A study from Carnegie Mellon reveals that online dating algorithms tend to favor more popular and attractive users. The researchers looked at data from over 240,000 users on a major Asian dating platform for three months.
They discovered that if someone is seen as more attractive by the platform, the algorithm is more likely to recommend them. This indicates a bias in the algorithm towards suggesting users who are already popular or considered attractive on the platform.
“Online dating has grown rapidly—especially during the COVID-19 pandemic,” the researchers explain. “Even though dating platforms allow users to connect with others, questions regarding fairness in their recommendation algorithms remain.”
Not in our interest
People join online dating platforms to find matches, but it’s important to remember that the companies running these platforms also need to make money. They do this through ads, subscriptions, and in-app purchases. Because of this, the platforms might prioritize keeping users active on their sites rather than just helping them find the perfect match.
The researchers created a model to figure out why platforms might recommend more popular users if their goal is to make more money or get more matches. They used an unbiased approach, where popular and unpopular users have an equal chance of being recommended, as a fairness benchmark to compare how likely popular and unpopular users are to find matches.
The results show that unbiased recommendations often lead to lower revenue for the dating platform and fewer matches. This is because popular users tend to boost engagement on the platform by getting more likes and messages, ultimately helping the platform make more money. Also, popular users contribute to more successful matches, as long as they don’t get too picky and exclude less popular users.
The study suggests that in the early stages of a platform’s growth, there might be less bias towards popular users. This is because having a high rate of successful matches can build a good reputation and attract new users. However, as the platform matures, its focus may shift towards making more money, leading to more bias towards popular users.
“Our findings suggest that an online dating platform can increase revenue and users’ chances of finding dating partners simultaneously,” the authors conclude. “These platforms can use our results to understand user behavior and they can use our model to improve their recommendation systems.”
