Redefining Fairness: Balancing Engagement and Well-Being in Social Media Algorithms

Social media algorithms prioritise engagement, often at the expense of user well-being, raising questions about fairness in predictive systems. This study investigates how to redefine fairness to balance engagement with well-being, aiming to mitigate harms like reduced productivity and attention spans. A mixed-methods approach was employed, integrating qualitative case studies of four major platforms (Facebook, Twitter/X, TikTok, YouTube), quantitative metrics from synthetic datasets and user surveys, and publicly available data from platform reports and peer-reviewed studies (2020–2025). Findings reveal significant engagement-driven harms, with 55% of Facebook users reporting distractions and TikTok’s average 90-minute daily usage linked to youth performance declines. Mitigation strategies, such as YouTube’s autoplay toggles, reduced session lengths by 10%, though trade-offs include revenue losses. Synthetic data modelled a 15% scrolling reduction with engagement caps. The proposed governance framework integrates technical interventions, ethical principles, and regulatory oversight (e.g., EU Digital Services Act), offering a scalable approach to fairer social media systems. These findings underscore the need for interdisciplinary solutions to align algorithmic design with user well-being.