Can Alexa and Facebook predict the end of your relationship?

We’ve quickly come to accept that brands know as much about us as we know about ourselves. Facebook serves you ads for cat food after you talk about getting a cat. Target knows you’re pregnant before you tell your friends and family. Even Instagram knows about your shameful predilection for Hallmark Christmas movies. So it stands to reason that fewer pics of you with your significant other on Instagram could signal to apps and brands that your relationship may be coming to an end.

But how much can Big Data actually tell you about your relationship? Can it predict, say, when you’re about to break up? And if you start to see pop-ups for ice cream, Kleenex, and dating sites, should you be concerned?

As it turns out, while experts say it isn’t being used, relationship prediction technology already exists to some degree. The question is, will brands take advantage of it? And more importantly, will you?

Getty Images

Relationships are notoriously difficult to predict. There are so many variables at play, from environmental context to biological attraction to personality compatibility to whether or not you share the same opinion on Jimmy Fallon, that mathematically speaking, determining a specific expiration date for a relationship with a great deal of accuracy is close to impossible. “Relationships are dynamic processes and complex systems,” Justin Garcia, Match Scientific Advisor at dating site Match, told Vox. “To be able to say, ‘You and Brian are going to break up in 2.5 months’ — I don’t think that’s very likely.”

That said, researchers have been studying relationship outcomes for decades, so there’s a significant (and fairly strong) body of work for developers to build on. Perhaps the most well-known research of this kind is John Gottman’s 1992 study of 52 newlywed couples, in which Gottman’s team interviewed them and observed their interactions, then asked them to fill out a questionnaire three years later. Gottman was able to develop a model predicting the likelihood of whether a couple would get divorced with more than 94 percent accuracy. He later published a book arguing there are seven traits associated with relationship outcome, such as whether couples express fondness or affection toward each other and how they deal with conflict.

Gottman’s research has gotten a great deal of media attention, but it also isn’t particularly surprising: If couples are arguing about money right after they’ve gotten married, it stands to reason they’ll continue to do so for the duration of the relationship. Further, the 1992 study also relied on oral history interviews with the couples, implying that the model only works if you can observe couples IRL.

But can you come up with an equally accurate model based on data alone?

Perhaps unsurprisingly, Facebook has also long been interested in the question of relationship outcome prediction, to the degree that it is embedded in the company’s DNA: According to the book The Facebook Effect, while Mark Zuckerberg was at Harvard, he developed an algorithm that could predict which of his friends would hook up with each other with 33 percent accuracy. (Apparently, this was something he did for fun, which speaks volumes about both Facebook’s origins and Zuckerberg’s college social life.)

In 2013, Facebook engineer Lars Backstrom and Jon Kleinberg of Cornell University co-authored a paper identifying a number of factors that contribute to long-term relationship success, such as whether a couple had lots of friends in common or whether they posted a lot of photos together. The researchers found they were able to determine with 60 percent accuracy whether a couple would break up. A subsequent study by Facebook data scientist Bogdan State analyzed Facebook relationship statuses from 2008 to 2011. He found that couples on Facebook were more likely to stay together once they hit the three-month mark, with their chances of success increasing the longer they stay together.

StayGo/iTunes store

The idea that big tech companies could use your data to predict something as intimate and emotionally charged as a breakup resonated with people, and not necessarily in a good way. (“Facebook can predict with scary accuracy if your relationship will last,” one headline read.) Nonetheless, a number of developers have since dabbled in relationship prognostication, with one such app, StayGo, launching in 2016.

Developed by a team of psychology and relationship experts and based on a mountain of relationship research, StayGo asks couples 20 questions that ostensibly determine long-term compatibility, from how satisfying their sex life is to how they handle money. Then it calculates an “SG score” out of 100. The app also allows couples to crowdsource feedback about their relationships from family members and from other users of the app, which sounds both extremely pragmatic and like an extremely easy way to endanger your relationship, whether your “SG score” is high or not. (It’s unclear exactly how StayGo’s algorithm works or how accurate it is. StayGo did not respond to my questions by press time.)

As technology has evolved and become further integrated into our private lives, so too has the amount of personal data we’ve made available to Big Tech, which has inevitably resulted in researchers getting more and more creative about studying our sex lives. A 2018 report from eHarmony in conjunction with the Imperial College Business School in London, for instance, found that smart home assistants that use voice recognition technology, such as Google Home and Alexa, could one day be used to predict breakups or even provide relationship counseling by listening to our conversations.

As study co-author Aparna K. Sasidharan recently explained, this insight was based largely on 2017 research that used speech recognition technology to analyze 134 couples’ conversations during marital therapy over the course of two years. The researchers analyzed data such as changes in pitch or how often someone would switch from “you” to “I” or “me,” and they developed an algorithm that was able to predict whether a couple would break up with 79 percent accuracy.

To be clear, eHarmony’s report does not state that companies like Google and Amazon are actually doing this, nor is there any substantial evidence that smart home assistants are listening to our conversations without consent. That said, Amazon recently filed a patent suggesting it may at least have interest in doing this. (In a statement sent to Vox, an Amazon spokesperson said: “We take privacy seriously and have built multiple layers of privacy into our devices. Like many companies, we file a number of forward-looking patent applications that explore the full possibilities of new technology. Patents take multiple years to receive and do not necessarily reflect current developments to products and services.”)

Sasidharan’s point is more that companies like Amazon and Google could do this if they wanted to. “We have a model that can predict the fate of a relationship fairly accurately, but no one has operationalized it or incorporated it into a device or a dating app and said, ‘Okay, use this,’” she says. “But if people are accepting of it, it can be done very soon.”

