Date of Award

2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Complex Systems and Data Science

First Advisor

James P. Bagrow

Second Advisor

Laurent Hébert-Dufresne

Abstract

Much of human socialization occurs online, and is mediated by telecommunicationsplatforms, particularly social media. These platforms both facilitate and restrict interaction in two ways: first, through the technical affordances they offer, such as conversation trees or direct messages or community self-moderation and voting; and second, through social policy, particularly regarding what content is permissible on a platform and how infractions are penalized. My work engages with platform influence over group social behavior through a series of case studies and through introducing purpose-built methodology.

I begin by examining the influence GitHub exhibits over open-source software de-velopment by contrasting the development practices of projects hosted on and off of the platform, showing how increased project discoverability and lower barriers to par- ticipation increase “drive-by” contributions from non-project-members, yet GitHub projects tend to have fewer active team members and shorter maintenance lifespans than their off-platform peers.

Next I study the impact of Reddit’s content policies regarding hate speech andharassment. My team examined behavioral changes after Reddit banned thousands of communities, illustrating how top power users and the broader community population change their activity and in-group vocabulary usage after such interventions. The heterogeneous results suggest that community-level bans are effective at disrupting only some kinds of communities, and are ineffective at curbing other hostile behavior. I pivot from platform influence on groups to how groups influence one another by introducing a metric for measuring group-level social centralization. This metric identifies how far a platform tends towards an oligarchy where the largest commu- nities are well-integrated with the platform and are involved with most users. This incorporates both the distribution of community sizes and their insularity. I describe a cumulative “disruption” metric, which removes communities largest to smallest and measures the impact on the remaining population. I demonstrate this metric on five real-world social and collaboration networks, and a variety of synthetic networks, showing how it distinguishes between different kinds of platforms.

Online social platforms exist in an ecosystem, where users and communities canmigrate between sites and technologies. Therefore, the affordances offered by and social policies instated on one platform can impact behavior on other platforms. In my final chapter, I propose a group linguistic fingerprint approach to identifying communities even as they migrate between platforms. Such a fingerprint would face a number of challenges, and this chapter is concerned with distinguishing between the vocabulary of group members and the vocabulary of people discussing a group.

Language

en

Number of Pages

203 p.

Share

COinS