Whether it’s Facebook’s News Feed or TikTok’s For You page, social media algorithms are constantly making behind-the-scenes decisions to boost certain content – giving rise to the “curated” feeds we’ve all become accustomed to.
But does anyone actually know how these algorithms work? And, more importantly, is there a way to “game” them to see more of the content you want?
Optimising for engagement
In broader computing terms, an algorithm is simply a set of rules that specifies a particular computational procedure.
In a social media context, algorithms (specifically “recommender algorithms”) determine everything from what you’re likely to read, to whom you’re likely to follow, to whether a specific post appears in front of you.
Their main goal is to sustain your attention for as long as possible, in a process called “optimising for engagement”. The more you engage with content on a platform, the more effectively that platform can commodify your attention and target you with ads: its main revenue source.
One of the earliest social media feed algorithms came from Facebook in the mid-2000s. It can be summarised in one sentence:
Sort all of the user’s friend updates – including photos, statuses and more – in reverse chronological order (newer posts first).
Since then, algorithms have become much more powerful and nuanced. They now take myriad factors into consideration to determine how content is promoted. For instance, Twitter’s “For You” recommendation algorithm is based on a neural network that uses about 48 million parameters!
A black box
Imagine a hypothetical user named Basil who follows users and pages that primarily discuss space, dog memes and cooking. Social media algorithms might give Basil recommendations for T-shirts featuring puppies dressed as astronauts.
Although this might seem simple, algorithms are typically “black boxes” that have their inner workings hidden. It’s in the interests of tech companies to keep the recipe for their “secret sauce”, well, a secret.
Trying to “game” an algorithm is like trying to solve a 3D box puzzle without any instructions and without being able to peer inside. You can only use trial-and-error – manipulating the pieces you see on the outside, and gauging the effects on the overall state of the box.
Even when an algorithm’s code is revealed to the public – such as when Twitter released the source code for its recommender algorithm in March – it’s not enough to bend them to one’s will.
Between the sheer complexity of the code, constant tweaks by developers, and the presence of arbitrary design choices (such as explicitly tracking Elon Musk’s tweets), any claims of being able to perfectly “game” an algorithm should be taken with a pinch of salt.
TikTok’s algorithm, in particular, is notoriously powerful yet opaque. A Wall Street Journal investigation found it uses “subtle cues, such as how long you linger on a video” to predict what you’re likely to engage with.
So what can you do?
That said, there are some ways you can try to curate your social media to serve you better.
Since algorithms are powered by your data and social media habits, a good first step is to change these habits and data – or at least understand how they may be shaping your online experience.
1. Engage with content you trust and want more of
Regardless of the kind of feed you want to create, it’s important to follow reliable sources. Basil, who is fascinated by space, knows they would do well to follow NASA and steer clear of users who believe the Moon is made of cheese.
Think critically about the accounts and pages you follow, asking questions such as Who is the author of this content? Do they have authority in this topic? Might they have a bias, or an agenda?
The higher the quality of the content you engage with, the more likely it is that you’ll be recommended similarly valuable content (rather than fake news or nonsense).
Also, you can play to the ethos of “optimising for engagement” by engaging more (and for longer) with the kind of content you want to be recommended. That means liking and sharing it, and actively seeking out similar posts.
2. Be stingy with your information
Secondly, you can be parsimonious in providing your data to platforms. Social media companies know more about you than you think – from your location, to your perceived interests, to your activities outside the app, and even the activities and interests of your social circle!
If you limit the information you provide about yourself, you limit the extent to which the algorithm can target you. It helps to keep your different social media accounts unlinked, and to avoid using the “Login with Facebook” or “Login with Google” options when signing up for a new account.
3. Use your settings
Adjusting your privacy and personalisation settings will further help you avoid being microtargeted through your feed.
Ad blockers and privacy-enhancing browser add-ons can also help. These tools, such as the open-source uBlock Origin and Privacy Badger, help prevent cookies and marketing pixels from “following” your browsing habits as you move between social media and other websites.
4. Get (dis)engaged
A final piece of advice is to simply disengage with content you don’t want in your feed. This means:
- ignoring any posts you aren’t a fan of, or “hiding” them if possible
- taking mindful breaks to avoid “doomscrolling”
- regularly revising who you follow, and making sure this list coincides with what you want from your feed.
So, hypothetically, could Basil unfollow all users and pages unrelated to space, dog memes and cooking to ultimately starve the recommender algorithm of potential ways to distract them?
Well, not exactly. Even if they do this, the algorithm won’t necessarily “forget” all their data: it might still exist in caches or backups. Because of how complex and pervasive algorithms are, you can’t guarantee control over them.
Nonetheless, you shouldn’t let tech giants’ bottom line dictate how you engage with social media. By being aware of how algorithms work, what they’re capable of and what their purpose is, you can make the shift from being a sitting duck for advertisers to an active curator of your own feeds.
- is a Senior Lecturer of Information Systems, School of Computing and Information Systems; and (Honorary) Senior Fellow, Melbourne Law School, The University of Melbourne
- This article first appeared on The Conversation