In the digital age, finding love has transitioned from chance encounters to calculated algorithms. More than two in five couples first met online in 2017, whereas only one in five met through friends. While it was sort of weird and embarrassing to admit to someone you met your partner online in the early 2000s, it’s now quite common. In fact, many young people don’t even know how else they could meet new potential romantic partners.
While online dating sounds great on paper, in reality, it can be a struggle for many — unless you’re popular.
Researchers from Carnegie Mellon University and the University of Washington have recently highlighted a huge bias in these digital cupids. Their study reveals a preference for the more popular and attractive users on online dating platforms, raising questions about fairness in digital matchmaking. At a glance, this seems obvious since people like attractive people. But this isn’t the users being biased — this is the algorithm.
By analyzing over 240,000 user profiles on a major Asian dating platform, the team discovered a clear trend: higher average attractiveness scores increased the likelihood of a user being recommended by the platform’s algorithm.
“Online dating has grown rapidly — especially during the COVID-19 pandemic,” noted Soo-Haeng Cho, Professor at Carnegie Mellon’s Tepper School of Business, who co-authored the study.
“Even though dating platforms allow users to connect with others, questions regarding fairness in their recommendation algorithms remain.”
The business of online romance
The core of the dilemma lies in the dual objectives of these platforms. On one side, there’s the stated goal of helping users find meaningful connections. Look at Tinder or Bumble’s marketing: their messaging revolves around finding the right romantic partner for you. On the other hand, the platforms must generate revenue through ads, subscriptions, and in-app purchases. This dichotomy can lead to a conflict of interest, potentially prioritizing user engagement over the likelihood of finding a perfect match.
This is, of course, nothing new to people who’ve been swiping on dating apps for some time. The idea that the game is rigged is rather pervasive. But what if there was an app that didn’t use engagement algorithms to decide who should be more visible over others?
The researchers developed a model to explore the incentives for recommending popular users, contrasting revenue maximization with match maximization. Their findings indicate that a hypothetical dating app that offers unbiased recommendations, with equal visibility to all users, results in lower revenue and, rather surprisingly, fewer matches. Popular users, it seems, are crucial in driving engagement and, ironically, successful matches, provided they remain within reach of the average user.
Interestingly, the study suggests that popularity bias in dating platforms might fluctuate with the platform’s life cycle. In the early stages, high match rates are vital for building a reputation and attracting new users. As platforms mature, however, the emphasis might shift towards revenue generation, intensifying the popularity bias.
Tinder has generated more revenue every year since Match Group launched as a public company in 2015. Paid users are offered features and tools that allow them to increase their visibility to potential matches. It made $1.79 billion in 2022.
This may explain why people who used to get a decent amount of matches a few years ago are now shocked to find barely anyone is paying attention to them. It’s not like they got ugly overnight, but rather the algorithm or ‘the game’ has changed. It’s a rich get richer and poor get poorer kind of scenario, where dating app users are increasingly forced to pay to play.
Of course, dating was never ‘fair’ even before dating apps. Some people are just naturally very attractive, so they command a lot more attention. However, there’s something to be said about how dating apps are amplifying this attractiveness gap in unnatural ways.
The researchers believe a middle ground is possible, one that’s good for business but also for the average user.
“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 when he led the study.
“These platforms can use our results to understand user behavior and they can use our model to improve their recommendation systems.”
Elina Hwang, Associate Professor at the University of Washington, emphasizes the broader implications of their work. The same model could potentially be extended beyond dating apps in other fields where there’s a system of incentives and extensive user interactions.
“Our research not only sheds light on fairness and bias in online dating but also proposes a new model to predict user decisions,” she says.
Although the study focused on one specific platform from Asia, the insights and models developed are applicable across various online matching platforms. The team calls for greater transparency in how dating algorithms work and stresses the need for more research into balancing user satisfaction, revenue goals, and ethical algorithm design.
The findings appeared in the journal Manufacturing and Service Operations Management.
