Elon Musk delivers on his promises; the X (Twitter) platform has just open-sourced its recommendation algorithm. As one of the largest content platforms in the world, let's take a look at how X's recommendation algorithm works.

Core Logic of X's Recommendation Algorithm

X's recommendation system mainly consists of three steps: candidate generation - scoring and ranking - filtering.

Core Scoring Criteria

The algorithm scores each tweet, and the higher the score, the greater the probability of appearing on the "For You" timeline. The main scoring factors are as follows:

Positive Factors

· Interaction: Extremely high weight. If you can spark discussions (users reply to you, you reply to users), the algorithm considers this very valuable. X's model predicts content that can elicit positive behaviors such as likes, retweets, replies, and shares based on the similarity of users' historical interaction sequences.

· Time Spent: If users click on your tweet, view long images, or stay there reading for two minutes, it indicates high content quality.

· X Premium (Blue Check/Gold Check): Paid users have significant traffic weighting in the algorithm, resulting in 2x - 4x exposure opportunities.

· Video/Image: Compared to plain text, content with native media is more likely to gain time spent, maintaining media diversity.

· Timeliness: Recently published posts have an advantage, as posts older than a certain time will be filtered out.

· Originality: This goes without saying; non-duplicate content is favored, which may also be a reason why some accounts get banned.

Negative Factors

· External Links: This is the biggest pitfall. The algorithm strongly dislikes directing users away from the X platform. Tweets with external links will be significantly downgraded (unless you are a highly influential account).

· Being Blocked/Muted: If someone blocks or hides your tweet, it will severely damage your Reputation Score.

· Marked as "Not Interested": Negative feedback carries a high weight.

· Triggering Negative Signals: Certain words that need to be banned, being unengaged, being reported, etc.

At the same time, X has a community-like concept called "SimClusters," which is the core part of X's algorithm. X categorizes all users and tweets into over 145,000 communities, and the algorithm determines which circle you belong to based on whom you follow and like (e.g., Crypto circle, AI circle, K-pop circle). Only when your content is deemed to belong to the "Crypto circle" and interacts with core users of that circle will it be recommended to more people interested in Crypto who have not yet followed you.

Version Analysis

Let's analyze a recent post with the highest views on the X platform, which hit the algorithm's recommendation points. Dan Koe posted an article titled "How to fix your entire life in 1 day" on January 12. It currently shows 160 million views, a historically rare high-traffic piece on the Twitter platform.

Dan Koe's account has 770,000 followers and is a blue check, so it is not a cold start, giving it some natural advantages in algorithm recommendations: posts from big accounts are more likely to enter fans' "For You" feed, with initial views coming from direct followers. Dan's previous posts also had positive interactions, and the historical signal feedback to the model is inevitably positive, indicating high interaction potential. Referring to the positive factors mentioned above, we find that it hits almost all of them.

· Topic + Timeliness: It's very inspirational, talking about changing life during the New Year, absolutely positive, with no sensitive words, and includes self-improvement language, making it a perfect piece to kick off the New Year.

· Retention Time: It's a long article; although it lacks multimedia forms like images and videos, long articles can still stand out from short tweets, increasing retention time.

· High Interaction + Community Spillover: High interaction comes from initial followers, and after achieving effects within the community, universally appealing topics are very suitable for recommending to non-follower users. Then continue to maintain high interaction, creating what is known as a flywheel effect.

How to Build a Presence in the Cryptocurrency Industry

So when writing cryptocurrency content on X, the worst approach is "Bitcoin is surging, check out this analysis: [link]," which is full of pitfalls and will not be recommended.

In terms of format, it's best to have Threads, meaning threads of tweets, and include images, as these are ways to enhance user retention time. In terms of content, you can guide user interaction and also leverage X's community circles, such as purchasing a blue check and commenting under core accounts in the circle. If you get a reply, it can also be marked by the platform, boosting your account's weight. Avoid behaviors like "following a large number of people in a short time" and "tweets with a lot of hashtags," which appear robotic.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。