Social Media Facebook Algorithms

How do social media figure out what we like?

You’ve probably already noticed that social networks tend to suggest stuff that might be interesting to you. Sure, sometimes you get some really weird suggestions from Wish, but you probably found some really cool pages and even groups this way on Facebook, some interesting profiles on Twitter and Instagram, and maybe some interesting companies on LinkedIn.

But how does all that work? And, most importantly, why are things like that? Why did social networks change from a place to keep in touch with your friends and relatives into a way to discover new stuff you like and find like-minded people? Let’s take a look.

The Facebook business model

As you know, Facebook is free for users. You can create a profile, create groups and pages, connect with people, among other stuff, with no limitations.

So, where does the money come from?

From advertisement.

If you’ve ever created a Facebook page, you may have noticed you get immediately bombarded with notifications about creating ads for your page. For a rather small sum, you can promote any of your page’s posts and select which kind of people you want to reach: age range, gender, preferences, political positions, among many others. You can easily advertise even the most niche of pages.

The algorithm

But where does all of that information come from? Well, part of it comes directly from the user: date of birth, gender, relationship status, family, locations they went to and all that. Even if that information is restricted to yourself or just your friends in your profile, Facebook still has access to it and uses it to direct ads too.

Information about preferences also comes from you, but not directly. Facebook employs a machine learning algorithm designed to analyze the pages you like, the groups you are in, your posts and even your photos to find out what you like and what you don’t like.

Of course, finding out people’s preferences from their Facebook likes is no rocket science. If you enter someone’s profile and find a lot of likes on pages about hiking, then they clearly like hiking a lot. However, as of 2020 Facebook has more than 2.8 billion accounts. Nobody could possibly go through every single one of those everyday to update every user’s preferences in Facebook’s database.

Instead, the algorithm keeps track of your information and compares it with the patterns it already knows, and the usage patterns that other users have, and assigns to you some of the several categories it has figured out, while also creating and assigning more of those categories as time goes on.

Although Facebook jump-started this technology, most other social networks have adopted it too, such as Twitter and Instagram, allow their user base to grow exponentially and, with that, attract investments from many different kinds of companies.

Results

Of course, the technology is good, but it isn’t perfect. If you go to your Facebook profile’s privacy settings, you can find your ad preference settings, and within it the list of categories that Facebook has assigned to you. Most of them will probably be spot on, but others will be nonsense and some will probably be outdated already. It may, however, explain the kinds of ads you’ve been getting. Keep in mind that Facebook also tracks you across many of the websites you visit!

In some way, one may also regard the technology as being too good. During the last few years Facebook has been facing scrutiny for using the algorithm to do social experiments, including only showing to users the kind of political posts which would already agree with their perspectives. And also, for allowing other companies to use all that data for their own ends, which allegedly has led to more political extremism. Remember the Cambridge Analytica controversy? Yeah, that kind of stuff.

These kinds of things show the power of technology today. Even an imperfect machine learning algorithm may be able to sway people’s opinion. Who knows what may come next?