Under The Hood X Report Analyzer
Negative Feedback Architecture Guide

How Blocks and Mutes Affect Post Recommendations

A block is one of the strongest negative feedback signals explicitly modeled by the current open-source X recommendation system. The Phoenix model includes a block_author prediction, and the ranking system includes a corresponding negative weight. This means X can predict whether showing a candidate post to a particular user could result in the user blocking its author. (X Algorithm — Phoenix)

But there is an important distinction: A block is not a universal penalty applied to the creator's account. The public code does not establish a rule such as “one block removes X points from your account.”

What Happens Conceptually When a Block Occurs?

The recommendation pipeline can be simplified as:

Candidate Post + Viewer Context
↓
Phoenix
↓
P(Block Author)
↓
Negative scoring
↓
Overall candidate score
↓
Ranking / Selection

The prediction is viewer-specific. If User A is likely to block the author, that candidate can have a different recommendation outcome for User A than for User B.

Does a block reduce your reach? Potentially, for recommendation to that viewer or similar recommendation decisions—but the public source does not establish a universal account-level reach penalty. The current architecture also contains filters for blocked authors. That is different from merely receiving a negative score: a candidate can be removed before or during recommendation processing depending on the applicable filter. (X Algorithm repository)

How Mutes Affect Post Recommendations

A mute is also explicitly represented in the current X recommendation architecture. Phoenix includes a mute_author prediction, while the ranking system includes a corresponding negative weight. (Phoenix source)

The model can therefore estimate: “How likely is this viewer to mute the author if this candidate is recommended?”

Mute vs. Block vs. Other Actions

Action Meaning
LikePositive interest
ReplyPositive interaction
RepostPositive redistribution
Not InterestedDoesn't want this content
MuteDoesn't want this author's content
BlockStronger rejection of the author
ReportPotential policy/safety concern

The recommendation system models these actions separately rather than treating all negative feedback as one signal. (X Algorithm — Phoenix actions)

Does One Mute Hurt an Account?

The open-source code does not prove a universal account-wide penalty. A mute provides information about a particular user's preference toward an author. It can also interact with candidate filtering. The current candidate pipeline documents filters involving muted authors and muted keywords. (X Algorithm repository)

So the better explanation is: A mute tells the recommendation system that this user does not want content from that author, rather than proving that the author has received a global reach penalty.

Under The Hood Takeaways

Blocks: Explicitly modeled as negative recommendation feedback, but one block should not be interpreted as a fixed penalty against an entire account.

Mutes: A negative recommendation signal that can also affect whether content is eligible for a particular user's feed through filtering.

How X Ranks and Recommends Posts

How “Not Interested” Affects Recommendations How Reports Affect Post Visibility How Blocks and Mutes Affect Post Recommendations How Clicks Affect X Reach How Quotes Affect X Reach How Reposts Affect X Reach How Replies Affect X Reach How Likes Affect X Reach Positive vs Negative Signals on X Which Engagements Matter Most on X? How X Predicts Engagement How X Scores Your Posts How X Decides What to Recommend

Official Open-Source References