We asked Grok—the AI built into X—to analyse our account data to determine if the “free speech” platform is suppressing our content, and the results confirm what we’ve long suspected. Caldron Pool, with nearly 35,000 followers on X, shows an 85-95% likelihood of algorithmic suppression, marked by low visibility, buried replies, and engagement levels so low they resemble those of accounts with fewer than 1,000 followers.

According to Grok’s review, we post between seven and ten times per day — mostly links to caldronpool.com and commentary on topics like religious freedom, cultural decay, and government overreach. The AI found no evidence of rule violations related to spam or harassment.

Yet, despite consistent posting and a sizeable audience, engagement is drastically below normal.

From 1 August–1 October 17, our 500 most recent posts averaged:

• 842 views (median 434) — some as low as 17 after 12 hours.
• 16.4 likes, with roughly one-third of posts getting five or fewer.
• 2.6 reposts (median 1).
• 1.1 replies per post, mostly from low-follower accounts.
• 0.5 quotes and 0.7 bookmarks on average.

By contrast, an account of similar size and activity should be averaging 1,000–3,000 views and 350–700 likes per post, based on standard social media analytics. Caldron Pool’s engagement rates sit at just 2–3% of that expected range — roughly 5–10 times lower than platform norms.

Even accounting for niche subject matter, the figures don’t match normal audience behaviour.
In fact, the average performance mirrors that of an account with under 1,000 followers.

A post with 17 views after half a day is statistically comparable to an account with hundreds of followers — not 35,000. High-performing posts, such as a video on free speech that drew 7,005 views and 187 likes, are rare exceptions that inflate the average but do not reflect consistent reach.

Visual posts (videos and memes) performed best, while article links — the backbone of Caldron Pool’s content — languished around 200–400 views. According to Grok, that’s a sign of “selective visibility,” where the algorithm boosts “safe” content while quietly burying posts on polarising topics.

Engagement also shows flat-low consistency across time and topic — a classic indicator of algorithmic throttling rather than organic audience response.

The low engagement is not easily explained by user behaviour. Caldron Pool doesn’t spam, overpost, or abuse hashtags. Its activity is steady and within normal limits for media outlets.

But certain patterns point toward algorithmic suppression:

• Buried Replies: Many responses appear hidden under “show more” prompts, reducing visibility and engagement.
• Flat-Low Metrics: Engagement doesn’t fluctuate with timing or topic — a sign that posts aren’t being distributed to follower feeds.
• Topic Filtering: Content about Christian nationalism, religious persecution, and government censorship may trigger X’s “divisive content” filters, reducing exposure under the platform’s “freedom of speech, not freedom of reach” policy.

For Christian and conservative publishers, Grok’s findings echo a familiar frustration — the sense that their speech is tolerated, but intentionally hidden.

Caldron Pool’s expected engagement (1–2% of followers for likes, 2–5% for views) would translate to hundreds of likes and thousands of views per post. Instead, the outlet is getting a fraction of that — 0.05–0.5% for likes and 2–3% for views.

Grok concluded that while niche content and audience behaviour play a small role, the consistent shortfall, hidden replies, and isolated high-performers indicate “algorithmic deboosting.”

What makes these results even more staggering is that Caldron Pool is subscribed to X Premium, a paid subscription that is said to boost the visibility of the user’s posts by giving them priority in search results, mentions, timelines, and recommendations.

Despite this, the AI estimates an 85–95% likelihood that X’s algorithm is suppressing the account, likely through visibility throttling tied to sensitive cultural or political themes.

In short: the content’s still there. Our audience just isn’t allowed to see it.