Chris McKinlay is one of many who successfully find love online — but the path he took wasn’t exactly traditional.
McKinlay was working on his mathematics thesis in June 2012 when he decided to check his OkCupid profile. Discouraged by what he called the “spammy” aspect of the site where men send the same message to many women, McKinlay figured there had to be a better way to get results. (Spoiler alert: Happy ending. But how McKinlay got there is completely unique, and where things get really interesting.)
OkCupid, founded by two Harvard math majors in 2004, relies on algorithms to make matches. Users answer a series of questions such as “How long will it take before you have sex?” and “Should gay marriage be legal?” and choose the importance of those topics to them. The site then aligns your answers with the answers of your potential matches and ranks a person’s potential to hit it off with you. However, the site can only do that if both respondents answer the same questions, of which there are thousands to choose from. So the chances of a potential match’s answering the same questions that you do, and answering those questions in a compatible way, are lower.
McKinlay, 35, was only getting high match percentages (in the 90th and higher percentiles) with fewer than 100 women out of thousands of potentials in the Los Angeles area, which didn’t seem right to him.
“I started thinking about it when I was in dissertation mode, so I was applying grad student mentality to everything back then,” McKinlay said.
Realizing he had the training to get better results from a site that functioned off the same skills he possessed as a mathematician, he put his research and programming abilities to use. Following a complicated series of coding, McKinlay essentially used robots to troll profiles, determining the questions most women were answering, and then answering those questions himself.
Using data from 22,000 women, McKinlay found seven distinct categories of women on the site, of which the five largest, with nicknames he assigned, are:
- Dog (19% of sample): Creative professionals looking to meet Mr. Right. Usually owned a dog. Median age: 33.
- Tattoos (18%): Two or more tattoos. Sometimes two or more jobs. Median age: 27.
- Mindful (17%): Often divorced. Sometimes with kids. Median age: 37.
- Green (16% ): Young and new to online dating. Median age: 25.
- God (11%): Strong religious and/or ethical views. Median age: 26.
Focusing on the two groups that appealed to him the most—dog and tattoo—McKinlay created two different profiles that played into these women’s different interests, and, most crucially, answered questions that they had answered and ranked important.
“I didn’t want to lie in any way. That felt wrong to me,” McKinlay said. “But I just feel O.K. with deciding whether or not to answer a question and how much quote unquote ‘importance’ to give it based on something more strategic.”
It’s now a month since he came up with the idea and had worked full-time on the project, and McKinlay sat back to see the results. And they were speedy. Hundreds of uninitiated messages poured in, followed by 88 dates in three months.
“All the math almost worked too well. I would sit down with someone and we’d totally hit it off. There almost always was some mutual attraction and it was a blast. It was almost addicting in and of itself to have flirty, interesting, light-hearted conversation with people seven times a week. So that was really successful,” McKinlay said. “The only unsuccessful part was once I got into it, I had a hard time saying ‘yes’ to one person in particular. I didn’t really know what I was looking for, so all I was able to do was say ‘no.’”
Until he met Christine Tien Wang, that is. “It was ineffable. Within 15 minutes, we just connected on a totally different level,” McKinlay said. “What was really great about this was saying ‘no, no, no’ to a bunch of people and getting kind of confused about what it was I was doing. And then all of a sudden meeting someone and having just ‘yes, yes, yes’ screaming from every part of my body.”
Two weeks after they met, McKinlay and Wang, 28, removed themselves from OkCupid and began exclusively dating, followed by a proposal a year later. Another match made in online algorithms.
Christian Rudder, co-founder and President of OkCupid, thought McKinlay’s approach was “pretty cool,” he said. “In general, whatever people need to do to make OkCupid work for them, we support. The point is to help people find dates–that’s our only goal. We’re totally happy for people to ‘hack’ us. As long as no one is being treated with disrespect or being tricked, which is doesn’t sound like he was doing, then we’re game for it.”
Now engaged, McKinlay thinks it was all worth it.
“On a basic level my experiment put me in front of a giant spotlight of female interest, at least according to the questions they answered. I figured out what they were looking for and stepped into the spotlight, not by misrepresenting myself, but by honestly answering the questions they thought were important,” McKinlay said. “But Christine was from the cluster of creative professionals, and this applied to a lot of women, but she wasn’t really exactly in the middle of that cluster of women and there are a lot of things about her that are completely different and don’t fit the stereotype. And that was true of a lot of women. Profiles aren’t people. So, all the fancy machine learning worked to game the system that’s based on machine learning, but it certainly doesn’t work to actually make a Google for love. There’s no computer in the world that’s going to replicate that experience.”
For those interested in applying McKinlay’s techniques to their online dating, you can buy his e-book on the takeaways from the experiment here.