The Algorithm for Love – How Tinder Has Improved its Matching Markets System
“You’ve got a match!”
From the outset, it is apparent that Tinder houses a matching market within its app, which supports over 57 million users as of 2021 (Stancheva).
A few years ago, Tinder released to the public that they used an Elo rating system for matches, the same method used to match chess players of similar skill levels (Tiffany). Through this system, every user would be given a score based on how many ‘right-swipes,’ or admirers, they had. Depending on the person, one right-swipe could be weighted more than another; for example, if a person with 500 right-swipes swiped right on you, this would boost your score more than a right-swipe from a person with just 10 right-swipes. Users would then be matched with other users with similar scores.
In March 2019, Tinder boasted a new technology that was an improvement from the previous algorithm. While they did not state clearly what this new technology was, it is likely that it resembles their competitor Hinge’s system, which uses a Gale-Shapley algorithm. The Gale-Shapley algorithm, or the stable matching problem, is often hailed for creating stable matches. In short, the algorithm consists of an interactive process in which one group of people proposes to people in another group, after which they can either be provisionally matched or rejected. If rejected, they can propose to the “next-best,” option, which may correspond [in Tinder’s case] to the person with the next highest Elo score. The algorithm ends when all individuals have been matched and there is no pair of people that mutually prefer each other over their current partner.
If it is true that Tinder has steered towards stable matching problems, then this sends the message that they value all of their users over a select few. Under just the ranking system, there is a possibility that those with low scores will reach a point where there are no individuals presented to them for them to match with, leading to them removing themselves from the app. However, the Gale-Shapley algorithm ensures more equity in the matching process, as low-scoring individuals can propose to high-scoring individuals, even if they get rejected. While it is unlikely that there will ever be a perfect matching, where no individual is left unmatched, due to the sheer volume of users on the app, Tinder’s changes minimize the amount of unmatched people and attempt to create the most connections between humans possible; in other words, they attempt to reach a maximum matching.
Sources:
https://techjury.net/blog/tinder-statistics/#gref
https://www.vox.com/2019/2/7/18210998/tinder-algorithm-swiping-tips-dating-app-science