How dating app algorithms push us together and drive us apart

The question of how algorithms might work in or against users’ favour has always been shrouded in mystery, particularly when it comes to dating apps.

In 2015, American author Apryl Williams was at a conference for dating platform OKCupid that seemed to confirm many people’s suspicions. When a member of the audience asked OKCupid’s chief executive, Christian Rudder, why their matches were “unattractive”, he responded bluntly: “If you think your matches are ugly, that’s probably because you’re ugly.”

In the introduction of her book, Not My Type: Automating Sexual Racism in Online Dating, Williams wrote: “He stated that matches reflect a mathematically generated score that is a combination of several factors: attractiveness scores, how often users send and respond to messages, and how much traffic a particular person generates on the app.”

In today’s world of online dating, it’s easier than ever to seek out love online – an internet connection is all you need to unlock a portal of potential mates. But convenience hasn’t necessarily translated to success. Rates of loneliness in young people are on the rise in Australia, while on social media, horror stories about online dating are common, with many swearing off apps entirely.

Globally, more than 337 million people use dating apps and, with nearly half of Australians aged 18 to 49 using them, it has become the most common way for Aussies to find partners, according to independent research by Choosi.

And while these platforms may promise a democratised world of dating, one in which everyone has an equal chance at finding love, it may be much more engineered than you think.

The architecture of dating apps

Like social media, most dating apps are designed to privilege appearance over personality. Photos are the first thing hopeful singles will typically see, and the process that serves up each match is governed by an elusive, invisible set of rules.

Without even knowing it, users begin interacting with an algorithm – the computing system behind social media apps, often involving AI – as they fill in their information and what they are looking for in a partner.

The first type of algorithm used by apps is content-based, which filters potential matches based on characteristics such as age, location and religious preferences.

Finding love in the matrix of algorithms can be difficult if it works against you.Credit: Stephen Kiprillis

But as Judy Kay, a professor of computer science at the University of Sydney who led a years-long project with a large online dating company, says, “most people don’t do a particularly good job” of truthfully filling out these fields.

“People lie, they lie about their height – we looked at statistics that tell you what’s plausible for the large population – men were saying they were taller than they are, but women were saying they shorter than they are. Women were much more likely to be blonde. Amazing, isn’t it?” she says.

The second type of algorithm, called collaborative filtering, refers to when an algorithm learns your preferences the more time you spend on a platform. In dating app terms, it’s more of the types of people you swipe right on, and less of those lefties.

“Matching the same behaviour in terms of who you like and didn’t like is much more powerful as soon as you’ve done a few of those clicks,” says Kay, noting the algorithm is also judging your performance in the battle of the swipes on other profiles.

“Our algorithm looked at predicting not just who you would like, but who would like back.”

It is in the best interests of dating apps to keep users on their platforms for as long as possible by getting as many matches as they can, particularly if the users aren’t paying for a subscription.

Dating apps also rank their users’ appearance and popularity to determine who to match with each other. “If you have some people who are very popular … you can’t afford to recommend your star-like people to everyone,” says Kay.

Still, dating app platforms are notoriously opaque about how their algorithms work. According to Tinder, potential matches are paired based on how active they are on the app, location, common interests, similar photos and how often profiles in their area have been “liked” or “noped”.

Tinder denies the tracking of users’ social status, religion, and ethnicity.

Hinge, which is owned by the same parent company as Tinder, says its algorithm is designed to show “people you are most likely to want to go on a date with and who is likely to want to go on a date with you ... based on the preferences you set … [and] the profiles you’ve liked or skipped, and how others interact with your profile”.

Sexual racism

Catherine D’Souza recalls the many unprovoked remarks men made towards her about her Indian race while she was on dating apps.

Catherine D’Souza says being called exotic can feel dehumanising.Credit: Kate Geraghty

“One word that springs to mind, was the word exotic. In whatever context, something like that can feel quite dehumanising,” says the 26-year-old, who met her current partner on an app.

“It goes from either being like, ‘I’m not good enough, how I look is not aligning to the [standards] of beauty that I’m seeing’, to then kind of feeling like you’re ... a fetish.”

She says many men would use her race as a pick-up-line, thinking it was flattering, when in reality, it felt like a cloaked racist remark. Or they would rule out dating her because they “preferred” other races.

“The first thing that that person sees about you when they see a picture of you is the colour of your skin.”

D’Souza, who co-created a film, I’d Never Date an Indian Girl, based on a real comment a man made towards an Indian friend, says people online are much more “brazen” than in person.

“There’s a difference between saying, ‘I want to date someone that aligns with my values,’ and ‘I’m not attracted to Indian people.’ Millions and millions of people are Indian. But what makes those people ... different?” she asks.

“There has to be a lot of prejudice attached to the decision to say that you’re not attracted to a whole ethnic group of people.”

