Understanding Political Communication and Political Communicators on Twitch

TL;DR

This paper examines political communication on Twitch by (i) identifying political streamers via the Twitch API and supervised learning to yield 574 actors, (ii) analyzing stream content through topic modeling of chat posts to uncover seven political topic categories and Twitch-specific emote usage, and (iii) mapping audience interaction with reference networks to show that a small set of users function as opinion leaders within many streams. The study demonstrates that Twitch's real-time chat and audience participation create distinctive patterns of political discourse and network structure comparable to other social media, while highlighting unique features such as emote-based signaling. By providing an open-end pipeline and a comprehensive empirical portrait of Twitch politics, the work advances understanding of how younger audiences encounter and interact with political content on real-time platforms. The findings have practical implications for political communication research, platform governance, and the study of micro-celebrity dynamics in live-streamed political content.

Abstract

As new technologies rapidly reshape patterns of political communication, platforms like Twitch are transforming how people consume political information. This entertainment-oriented live streaming platform allows us to observe the impact of technologies such as ``live-streaming'' and ``streaming-chat'' on political communication. Despite its entertainment focus, Twitch hosts a variety of political actors, including politicians and pundits. This study explores Twitch politics by addressing three main questions: 1) Who are the political Twitch streamers? 2) What content is covered in political streams? 3) How do audiences of political streams interact with each other? To identify political streamers, I leveraged the Twitch API and supervised machine-learning techniques, identifying 574 political streamers. I used topic modeling to analyze the content of political streams, revealing seven broad categories of political topics and a unique pattern of communication involving context-specific ``emotes.'' Additionally, I created user-reference networks to examine interaction patterns, finding that a small number of users dominate the communication network. This research contributes to our understanding of how new social media technologies influence political communication, particularly among younger audiences.

Figures (21)

• Figure 1: Screenshots of Twitch stream (Captured on 22/03/20)
• Figure 2: Venn Diagram of Twitch Streamer Data
• Figure 3: Frequency of streamings by political streamers
• Figure 4: Number of chat posts. The x-axis is log transformed while labels show the raw values.
• Figure 5: Number of unique users in a channel.The x-axis is log transformed while labels show the raw values.