Popularity trumps compatibility: Why dating apps recommend so many bad matches

PITTSBURGH — If dating as an adult has you feeling like you’re back in high school, blame the algorithm. Researchers from Carnegie Mellon University and the University of Washington have found that the algorithms which control most online dating platforms these days prioritize both popularity and attractiveness over compatibility when making user recommendations.

The study authors evaluated data from over 240,000 users on a major online dating platform in Asia over the course of three months. This led to the determination that a user’s chance of being recommended by the platform’s algorithm increased by a notable margin as their average attractiveness score went up. This indicates the algorithm is biased toward users who are either more popular or considered more attractive on the platform.

Online dating has grown rapidly – especially during the COVID-19 pandemic,” notes study co-author Soo-Haeng Cho, IBM Professor of Operations Management and Strategy at Carnegie Mellon’s Tepper School of Business, in a media release. “Even though dating platforms allow users to connect with others, questions regarding fairness in their recommendation algorithms remain.”

People, of course, join online dating platforms to find matches – but the companies behind these platforms are after revenue above all else. The companies generate money through ads, subscriptions, and in-app purchases. So, researchers speculate platforms may seek to keep users engaged rather than maximizing their chances of finding the perfect person.

Online dating: Man deciding whether or not to swipe right
(© Kaspars Grinvalds – stock.adobe.com)

To study this topic, researchers built a new model for analyzing the incentives for platforms to recommend popular users more frequently when their goal is to maximize revenue or maximize matches. According to the model, they adopted an unbiased approach (meaning both popular and unpopular users had equal chances to be recommended to others) as their benchmark for fairness in order to compare popular and unpopular users’ matching probabilities.

Revealingly, the analysis showed unbiased recommendations usually result in significantly lower revenue for the dating platform, as well as fewer matches. This is likely because popular users help the platform generate more revenue by boosting users’ engagement (through more likes and messages sent). Moreover, popular users generate more matches as long as they do not become so selective that they are considered out of reach to most users.

Researchers also note popularity bias may be low when a dating platform is in its early stage of growth since a higher match rate can help build a platform’s initial reputation, generate buzz, and bring in new users. However, as these platforms grow and mature, the focus always shifts over to maximizing revenues, leading to more popularity bias.

“Our findings suggest that an online dating platform can increase revenue and users’ chances of finding dating partners simultaneously,” explains Musa Eren Celdir, who was a Ph.D. student at Carnegie Mellon’s Tepper School of Business while leading the study. “These platforms can use our results to understand user behavior and they can use our model to improve their recommendation systems.”

“Our work contributes to the research on online matching platforms by studying fairness and bias in recommendation systems and by building a new predictive model to estimate users’ decisions,” concludes Elina H. Hwang, Associate Professor of Information Systems at the University of Washington’s Foster School of Business, who also co-authored the study. “Although we focused on a specific dating platform, our model and analysis can be applied to other matching platforms, where the platform makes recommendations to its users and users have different characteristics.”

The researchers suggest online dating platforms be more transparent with their users regarding how their algorithms work. They add that additional research is warranted regarding how best to balance user satisfaction, revenue goals, and ethical algorithm design.

The study is published in the journal Manufacturing and Service Operations Management.

You might also be interested in: 

YouTube video