The Science of Online Dating

Is there a formula for calculating compatibility and finding love online? JV Chamary finds out whether meeting your perfect partner is a problem that can be solved with an algorithm

Published: February 14, 2011 at 2:00 pm

Our busy modern lives make it hard to meet new people, so more and more of us are turning to technology to find that special someone. Once the last resort for lonely hearts, online dating has become socially acceptable, even widespread – 4.7 million British people visited a dating site during 2008.

But popular as such sites may be, finding your ideal match online can take a lot of time, as traditional dating sites force users to read dozens or even hundreds of profiles. So some sites are helping people narrow the field by using an algorithm – a set of logical instructions for solving a problem – to find love online.

These algorithms take personal information, such as your interests, and push the data through a computer to calculate a couple’s degree of compatibility (or lack of).

Services like GenePartneroffer DNA tests to check for genetic compatibility, while OkCupid’s matching algorithm is powered by a user-generated personality test. eHarmony, the second-biggest dating site, carries out scientific studies to improve its algorithms.

But can love really be found by crunching numbers in a computer?

PSYCHOLOGY: Matching personalities for long-lasting relationships

Online matchmaking services like eHarmony and Chemistry.com are designed for those who are serious about dating. Unlike traditional sites, they don’t let people browse a database. Instead, members take a personality test upon signing up with the service, and their answers are entered into a matching algorithm to calculate compatibility with potential partners.

eHarmony has a reported 33 million members worldwide and even has its own lab, where psychologists question and observe couples and use the results to continuously refine the matching algorithm. “We see which parts of the compatibility model are the strongest, and how we can revise the model to make better matches,” says Dr Gian Gonzaga, director of research and development at eHarmony Labs in Santa Monica, California.

eHarmony was founded in 2000 by Dr Neil Clark Warren, a clinical psychologist and marriage counsellor who, after spending over 35 years trying to stop people from getting divorced, realised that the best way to build long-lasting relationships was to make sure people were compatible to begin with.

His researchers came to the conclusion that a person’s key character traits and values are the factors that best predict future relationship success.

Using this information, they created a personality test that was taken by thousands of married couples, who also completed a test to measure their relationship satisfaction. By cross-referencing the responses from these two questionnaires, Warren worked out which characteristics two people would need to have in common, in order for them to be compatible. Applied to single people, this became eHarmony’s first matching algorithm.

Today, the personality questionnaire has about 300 questions (the original research used 700), but the algorithm is much more complicated. If two people get 90 per cent compatibility, for instance, it doesn’t mean they answered nine out of 10 questions the same – whether you’re an extrovert or an introvert carries more weight than your shared hobbies, for example. The result is a shortlist of compatible people, and users are shown four or five profiles every day.

Love isn’t cheap (eHarmony members pay at least £10 a month), but it may be worth it: according to market-research firm Harris Interactive, eHarmony created 4.77 per cent of US marriages between January 2008 and June 2009, an average of 542 new marriages every day.

BIOLOGY: Testing for sexual chemistry through DNA dating

“People don’t want to bring science into love,” says Dr Tamara Brown of GenePartner, a DNA analysis service that helps people find their perfect match. “This is not really love, it’s sex.” For $99, GenePartner will test your DNA, sequencing key genes and examining those in your major histocompatibility complex (MHC).

These are genes which produce proteins that sit on the surface of cells so that the immune system can identify foreign invaders. These results can then be used either to check compatibility with your current partner, or to help you find a suitable mate by taking the data to one of several dating services (both on and offline) that offer a ‘DNA dating’ service.

We usually have nine MHC genes, each with many variants, so there are hundreds of possible combinations. When parents have different MHC genes, their offspring will have a wider repertoire of MHC genes, making them better able to recognise invaders. MHC genes were made famous in 1995 by Swiss scientist Claus Wederkind’s ‘sweaty T-shirt experiment’, in which women had to smell T-shirts worn by men and rate the owner’s attractiveness on the chemical odour alone. Wederkind showed that women subconsciously selected potential mates based on whether their genes would produce future children with stronger immune systems.

