How science can help you find love
Here is a formula for love that you can use right now:
Silly as it sounds, I find this formula much more precise and comprehensive than most of the ones you can find out there. Don’t believe me? Go ahead and Google “scientists find formula for love”. I’ll wait…
See. There is a new formula for love everyday. How likely is it that this new formula is the one? Well, not very
a love equation is improbable
Here’s the thing: It is highly improbable that a universally applicable equation for love even exists. It is way too often people that we see similar fallacies in fields that study human behaviour from psychology to economics. When we know that people rarely behave in logical ways, any devised formula to generalise love is bound to fail. Even when an equation is simple enough to be accurate for a large number of people (just as the one that we present), there will always be plenty of individuals who will prove it wrong.
In addition, the very own definition of love is highly contextual. Different people, different cultures, different definitions and expectations on love. If you don’t believe me, try reading about the Mosuo people. It is likely that they understand love differently to what you and I do. People are different, so are their understanding of love and relationships.
Because love may not be defined by a universal mathematical formula, it does not mean that people cannot get help from a scientific approach to dating. Now, that does not mean that online dating can promise to match you with your soulmate. Regardless of how much “science” it says that does, real science has yet to back that claim. Looking at the science on both psychology studies and computer science studies, there are two contradicting, and sort-of complementary views.
Psychologists tend to side with the view that online dating cannot predict future happiness in a relationship, and that any compatibility algorithm are not much better than random. See #1 below for a comprehensive study.
On the other hand, Computer scientists (including me) view dating as a mathematical problem. This means that we look into improving the odds of people finding a potential partner online.
Here is where it gets tricky. Unless you told your online dating service, they will not know if you married someone after finding them online. They will also not know how long you will stay married and whether that marriage is a happy one or not. So any studies that analysed people’s personality and their compatibility is limited and unlikely to apply to a wide pool of users.
no formula, no data, no science? not quite
So, how do you know that people are going to be good as partners when psychology says there is no such a formula and computer science say there is not enough data?
We don’t know. But we can make assumptions. For instance:
- If you don’t meet someone, you are not going to be each other’s partner.
- You both must have a certain degree of mutual interest in each other. If one doesn’t like the other, it is not going to work out.
These two assumptions are quite reasonable and are the driving motivation for technological approaches to dating.
The goal of online dating is to help the user weed out potential time wasters and other bad matches so that your experience in the offline dating is enjoyable and more likely to lead to you finding a partner.
So there is a formula for love?
Yes, but it is not a single equation.
We compute a different formula for each individual online dating user. We understand that some users have different interests than others and value different things. So for instance, a user may value people who care about health while others may have a strong preference towards the person’s cultural background.
The use of individual formulas for love can help to measure if two people will like each other. This is what we call reciprocal recommendation, where recommendations are given on the basis of how much you like someone and on how much that someone likes you back.
To have an idea how well this works, have a look at the following graph with data from one of my scientific papers #1.
The graph shows how often two users like each other when one finds the other by browsing the website, or when they use recommender systems. What you can see in the graph is that the performance of a non-reciprocal recommender is very similar to the users themselves when browsing. This is expected as the non reciprocal recommender is based on the behaviour of the users themselves. Now, the reciprocal recommender outperforms the users themselves as it knows more. It knows how likely users are to respond positively to each other. Thus, helping user to find real potential dates.
Even thought online dating cannot guarantee to find your soulmate, computer science can help you find a potential partner in a much easier, convenient and quicker way than at the bar or at the club.
- Eli J. Finkel, Paul W. Eastwick, Benjamin R. Karney, Harry T. Reis and Susan Sprecher.
Online Dating: A Critical Analysis From the Perspective of Psychological Science
Psychological Science in the Public Interest, January 2012 13: 3-66
- Luiz Pizzato, Tomek Rej, Thomas Chung, Irena Koprinska and Judy Kay.
RECON: A Reciprocal Recommender for Online Dating.
RecSys ’10: Proceedings of the Fourth ACM Conference on Recommender Systems.
Barcelona, 26-30 September 2010