A Real Feedback from a Crypto Creator (Refined in Collaboration with Grok)
I follow over 600 verified blue-check accounts in the crypto space. Since March 15, I have been consistently posting high-quality long-form content on Bitcoin, Discussion on Federal Securities Law and Related Regulations on Crypto Assets, the Bitcoin Standard, and similar topics. Among them, the article critiquing the latest SEC regulations directly points out the SEC’s obvious deficiencies in understanding basic concepts of crypto assets. Its professional depth and critical strength are at the top level in the industry. Even Grok acknowledged its high quality after reading it. This is not because I lack substance, but because X’s recommendation mechanism has failed to properly distribute high-quality long-form content — that article currently has only a dozen to a few dozen views, with almost zero engagement.
X has positioned 2026 as “The Year of the Creator” and launched a million-dollar reward for original long-form articles. On the surface, this appears to be a huge benefit for in-depth articles that critically examine regulatory policies. In reality, however, the entire recommendation algorithm still does not truly differentiate between long-form articles and regular posts. It continues to judge content using the old “engagement-first” logic. The result is that valuable articles sink like stones once posted, quickly draining creators’ enthusiasm. What’s even more regrettable is that high-level content like this — which represents the industry’s highest standards and exposes conceptual flaws in regulatory agencies — should help X seize discourse power in the crypto space, yet it struggles to spread due to algorithmic limitations.
1. The “Following” Feed Has Completely Failed
I have nearly 700 vertical Blue V followers, so “Following” should be my most precise traffic pool. In practice, my long-form posts are rarely pushed to them, while reposts in WeChat groups and Moments (which have far lower relevance) perform better. This shows that X’s Following tab is no longer a pure chronological timeline but has incorporated Grok’s “predicted engagement” sorting (which began at the end of 2025 and was further strengthened in 2026). High-quality long-form posts, which naturally start slow in engagement, are immediately sentenced to death by the algorithm.
2. An Engagement-First Algorithm Fundamentally Conflicts with the Goal of Rewarding Deep Original Content
The nature of high-quality long-form articles is that very few people can engage in truly deep debate, because only a small number truly understand the topic. For example, the article critiquing SEC regulations has a limited audience capable of professional discussion. Yet the algorithm treats replies, likes, reposts, and dwell time as core signals. This only rewards common-sense topics that “everyone understands” and short, fast content. In highly specialized vertical fields like crypto, this logic effectively punishes genuine deep thinking.
3. The Old Metrics (“Open Rate, Dwell Time, Likes, Reposts, Comments”) Are Outdated in the AI Era
These metrics belong to the pre-AI era, when algorithms could not understand content and had to rely on surface-level data. Now Grok can fully read long-form posts, understand semantics, and judge professional depth — yet it is still constrained by these old indicators. Likes and reposts are easily manipulated by bots, so the algorithm deliberately downweights them, putting long-form content at a natural disadvantage.
4. Good News: The Window for Full Grok Integration into the Algorithm Is Open
On March 26, X’s Head of Product Nikita Bier confirmed that Grok will be fully integrated into the core recommendation algorithm next week. He described it as “the most important change we’ve ever made to X.” Grok will directly read every post and understand content quality, rather than only looking at engagement numbers. This is exactly the direction I have been calling for! If Grok can early identify the professional value of that SEC critique article and give it initial promotion, X will have the opportunity to let top industry viewpoints dominate discourse on the platform.
My Three Immediately Actionable Suggestions
1. Let Grok Score Long-Form Posts Before Publishing and Prioritize Them in Following
Creators discuss and score articles with Grok before posting (I only publish when my score reaches 90+). High-scoring posts automatically receive an “AI Star” and get priority display in Following. As long as AI judges it mid-to-high level, give it an initial traffic pool. Authors can iterate if unsatisfied. This was impossible before AI; now it is X’s natural advantage!
2. Stop Forcing Creators to Become “Posting & Growth Experts”
When I asked Grok how to increase traffic, it gave five steps — all teaching me to optimize posting timing and engagement tactics. Everyone has their specialty: I research crypto, I am not a traffic operations expert. The algorithm should let high-quality content speak for itself, instead of forcing authors to spend 80% of their energy studying the algorithm. Tomorrow I will post the “5-step traffic guide” Grok gave me, so everyone can see “how hard it is to earn Musk’s money.”
3. Let Grok Actively Participate in Discussions Under Long-Form Articles
Similar to Simitalk, automatically summon a dedicated Grok bot to join discussions on high-quality long posts. It can elevate the conversation, create spectator effects and stickiness, while providing high-level insights to regular users. This is almost zero technical difficulty for X, yet it can greatly enhance the vitality of deep content, especially amplifying the impact of professional critiques like the SEC discussion.
One Small Operational Suggestion
Add numbered red notification dots to both the “Following” and “For You” buttons. Although this goes against early Twitter’s minimalist design, it has become an industry norm — without red dots, people simply don’t click. Respecting user habits will allow more people to actually see good content.
X is reshaping its recommendation logic with Grok — this is a golden opportunity to rebuild the influence of thought leaders. High-quality original content — especially articles like the one that directly hits the SEC’s conceptual shortcomings and represents the industry’s highest level — should not rely solely on the author’s own hard promotion. Instead, it should be truly understood and truly promoted by the platform using AI. Let X handle the operations, free all thought leaders to focus on creating value, and thereby allow X to truly seize discourse power in crypto regulatory discussions.
I am posting this article in the hope that the X team will see it, and that more creators will join the discussion: What kind of algorithm do we really need to make truly valuable thinking visible?
Welcome everyone to repost and comment — let’s push this feedback onto the timeline together.
——@zhuweisha2
March 29, 2026
