Swipe dating app intelligence

While you’re out mining dating apps for love this Valentine’s Day, these platforms are doing the same to your data. That’s because these apps and sites’ business models rely on the information you provide, to determine things like the matches they suggest and the ads they show you as you swipe.

But in a sea of strangers’ profile pictures, it can be hard to tell how, exactly, services like Tinder and OkCupid choose the suggested matches for you that they do. After all, the algorithms that power these platforms are proprietary, and companies have no interest in dishing out intimate details about how they work, neither to us nor their competitors.

Still, the information these companies have volunteered (and what they’ve disclosed thanks to data privacy laws like the European Union’s General Data Protection Regulation) can give us a good idea of how they generally work. As to whether these algorithms are actually better than the real world for finding love? That’s still up for debate, though that hasn’t stopped 30 percent of US adults from trying one of these platforms at least once in their lives.

What types of data do dating sites track, and who can get it?

First and foremost, whatever data you explicitly share with a dating app or site, the platform now has it. Depending on the platform you’re using, that can mean your gender, sexual orientation, location data, political affiliation, and religion. If you’re sharing photos or videos through a dating app, yes, the company has access to those. And they might be screening them with AI too; Bumble uses such tech to preemptively screen and block images that might be lewd.

But a dating platform can also have access to data about your activity on social media platforms if you connect them to your dating profile. As journalist Judith Duportail recounted in the Guardian, the dating app platform Tinder had maintained at least 800 pages worth of information on her that included info from her Facebook and Instagram accounts (including her “Likes” and the number of Facebook friends she had) and the text of conversations she had with every single one of her matches on the app. (You too can try requesting some ofyour Tinder dating app data, if you’re curious.)

So whatever service you’re using, be it an app-based platform like Hinge or a website-based service like Match.com, it likely has a bunch of your data. And these platforms work with third-party services that can also receive information about you.

For instance, a website data tracker can pick up the URLs you visit while you’re on a dating site and use that information to gather analytics or target ads at you, as we explained earlier this week. Your data could also be shared with third-party companies that your dating app might work with for the purpose of studying their site usage and to help target ads.

Some of these dating-sharing processes are questionable. For instance, back in 2018, Grindr was forced to admit that two companies it had paid to study its app usage were ultimately able to access information about its users’ HIV status (that practice has since been stopped). The Android versions of OkCupid and Tinder, which are both owned by the Match Group — which, yes, also owns Match.com — have reportedly shared users’ data, including information about their political views, ethnicities, and location, with a customer engagement service called Braze, according to research from consumer protection agency the Norwegian Consumer Council earlier this year. (Responding to this report, Match said that it does not use “sensitive personal information whatsoever for advertising purposes,” and that it uses third parties to “assist with technical operations and providing our overall services.”)

Though they share user data with third parties, dating companies generally claim that they’re not sellingusers’ personal data. But that doesn’t mean they can’t have security vulnerabilities. Here’s just one concerning example: A bug in the chat feature on the dating app Jack’d made it possible to view users’ images sent as “private” on the public internet, as reported by Ars Technica last year. And on Tinder, a security flaw caused by issues on both the Facebook platform and Tinder’s login system allowed researchers to take over accounts on the dating app with just a user’s phone number (the problem, which was raised in 2018, was quickly fixed).

Another privacy consideration: There’s a chance your private communications on these apps might be handed over to the government or law enforcement. Like a lot of other tech platforms, these sites’ privacy policies generally state that they can give your data when facing a legal request like a court order.

How do the algorithms use my data to suggest matches?

While we don’t know exactly how these different algorithms work, there are a few common themes: It’s likely that most dating apps out there use the information you give them to influence their matching algorithms. Also, who you’ve liked previously (and who has liked you) can shape your future suggested matches. And finally, while these services are often free, their add-on paid features can augment the algorithm’s default results.

Let’s take Tinder, one of the most widely used dating apps in the US. Its algorithms rely not only on information you share with the platform but also data about “your use of the service,” like your activity and location. In a blog post published last year, the company explained that “[each] time your profile is Liked or Noped” is also factored in when matching you with people. That’s similar to how other platforms, like OkCupid, describe their matching algorithms. But on Tinder, you can also buy extra “Super Likes,” which can make it more likely that you actually get a match.

You might be wondering whether there’s a secret score rating your prowess on Tinder. The company used to use a so-called “Elo” rating system, which changed your “score” as people with more right swipes increasingly swiped right on you, as Vox explained last year. While the company has said that’s no longer in use, the Match Group declined Recode’s other questions about its algorithms. (Also, neither Grindr nor Bumble responded to our request for comment by the time of publication.)

Hinge, which is also owned by the Match Group, works similarly: The platform considers who you like, skip, and match with as well as what you specify as your “preferences” and “dealbreakers” and “who you might exchange phone numbers with” to suggest people who could be compatible matches.

But, interestingly, the company also solicits feedback from users after their dates in order to improve the algorithm. And Hinge suggests a “Most Compatible” match (usually each day), with the help of a type of artificial intelligence called machine learning. Here’s how The Verge’s Ashley Carman explained the method behind that algorithm: “The company’s technology breaks people down based on who has liked them. It then tries to find patterns in those likes. If people like one person, then they might like another based on who other users also liked once they liked this specific person.”

It’s important to note that these platforms also consider preferences that you share with them directly, which can certainly influence your results. (Which factors you should be able to filter by — some platforms allow users to filter or exclude matches based on ethnicity, “body type,” and religious background — is a much-debated and complicated practice).

But even if you’re not explicitly sharing certain preferences with an app, these platforms can still amplify potentially problematic dating preferences.

Last year, a team supported by Mozilla designed a game called MonsterMatch that was meant to demonstrate how biases expressed by your initial swipes can ultimately impact the field of available matches, not only for you but for everyone else. The game’s website describes how this phenomenon, called “collaborative filtering,” works:

Collaborative filtering in dating means that the earliest and most numerous users of the app have outsize influence on the profiles later users see. Some early user says she likes (by swiping right on) some other active dating app user. Then that same early user says she doesn’t like (by swiping left on) a Jewish user’s profile, for whatever reason. As soon as some new person also swipes right on that active dating app user, the algorithm assumes the new person “also” dislikes the Jewish user’s profile, by the definition of collaborative filtering. So the new person never sees the Jewish profile.

If you want to see that happen in action, you can play the game here.

Will these apps actually help me find love?

A couple of respondents to our call-out (you, too, can join our Open Sourced Reporting Network) wanted to know why they weren’t having much luck on these apps. We’re not in a position to give individualized feedback, but it’s worth noting that the efficacy of dating apps isn’t a settled question, and they’ve been the subject of extensive debate.

One study last year found connecting online is now the most popular way to meet for US heterosexual couples, and Pew reports that 57 percent of people who used an online dating app found it to be at least a somewhat positive experience. But these apps can also expose people to online deception and catfishing, and Ohio State researchers suggest that people suffering from loneliness and social anxiety can end up having bad experiences using these platforms. Like so many tech innovations, dating apps have trade-offs, both good and bad.

Still, dating apps are certainly helpful tools for landing a first date, even if their long-term success isn’t clear. And hey, maybe you’ll get lucky.

Open Sourced is made possible by Omidyar Network. All Open Sourced content is editorially independent and produced by our journalists.