Some Caution on Online-Dating Algorithms

by GK on February 15, 2012 · 4 comments

Regarding the study of attraction and dating, I’ve always been one to play Biff and push around the George McFlys of the academic world. But sometimes, those Buttheads get my attention, and nowhere are their virginal-yet-exact methods more useful than for online dating.

Perhaps I like this particular dissection of online dating because, in a way, it contradicts science. Or, at least, the half-assed science of online-dating algorithms.

online datingYou can read a much shorter explanation in the New York Times, but I took one for the team and read the entire 64-page article written by Eli J. Finkel and four others, which is to be published in the journal Psychological Science in the Public Interest.

Finkel and the gang examine the validity of online dating, and I was especially interested in its critique of matching sites — namely eHarmony — that use algorithms to predict not just compatibility with another partner, but also long-term relationship success with a “soulmate.” (Sorry, I can’t help but put quotations marks around the word, since I consider the idea of the perfect partner one of mankind’s worst and most harmful inventions. But that’s for another post.)

In short, the authors don’t conclude that mathematical matching sites are any worse at producing lasting relationships than the old-fashioned way of saying hello to that honey in the bar. But after examining the facts, they don’t consider them any better, either, even though users are paying a lot more for those sites than $10 for a vodka gimlet, and spending many hours chained to their desks.

The article’s reasoning includes the following:

  • Although nobody knows exactly how the sites’ algorithms work — something the authors also object to — we do know that they can’t possibly measure how two people get along after they meet each other (such as resolving disagreements), as well as what external forces (finances, stress, etc.) will arise. Yet these latter factors have far more effect on how a relationship ends up.
  • The algorithms use similarities in personality and attitude to match people. And while such similarities do appear to foster attraction, their effect on long-term relationships is practically zero. Introverts do just as well with extroverts, for example.

But there is a twist: the authors suggest that to succeed at online dating, singles are better off with a screening site like eHarmony over a feeding-frenzy site like OK Cupid, just because of the illusion of greater compatibility:

A second process that may emerge as a result of users’ faith in the validity of a site’s algorithm is the placebo effect … A user’s belief in the validity of the algorithm used by a dating site may cause him or her to view a match as compatible.

To those of you familiar with the seduction community, this is online dating’s version of social proof. And I can see their point, the same way I know that a beautiful woman is more likely to choose one of two guys at a private party than one of 20 dudes at a bar. When we’re loaded with options, as on many dating sites, we’re less likely to pick one. Or we pick one for the wrong reasons.

I can’t deny that I have a bias in all this, considering how much time I’ve spent teaching guys how to meet women in person. And the report only confirms what I found in my experiment with online dating: the one advantage to it is easy, low-risk access to singles, which puts me at a disadvantage. Offline, I’m one of the few guys with the guts to approach an attractive woman.

That said, I am an emo-boy romantic, and if computers could match us all with someone uniquely meant to last with us, I would be all for that. It’s just that it won’t happen anytime soon. So I’d rather spend my leisure time in the analog glory of the outdoors.

{ 4 comments }

1 Gal @ Equally Happy February 15, 2012 at 4:57 pm

I met my wife through eHarmony. It has both advantages and disadvantages compared to meeting women offline (which I’ve also done).

Advantage – Greater access to many singles and the ability to filter this “candidate” pool. You know something about the woman before you actually go out with her. For example, I knew I wanted an atheist who wants kids and eHarmony allowed me to filter out people who didn’t match that requirement.

Disadvantage – There’s a lot of pressure right from the start to get in a relationship. You both know that you met each other on a site because you were looking for something long term (this might be different on sites other than eHarmony) and that means you’re checking each other out for marriage potential from the first date. This interferes with the whole getting to know each other and becoming friends process.

Still, it’s a good alternative for those of us who are a bit too busy to go out much and who have already dated most of the people in our friends and family circle.

2 GK February 15, 2012 at 6:18 pm

Thanks a lot for the insight, Gal. I’m going to a wedding in April where my friend met his fiancee through eHarmony. I’d considering trying it as a supplement, though as you suggest, I’m a little cautious about how seriously the daters take it. I’m glad it worked for you!

3 Anonymous February 21, 2012 at 1:11 pm

On CBC Radio here in Canada, the show The Current did a Valentine’s Day special on online dating (there was also a special on sex robots but that’s a whole other ball game…) and a recurring theme with numerous guests – including online daters, the owners of several online dating sites, and independent researchers who have studied online dating – was the choice paradox, the idea that the more options a person has the less likely they are to actually make a choice. For this reason, most of the guests made explicit in one way or another that the most effective online dating sites are ones that use algorithms to match people up, simply because they give you a narrower field of options. The “feeding frenzy” sites like POF are less effective because a user’s only option is to scroll through thousands of profiles, so people using those sites are less likely to commit to one individual enough to actually meet in person. Conversely, on sites where people are given a smaller pool of potential mates based on a compatibility algorithm, they are more likely to be able to choose one or two people to go on dates with simply because they are not overwhelmed with options

Sensory overload can be paralyzing…

Another aspect of these feeding-frenzy style sites was that because users had so many options, they were more likely to write people off for arbitrary reasons. In real life, you’d never write someone off for being 5’11 instead of 6’… in fact, you wouldn’t even be able to tell the difference. But on these open-ended, mass-option dating sites, a person might look at someone’s vital statistics, see that they are 5’11, write them off completely and not bother to look at the rest of the profile simply for being under 6 feet tall.

This apparently happens less on compatibility oriented sites because not only are people more “forgiving” of small things when their options aren’t so wide, but they are also more likely to ignore things like height because they can see that someone is 93% compatible with them or something.

It was pretty interesting stuff and reinforced my desire to stay away from online dating. The wide-open feeding frenzy sites encourage (and even create) frivolity, and the algorithm-based sites cost money. Plus, I really don’t want to trust a computer running some math equations to determine who I’m going to be compatible with… human relationships are way too fluid and often what I think I want is different from what I actually want and what works. I’d rather meet someone in person and go with my intuition than have a computer tell me who I “should” connect with.

4 GK February 21, 2012 at 4:21 pm

Online dating … sex robots … the Canadians uncover it all! Thanks for letting me know about that report … it pretty much echoes my own conclusions. Again, online dating is a sensible alternative, but it’s not what I want to depend on.

Comments on this entry are closed.