Breaking up is never easy, particularly when you’re confronted with memories of happier times. A smell, an old photograph, a note somebody left you — weeks or even months after a break-up and you can still be reminded of your ex-partner, whether you like it or not.
On social media, this can be even worse. If you’re still friends with your ex, you’re likely to still see their posts on your feed; if you’re not, you can still rub salt into the wound by checking their profile anyway. ‘On this Day’ features are also notoriously bad for bringing up unhappy memories at the worst possible time.
According to a new study published in Proceedings of the ACM on Human-Computer Interaction, we also see our exes so much because of the so-called “social periphery” — the networks of people we know tangentially through our ex-partners. So why not design an algorithm that causes us less pain? The new work suggests that this could be the answer to our online break-up woes.
The study, conducted by Anthony Pinter and colleagues at the University of Colorado Boulder, focused on 19 adult Facebook users based in the US. Semi-structured interviews were held with each of the users on their feelings around break-ups and social media. Each had been in a relationship prior to the interview — either dating, cohabiting, or marriage — and were aged between 19 and 46.
Participants described a range of experiences in which they came into contact with their ex-partners online, from anything between six days to five years after the break-up. They were then asked to focus on specific features that could stop them from coming across their ex — unfriending or unfollowing, for example, or changing the way they view their newsfeed.
Unsurprisingly, emotions ran high. Participants reported feeling pained by seeing content involving their ex-partners, whether that was new information (such as an ex’s new relationship status) or past memories (such as anniversary posts or photographs). “The most upsetting thing on Facebook is On This Day,” one participant said. “It said I was the best husband ever and she loved me the most… I remember that, and obviously not physically being hurt, but just feeling an emotional wallop of like ‘Fuck, that wasn’t that long ago’”.
This was all fairly unexpected: unwanted contact with an ex-partner is obviously going to be difficult in some regard. But while the problem may be well-established and familiar, there could still be a novel response.
The problem, the authors argue, is that machine learning has focused on methods that “fail to capture social nuances, relationships and other human-centred concerns” — in other words, that the algorithms present to us an unnatural or unhelpful model of our social relationships.
There are workarounds when it comes to existing platforms — unfriending, unfollowing or blocking ex-partners, or opting out of features like ‘On This Day’. But because of the social periphery, distant connections still linger after a break-up: one participant talked of their ex-partner’s mother’s frequent appearance on their feed.
Being clear about what will happen when you mute or block someone is a good first step. But such fixes, the authors believe, are far from perfect. It’s the algorithms themselves that need changing, taking into consideration our complex social peripheries as well as our one-to-one connections.
Currently, algorithms mainly take notice of binary connections — how much or little we choose to see from one particular person. By tweaking these algorithms to take into consideration not only peripheral relationships but also events, interests, photos and groups could mean our social periphery is both better represented online and easier to evade post-break up.
The complexities of such encounters should also be taken into consideration. It’s unlikely to matter if an ex has clicked ‘attending’ on a large event that spans multiple days or takes place in multiple locations, so seeing that they’ve done so may cause unnecessary pain. Knowing they’re likely to attend a small gathering of friends, however, may be more useful information if you’re keen to avoid an awkward meeting.
When, or if, algorithms become more human-focused, we may find ourselves having less stressful interactions with our ex-partners online. Blocking and unfriending might not be perfect, but at the moment may be the next best thing.