X Algorithm Open Source: Complete Guide to the For You Feed

Introduction

X just open-sourced part of their recommendation algorithm in January 2026, revealing exactly how the "For You" feed decides what to show you.^[1] This isn't just another vague platform update. It's a detailed look at the scoring systems, filters, and ranking mechanisms that determine whether your posts reach thousands or disappear into the void.

If you're writing on X, understanding how this system works is essential. Your posts aren't floating around hoping to get noticed. They're competing in a specific process, and knowing that process helps you write better content.

In this guide, we'll cover all the changes and their impact, then provide practical advice on how to write better for the new algorithm.

How the X Algorithm Works: The Basic Pipeline

Every time someone opens their "For You" feed, X runs the same process. Understanding this pipeline is crucial because each stage represents an opportunity for your content to either advance or get filtered out.

The Four-Stage Process

1. Candidate SourcingThe system gathers potential posts from two places: "In-network" content from accounts you follow, and "Out-of-network" content from accounts you don't follow, discovered through machine learning.
2. Feature AdditionThe system fetches additional details about each post, including video duration, engagement numbers, and other metadata needed for ranking decisions.
3. FilteringThis is where many posts get eliminated. Blocked accounts, muted keywords, already-seen content, and posts that are too old all get removed before scoring.
4. Scoring & SelectionSurviving posts are scored using a neural network with approximately 48 million parameters. The highest-scoring posts make it to your feed.^[2]

What the Algorithm Predicts About Your Posts

The algorithm doesn't guess whether your post is "good." It predicts specific actions that each viewer might take. This is a crucial distinction because your post gets a different score for every potential viewer.

Engagement Actions and Their Weights

Based on analysis of X's open-source code, here's what we know about how different actions are weighted:^[3]

The algorithm uses a scoring formula that looks something like this:

Score = 0.5 × P(Like) + 1.0 × P(Retweet) + 0.3 × P(Reply) + 0.15 × P(Profile_Click) - 1.5 × P(Report) - 3.0 × P(Block)

Notice the negative weights for adverse actions. X explicitly penalizes content that users actively dislike. A single block or report can significantly damage your post's reach.

Higher-Value Actions

The system also predicts quote tweets, video views, shares (including copying the link or sending via DM), and whether someone will follow you after seeing the post. These are strong signals that your content provides genuine value.

The Filters That Kill Posts Before They Get Scored

Even if your post would score perfectly, it can get filtered out for certain viewers before scoring even happens. Understanding these filters helps you avoid invisible reach killers.

Key Filters to Understand

• Viewer Blocks and Mutes: If someone has blocked or muted you, your posts don't get considered for them at all.
• Muted Keywords: If your post contains a keyword someone has muted, it's filtered out for them. Packing posts with controversial or commonly muted terms can kill reach.
• Already Seen Posts: The system checks for exact duplicates, similar reposts, and posts within the same conversation thread.
• Age Cutoff: Posts older than a certain threshold get filtered out completely. This is a hard cutoff, not a gradual decay.^[4]
• Subscription Gates: Subscribers-only content won't be recommended to non-subscribers.
• Visibility Filtering: Policy violations and sensitive content get filtered or downranked.

In-Network vs Out-of-Network Content

The feed blends two types of content, and they're treated differently. Understanding this distinction is crucial for your content strategy.

In-Network Content

These are posts from accounts someone follows. They're retrieved from a post store tracking recent content. If you have followers who regularly engage with you, you start with a built-in advantage for reaching them.

The most important component for in-network ranking is the Real Graph model, which predicts engagement likelihood between two users. The higher this score between you and someone who follows you, the more of your posts they'll see.

Out-of-Network Content

These are posts from accounts the viewer doesn't follow, discovered through machine learning. The system tries to match posts based on topic, engagement patterns, and semantic meaning.

X uses SimClusters, which discovers communities anchored by influential users. If your content resonates with a specific community, you have a better chance of reaching non-followers who share those interests.

Author Diversity Scoring

Even if you post amazing content, the algorithm limits how often you appear in any single feed refresh. This "within-response diversity" means your posts compete against your own previous posts, not just other creators.

What We Don't Know (and Why It Matters)

The open source release explicitly excluded some critical pieces:

• Exact weights are missing: We know the system combines predicted actions, but exact multipliers aren't confirmed.
• Thresholds are hidden: The exact age cutoff, top selection count, and other key settings aren't public.
• Model weights not included: The architecture is public but the trained weights that make predictions aren't.
• Some services excluded: Model serving, caching, and additional features aren't in the public code.

This means everything we recommend should be treated as informed best practices, not guaranteed formulas.

10 Strategies to Optimize for the X Algorithm

Now that you understand how the system works, here's how to write content that performs.

Practical Writing Tips for the New Algorithm

Write for Specific Actions

Your post isn't just competing for views. It's competing on predicted engagement. Make it obvious what action someone should take:

• Ask clear questions: Questions naturally invite replies. Don't make them rhetorical.
• Prompt specific responses: A strong stance with reasoning makes people want to agree, disagree, or add nuance.
• Use two-option framing: "Would you rather A or B?" gets more replies than open-ended questions.
• End with an invitation: "What's your experience with this?" explicitly tells people to engage.

Understand the Time Window

Your post has a limited window to prove itself:

• Early engagement drives everything. If your post doesn't get traction in the first hour or few hours, it ages out.
• Don't expect late success. Starting a new conversation usually beats hoping an old one revives.
• Use scheduling tools to post at optimal times when you can't be online.

Handle Threads Strategically

• The first post needs to hook. If your thread starter doesn't perform, the rest won't get distribution.
• Each reply in a thread is a separate candidate. Make each tweet valuable enough to stand alone.
• The system picks the "best branch" of a conversation to promote.

What Not to Do

Some things are actively counterproductive:

• Don't pack posts with trending keywords just because they're trending. If those keywords are commonly muted, you're filtering yourself out.
• Don't spam the same content repeatedly. Duplicate detection exists specifically to prevent this.
• Don't expect subscription content to get broad reach. The filter is explicit.
• Don't post important content when your audience isn't around. Timing matters because of the age cutoff.
• Don't chase negative engagement. Blocks and reports actively hurt your score.

Frequently Asked Questions

Conclusion

The X algorithm is ultimately designed to serve high-quality content to the right people. Your job is to make it easy for the system to match your posts with interested viewers, predict high engagement, and survive filtering.

Focus on clarity of communication and writing to provide value. When you create content that genuinely helps or entertains your audience, the algorithm becomes your ally rather than your obstacle.

The key principles are simple: post when your audience is active, write for replies and shares not just likes, make your topic obvious, and serve your existing followers well. Do these consistently, and the algorithm will help your content reach the right people.