Why online dating goes wrong
In autumn 1965, Peter Lake filled out a survey that changed the course of his life.
Lake, who now works in real estate in Marblehead Massachusetts, signed up to Operation Match, a computer dating service started by three Harvard undergraduates and one Cornell University dropout.
"I was going to Boston University and it was such a deal, you couldn't turn it down," Lake says.
"For three dollars they would give you three matches at least. They would give you as many as you got, but at least three - or they would give you your money back."
He mailed the survey back to Operation Match and was matched with a dozen women. With the exception of one woman who lived too far away in Maine, he met all of the women, marrying the eleventh.
"The last one I met was a student at Wellesley College. She and I talked on the phone a few times and then we had some coffee and I just fell in love her then. Boom! We started dating immediately."
Fast forward almost 50 years and computer dating has graduated from paper-based surveys directed at horny students to a become multi-billion dollar global industry.
US market research firm Marketdata estimates that the online dating industry is worth around US$4 billion.
While many sites allow people to freely roam through lists of potential mates, niche services promise to match you with that special someone.
The punch card technology that united Peter Lake with his future wife has been replaced by patented online personality tests devised by psychologists and anthropologists.
eHarmony, for example, uses a "scientific Relationship Questionnaire" of 400-plus - 100 questions if you're using the mobile app - to match clients with the man or woman of their dreams.
eHarmony's resident Dr Love is Dr Gian Gonzaga. Gonzaga, whose formal job title is Senior Director of Research & Development at eHarmony Labs, was lured to the position after completing his PhD at the University of California on how love promotes commitment in long-term relationships.
He and his team undertake continuous research of couples in committed, long-term relationships to find out the shared personality characteristics and values that best predict successful relationships.
"What we're doing, in essence, is saying we have some ideas about what could predict relationship satisfaction, but we're going to actually poll couples who have been together and figure it out, what is it that empirically predicts relationship satisfaction," says Gonzaga.
Not everyone is sold on the science. In a recent issue of the journal Psychological Science in the Public Interest, a team of five psychologists pour cold water on the scientific claims of eHarmony and similar sites like Match.com, noting that none have ever subjected their algorithms - their secret sauce that matches couples - to peer scrutiny.
They also question whether their algorithms are effective at positively matching people or whether they just exclude unsuitable suitors.
eHarmony, for example, excludes people if their survey responses suggest that they suffer from underlying emotional or mental health problems like depression.
They also suggest that "selection bias" - a statistical bias that occurs when your sample population is different from the norm - may be at work.
The reasoning is that people who use matching sites are different from the average Joe or Jill.
For starters, they're likely to have a higher disposable income and, given that they sit through a 400 question survey, more highly motivated than the average dater. The claimed success of matching sites may have more to do with narrowing the pool of eligible daters, rather than psychological tests or computer science.
One of the co-authors of the paper, Associate Professor Paul Eastwick from Texas A & M University says that the sites claim to do much more than weed out Mr or Ms Wrong.
"The sites promise to find you someone who is especially compatible with you - your soulmate. That's a very different promise that they cannot fulfill," says Eastwick.
That's not to say that Eastwick and his co-authors are saying that the algorithms have no effect whatsoever.
On the contrary, they suggest that the algorithms may exert all kinds of influences on the dateless - just not the ones advertised by online matching services.
For example, Eastwick and his colleagues argue that there may be a placebo effect at work.
Just as placebos work because of the aura of authority around the person prescribing the "drug", rather than its inherent medicinal value, so online matching services may work because the couple believe their coupling has been validated by relationship experts using complex computer science.
As the authors of the paper write "having a purportedly authoritative source claim to use science to select putatively ideal or highly compatible matches could predispose people to be more accepting of these matches, at least initially, than they might otherwise be."
In reply, eHarmony's Gian Gonzaga says "It's an interesting hypothesis, but I don't know of any data that would support or refute it. That's still an open question."
Gonzaga also flatly rejects suggestions that eHarmony's technology lacks scientific credibility.
"We based our matching system on years of research, both clinical and empirical. We've taken what we can discover about personality and values and how those similarities predict relationship success," he says.
"That's the empirical knowledge, the scientific knowledge we've used to build our matching system. We've done years of research with thousands of couples to empirically come up with our algorithm."
Drawing on psychological studies to create matching software is one thing. It's quite another thing to say that you have successfully transferred the findings of those studies into software that can successfully match people with their romantic ideal.
Gonzaga does say that the algorithm has its blindspots, especially if your survey responses are overly nuanced. For example, if you reply that you're an extrovert in some situations and an introvert in others, it may be that you're too complex for the algorithm to figure you out.
"Because our system is based, in part, on similarity, we then don't know what to do," says Gonzaga. "Do we match you with people who are extroverted or introverted? The system just breaks down."
But Gonzaga is adamant that algorithms can help people find a long-term partner, cutting the pool of potential mates down to a manageable number.
"I think the upside - and it's a big upside - to using algorithms in making choices, it can help you filter and make that process easier. It can empower you to make those decisions."
Even if you believe the promise that algorithms can find you true love, the digital Cupids cannot guarantee living happily ever after.
Peter Lake and his wife, two of computer matching's first success stories, divorced after 11 years of marriage, although they remain in contact. Lake has returned to computer dating since, using both Match.com and eHarmony, but the barrier to finding true love has turned out to be more geographical than technological.
"I met a really nice dentist in New Hampshire - but she lived in New Hampshire. It's an hour and a quarter away and I thought, God, this is just too far with my advanced age," says Lake.
"Eventually I realised unless they live down the street, I'm really not interested."
Lake has abandoned algorithm-assisted online dating in preference of online chat rooms and forums. "Now if I want to meet somebody, I just go online and find 'em and chat 'em up."
Matching software, it seems, is no match for a good chat up line.
- Daily Life