Recent research from Carnegie Mellon University and the University of Washington has shed light on a significant bias in the algorithms of online dating platforms. Published in the journal Manufacturing & Service Operations Management, the study shows that these algorithms tend to favor more popular and attractive users over those who are less popular or considered less attractive.
The research team analyzed data from over 240,000 users on a major Asian online dating platform over three months. They discovered a clear trend: the higher a user’s average attractiveness score, the more likely they were to be recommended by the platform’s algorithm.
Soo-Haeng Cho, IBM Professor of Operations Management and Strategy at Carnegie Mellon’s Tepper School of Business, co-authored the study. He pointed out the rapid growth of online dating, especially during the COVID-19 pandemic, and raised concerns about fairness in these platforms’ recommendation algorithms.
The study explored the balance between user satisfaction and revenue generation for dating platforms. Companies running these platforms need to make money through ads, subscriptions, and in-app purchases, which can sometimes conflict with the goal of helping users find their ideal match.
The researchers developed a model to analyze the incentives for platforms to recommend popular users more frequently, using an unbiased approach as a benchmark. Their findings suggest that while unbiased recommendations can lead to fewer matches and lower revenue, popular users boost platform engagement and revenue through likes and messages, contributing to more successful matches.
Interestingly, the study found that popularity bias tends to be lower in the early stages of a platform’s growth, where a high match rate can build reputation and attract new users. However, as platforms mature, there’s a shift towards maximizing revenues, leading to increased popularity bias.
Lead researcher Musa Eren Celdir, a Ph.D. student at Carnegie Mellon’s Tepper School of Business at the time, emphasized that platforms could use their model to improve their systems, balancing revenue generation with fair user chances of finding partners.
Elina H. Hwang, Associate Professor of Information Systems at the University of Washington’s Foster School of Business and co-author of the study, highlighted the broader applicability of their model and analysis to other matching platforms.
The researchers recommend greater transparency from online dating platforms about their algorithmic operations. They also call for further research into balancing user satisfaction, revenue goals, and ethical algorithm design in the digital dating arena.
