What Is It About True Love That Dating Apps Can’t Get Quite Right?
If you’re interested in sharing our artwork, please use the following citation: Published by Syracuse University’s Online Master in Computer Science program.
Maybe you’ve been there.
You spend hours on your phone swiping left and right on Tinder or Bumble. You go on date after date only to discover that the initial spark from a few profile pictures and text-flirting dissipates quickly when you finally meet someone in person. You’re frustrated, but you continue swiping or scrolling, hoping technology will fix all your relationship problems.
Like many recent tech innovations, including Uber, Facebook and Mint, dating apps use algorithms to make life easier—in this case navigating the dating minefield to reach your soulmate. So why is it so much harder to design an algorithm to help you find love than one to find you a ride? The short answer: chemistry.
How Do Dating Apps Work?
Dating apps run on algorithms, which “are like a recipe,” according to Professor Sucheta Soundarajan, who teaches in Syracuse University’s Master of Computer Science program. In order to bake a cake, for example, you need to have certain ingredients like flour and butter, and you must combine the ingredients in a specific sequence. If you replace the flour with powdered sugar, you’re likely to end up with something inedible.
Like any good recipe, an algorithm requires specific ingredients applied in the right sequence. Proper algorithm design, according to Soundarajan, includes:
- Identifying the real problem that needs to be solved.
- Finding an efficient means of solving that problem.
- Using the right data in order to solve the problem.
For computer and data scientists who work on dating apps, that translates roughly into helping users find love, using mathematical reasoning to identify prospective matches, and asking users to input location, biographic, demographic and psychographic information.
3 Examples of Proper Algorithm Design
- Problem: Wanting to bake a delicious cake
- Solution: Combine the ingredients and cook in an oven
- Data: Flour, sugar, eggs, butter, and vanilla extract
2. Rideshare Apps
- Problem: Getting a safe, reliable and affordable ride
- Solution: GPS matches ride requesters with nearby drivers
- Data: Location and demand
3. Dating Apps
- Problem: Finding love
- Solution: Match users with one another
- Data: Location, interests and photos
For example, Tinder, used by more than 50 million people worldwide, helps users find love by presenting them with prospective matches in the form of profiles containing a few pictures and some personal information—age, profession and a short bio. Users can then swipe right if they’re interested in the prospective match, or left if not.
Tinder representatives didn’t respond to an email seeking comment, but the company’s vice president of technology, Dan Gould, told Buzzfeed External link that aside from age and gender preferences, distance (proximity between users) and recency (when a user last opened the app) are among the most crucial components in Tinder’s algorithm design. Fortune reports that Tinder’s algorithm also uses a technology called “Smart Photos” to identify which of a user’s photos will most likely result in a right swipe.
All dating apps are not created alike, but they are united by common purpose: helping users find love. That’s a tall order, something Cupid’s been attempting since the time of the Greeks and Romans, millennia before algorithms existed.
Leora Hoffman, president of Leora Hoffman Associates matchmaking company and a modern-day Cupid, said love is about “an attraction between people, and a comfort level.” She said a successful match depends on factors such as shared goals and values, intellectual capability, spiritual capacity and physical attributes.
To help clients find someone who checks all the boxes, Hoffman spends months getting to know her clients, often asking about everything from professional goals to family history. When she’s gathered enough information, Hoffman makes educated guesses about the kind of person her clients might click with. But ultimately, she conceded, even after gathering abundant information, she can’t always predict chemistry.
Like Hoffman, Callie Harris, a matchmaker for Three Day Rule, relies heavily on factors like personality type and body language, as well as her own intuition, when assisting clients in their search for love. Harris agreed that the most difficult part of matchmaking is trying to determine chemistry.
Dawn Maslar, a TEDx speaker who also goes by “the Love Biologist,” said the chemistry that rules our attraction to each other springs from our physiology. Studies show, for example, that humans are more attracted to partners with different immunity genes (so that hypothetical children would have stronger immune systems). Other studies suggest that women are often attracted to men who have high testosterone levels, while men are attracted to women with high copulin levels. So, if a date isn’t going so well, and things just aren’t clicking “you can’t take it personally,” said Maslar. “It’s subconscious.”
When sparks do fly, there are actually two different chemical reactions in the body. In a promising first-time encounter, the heart begins to beat faster, the liver releases glucose, and the brain makes a split-second “fight or flight” decision.
The second chemical reaction, which can occur quickly after the first, said Maslar, involves the senses. Each sense seeks out certain characteristics and votes “Yes” or “No.” For example, our eyes scan for evidence of good health, our ears listen for a certain acceptable vocal pitch, and our noses search for traces of certain aromas. If any of the senses vote “No,” the chances of a chemical match shrink.
Inherent Flaws in Design
So, is it possible for dating apps to replicate even a human matchmaker’s attempt at predicting chemistry? Sebastien Koubar believes it is. He’s the co-founder and CEO of Meetwo, a dating app launching in February that promises, in its slogan, to change the way people meet online: “By using chemistry. Not [an] algorithm.”
“A computer is a computer, and it will never understand how people behave, and think, and especially love,” said Koubar. “Love is something that even people can’t describe and understand, so how could a computer do it?”
Meetwo is attempting to solve that problem by asking people who sign up for an account a random series of yes-or-no questions about themselves designed to elicit real stakes in the responses. Questions will range from “Do you think women should always do the dishes?” to “Do you think the man should always pay the bill on a date?” to “Should children have a smartphone at an early age?”
“When you go on a first date, you’re putting your clothes on, you’re feeling excited,” said Koubar. “And when you first meet, it’s a mix of excitement and fear. This is the same feeling when using our app. When you’re passing the test of someone else, and you know you can’t fail or you’ll never get to meet this person, it’s the same emotions. We’re telling a dating app to reproduce a real-life dating situation.”
Sam Yagan, CEO of the Match Group, which oversees online dating sites including Match, Tinder and OKCupid, isn’t holding his breath until the day that technology can approximate dating chemistry.
“Picking the most likely and the least likely, that’s something that an algorithm can do really well,” Yagan told Boston’s NPR station WBUR. “Predicting chemistry, ‘who is the one,’ that’s something that we’re probably decades away from being able to do online.”
A dating app can learn certain user preferences over time, but unlike a human matchmaker, it doesn’t spend months getting to know its subjects. And neither algorithms nor matchmakers can predict pheromone levels or gauge someone’s reproductive health. So, why bother?
“At the end of the day, it’s just an app, and it can’t do the job of real dating,” said Meetwo’s Koubar. “But apps can start the spark for when two people meet.”
And that, as matchmaker Harris says, is “a cause that’s worthy.”
Share Our Story
We encourage you to share or cite our story. If interested, please use the following citation: Published by Syracuse University’s Online Master in Computer Science program.