OKCupid Has Been Conducting Experiments on You for Funsies

A good rule of thumb for internet denizens: if an online service is offered to you for free, then you are the product. At the very least, you're a data point, and, therefore, a test subject.

Fortunately, not all tests involving internet users are as insidious and creepy sounding as the one Facebook conducted earlier this year, the one that bombarded users with sad stories in an attempt to manipulate their emotions. But they're all still kind of creepy. Especially because the very nature of the experiment mandates that users don't know they're emotional lab rats.

OKCupid, for example, has run a few tests on its users. Maybe they ran a test on you! Maybe you're being OKCupided right now.

Advertisement

In one of three experiments detailed in a worth-the-read blog post, the site removed all user pics for what it called a "Love Is Blind Day" last January. With no fap or superficial brag fodder to be found, traffic to the dating site cratered. But folks behind the scenes at OKC were heartened to find that while the site didn't show pictures, users who connected were more likely to respond to messages and more likely to have conversations that went deeper. Of course, this could be attributed mostly to the fact that the users who fled OKCupid on that random tuesday in January were probably the sort of people who were more interested in the superficial aspects of what OKCupid has to offer in the first place.

But! When pictures were restored to users at the end of the day, people involved in the aforementioned deep conversations abandoned them en masse. OKC blogger Christian Rudder compared it to the lights turning on in a bar at the end of the night combined with the fact that people are sort of assholes.

In the second experiment, analysts dug up data from back when the site had users review each other on two metrics — looks and personality. Predictably (and depressingly), the "personality" ratings folks doled out matched the corresponding "looks" ratings almost exactly. In other words, good looking people were assumed to also have great personalities, even if there was no text contained in the user's profile. Hot people were automatically assumed to be, uh, cool despite the fact that, as a general rule, incredibly hot people are the very worst.

OKCupid's third experiment is the ickiest, but also arguably the most interesting. In this case, data tinkerers wanted to find out how much the "power of suggestion" bore on people's compatibility. So they lied to a few people, and either inflated or deflated the "compatibility score" users see on profiles they visit. They found that a high compatibility score positively influenced the likelihood that users would respond to messages and that a low score would negatively influence users' likelihood of responding. However, chemistry between the disingenuously matched users was more likely to fizzle than between those with actual compatibility. So here's hoping that two people, spurred on by a falsely inflated 94% compatibility score, did not at some point in the recent past end up wasting years on each other, buoyed by the hope that a computer algorithm must have seen something in the other that they did not.

Conclusion: people are simultaneously both better and worse than you'd hope. Secondary conclusion: get off the internet and walk around outside.