X is rolling out an algorithm tweak that will boost the visibility of posts and replies from mutual followers — people you follow who follow you back — restoring a basic function of social media that the platform's own AI had quietly discarded. For ordinary Americans, this means less shouting from anonymous strangers and more actual conversation with people you chose to hear from.
Why it matters: For years, the major platforms optimized for engagement, which really meant outrage. An angry reply from a stranger generated the same algorithmic signal as a thoughtful response from a friend. X's head of product, Nikita Bier, put it plainly: the algorithm was missing the programming to prioritize mutuals, and "this resulted in the reply section feeling more like a battleground with people you don't recognize." That battleground wasn't a bug for the old regime — it was the business model. Keep people agitated, keep them clicking, keep them tracked.
The fix sounds simple — show people content from people they actually know — but the backstory reveals how broken the system was. Digital Trends reported that X's own published algorithm code already contained tools for understanding mutual relationships, added in a May release that included mutual-follow graphs and corresponding scores. The data was there. It just wasn't reaching the ranking process, or wasn't weighted properly. X hasn't explained the disconnect — whether it was an omission, a weighting problem, or something else. Either way, the infrastructure existed and sat unused while strangers filled your mentions.
X's Grok-based model predicts whether a post will earn replies or clicks. Those predictions can surface content that keeps people active, but activity and genuine conversation are not the same thing. An algorithm can learn that arguments generate reactions without understanding that everyone involved is miserable. The tweak to prioritize mutuals is an acknowledgment that predicted engagement is a poor proxy for actual human connection.
Engadget framed the change as X hoping to make exchanges "less toxic" and lead to "more meaningful conversations." That's one way to put it. Another: X is reversing years of incentive design that rewarded pile-ons and bad-faith replies because they generated metrics. The old social networks — the ones that felt like communities — understood that who you hear from matters as much as what they say.
Context worth noting: X shut down its Communities feature in April. Bier said it was "used by less than 0.4 percent of users — yet contributed to 80 percent of spam reports, financial scams, and malware on X." The feature was eating team resources while the rest of the app suffered. The mutual-boost tweak is a leaner approach — use the existing follow graph rather than building separate infrastructure that bad actors exploited.
What X hasn't disclosed is the strength of the boost, how far it extends, or how success will be measured. The company has described the desired result without demonstrating it. The open question is whether a "small tweak," as Bier called it, can overcome the massive weight that engagement-optimizing algorithms have built up over years — or whether the strangers still get the microphone when the metrics say they should.
While Meta scrapes your data without consent and Google builds surveillance tools, X is at least trying to point AI toward the conversations people actually want to have. Whether the tweak is enough to break the outrage loop — or just a gesture toward it — is what matters next.