X Adjusts Feed Logic to Surface Mutual Connections in Replies
The platform's head of product says the shift aims to reduce the "battleground" feeling in comment threads by prioritizing users who follow each other.
A Small Change with Familiar Goals
X rolled out a modification to its recommendation system this week that increases the visibility of posts from mutual connections, users who both follow and are followed back. The adjustment targets reply threads specifically, where head of product Nikita Bier acknowledged the platform had been omitting data on mutual relationships. The result, according to Bier, was comment sections that felt "more like a battleground with people you don't recognize."
The terminology is revealing. At DailyTechWire, we've tracked how platforms across the region and beyond grapple with the tension between virality and community cohesion. Algorithmic feeds optimized purely for engagement often surface the most incendiary or high-reaction content, regardless of whether the speaker has any relationship to the viewer. For a platform trying to retain creators and reduce churn, that dynamic becomes a liability.
Bier framed the change as correcting a data gap rather than a strategic pivot, but the timing aligns with X's recent efforts to position itself as a hub for original content and sustained creator activity. Earlier this year, the company overhauled its revenue-sharing model to favor accounts producing original material over those aggregating or reposting. A video editing suite launched earlier this month further signals that the platform sees in-app content production as a retention lever.
The Mechanics of Mutual Weighting
The technical implementation remains opaque. Bier did not specify whether the mutual-follow signal is now a discrete ranking factor, a multiplicative boost applied to existing engagement scores, or a threshold filter that promotes replies from mutuals above a certain percentile of strangers. What is clear is that the platform had been underweighting or ignoring mutual relationships in reply ranking, a design choice that prioritized reach and novelty over familiarity.
In practice, this means a user scrolling through replies to their own post, or to a post from someone they follow, should now see more comments from accounts they have a bidirectional relationship with. The change does not appear to affect the main timeline or the "For You" feed, which continues to blend follows, recommendations, and promoted content according to X's broader engagement models.
Bier also suggested the tweak would "help clusters form around interests more easily," a phrase that gestures toward network effects but stops short of describing how mutual visibility in replies translates to cluster formation. If users begin recognizing the same handles in comment threads, the theory goes, they may be more inclined to engage repeatedly, forming tighter subcommunities around shared topics. Whether that emerges organically or requires additional product scaffolding, such as group-like features or interest tags, remains an open question.
Competitive Context and the Threads Parallel
The move comes as rival platform Threads, operated by Meta, has been iterating aggressively on algorithmic transparency and user control. Last month Threads introduced a feature allowing users to privately tune what appears in their feed, a direct response to complaints that Meta's recommendation engine was surfacing too much content from strangers and not enough from chosen follows. Threads also recently crossed 500 million monthly active users, a milestone that underscores its rapid growth since launch.
X's adjustment can be read as a response to that competitive pressure, though the two platforms are optimizing for somewhat different outcomes. Threads has leaned into opt-in curation and explicit user control, giving individuals sliders and toggles to adjust their own experience. X's approach, at least in this instance, is more paternalistic: the platform identified what it considered a flaw in the algorithm and corrected it without surfacing new controls to users.
Both strategies reflect a broader industry reckoning with the side effects of pure engagement maximization. Platforms that grew by amplifying viral content and algorithmically curated feeds are now facing user fatigue, creator burnout, and regulatory scrutiny over the social costs of those design choices. The pendulum is swinging, however incrementally, toward features that privilege familiarity and intentional community building over raw reach.
Creator Economics and the Retention Problem
X's recent product decisions, including the mutual-follow tweak, the revised compensation model, and the video editor, all point to a single underlying concern: creator retention. Platforms live or die by the health of their creator economies. If the most active users, those producing original posts and driving conversation, conclude that the environment is hostile or unrewarding, they migrate. Threads, Bluesky, Mastodon, and a constellation of smaller networks are all competing for that same cohort.
The battleground metaphor Bier used is apt. When reply sections feel adversarial, when every post invites a flood of criticism or mockery from accounts the poster has never interacted with, the psychological cost of posting rises. Some creators thrive in that environment, but many do not. By surfacing mutuals more prominently, X is attempting to create pockets of relative safety or at least familiarity within the broader platform, reducing the friction of participation.
The revenue-sharing changes earlier this year had a similar aim. By tying payouts more closely to original content rather than aggregation or reposting, X incentivized the kind of labor-intensive work that sustains platform value over time: reporting, analysis, creative production. The video editor, meanwhile, lowers the technical barrier to producing that content natively, keeping users inside the X ecosystem rather than forcing them to edit elsewhere and upload.
Limits and Unknowns
It is too early to measure whether the mutual-follow adjustment will materially change user behavior or sentiment. Algorithmic tweaks often produce subtle effects that are difficult to isolate from other variables, such as seasonal usage patterns, external events, or concurrent product changes. X has not shared any internal metrics on reply engagement, creator retention, or user satisfaction, so outside observers are left to infer impact from anecdotal reports and public discussion.
There is also the question of unintended consequences. If the algorithm now favors mutuals in replies, users with smaller or more closed networks may see their reply sections become echo chambers, dominated by a handful of familiar voices. That could reduce the serendipity and cross-pollination that makes social platforms valuable in the first place. Conversely, users with large mutual networks may see little change, since they already had a high baseline of familiar faces in their replies.
The platform's ongoing challenges with moderation, misinformation, and advertiser relations also complicate any product-level intervention. A friendlier reply section does not address the structural issues that have driven some users and advertisers away from X over the past two years. Algorithmic adjustments are relatively low-cost to implement, but they are not a substitute for the harder work of content policy enforcement and trust-building.
What Comes Next
If X continues down this path, the next logical steps might include more granular controls over reply visibility, the ability to filter replies by mutual status, or features that let users curate their own reply audiences more actively. Other platforms have experimented with reply gating, where the original poster can limit who can comment based on follower status, account age, or other criteria. X already offers some of these controls, but they are not widely adopted, possibly because they require users to opt in for each post.
The mutual-follow signal could also be extended beyond replies into other parts of the product: direct messages, notifications, search results, or even the main timeline. Each of those surfaces presents a different trade-off between discovery and familiarity. The challenge for any platform is calibrating that balance in a way that serves diverse user needs without fragmenting the experience into incompatible modes.
For now, the change is modest in scope and ambition. It addresses a specific pain point, reply threads dominated by strangers, without overhauling the broader architecture of X's recommendation systems. Whether it succeeds in making the platform feel less like a battleground and more like a community will depend on execution details we cannot yet see and on user behavior that is inherently unpredictable. What is clear is that X, like its competitors, is searching for a formula that reconciles the demands of engagement, creator economics, and user well-being. The answers remain elusive.
