There are increasing examples of how technology and social media influence collective behavior. For instance, the Newtown, Connecticut tragedy inspired an outpouring of gifts from over half a million people, overwhelming the small town. One explanation for this gift overload could be that social algorithms on Facebook amplified public compassion, creating a feedback loop that smothered the grieving community. Similar feedback loops potentially drive escalations of inter-group conflict, polarization, and harassment, as well as generosity in online networks. For good and ill, social scientists and computer scientists have struggled to predict and manage these cycles of human and algorithm behavior.

J. Nathan Matias and his team at Citizens and Technology (CAT) Lab at Cornell University have been exploring ways to develop the basic scientific knowledge needed to forecast and guide human-algorithm behavior for the common good. This project aims to monitor algorithm behavior, forecast influxes of collective human-algorithm behavior, and test interventions to direct that behavior in beneficial ways.

To achieve this, the team will create a scientific roadmap across computer science and the social sciences that reviews and summarizes the state of knowledge on human-algorithm behavior and outlines directions for future research. 

They’ll also forecast online harassment, and prosocial responses to it based on monitoring adaptive algorithms, such as “recommender systems” that communities hypothesize drive harassment.

In collaboration with community leaders, prosocial community approaches to managing the effects of adaptive algorithms on communities on Reddit, Wikipedia, and other platforms will be developed.

The team also plans to engage leading social and computer scientists to develop shared theories and breakthrough ideas in the study of human-algorithm behavior. They’ll collaborate with scientists and community leaders to envision and fundraise research infrastructures related to human-algorithm behavior, and hold a "Citizens Agenda" convene to develop study ideas that will inform future research on this topic.