Of course, this brings up two very important questions: Will people ever accept it? And why, exactly, would companies want to know about the fate of your romantic relationships to begin with?

Netflix

“Hang the DJ,” the fourth episode of the fourth season of the dystopian BBC series Black Mirror, attempts to answer that first question, albeit in a somewhat oblique way. In the universe of “Hang the DJ,” couples are matched by the Coach app, which tells you how long your relationship will last. When a couple falls in love despite their relationship having an expiration date, they try to beat the shadowy forces that control their ecosystem (the ominously titled “System”) to be together.

Unlike the majority of the Black Mirror canon, “Hang the DJ” isn’t necessarily an indictment of our over-reliance on technology; in fact, in some ways, it is an endorsement. The couple’s subversion of Coach turns out to be a function of a larger computer simulation, which determines their compatibility based on how many times their avatars beat the System (998 times out of 1000, as it turns out, making them 99.8 percent compatible.)

Even more surprisingly, Black Mirror fans weren’t necessarily creeped out by the concept of “expiration dating.” In fact, they immediately began to speculate whether such technology would one day be available, prompting Netflix to release an ersatz version of the app for Valentine’s Day. In an article for the Washington Post, Lisa Bonos even argued that the System would be an improvement on the swipe-filled drudgery of the real-life dating world.

“When the dangers of online dating are discussed in real life, the paradox of choice comes up. This is the idea that, faced with an abundance of choices, be it on Tinder or brands of cereal, we’ve become not freer and happier but more paralyzed and dissatisfied,” Bonos writes. “The System aims to offer the best of both worlds: Lots of options, and at the end of it you get the best one.”

For single millennials who have slogged through one bad Tinder date after another, expiration dating has obvious appeal: Why waste time arranging a coffee date or asking the obligatory questions about jobs and siblings if there’s a deadline hanging over your head?

Garcia, the data scientist at Match, has not seen “Hang the DJ,” but he agrees that people would be interested in using technology to predict relationship outcome. “Romantic love is one of life’s greatest prizes. If we could use people’s user behavior [to find love], my hunch is there are plenty of people who would want to do that,” he says. And using data to predict the long-term success of a relationship with a prospective match could be one of many possible ways to do that.

It’s also not impossible to imagine a world in which people would be willing to relinquish such data in the name of finding love. As much as Americans purport to care about privacy and social media platforms accessing their data, most don’t care enough to actually leave said platforms. It’s safe to assume they’d care even less if these companies started using their data to help them get laid.

That said, the question remains: Why, exactly, would companies like Facebook and Google want to know the details of your breakup, long before either you or your partner decide to pull the trigger? While it’s tough to answer this question with certainty — most big tech companies are understandably tight-lipped about the specifics of collecting user data — it probably wouldn’t be for altruistic reasons.

It’s a well-established fact in the marketing world that people are more likely to spend lots of money after a bad breakup. “When a relationship ends, people go shopping,” Sasidharan says. It’s also a well-established fact that companies like Facebook track this information — per their own data, users spend 25 percent more on travel-related purchases after changing their relationship statuses — and share it with marketers looking to brush up on their targeted advertising practices.

With this in mind, Sasidharan predicts that companies will use that data to serve you ads, “knowing what you’re going to buy, and where you’re going to go.” It’s not tough to imagine a world in which Facebook or Instagram would see you posting fewer photos with your partner and then serve you ads for an airline, a hotel brand, or even a dating app. And now that Facebook has officially moved into the dating space by launching its own dating app in Canada and Thailand, it will gain access to even more valuable (not to mention intimate) user data. (Facebook says that it currently doesn’t show ads or plan to show ads within Facebook Dating, and says that the company does not “use information on how people use the dating experience” to target ads across Facebook’s products.”)

The question of whether Big Data would embrace the opportunity to monetize heartbreak aside, however, let’s assume for a moment that people would want to have access to this information, and that we could predict, with a strong degree of accuracy, whether or not a relationship would be successful. Garcia says that maybe it’s worth asking ourselves what defines a successful relationship to begin with.

“When we talk about successful relationships, my measure of success is not longevity,” he says. “It’s ‘Did you have personal growth? Did you laugh? Was the sex good?’ When we think about technology to predict our relationships, we have to think about what we’re trying to predict.”

Want more stories from The Goods by Vox? Sign up for our newsletter here.

Will you support Vox’s explanatory journalism?

Most news outlets make their money through advertising or subscriptions. But when it comes to what we’re trying to do at Vox, there are a couple of big issues with relying on ads and subscriptions to keep the lights on:

First, advertising dollars go up and down with the economy. We often only know a few months out what our advertising revenue will be, which makes it hard to plan ahead.

Second, we’re not in the subscriptions business. Vox is here to help everyone understand the complex issues shaping the world — not just the people who can afford to pay for a subscription. We believe that’s an important part of building a more equal society. And we can’t do that if we have a paywall.

So even though advertising is still our biggest source of revenue, we also seek grants and reader support. (And no matter how our work is funded, we have strict guidelines on editorial independence.)

If you also believe that everyone deserves access to trusted high-quality information, will you make a gift to Vox today? Any amount helps.

One-Time Monthly Annual

$5/month

$10/month

$25/month

$50/month

Other

$

Yes, I'll give $5/month

Yes, I'll give $5/month

We accept credit card, Apple Pay, and Google Pay. You can also contribute via