How Likes Affect X Reach
A like is one of the engagement actions that X's recommendation system predicts when evaluating a post. But a like is not simply a fixed number of points added to your post's reach.
The current open-source X algorithm includes favorite as one of the predicted actions used by the ranking system. Phoenix predicts the probability that a viewer will take that action, and RankingScorer combines that probability with its corresponding weight. (GitHub)
A Like Is a Prediction, Not a Reach Multiplier
The important distinction is:
The system is closer to:
The current documentation describes the ranking formula as: Final Score = Σ (weight × P(action)) where P(action) is the model's predicted probability of a particular action. (GitHub)
Why Your Likes Can Matter (Personalization)
Likes can indicate that a viewer found a post interesting or valuable. But the recommendation system isn't simply asking: “How many people liked this post?” It's also asking: “How likely is this particular viewer to like this post?”
Phoenix uses the viewer's recent engagement history as a major input when making these predictions. That means the same post can have different predicted like probabilities for different people.
Example Scenario
Post: "10 new AI tools for developers"
→ Higher predicted Like probability
→ Lower predicted Like probability
The post doesn't change. The viewer does.
Likes Aren't the Whole Story
The current Phoenix model predicts many other actions alongside likes, including replies, reposts, quotes, clicks, shares, dwell, follows, and negative feedback. These predictions are combined during ranking. So a post with many likes isn't automatically the highest-ranked post for every user.
Likes can contribute to recommendation ranking, but they don't guarantee additional reach. The post still has to go through the rest of the recommendation pipeline, including candidate retrieval, filtering, scoring, ranking and selection.
Under The Hood Takeaway
A like is a positive predicted action, not a fixed reach multiplier. The important factor is not simply how many likes your post has accumulated, but how likely a particular viewer is to engage with it and how that prediction contributes to the overall ranking score.
How X Ranks and Recommends Posts
Official Open-Source References
- • X / xAI — For You Algorithm repository — current architecture, Phoenix, Thunder, filtering, scoring and candidate pipeline. (GitHub)