Facebook’s Influence on Italy: Tracking Exposed Experiment

In an experiment conducted by Fabio Chiusi and Claudio Agosti during the 2018 election season and set out in detail in their report for Tactical Tech, the duo sought to investigate the Facebook algorithm that powers users’ news feed and the algorithm’s treatment of political content. One of the experiment’s goals was to observe how perception of reality is algorithmically shaped in the context of an electoral campaign. In order to do this, Chiusi and Agosti created six bot accounts on Facebook which mimicked the behaviour of users with different ideologies. Their experiment revealed that:

  • The far-right bot was both exposed to a lack of content variety compared to its ideological counterparts and the highest number of repeat posts and content. At the same time, it was exposed to a disproportionate amount of visual based content.
  • Center-left content was repeated across all bot News Feeds more than any other ideological orientation and received a better exposure ratio compared to all others as well. Yet it only came third when absolute numbers of content posting were compared.
  • When different news sources are compared, similar disparities in post number vs. post exposure exist in News Feed. For example, progressive leaning La Repubblica’s posts surfaced more frequently in all bot News Feeds than conservative Il Giornale, despite the latter’s higher number of posts.

While the duo concluded that it is impossible to understand what accounts for these disparities without more insight into Facebook’s data, they also claimed that it can be convincingly argued that algorithmically-steered sources of information impact users’ perceptions of political reality.