Why social media algorithms make you angry

Social media algorithms can make you angry because they rank your feed by predicted engagement, not by time, and content that provokes strong emotion, especially outrage, tends to get more replies, reposts, and attention, so the system pushes it to the top. Newsreel is built the other way: a news platform where you follow journalists you choose and read their work in the order they publish it, with no engagement-ranking algorithm and no ads.

How an engagement-ranking feed works

A social feed does not show you posts in the order they were written. It shows you the posts a computer program predicts you are most likely to react to. That program is the algorithm, and the useful thing about X publishing its code is that you no longer have to guess how it behaves.

X first open-sourced part of its recommendation system in March 2023, and later released a fuller version of the "For You" ranking code in a public repository called xai-org/x-algorithm. The code lays out the feed in two broad stages. First, the system gathers a large pool of candidate posts, some from accounts you follow and many from across the platform. Then a ranking model scores each candidate and sorts them.

For every candidate post, the model predicts the probability of each way you might react: that you will like it, reply to it, repost it, or stop and read it. It then combines those predictions into a single number, roughly a weighted sum of the predicted actions, and the posts with the highest numbers go to the top of your feed. The system is also tuned to notice signals that you did not like something, such as marking a post "not interested." But the center of the machine is a straightforward idea: predict what will get a reaction, then show you that first.

Why outrage rises to the top

If a feed is ordered by predicted engagement, then whatever gets the most engagement wins the top spots. And the kind of content that reliably gets a reaction is not calm, balanced, or nuanced. It is content that makes you feel something strong, fast.

Researchers who study social media have found the same pattern across platforms: posts carrying moral and emotional language, and outrage in particular, tend to spread further than neutral posts. Anger drives replies, because people argue back. It drives reposts, because people share what makes them furious. It holds attention, because it is hard to look away from a fight. Every one of those is exactly the behavior an engagement model is built to predict and reward.

The accurate reading of the code is not that engineers set out to make you angry. There is no line that says "maximize rage." The systems are built to maximize engagement, and anger turns out to be one of the most engaging things a human can feel. The outrage is a side effect of the goal, not the goal itself. That distinction matters, because it means the problem is structural. Any feed that ranks by reaction, run by anyone, will tend to surface the content that provokes the strongest reaction, and that content skews angry.

Researchers have documented a further wrinkle: outrage travels fast even when the underlying claim is thin or false. Studies of misinformation have found that posts from unreliable sources draw more angry reactions than happy or sad ones, and that people reshare outrage-provoking content without reading past the headline. A feed sorted for engagement does not check whether a claim is true. It checks whether you will react.

What you can do about it

You cannot rewrite a platform's ranking model, but you can change your relationship to it.

• Notice the feeling before the action. If a post makes you instantly furious, that is often the signal the algorithm selected it. Pause before you reply or repost.
• Check the source before you share. Outrage spreads fastest when no one stops to ask whether the claim is real. A quick look at who is reporting it is the cheapest defense there is.
• Prefer feeds you control. A chronological feed, a set of newsletters, or a following-based reader hands the ordering back to you instead of to an engagement model.
• Follow people, not the algorithm. When you choose the voices you trust and read them directly, the incentive to bait you disappears.

A feed built the other way

This is the problem Newsreel is built to answer. Newsreel is a news platform, on the web and in an app, for people who have given up on the news. You follow the journalists and experts you choose, and you see their work in the order they publish it. There is no engagement-ranking algorithm sorting your feed for outrage, and there are no ads pulling to keep you scrolling. Every contributor is a vetted journalist or a vetted expert, not an anonymous account chasing a reaction.

When a story is contested, a feature called Perspectives clusters the actual arguments people are making about it in real time, so you can see more than one side laid out instead of only the angriest take. Newsreel was named the 2025 Resource of the Year by the National Association for Media Literacy Education. We make Newsreel, so we have said this plainly: the point is a feed that informs you rather than one tuned to keep you upset.

Frequently asked questions

Why do social media algorithms make you angry?

Most social feeds are sorted by predicted engagement, not by time. The system estimates how likely you are to reply, repost, like, or keep looking, then shows the posts it expects to get the most reaction. Content that provokes strong emotion, especially anger and outrage, tends to draw more of those reactions, so an engagement-ranked feed pushes it toward the top. The anger is a side effect of optimizing for engagement, not usually a deliberate goal.

Is the algorithm designed to make you angry on purpose?

Not usually. The accurate way to describe it is that these systems are built to maximize engagement, and anger happens to be one of the most engaging emotions. Researchers who study social media have found that moral and emotional content, including outrage, tends to spread further than neutral content. When a ranking system rewards whatever gets the most reaction, outrage rises even if no one set out to make people angry.

What did the open-sourced X algorithm reveal?

X, formerly Twitter, first open-sourced part of its recommendation code in March 2023, and later published a fuller version of the For You ranking system in the xai-org/x-algorithm repository. The code shows the feed working in two broad stages. First it gathers candidate posts from accounts you follow and from across the platform. Then a ranking model predicts the probability of each engagement action, such as a like, reply, or repost, and combines those predictions into a single score that decides the order. It is a plain look at how an engagement-ranking feed is built.

How is Newsreel different from an algorithmic feed?

On Newsreel you follow the journalists and experts you choose, and you see their work in the order they publish it. There is no engagement-ranking algorithm sorting your feed for outrage, and there are no ads. Every contributor is a vetted journalist or a vetted expert. A feature called Perspectives clusters the arguments on a contested story in real time so you can see more than one side. We make Newsreel, so we have said this plainly.