Sexual racism refers to an individuals’ choice to date or not date a person purely based on their race. Dr Gene Lim, a research officer at La Trobe University whose PhD looked at sexual racism experienced by Asian men in the gay community, echoes D’Souza’s comments, explaining sexual racism is often couched in the language of preferences.

And he points out that sexual racism isn’t just rejection – it can be fetishisation, too – and can put undue pressure on those it targets.

“For a lot of people, let’s say from African or Middle Eastern backgrounds, the abiding racial stereotype is of hypermasculinity,” he says. “People from those two backgrounds may feel pressure to embody that stereotype.”

Conversely, some people may feel pressure to defy racial stereotypes. Asian men, often seen as more effeminate, might attempt to present in a more masculine way to appear more desirable, Lim says.

Dating apps reflect racial biases in the real world, but their infrastructure can also perpetuate them. And when the algorithm acts upon the behaviour of a user, it can begin to recommend matches based on race.

“Once [the user has] clicked on a few people all of the same race, that’s going to push the algorithm, even if you don’t ask about race,” says Kay.

Dating app Bumble says it actively enforces policies against identity-based hate and harassment, saying they deem “both non-consensual fetishisation and body-shaming as sexual harassment”.

A 2018 study from Cornell University argued that simple design features, such as filters and labels, can create bias against marginalised groups. The paper points to research that shows men who use dating apps heavily are more likely to view multiculturalism as less favourable and sexual racism as acceptable.

Some platforms have taken steps to attempt to reduce racial discrimination, such as Grindr, which removed its “ethnicity” filter in 2020.

But others, like Hinge, still allow users to stipulate racial preferences – a decision they told this masthead “is available to support our users of colour in finding a partner with shared cultural experiences and background”.

Beyond the algorithm, racism is dealt with by moderators who Lim says, “don’t necessarily take complaints relating to racial abuse very seriously ... They don’t seem to ban people for that sort of infraction, especially if they [the user] rationalise it as ‘Oh it’s just my preference.’ ”

Body image

David Milano, a 38-year-old from Melbourne, has struggled with body image since his teens. Milano, who is gay, says men face certain appearance-based pressures that aren’t always spoken about.

“There’s a lot of stigma, and there’s a lot about men talking about the way they look, how much they weigh, and whether they have big muscles. And that was me. Since I was 20, I can’t tell you the number of diets I was on because I felt like I needed to look a certain way.”

Negative body image, which is associated with depression, low self-esteem and the development of eating disorders, disproportionately affects queer men, heterosexual women and trans people.

Milano regularly posts on social media about the importance of mental health and being comfortable in one’s own company. But he says the photo-centric nature of most apps (Milano mainly uses Hinge and Tinder), can make it hard to find a partner who sees beyond the surface.

“I want people to get to know me, not for what I look like on the outside ... but it’s hard to portray that message on a dating app.”

“I’m actually so much happier now,” he says. “And I’m 10 kilograms heavier.”

David Milano, 38, says he struggled with body dysmorphia in his teens but is finally in a good place.Credit: Justin McManus

A recent systematic review from Flinders University analysed 45 studies from around the world to determine the effects of dating apps on body image, mental health and wellbeing.

Strikingly, over 85 per cent of studies reported dating apps had a negative effect on body image, while nearly 50 per cent said dating apps worsened mental health.

PhD candidate Zac Bowman, who led the Flinders University review, thinks dating app platforms should create a way to make the “non-physical attributes” of more importance to users.

Kay says these results are concurrent with the effects of social media on body image and mental health. “A lot of people use social media, especially the sort of people who are looking to use dating apps. I don’t know how you can separate the effect of all of that.”

Lim connects this growing body dissatisfaction to the phenomenon of “looksmaxing”, a trend in which young men in particular are encouraged to maximise their physical appearances.

“The body is seen as a never-ending project of improvement ... this resonates with a lot of the ways people approach dating, by trying to control all these variables in the hope that this amounts to a more attractive prospect on dating apps,” he says.

This, in turn, is fuelling demand for cosmetic enhancements.

“There’s a reason we are seeing a huge uptick in plastic surgery, cosmetic procedures and in things like anabolic steroids. All of this feeds into a lot of insecurity around desirability and access to intimacy,” says Lim.

Some new Australian dating apps, like Hello Darling and Tribal, are attempting to go beyond the appearance-obsessed nature of traditional platforms.

Scott Griffiths, a psychologist at the University of Melbourne who specialises in body image, sees these efforts as a positive step, saying more is needed in addressing the root of the problem by reducing “our entire cultural emphasis on appearance and how we look as being the arbiter of dating decisions”.

D’Souza says dating apps can make people feel less accepted in society because of the way they look.

“I hope that for the people who experience what I experience, which is not feeling accepted by society because of things that are out of your control, I hope that they feel seen.”

Make the most of your health, relationships, fitness and nutrition with our Live Well newsletter. Get it in your inbox every Monday.