587, 2007), we simulate communities of agents who generate evidence, share it with their neighbours, and then update their beliefs. After recapping previous results, we report two novel findings. First, in variations on the mistrust models developed by O’Connor and Weatherall (O’Connor and Weatherall, European Journal for Philosophy of Science 8:855-875, 2018), discounting evidence received from neighbours delays convergence to the truth. Second, ignorance persists differentially in networks of different shapes, even when they have the same overall connectivity. These results shed light on the structural causes of ignorance that can be exploited by those engaged in disinformation campaigns; and they constrain the space of knowledge-conducive responses.