Can You Solve For Love?

By , February 14, 2013 12:13 pm

This title already sounds like the sequel to ‘It Felt Like Love‘ aka the most depressing movie ever made (shout out to The Sundance Six for not committing seppuku directly following the film’s screening).  Thankfully, I have no interest in exploring the depths of loneliness experienced by a 14 year old girl desperately seeking to be noticed in the world.  I’m interested in something much more … pleasant?  I want to know if we can write a mathematical formula to explain what makes someone do this:

Ok, ok, yea, modeling THAT will not be easy. There’s a whole lot going on there, some of which would be best left between a man and his mirror. However, there are attempts to understand what chemical reactions and brain activities are associated with the feelings that Sam has put so well into dancing and mock dialogue. At a panel discussion on ‘Lust and Love in the Animal Kingdom,’ we were able to see what our brains do when we experience the feeling of love for another person (we also learned how slutty ‘monogamous’ animals are and theories on how dinosaurs had sex … too much information). With more of these types of studies, in a few years we should have enough data that we can start to actually answer the question ‘What is Love?’ (baby don’t hurt me, don’t hurt me, no more).

So if modeling love is currently out of our reach, what about trying to figure out what type of person gives us butterflies in our stomachs (or the impulse to buy a Parisan night suit)? What qualities uniquely, or not so uniquely, attract us to a person? And more importantly, can we model mutual attraction? It doesn’t matter how many times you ‘like’ a person’s profile picture if they don’t like you back (there are probably some prescribers of the ‘wear them down’ technique who might disagree with this statement).

This is a problem that online dating sites are attempting to solve. Sites like Match, eHarmony, and OKCupid are all working on different approaches to collect data from people, and use that data to effectively match people with other people they may like AND will like them too. It’s the ‘proposal aunty’ in a digital format (note to self: buy domain name proposalaunty.com for new south asian dating site). OKCupid has a pretty simple algorithm for how they do this, which you can see here:

The first step is to simply to answer questions and then using those answers, start to match people together. This works really well for things like personal interests, but can fail miserably for things like personality types. In a relationship, for some qualities you need a Yin for your Yang. OKCupid solves this by also asking what a person would want the OTHER person to answer. This effectively allows you to match interests where you want to and find balance where you need to.

The next thing they do is allow the person to weigh how important the question is to them. People care about different things so understanding what is individually important creates priority in matching characteristics. It would be a whole lot easier in life if people had exactly the same priorities. Marriage vows would turn from “I do” to “You’ll do” which is kind of where most post-30 year old women are headed anyway (half-kidding). OKCupid’s actual numerical weights seem a bit suspect, but they fit the idea that priorities are not linear or in other words, some things are MUCH (exponentially) more important than others.

Now comes the actual number crunching. The video does a real good job of describing how they come to the final match percentage, so I won’t reiterate that here. The geometric mean seems to make sense when common questions are small, but it probably becomes irrelevant at some point (i.e. when individual %s are very high or common questions are very high). Is there a real difference between 98% compatible versus 99%? It could be in this threshold difference where true love is found. Or maybe the sweet spot is between 76% and 79% (after all, a bit of conflict can make a romantic relationship more exciting and differentiate it from a great friendship). These are the more interesting questions that OKCupid could answer by regression testing their users, in particular successful couples.

I think there is a lot more complexity that needs to be modeled to figure this out well, mainly behavior and timing. One thing that behavioral analysis could bring to the table is actually identifying characteristics that people demonstrate they like rather than stating they like. I don’t know how many times I’ve seen girls list ‘sushi’ and ‘reading’ as things they like, when all they do is eat bacon and watch reality TV. The truth is that there are plenty of guys out there who like bacon and reality TV, so listing what you DO versus what you WANT TO DO or what MIGHT LOOK IMPRESSIVE would probably yield better matches.

Timing is another tough element to model. Being ‘available’ is almost a minute to minute thing, unless well disciplined by the person. And I don’t mean ‘available’ in a ‘free of a relationship’ sense necessarily. Just being receptive to someone seems to fluctuate widely based on all sorts of stimuli. A 99% match might never be pursued simply because you had a bad day at work, and a 65% match might have a dinner date because your favorite team won the Superbowl (I think I just made the argument for the ‘wear them down’ approach). People do a huge disservice to themselves by not separating their current, temporary emotional state from the task of finding a long term soulmate. Embrace opportunity no matter how crappy you feel! Finding the right person is already a low frequency event as it is.

Okay, I’m going to end this rather abruptly. This post is dedicated to my crushes over the years: Kristen DiMaggio (kindergarten), Tatiana Thumbtzen – the slinky blue dress girl in ‘The Way You Make Me Feel’ video (middle school), Jennifer Love Hewitt (high school), Aishwarya Rai (college), Saira Mohan (early 20s), David Wright (mid 20s), and Melanie Kannokada (seems as good as anyone in a post-crush era). It’s a … self preservation thing, you see.

Lastly, awesome pick-up lines for the day:

“You know, there’s a 100% chance that we are dating in some alternative universe, so why don’t we date in this one?”

“1 …. 2 …. 3 …. 5 …. 7 … 11 …”
“What are you doing?”
“I’m priming you up.”

“If you give me your number, I’ll tell you its 3rd derivative.” – actually said by a real life girl.

Leave a Reply

Panorama Theme by Themocracy