Dating websites love data. Mostly this is because they have loads of it, and having loads of data appears to make it meaningful. The secret of compatibility is there for the taking so long as we have enough information, or so the thinking goes. eHarmony has taken this to a new level, employing its own in-house analytics team and amassing up to 25 terabytes of data about its customers’ behaviours and preferences (that’s equivalent to more than twice the information in the printed collection of the US Library of Congress).
But data is only useful if what you collect is relevant to the questions you’re trying to answer. Very often the information gathered by dating websites reveals not the secret to a great match, but how poorly their matching algorithms are working. In a well-reported experiment last year, OKCupid duped certain users into thinking they were a good match for each other when, according to its algorithm, they weren’t. It turned out that for an ill-matched couple who believed they were a near-perfect match, the chance of them striking up a conversation was 17%. But the chance of a well-matched couple striking up a conversation was not much better, just 20%.
Christian Rudder, co-founder of OKCupid, claimed this 3% improvement was evidence that “OKCupid definitely works”. Hardly! What it actually shows is that the company’s strategy, which involves pairing people based on what they think they want in a partner, is flawed. The fact is we don’t know what we want until we find it, even though we think we do. Plenty of people end up with someone they didn't see coming; hardy anyone marries their fantasy.
Maths or chemistry?
The world is abuzz with behavioural data, yet we’re not much closer to knowing the personal ingredients that make a good relationship tick. Perhaps this is because personal ingredients – the traits, beliefs, interests, talents and preferences that eHarmony, OKCupid and others value so highly – are not the full story. The psychologists John and Julie Gottman, who have studied thousands of couples over the past four decades, have found that relationship success depends a great deal on how two people communicate, how they manage the negative issues that crop up day to day, and how kind they are to each other.
At 21Pictures we don’t have too much data to play with yet, which is hardly surprising since we only launched a few weeks ago. We’re a little more sceptical about the magic of numbers than some of our competitors, though it would be fascinating to know, for example, which kinds of pictures prompt the most interactions, or how our users’ decision-making is affected by them having a limited choice of dates (previous studies suggest it should benefit them significantly). We’re more interested in creating an environment where people can make the best use of their intuitive social intelligence to make up their own minds (for more on how we do that, see here).
Patterns of love
Yet mathematics can be beautiful, and, when it demonstrates patterns in human behaviour, highly instructive. If you agree, you may care to take a look at The Mathematics of Love by the mathematician Hannah Fry, published last month by Simon & Schuster / TED. The book claims to pull back the covers, as it were, on the hidden patterns behind the rituals of love.
Here are three findings from Hannah's book you might find useful:
P(r) = (r-1)/n n∑i=r 1/(i-1)
Finally, you won’t do better than this gem of advice, hidden in the book’s prologue:
Being good on paper doesn’t mean anything in the long run. There’s no point in restricting your search to people who match everything on your checklist, because you’re just setting yourself an impossible challenge. Instead, pick a couple of things that are really important and then give people a chance. You might just be pleasantly surprised.
We couldn't agree more.