Stephen McDermott

THE ALGORITHM WON’T stop showing me videos of Alysa Liu, the double gold-winning American Olympic figure skater.

It’s just over two weeks since Liu’s achievement at this year’s games, which catapulted her to mega-stardom in her home country and into social media feeds around the world.

During that time, X has been feeding me a steady stream of posts about her from accounts I don’t follow in its ‘For You’ tab.

For those unfamiliar with them, ‘For You’ tabs are personalised, algorithm-driven feeds which push recommended content based on an individual user’s engagement history, location, and personal interests.

They aren’t unique to X (platforms like TikTok and BlueSky also have their own versions) and are separate to ‘following’ tabs that only display posts from accounts a user follows in chronological order.

In my case, my ‘For You’ tab on X tends to be tailored to me as a man in his mid-30s who, as The Journal’s FactCheck Editor, primarily uses the platform for work and occasionally to get insights into live sport and reactions to big cultural moments.

The posts about Liu first grabbed my attention because her distinct halo hairstyle stood out in my feed, something that’s generally populated by snooker clips, culture war discourse and Irish news content.

I had never used X to watch figure skating, so it was mildly perplexing to be inundated with so many videos and images of Liu specifically.

Far less surprising was how few of the posts had to do with skating; instead, they ranged from minor expressions of attraction to full-on fetishisation and lascivious – or dismissive – commentary about her looks.

It’s not the first time this type of content has been pushed into my ‘For You’ feed by X’s algorithm.

The same pattern has previously played out with the actor Sydney Sweeney and singer Sabrina Carpenter, neither of whom I have a particular interest in but both of whom X has decided I want to see images and videos of over several months.

Alysa Liu with her individual gold medal Alamy Stock Photo Alamy Stock Photo

The persistence of this trend shows how social media has evolved over two decades, and how platforms are now pushing sordid content into people’s feeds to keep them engaged, even when users haven’t been looking for it.

Liu’s individual gold in Milan, which saw her become the first woman to do so for the United States since 2006, was a case study in how algorithms work to recommend content.

When people start posting about or searching for someone like Liu a lot, platforms begin digesting posts that provoke engagement, then feed them and similar posts back to other users.

In Liu’s case, her popularity spiked the day after she won her second gold, prompting X’s algorithms to push its users more content about her in the days afterwards.

The origin story of social media is, in large part, the story of the male gaze; so it was grimly predictable that within days of Liu going viral for her remarkable performances, she became the subject of sexualised commentary on X.

There were posts about Liu making silly faces with her gold medal with captions describing her as “goonbait” (internet slang that basically means sexual clickbait).

Multiple posts from an account with the handle @celebfaceporn shared images of Liu with her arms aloft with the caption “Alysa Liu armpits” (which, from Google searches, appears to cater to a specific type of fetish known as axillism).

There were, typically, posts which described her less flatteringly as “America’s beautiful mid” (someone whose looks are deemed to be average or mediocre) and which said that she “would be a legit 10/10 without that stupid shit in her teeth”, a reference to her frenulum piercing.

X’s algorithm had recognised that the initial, more innocuous posts about Liu were attracting a lot of engagement, and presumably pushed them into the feeds of licentiously minded men.

The posts shared by those men then began attracting their own engagement, prompting the algorithm to amplify them into the ‘For You’ feeds of other men, including mine.

As a sign of how far the trend has come in less than two weeks, one post on Friday racked up millions of views within hours: it showed an AI-generated sexualised image of a woman resembling Liu, sitting on a locker room bench, wearing her gold medal and skating costume with her bare feet prominently displayed.

The AI-generated clip of Liu that went viral on Friday

This stuff is red meat to X’s algorithms and the ‘For You’ tab, which operate on a simple premise: show people content that is most likely to make them stop scrolling.

Sexualised and divisive posts are particularly effective at this, because they invite waves of approval or backlash that give users a reason to read, share or respond to them.

The process is self-perpetuating: if a user hovers over or clicks into a certain type of post, they begin to get more of that type of content recommended to them, which they may then click into again.

There is also a financial incentive for some accounts to go viral no matter what they churn out.

Under X’s revenue-sharing scheme, users who subscribe to X Premium can earn money when their posts generate large amounts of engagement, which means their content can translate directly into payment.

It shows how extreme discourse on social media seeps into conversations in the media and wider society that don’t always represent widely held views in the real world.

And in this specific instance, it’s easy to see how algorithms feed the popularity of the manosphere, the loose network of online male communities that thrive on misogyny and anti-feminist beliefs.

Social media has always been driven by strong reactions, but the increasing prominence of recommendation tabs driven by algorithms mean that platforms are no longer simply hosting conversations – they are actively curating them.

If I were to rely solely on what X has fed me, my entire knowledge of Alysa Liu would amount to what other men think about how she looks. 

I presume – though can’t be certain – that part of this is because I use X to monitor extreme content online; either way, the platform has given me an entirely different view of Liu compared to others who aren’t in my demographic or tend not to seek out that stuff.

It’s an outcome that says less about what I want to see as it does about how X wants us to see the world to keep us engaged.