Can mathematical models predict whether your relationship will last?

Modern relationships are governed by algorithms and big data: could we learn something from this?

Author: Dan Jacobson
Editor: Amalia Houpi
Artist: Daniela Mendelevic


Question: If you were offered the opportunity to find out who you are most compatible with, would you take it? 

Answer: Yes, please save me from awkward pints and the politics of bill-splitting and let’s get straight to the point.

We are in the midst of a paradigm shift in how relationships are formed and conducted, one governed by algorithms and data. As of 2018, Tinder registered 57 million users worldwide, processing 1.6 billion swipes per day, whilst dating website eHarmony, in 2016, was estimated to hold 120 terabytes of data about its users. Whilst Tinder matches individuals mainly on location, and previously a “likeability” score, eHarmony is known to use predictive models, the details of which they keep fairly close to their chest. But how close are we to developing models that could be useful in explaining, and predicting, compatibility, based on more than a dating profile?

From my perspective, the most interesting research into individual compatibility takes a game theory approach, using knowledge of how people respond strategically in times of conflict to predict whether a relationship will succeed. Initially, this seems both counter-intuitive and pessimistic, as this might assume that a relationship’s success rests solely on conflict. However, this could provide guidance for understanding how a relationship works. For example, if Person 1 is dissatisfied in a relationship, they can either confront Person 2 and potentially upsetting them, or move on. If confronted, Person 2 can accept blame, or reject it and potentially resulting in fighting. In this way, the risks and rewards of cooperation and conflict can influence a relationship’s success.

The drawback of this method is that it is mainly theoretical. In 2004, building from these ideas, psychologist John Gottman, who has pioneered research into marital stability, collaborated with applied mathematicians James Murray and Kristin Swanson. They developed a model to describe how partners react during a conversation based on general mood, current mood, and their partner’s influence on them. This two-equation model has been shown to predict divorce with 94% accuracy, using only a few minutes of their interactions.

The researchers found that the most telling feature in determining marital longevity was the “influence” that each individual has on their partner, measured as the threshold at which one person triggers a response from the other, especially negatively. Whilst it might be assumed that a large threshold, indicating a high ability to compromise, may be indicative of long-term romantic success, the opposite was found to be true. Couples who chose not to let issues go unmentioned tended to give room for discussion about their relationship, allowing emotions to be presented in a healthy environment, rather than remain hidden.

This article stems from my ongoing fascination in the intersection between mathematical models and relationships. But is there any real value in studying this subject? Whilst I personally do not see the value of applying mathematics to predict success within individual relationships, simply due to sheer variation, I do believe that, someday, mathematics will have the power to reveal patterns that are indicative of successful relationships across a population. My hope is that these findings could influence our behaviour, allowing us to create more meaningful connections with each other. Currently, the way we conduct ourselves in relationships is influenced by all sorts of factors, from past experiences to Jane Eyre. Why not include data?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s