Generally, it might be thought that there are two factors at work. The first is the impact of depersonification. In a number of experiments, it has been noted that subjects who cannot see or hear the person to whom they are reacting tend to inflict far more extreme punishments upon one another. The most famous study demonstrating this is Milgram (1963). In this, most subjects were willing to give electric shocks, which they had some grounds for believing could be dangerous and possibly lethal, to another person when that person was unseen and in another room. Most were not willing to do so when seated right next to the recipient of the shocks. Clearly, anything which increases contact between parties will reduce this tendency.

XREF: Connects to Milgram's obedience experiments and broader psychology of deindividuation, also relevant to online harassment/anonymity research. QUESTION: How does depersonification's effect interact with accountability structures, social norms, and digital interfaces today?

Diego Gambetta, Trust - Making and Breaking…, loc. 699