Upvotes? Downvotes? No Votes? Understanding the relationship between reaction mechanisms and political discourse on Reddit

Title of the Paper

Upvotes? Downvotes? No Votes? Understanding the relationship between reaction mechanisms and political discourse on Reddit

Paper Information

• Subject Area: Social Media Design and Political Discourse Analysis
• Keywords: Voting Mechanisms, Reaction Mechanisms, Platform Design, Political Discourse, Political Communication

Research Background and Problem

• Identified Problems or Challenges:
  • It remains unclear how the design of social media platforms influences the quality of political discourse on a global scale.
  • Online platforms are not merely digital representations of offline political activities but also directly impact the expression, organization, and implementation of political actions.
  • There is uncertainty regarding how the design features of social media platforms affect user interactions and the quality of discourse.
• Significance:
  • As social media becomes a critical tool for electronic political communication worldwide, understanding how design features shape discourse quality is essential for policymaking.
  • Understanding the relationship between different reaction mechanisms on social media and political discourse can provide a foundation for fostering more respectful and civil user interactions.
• Research Motivation and Related Work:
  • This study integrates decades of political discourse theories on deliberation, democratic civic engagement, and incendiary rhetoric to conduct an in-depth analysis of political rhetoric styles in social media discussions.
  • It explores the impact of social media platform design features on users' political discourse, complementing existing research on social media discourse.

Solution

• Proposed Methods or Solutions:
  • Using supervised learning techniques, confirmatory factor analysis, and a difference-in-differences (DID) model, the study analyzes political discussion data from 155 subreddits on Reddit between 2010 and 2018.
  • The data is categorized based on the core rhetorical elements of deliberative discourse, civic discourse, and incendiary discourse theories.
  • The study examines how reaction mechanisms on the platform (e.g., upvotes, downvotes, and no reaction mechanisms) influence the rhetorical characteristics of political discourse.
• Innovations:
  • Proposes a "minimal conceptualization" framework that operationalizes political discourse theories using core rhetorical elements for empirical research.
  • Develops a theory-based supervised classification model to quantify the core characteristics of different types of political discourse.
  • Combines traditional political theory with data-driven social media research, offering an empirical method to analyze the impact of platform design changes on discourse types.
• Implementation Steps and Key Techniques:
  1. Collect 1.55 billion user comments created in 155 political subreddits between 2010 and 2018.
  2. Annotate the rhetorical components of 4,500 comments and train a large language model, XLNet, to classify comments.
  3. Use "Confirmatory Factor Analysis" (CFA) to evaluate the alignment between theory and Reddit discourse.
  4. Apply the difference-in-differences (DID) method to analyze the causal relationship between different reaction mechanisms and discourse quality.

Research Findings

• Specific Findings:
  • The core rhetorical components of political discourse were found to be widely present in Reddit user discussions, particularly in deliberative, civic, and incendiary discourse.
  • Experiments showed that removing the downvote mechanism significantly increased the proportion of deliberative and civic discourse while reducing incendiary discourse.
  • When subreddits removed all reaction mechanisms, the proportion of incendiary discourse significantly increased, while deliberative and civic discourse decreased.
• Comparison with Existing Solutions:
  • The study confirms the nuanced relationship between social media design features and the quality of political discourse, providing important empirical evidence on how platform characteristics influence political communication.
  • It offers additional academic and policy-level insights to support the optimization of platform design.
• Limitations and Future Directions:
  • The causal relationships in the results require further experimental validation; differences in user demographics across social media platforms may limit the generalizability of the findings.
  • Future research is encouraged to explore how platform social dynamics and technical design features jointly influence discourse quality.
  • Calls for the inclusion of data from more platforms and diverse user groups to support broader policy recommendations.