“It’s not wrong, but I’m quite disappointed”: Toward an Inclusive Algorithmic Experience for Content Creators with Disabilities

Document Title

“It’s not wrong, but I’m quite disappointed”: Toward an Inclusive Algorithmic Experience for Content Creators with Disabilities

Document Information

• Subject Area: Accessibility Design and Algorithmic Interaction
• Keywords: Algorithmic Experience, People with Disabilities, Content Creators, YouTube, Inclusive Design

Research Background and Issues

• Issues and Challenges: Although social media platforms provide opportunities for self-expression and sharing life experiences for people with disabilities, their algorithms may lack inclusive support for disability-related content, thereby limiting creators' ability to achieve their goals. Personalized distribution algorithms, in particular, may fail to effectively deliver disability-related content to a broader audience.
• Research Significance: Studying how inclusive algorithmic design can improve the experiences of creators with disabilities not only enhances public awareness of disabilities but also promotes the expression of societal diversity.
• Research Motivation: Current algorithms may carry biases or lack sensitivity, such as filtering out content containing the keyword "disability" or reducing the monetization potential of such videos. This situation could exacerbate feelings of frustration and distrust toward the platform among creators with disabilities.

Solutions

• Research Methodology: Conducted semi-structured interviews with 8 YouTube creators with disabilities from South Korea to explore their algorithmic experiences (AX) and coping strategies.
• Innovations:
  1. Investigating algorithmic interaction and content recommendation issues from the creators' perspective.
  2. Identifying the shortcomings of algorithms in achieving inclusivity.
  3. Proposing specific design recommendations for inclusive algorithms.
• Implementation Steps:
  1. Content Analysis: Detailed classification of participants' YouTube channels and uploaded video content, identifying key creator strategies.
  2. Interview Analysis: Guided interactions to extract creators' experiences and strategies regarding algorithmic distribution.
  3. Thematic Analysis: Using ATLAS.ti, coding interview data to summarize major challenges faced by creators and their overall perceptions of algorithms.

Research Findings

• Key Findings:
  1. Creation Goals: Participants aimed to use video content to raise public awareness about disabilities and build connections with their communities. They also sought to earn revenue from platform advertisements, but algorithmic limitations hindered their ability to attract a diverse audience.
  2. Challenges:
     • Identity Expression Dilemma: Creators grappled with whether to highlight or conceal their disability identity in their content (e.g., avoiding disability topics in some videos to reach a broader audience).
     • Content Filtering: Algorithms might filter out certain themes due to a lack of adaptability to disability contexts.
     • Algorithmic Opacity: Platforms provided no clear explanation for why content was restricted or distribution strategies were altered.
  3. Perceptions of Algorithms: Creators generally believed that YouTube's distribution algorithm did not discriminate against disability topics but also did not adequately support their unique needs.
• Strengths and Weaknesses:
  • Strengths: Compared to traditional media, video platforms significantly empower people with disabilities to voice their perspectives. Creators can directly showcase their diverse lives, rather than being portrayed as mere "objects of sympathy" in traditional media.
  • Weaknesses: Personalized recommendation algorithms tend to cater to existing audience interests, making it difficult for disability-related content to reach new audiences, thereby hindering broader public understanding of disability topics.
• Experiments and Solutions:
  1. Adjusting Tags and Video Titles: Creators experimented with adding non-disability-related keywords to video titles or tags to attract potential viewers.
  2. Diversifying Topics: Some creators attempted to diversify video topics (vlogs, beauty, gaming, etc.) to balance the needs of existing subscribers and new audiences.
• Limitations and Future Directions:
  • The current study is limited to South Korean YouTube creators, making it challenging to generalize the practices and perceptions of creators with disabilities worldwide.
  • The content focuses on the production and distribution stages but does not delve deeply into algorithmic performance in audience interactions (e.g., comment engagement).
  • Future research should examine a broader range of disabilities and the needs for AX inclusivity in different cultural contexts, expanding sample diversity.

Design Implications

• Support for Creators:
  1. Provide more transparent algorithmic feedback, clearly explaining decision-making processes in video review and distribution.
  2. Create more flexible distribution settings, allowing creators to define target audiences and highlight their value goals.
• Platform Improvements:
  1. Enhance the collection and training of disability-related data to improve the sensitivity of content review and distribution algorithms to disability contexts.
  2. Encourage viewers to annotate and evaluate the unique value of content through comments, thereby transmitting implicit feedback to creators.