Paper short abstract:

Drawing upon troubles experienced by Google Flu Trends, this paper explores how algorithmic tracking aims to turn a messy, connected world into a searchable space. It pays particular attention to tracking’s relationship to the otherness of viral life, including its social manifestations.

Paper long abstract:

In August 2015, Google shut down Flu Trends, its flagship predictive analytics system specialized in the anticipation of influenza outbreaks. Launched in 2008, Flu Trends used an algorithmic model to track influenza activity based on people's web searches. However, it had its troubles. Put roughly, Flu Trends was susceptible to picking up false signals and overestimate influenza activity. This was particularly the case in times of epidemics, when unanticipated changes in search behavior were triggered by heightened exposure to disease-related information. As a result, whole populations started Googling symptoms they did not have, provoking Flu Trends' overestimation. In times of epidemics, Flu Trends did not reveal the sort of "collective wisdomˮ often associated with online behavior, and algorithms. Drawing upon these troubles, this paper explores algorithmic tracking, and how it turns a messy, connected world into a searchable space. On the one hand, Flu Trends' failure may be seen as one in accounting for the incommensurable otherness of emergent viral life. On the other hand, a lesson to be drawn from Flu Trends' failure is that how it did not have direct access to viral life, and compensated for this, was as important as the lack of access itself. This paper does not seek to elaborate an argument to be applied to all algorithmic tracking systems, but rather aims to examine how Flu Trends speaks to wider relationships — of coconstitution, but also tension, and disjunction — between sensuous participation to collective life, and emergent forms of digital abstraction.