Abstract: Our social interactions are increasingly mediated by digital platforms. On the one hand, these platforms leave behind detailed digital traces of interactions among millions of people allowing us to analyze social behaviors at an unprecedented scale and granularity. On the other hand, these systems are human-designed, malleable environments, making them ideal for experimentation and creating opportunities to study the nuanced causal effects of design decisions. In this talk, I will share research that demonstrates how we can use computational approaches and interventions to reduce political polarization and improve the health of conversations online. First, I will discuss the results of a series of randomized experiments that provide actionable insights into how the media organizations, the platforms, and the users can act to reduce political polarization online. Second, I will present a large-scale analysis of online conversations that investigates the relationship between conversation structure and toxic behaviors and informs design decisions on how to display conversations to reduce toxicity. Finally, I will discuss the broad methodological challenges and opportunities in estimating causal effects in social systems where people are connected and affect each others’ behaviors.
Bio: Martin Saveski is a Postdoctoral Scholar at Stanford University. He received his Ph.D. from the Massachusetts Institute of Technology in 2020. His research develops tools for analyzing large-scale social data, aiming to provide a better understanding of social structure and behaviors online while also impacting the design of digital social systems. His work typically appears in venues such as ICWSM, WWW, and KDD, and has included collaborations with researchers at Twitter, Facebook, and LinkedIn. His research has been awarded a best paper honorable mention at WWW ’18 and has been featured in popular media outlets, including The New York Times, NPR, and the MIT Tech Review.