Dynamics of the Early Life of Online Content

The growing presence and impact of social media has changed the way people communicate with one another, interact with news, and view the world. Therefore, online social media presents a unique opportunity to examine human behavior in a collaborative environment. The rapid adoption and spread of various social media platforms allows for social interaction beyond the previous geographical limits, connecting people from different countries, time zones, and backgrounds. Studying information adoption and the factors influencing content popularity provides insight into the motivations and decisions of individual users. Also of interest is the role of communities in these environments -- not only as agents of diffusion, but as diffusers of content. These online communities serve to organize both social media content, and the users that produce and consume it, leading to unique community identities.

In this dissertation I present several studies undertaken to understand and describe user behavior within online social media, with a specific focus on the dynamics of the early life of online content, and the role of user collaboration and communities in these processes. The central focus of this work is analysis of Reddit, a popular social media platform for anonymous users to create and interact with content organized into communities called subreddits. With easily accessible data, Reddit allows for comprehensive large-scale modelling and natural experiments. We find that leveraging community history allows for the prediction of comment threads on Reddit without observed activity, and that text summarization highlights the unique language found in each subreddit. Reddit communities also interact via direct subreddit links, inducing changes to community activity levels and spurring the creation of new communities. We also investigate software library adoption on GitHub, which reveals the cognitive cost of knowledge acquisition within a collaborative environment. Taken together, the work presented here underscores the complexity and nuance of user behavior and interaction within online social media communities, along with the challenges associated with predicting and understanding early content dynamics.