Millions of people all over the world are searching for their romantic partners online, using dating apps. To find a match, dating apps use algorithms to fish through hundreds of profiles so you don’t have to. But, like all algorithms, they aren’t perfect. Here to expose the pitfalls of online dating is MonsterMatch.

In MonsterMatch, you design a monster and their dating profile. Then the simulated dating app experience begins, in which you can swipe left or right and “message” with other monsters. Just like real dating apps, MonsterMatch uses an algorithm called collaborative filtering to decide which profiles to show. Collaborative filtering works by taking your data - a left or right swipe - and matching it to data from previous users. The app will  then show you another profile that was popular with people whose swipes agreed with yours. 

The problem with collaborative filtering is that it is heavily influenced by the first users. In the game MonsterMatch, this algorithm assumes that you like and dislike the same monsters as some of the early players. The monsters it shows you will start to be very similar to each other - a selection from the most popular monsters chosen by previous players.

In dating apps, the algorithm assumes you like and dislike the same people as previous users. Your first swipes can effectively pigeonhole you into a clique of users. If the clique says “No, we don’t like this profile,” then you will never be shown that profile. The clique filters which profiles you see, and therefore which people you date. This seems rather restrictive, given that dating is a tricky science. 

So the next time you open that dating app, consider how collaborative filtering influences whose profiles you view. To maximize your dating app success, notice when the profiles you are being shown lack variety - the collaborative filtering algorithm may have pigeonholed you. Try to reintroduce variety into your online dating search by experimenting with other available app features.  

You can play MonsterMatch here. In just a few minutes of swiping, you will discover the patterns, biases, and pitfalls of collaborative filtering for yourself.