A generation ago, most young men would have considered happy hour at the Chainsaw Sisters Saloon a target-rich environment. The drinks were cheap and the place was packed. Most importantly, while the odds of “getting lucky” were low, they were nonzero. So even if she said, “You’re more likely to get struck by lightning than to go home with me,” he could answer, “Awesome! You’re saying I have a chance to go home with you?”
Millennials empirically know that bar crawling is for recreation – not for archaic, time-wasting, low-percentage mating rituals. If you want to meet someone, there are any number of big dating sites and apps available.
The Space Is Crowded
For romance, the major big dating players include Match.com, Chemistry.com and eHarmony – all promise long-lasting relationships. Niche sites like OurTime.com (for serious daters over 50), ChristianMingle.com (for Christians seeking singles with similar values), BlackPeopleMeet.com (for African American people to connect) and JDate.com (for Jewish singles) offer eponymous consumer value propositions.
In the mobile first arena, Tinder is the undisputed leader. No other app comes close to its market share, but there are plenty of other offerings. Hinge, OKCupid, and Zoosk are all players, and niche apps such as JSwipe (Jewish Tinder), Happn (location-based dating), Bumble (women have to be the ones who initiate the conversation) and The League (“curated” members have to be selected to join) have all found an audience.
The Numbers Are Compelling
According to an infographic entitled Big Data Seeks Online Love by the Berkeley School of Information, one in 10 Americans has used a dating site or mobile app, and 23 percent have met a spouse or long-term partner through these sites. In fact, 11 percent of American couples who have been together for 10 years or less met online.
The matching has improved. In 2005, 47% of people agreed that online dating allows you to find a better match; in 2013, that number went up to 53%. Is online dating a good way to meet people? Forty-four percent said yes in 2005, while 59% said yes in 2013.
Does Data Science + Big Data = Love?
Not according to experts cited in a recent San Francisco Chronicle article entitled Critics Challenge The ‘Science’ Behind Online Dating by Kristen V. Brown. She reports: “‘There is no evidence that dating sites do anything much more than increase the pool of potential partners,’ said Eli Finkel, a psychologist at Northwestern University who studies relationships. Or, as Stanford sociologist Michael Rosenfeld put it, ‘The algorithms for matching at dating sites are mostly smoke and mirrors.’”
Thinking Things Are Not Reliably Predictable
I have heard many different data scientists describe their strategic approaches to big dating algorithms. Thod Nguyen, CTO of eHarmony, describes its approach as a compatibility matching system consisting of a “very sophisticated three tier process.” A compatibility matching model identifies potential matches based on a proprietary 29-dimensional array. eHarmony’s affinity matching model predicts the probability of communication between two people, and finally, the match distribution model helps ensure that eHarmony delivers “the right matches to the right user at the right time and to deliver as many matches as possible across our entire active network.”
While this sounds interesting and may actually work as a matching strategy, the inherent problem is bi-directionality. When Amazon recommends a camera for you, the camera has no say in the matter. This is not true with human beings. Someone may be your perfect match, but there are any number of reasons the feeling might not be mutual.
That said, there is an axiom working in favor of all big dating algorithms: boys and girls are genetically predisposed to be attracted to one another and attempt to reproduce (otherwise none of us would be here).
Some Number of Nos Equal a Yes
The problem at the Chainsaw Sisters Saloon was not the very low odds; it was the extended investment of time required to achieve success. If you had a one-in-100 chance (1/100) to take someone home, you’d need 100 trips to the bar, on average, to accomplish your mission. This adds a bit of a twist to big data’s role in big dating. Sure, you can answer the 150 questions on Match.com and hope to be matched to your soul mate, or you can just play the numbers.
Data Science + Big Data = Exponential Increase in Deal Flow
Tinder saves time. It offers an exponential increase in opportunities over bar crawling. Even so, motivated programmers have created dozens of Tinderbots to increase their efficacy. Some Tinderbots use game theory and others use brute force, but my favorite uses data science to achieve its goal.
Eigenface example Source: Justin Long’s blog post, Automating Tinder with Eigenfaces
Using Data Science to Date the Perfect Model
On his blog, crockpotveggies.com, Justin Long provides the code for “Tinderbox,” a Tinderbot that taps into Tinder’s APIs and uses Eigenfaces to build an invariant model of the face you’re most likely to “swipe right.” You can think of it as your “perfect model.” A model with all of the characteristics you love most. He also uses Stanford NLP to help the bot analyze the sentiment of chat responses. After about 60 manual swipes, the program has learned enough to start making choices for you – at a speed you could not possibly replicate. Read his blog post – it will make you smile. If you have the time (and inclination), go ahead and build a Tinderbox for yourself. You may never visit the Chainsaw Sisters Saloon again.