If you've ever gone a date with someone you met online and they rejected you by saying, "it's not you, it's me," we have good news: they weren't lying, and it might not have been your fault.
In fact, the authors of an August study have determined that dating app algorithms basically can't predict compatibility at all.
The authors of the Psychological Science study tested their own questionnaire-based attraction algorithm and discovered that their well-educated guesses couldn’t predict anything about what would happen after two strangers actually met for the first time.
To find out whether algorithms could predict mutual attraction, the researchers used 100 self-reported traits and partner preferences (for instance, "I enjoy binge-watching Game of Thrones") to predict a degree of variance in the choices of two strangers who then met in real life for four-minute speed dates. While the researchers didn’t specify whether the algorithm was based on a particular dating app, it sounds pretty similar to the one used by OKCupid, which uses a complex set of data to determine your compatibility with another user.
Using the statistical model, the researchers were able to predict fairly well the degree to which someone desired another person, and they were also able to somewhat accurately predict the degree to which someone else liked them. However, after researchers talked to both parties after the date ended, their responses were completely unpredictable, especially when it came to determining whether they were interested in pursuing a relationship with each other.
Dating apps and websites often boast about the efficacy of these secret sauce algorithms. But when you actually compare an algorithm’s predictions and speed daters' actual assessments of each other, it's clear that all the swiping in the world isn't all that effective. So basically, if you really want to use dating apps, don’t put too much stock in personalised questionnaires that promise to find your Mrs. or Mr. Right.
This article originally appeared on Men's Health US.