Extending this work, GenePartner researchers discovered that certain combinations of MHC genes are better for immunity than others, and the company has since included this consideration in its matching algorithm.

Brown believes that this DNA dating adds an additional filter – along with profiles and photos – for finding a compatible mate online. It also speeds up the process. “A filter for chemistry is something people do naturally,” she says. “The butterflies in the stomach and adrenaline in the blood will all still be there – it’ll just be there sooner.”

TECHNOLOGY: Let love find you with smartphones and social networks

StreetSpark is a smartphone app that helps you meet potential partners based on location. Using GPS and Wi-Fi chips to pinpoint where you are, it matches you with people nearby. Users download the app and register for the free service, create a profile with their lifestyle and interests, and provide the same information about their ideal partner. The matching algorithm then finds people whose interests overlap with yours.

“We’re like the host at a massive party,” says StreetSpark founder Anthony Erwin. “If you’re a good host, you introduce two people and point out the things they’ve got in common – it acts as an icebreaker.”

When a compatible person is nearby, StreetSpark sends an alert that pops up on the phone’s screen to let you know there’s a potential date just around the corner. Users can then either extinguish or ignite this ‘spark’. When two people ignite each other, they’re then able to see each other’s profile and details of how they match.

As the app only shows rough distances and both parties have to ignite a spark to show interest, this stops cyber-stalkers becoming real ones. Once a spark’s extinguished, a user disappears from the match results.

StreetSpark’s algorithm will soon incorporate ‘social matching’, which will aggregate data from social networks to paint a picture of who you are – without having to fill out a long profile. Location-sharing sites like Facebook and Foursquare can tell the world when you regularly ‘check-in’ at a particular pub, for example, and StreetSpark might match you with someone who also frequents the same local.

Once you find and ignite a spark, what happens next is up to you. Some people chat online first, while others might be more spontaneous and arrange to meet immediately for a coffee. Erwin won’t disclose the number of users, but he says the service generates about 700,000 sparks a month.

StreetSpark is currently available in the UK and US, centred on London and New York, and will roll out to other countries soon.

MATHS: Harness the power of stats to build a perfect profile

What do four Harvard-educated mathematicians know about dating? “Ours may not be the obvious background,” smiles Sam Yagan, CEO of OkCupid. “But we have a very analytical approach and we think we can build a system that lets you express your preferences and find the best people for you.”

OkCupid matches its 10 million users with potential partners based on their values and opinions, and a person’s profile includes answers to questions generated by the site’s users. Yagan says that, unlike the bigger matchmaking sites, their questionnaire isn’t a personality test. “Too many dating sites make it feel you’re visiting your shrink, and that’s not how dating works. Online dating should feel more like going to a bar than going to your psychologist. People have a good sense of the kind of person they want to date.”

The crowdsourced questions can be anything from ‘Have you ever committed a crime?’ to ‘Do you prefer Pepsi or Coca-Cola?’ Users don’t only answer questions for themselves, they also rate their importance, which determines how much weight they should carry in the algorithm. The best questions strike a balance between what people care about and what’s informative – few users want to date a criminal, but relatively few are actually criminals, whereas you can drink Coke and not mind if your partner prefers Pepsi. OkCupid promotes the most informative questions to the top of its profile quiz and the algorithm weights them more heavily when calculating compatibility.

The site also publishes OkTrends, a blog about data it collects. If mathematicians wrote an advice column on how to get a date, it would look like this. The site’s users can learn from the stats to build a better online profile. For example, it’s best not to smile in pictures: women who made a ‘flirty face’ with eye contact received more messages, while men should sport a serious expression and look away.

The blog also aims to remove any stigma about online dating. Yagan says, “Think about shopping: nobody is an online or offline shopper. It’s the same with dating: you shouldn’t be an online dater or an offline dater, you should be both.”

Follow Science Focus onTwitter,Facebook, Instagramand Flipboard