Who Gets Written In? Gender, Identity, and Moderation in AO3’s Celebrity Fanfiction

Paper Title

Who Gets Written In? Gender, Identity, and Moderation in AO3’s Celebrity Fanfiction

Publication Info

• Topic area: Gender representation and feminist HCI in online fanfiction communities.
• Keywords: AO3, fanfiction, feminist HCI, Real Person Fiction (RPF), gender representation, content moderation, identity, digital storytelling, inclusivity, cultural norms.

Background and Problem

• Problem / challenge: Gender and racial imbalances persist in online fanfiction communities, even on platforms like AO3 that are designed with feminist principles. Representation of women and marginalized groups remains limited, and ethical concerns arise from controversial content such as underage sexual themes and gendered tropes.
• Significance: Understanding these dynamics is essential for fostering inclusivity, equity, and ethical engagement in digital creative spaces. It also provides insights into how feminist HCI principles are operationalized in participatory platforms.
• Motivation and related work: Prior research has explored feminist HCI principles and fanfiction as spaces for marginalized voices. However, there is limited quantitative analysis of how these principles manifest in practice, particularly in Real Person Fiction (RPF), which raises unique ethical and representational challenges.

Solution

• Proposed approach: A large-scale computational analysis of RPF on AO3, focusing on gender representation, occupational portrayals, thematic content, and moderation practices, grounded in feminist HCI principles.
• Novelty:
  1. Quantitative analysis of gender and racial representation in RPF, revealing imbalances and stereotypes.
  2. Examination of occupational clustering and thematic diversity in RPF content.
  3. Analysis of content moderation and its impact on reader engagement and ethical considerations.
  4. Comparison of real-world personas with their portrayals in fanfiction, highlighting transformative gender tropes.
• Procedure and key techniques:
  • Data collection of 1,027,523 RPF works from AO3, focusing on English-language content.
  • Entity extraction and annotation of celebrity names and occupations using ChatGPT and manual verification.
  • Analysis of co-appearance networks, gender relationships, and thematic content using tools like BERTopic and LIWC.
  • Examination of moderation tags and engagement metrics (hits and kudos) to assess reader interaction and ethical concerns.

Results

• Concrete findings:
  • 82.93% of characters are male, and 70.39% are white, reflecting significant gender and racial imbalances.
  • Male-on-male relationships dominate RPF, particularly in sports and actor communities, while female characters and relationships are underrepresented.
  • Content tagged with underage sexual themes receives high hits but low kudos, indicating curiosity-driven engagement but limited endorsement.
  • Gender-swapping and Omegaverse tropes are prevalent, with 90.10% of gender-swapping involving male-to-female transformations.
• Advantage over baselines:
  • Provides a comprehensive, data-driven analysis of RPF content, bridging gaps in feminist HCI research by quantitatively examining representation, moderation, and thematic diversity.
• Experiments / evaluation:
  • Co-appearance network analysis of 29,086 nodes and 49,614 edges to explore celebrity interactions.
  • Topic modeling using BERTopic to identify thematic patterns across occupational communities.
  • Moderation analysis using a multi-label classifier with 0.75 accuracy and F1 score.
• Limitations and future work:
  • Excludes multilingual dynamics and user interactions like comments.
  • Ethical concerns around underage sexual content and gendered tropes require further exploration.
  • Future research will investigate AI-generated content and its impact on fanfiction communities.

Summary

This study investigates gender representation, thematic diversity, and ethical challenges in Real Person Fanfiction (RPF) on AO3, a feminist HCI platform. The analysis reveals significant gender and racial imbalances, with male-on-male relationships and white characters dominating the narratives. Content moderation highlights ethical concerns, particularly around underage sexual themes. Despite its feminist ethos, AO3 reproduces some traditional stereotypes while fostering creative freedom and inclusivity. Future work will explore multilingual dynamics, user interactions, and the impact of AI-generated content on fanfiction communities.