Blog Post

OkCupid lied to users about their love matches, calling it an “experiment”

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

On Monday, OkCupid co-founder Christian Rudder told the world – surprise! – just like Facebook, the company has been experimenting on its users. The gist of the post was, “You know how mad you were at Facebook? Well we did it too!”

In excruciating detail, Rudder laid out some of the behavioral tests the website has run on its unwitting daters. The first two are harmless: for example, the company removed all user images for a day to celebrate the launch of its blind dating app. It was a test run to see whether people were willing to take a chance on potentially ugly people. Spoiler alert: Nope. The second one tested how profile text impacted users’ judgments of each other. Spoiler: Very little. We’re shallow creatures who care more about photos.

The most concerning experiment Rudder shared was the last one. OkCupid lied to users for a period of time about their match compatibility. It told two users who weren’t a match that they were, and vice versa. Lying to users, particularly about something as sensitive as romantic matching, is ethical gray area.

When OkCupid users have a high match score with another user, they know what that means: This person probably has similar tastes, religious background, educational experience, or whatever other traits they’ve ranked as important to them after filling out a lengthy time-consuming profile. In other words, OkCupid potentially schlepped unwitting Republicans and Democrats off on a date with each other, which sounds like the premise for a reality show.

Rudder explained in the blog post, “In the back of our minds, there’s always been the possibility: maybe [our matching formula] works just because we tell people it does. Maybe people just like each other because they think they’re supposed to?” When Ok Cupid lied about a duo’s compatibility, the two were more likely to carry on an extended conversation with each other.

The confession is eerily similar to Facebook’s recent “sentiment” testing on users, where the company played with positive and negative newsfeed items to see whether people’s emotions could spread virally the way they do in the physical world.

When it comes to social media A/B testing, such companies might want to develop more rigorous ethical standards for such “experiments.”

There’s a difference between changing fonts and layouts — or, say, removing dating profile pictures — and toying with users’ emotions and relationships. It’s a fine line to straddle, but on one side is standard product experimentation and on the other side is a creepy futuristic novel waiting to be written about social media companies.

In any event, it’s a strange admission for OkCupid to make. Perhaps it’s an attempt to preempt the ire of the internet, in case some investigative reporter dug up the information? Of course, it might also just be a plug for Rudder’s new book Dataclysm, since it’s advertised at the bottom of the post.

8 Responses to “OkCupid lied to users about their love matches, calling it an “experiment””

  1. datinginsider

    It’s brilliant almost-free PR. Last time they wrote a few blog posts they were acquired for $90 million. Everything happens there for a reason, whether we understand it or not.

  2. Gwen E Mugliston

    I use OKC and while I enjoy reading the profiles, I have found profiles useful only for me to determine if the man is literate or cares enough to try to be literate and coherent. Some profiles are almost lyrical and some are sexual come-ons. I really enjoy the lyrical voices I hear.

    The photographs are often grossly disgusting. Some men go to a great deal of trouble to post interesting and decent photos. I truly appreciate that.

    The most interesting part of OKC are the Q & As. I I have answered some 3500 questions. If a man answers 100 to 300 of them thoughtfully with explanations for the more sketchy questions and we have a 90% “Match” and less than 20% “Enemy” we get along quite well. I have enjoyed meeting each of those men. I can almost guarantee if the man is seriously conservative politically I will have nothing to do with him.

    Unfortunately there are areas where OKC is sadly lacking. In my age group (70’s) it seems the “appropriate” men I actually meet are very often in unfinished relationships or perhaps should never have been in one in the first place. Unrequited loves, unfinished mourning, fury about divorces, basic dislike of mother who mistreated him transferring to other women, inability to commit and, last but not least, the serial monogomist. I don’t really care if they have any one of those problems, I would like to be warned of that before I spend money going to meet them

    I did meet one man I would have married. Weeks after I had spent lots of money visiting him it became clear he had a “Best Friend” of some 50 years whom he loved and she was not able to commit to him except as a friend. Now I know to ask, “Who is your best friend and how long have you known her?”. Apparently mate poaching/emotional infidelity are not well explored scientifically but I have learned to my regret to not involve myself with someone in that situation.

  3. Joseph Bielski

    oh please, get the F over it! Matching up Dem’s & Republicans… HOW DARE THEY!!! Not like these differences or “mismatches” wouldn’t come to light if the people actually start talking to each other. Do you think they’re going to propose and commit to something just based off the match score? And so what if they’re “different” – haven’t you ever heard “opposites attract”??? The bottom line is attraction is not rational, nor is it scientific, and quantifying it is probably an exercise in futility. While sites like OKC TRY to quantify it, it is just a GUESS, not a certainty. It will be completely different from one person to another, and suggesting people to each other who may have never tried contacting each other just because the ALGORITHM didn’t think they would be right for each other as an experiment is NOT ethically “concerning”. The people don’t HAVE to talk to each other, that is voluntary. They don’t HAVE to meet up, that is also voluntary. They also don’t HAVE to trust OKC’s numbers, or use the site at all! Using it is implicitly trusting their algorithm anyway, even if its garbage – who is to say how good it is now vs. when they did the experiment! Jesus, I’m shocked by how people are getting up in arms about this. It has to be the fringe of society who don’t actually understand human behavior or interaction. Get a life, and go complain about something that is REALLY concerning.

  4. snuggles

    I don’t know why OKC would do this. I really don’t. If the idea is to change things up and have a “mix with opposites” then allow users to opt in. With dating sites you tend to get tunnel vision and focus on numbers and don’t want to tweak search criteria too much otherwise the signal/noise ratio becomes out of control. I think there’d be a subset of the population who’d give it a try. But that note? Sounds like an awkward mea culpa before they got busted by a journalist, and managed to come out even more awkward than if they were busted.

    Oh, and will there be an over/under for the first class action lawsuit to be filed? I give it until August 23rd. Certainly if I was a premium customer and OKC decided to try this out.

    (Disclosure: I’ve used OKC in the past and it didn’t work out well. I blame my third eye.)