This Dating App Exposes the Monstrous Bias of Algorithms

Ben Berman thinks there’s a problem with the way we date. Not in real life—he’s happily engaged, thank you very much—but online. He’s watched too many friends joylessly swipe through apps, seeing the same profiles over and over, without any luck in finding love. The algorithms that power those apps seem to have problems, too, trapping users in a cage of their own preferences.

So Berman, a game designer in San Francisco, decided to build his own dating app. Sort of. Monster Match, created in collaboration with designer Miguel Perez and Mozilla, borrows the basic architecture of a dating app. You create a profile (from a cast of cute illustrated monsters), swipe to match with other monsters, and chat to set up dates.

But here’s the twist: As you swipe, the game reveals some of the more insidious consequences of dating app algorithms. The field of choice becomes narrow, and you wind up seeing the same monsters again and again.

Ben Berman

Monster Match is not really a dating app, but rather a game to show the problem with dating apps. I recently tried it, building a profile for a bewildered spider monstress, whose picture showed her posing in front of the Eiffel Tower. The auto-generated bio: “To get to know someone like me, you really have to listen to all five of my mouths.” (Try it for yourself here. I swiped on a few profiles, and then the game paused to show the matching algorithm at work.

The algorithm had already removed half of Monster Match profiles from my queue—on Tinder, that would be the equivalent of nearly 4 million profiles. It also updated that queue to reflect early “preferences,” using simple heuristics about what I did or didn’t like. Swipe left on a googley-eyed dragon? I’d be less likely to see dragons in the future.

Berman’s idea isn’t just to lift the hood on these kinds of recommendation engines. It’s to expose some of the fundamental issues with the way dating apps are built. Dating apps like Tinder, Hinge, and Bumble use “collaborative filtering,” which generates recommendations based on majority opinion. It’s similar to the way Netflix recommends what to watch: partly based on your personal preferences, and partly based on what’s popular with a wide user base. When you first log in, your recommendations are almost entirely dependent on what other users think. Over time, those algorithms reduce human choice and marginalize certain types of profiles. In Berman’s creation, if you swipe right on a zombie and left on a vampire, then a new user who also swipes yes on a zombie won’t see the vampire in their queue. The monsters, in all their colorful variety, demonstrate a harsh reality: dating app users get boxed into narrow assumptions and certain profiles are routinely excluded.

After swiping for a while, my arachnid avatar started to see this in practice on Monster Match. The characters includes both humanoid and creature monsters—vampires, ghouls, giant insects, demonic octopuses, and so on—but soon, there were no humanoid monsters in the queue. “In practice, algorithms reinforce bias by limiting what we can see,” Berman says.

When it comes to real humans on real dating apps, that algorithmic bias is well documented. OKCupid has found that, consistently, black women receive the fewest messages of any demographic on the platform. And a study from Cornell found that dating apps that let users filter matches by race, like OKCupid and The League, reinforce racial inequalities in the real world. Collaborative filtering works to generate recommendations, but those recommendations leave certain users at a disadvantage.

Beyond that, Berman says these algorithms simply don’t work for most people. He points to the rise of niche dating sites, like J-Date and Amo Latino, as proof that minority groups are left out by collaborative filtering. “I think software is a great way to meet someone,” Berman says, “but I think these existing dating apps have become narrowly focused on growth at the expense of users who would otherwise be successful. Well, what if it isn’t the user? What if it’s the design of the software that makes people feel like they’re unsuccessful?”

While Monster Match is just a game, Berman has a few ideas of how to improve the online and app-based dating experience. “A reset button that erases history with the app would go a long way,” he says. “Or an opt-out button that lets you turn off the recommendation algorithm so that it matches randomly.” He also likes the idea of modeling a dating app after games, with “quests” to go on with a potential date and achievements to unlock on those dates.