Love Equation #9

How to find the perfect partner with a little help from computer science

Columns | Linda Eid, Joe Telki | October 2009

From 5 to 122… This is your window of opportunity! Your chance to find the perfect mate! Unless it takes the whole 122… In some cases, people have actually spent a lifetime looking. Many tell happy tales of success, but others… well, when it was their turn, let’s just say fortune was looking in the other direction.

Why start at five? Well, this is the age when children begin to develop social affections, when the identification with the mother is broken and a child can build focused friendships. It is also the time of the first puppy-love... Oh, those simple and exquisite romances of early age!

And why 122?  Simply because this is the longest anyone has ever lived. Still, it may not be enough. Fortunately, there is a scientific formula that can ease the way!

The following is an attempt to explain with the full weight of scientific presumption the criteria by which people form couples and unveil the "WHY" that makes us choose our soul mate. And by "scientific," I am not referring to psychology or sociology, which would logically seem the most relevant to relationships. Not even politics or diplomacy – extensions on the international level. No! This is an innovative approach to "finding the perfect match" with the help of computer science. Matchmaking for nerds! Despite their apparent divergence, both human and computer-based approaches reflect on the same underlying concept of establishing "similarities" around which relationships revolve.

People identify their "perfect match" by evaluating compatibility according to criteria that are important to them, but of course, vary from prospective mate to prospective mate. Computers are also able to compare human entities as they do webpage activities, semantic concepts, etc. Now don’t get defensive. We all know you are not an object. It is simply that attributing a numerical value to each criteria in the selection process of our "other half" lifts the probabilities to another level of precision – and thus efficiency!  The "added value" here is that a computer-based method not only enables crunching the numbers, but also fearlessly bush-whacking your way through a jungle of options with a certain philosophical detachment from the tortures of self doubt that would otherwise accompany this bold venture into the unknown.

To develop the formula that could brighten that anti-chamber where many wait for the call to their "happily ever after", we have to clarify some crucial definitions.

To start with, we will consider humans to be ‘ontology-free’; meaning that we presume no prior knowledge of the relations (social, political, etc.) between them. Even though relations do exist and might have a determinant effect on finding the perfect match, simplicity requires we discard such cases. Another definition we need is the ‘human attribute’; which is any feature of a person and which can be of numeric type – between zero and 1 – for instance: IQ= 0.9, beauty= 0.8, attractiveness= 0.7. Note that attributes of textual type, example: hair color= brown, are also to be disregarded. With that, it all comes down to comparing two numerical values of similarity:

Sim (Att1, Att2) = 1-

After comparing attributes and determining similarities that matter to you, we then produce a combined score. Among several methods of combining outcomes are the ‘maximum’, ‘minimum,’ ‘average’ and ‘weighted sum functions.’ It does not matter which function you choose, as long as you are consistent.

Now buckle up and let’s get straight to the findings!

To find one’s perfect match, the most basic approach is to test affinity against all possible candidates. In computer science terms, this is simple. In human terms unfortunately, this proves to be impossible. In fact, by this logic, each person would have to date (or at least somehow know) 3 billion candidates. Thus, we sadly conclude that despite being 100% effective, this method is unfeasible. It is not practicable and at 122 years, our time is limited!

The human approach, faced with the impossibility of testing all people for a perfect match, tends to narrow down the search by defining a ‘minimum affinity threshold’ based on things like social groups, educational level, moral values etc… to weed out candidates. Alternatively by specifying a ‘virtual cognitive clustering’ we can search for candidates with only certain particular traits. So even though the human method does not guarantee perfect results, it is more practical than the computer-based approach.

Human versus machine, it is an infinite debate. But then again, we do not usually look for THE perfect match, but rather for the ‘best match possible’, considering the constraints at hand. And for the sake of that aspiration some humble words of wisdom: Do not lose faith in guidance from above, do not underestimate the element of chance, but most importantly, and whatever you do, DO NOT wait for a computer to figure this one out for you. Knowing and understanding yourself is the best – and only – hope.

As painter Joshua Reynolds said: "He who does not know himself does not know others."

Other articles from this issue