Skip to content

Futures in Research

Social media, algorithms and the problem of non-representativeness

William Brady
Northwestern University

Many have observed that social media communication skews toward non-representativeness; that is, in many contexts a minority of users dominate conversation such that their voices are represented as a majority in the information environment. Many questions remain as to what best explains this ‘problem of non-representativeness'. This working group will critically examine the problem of non-representativeness with a focus on the role of algorithms (e.g., feed-ranking algorithms) in social media contexts. Specific questions include: (1) How do we define non-representativeness (i.e., non-representative relative to what), and is it a problem at all? (2) How do features of human psychology and algorithms interact to produce non-representative information environments? (3) Can humans be trained, or algorithms be re-designed to solve this issue? Our goal is to develop a better theoretical understanding of the problem as well as map out concrete paths for interdisciplinary collaborations to study it holistically.