Intermediated Social Preferences: Altruism in an Algorithmic Era

Abstract

What are the consequences of intermediating moral responsibility through complex organizations or transactions? This chapter examines individual decision-making when choices are known to be obfuscated under randomization. It reports the results of a data entry experiment in an online labor market. Individuals enter data, grade another individual’s work, and decide to split a bonus. However, before they report their decision, they are randomized into settings with different degrees of intermediation. The key finding is that less generosity results when graders are told the split might be implemented by a new procurement algorithm. The asocial treatment results in less generosity relative to those whose decisions are averaged or randomly selected among a set of human graders. These findings relate to “the great transformation” whereby moral mentalities are shaped by modes of (a)social interaction.