X reply deboost and the Show more replies fold

On a busy post, X does not render every reply in a single flat list ordered by time. It ranks them. Replies from people you follow, replies with engagement, and replies the systems consider relevant tend to stay up. Replies that look like spam, that come from accounts you have no relationship with, or that simply have no engagement yet tend to sit in an extra group behind a control labeled Show more replies, or a close variant of that label.

Most replies on a popular post live in that lower group. That is ranking, and it happens to accounts that nobody would call banned. If your test of 'am I deboosted' is 'I replied to a huge account and I was not first,' you will diagnose yourself with a ban every time. The lower group is the default for strangers.

X has also described downranking a post in replies as an explicit restricted-reach action, alongside removing posts from search or limiting them to the author's profile. So the same screen covers two different stories: ordinary ranking, and a downrank applied because a label or a spam score told the product to push you down. You cannot see the score. You can see whether the downrank is a pattern.

Authors are a bad judge of the default view. Your own reply is often pinned into your personal view of the thread so you can find what you just wrote. Log out before you decide where strangers will find it. The steps are the reply half of the manual shadowban test.

The pattern worth acting on looks like this. You reply, with a normal sentence and no link, to several smaller public posts on the same day. Logged out, each of those replies is absent from the first screen and present as soon as you expand more replies. Meanwhile other fresh replies from unrelated accounts are on the first screen. Your reply is not missing. It is consistently demoted.

Contrast that with a ghost. A ghost ban means the expanded list still does not contain you, even though your profile shows the reply. If you find yourself after one click, do not use the ghost label. The cleanup is similar, because both are reply-spam patterns, but the test you repeat later is different. For deboost, success means the reply can appear in the primary list again on a quiet thread. It does not mean you will rank above the author's mutuals on a viral post.

• One reply under Show more on a viral post: expected ranking.
• The same placement on several small posts, while other new replies stay visible: a deboost pattern.
• Missing even after Show more, present on your profile: ghost pattern, different guide.
• Missing from your own logged-in view: the reply did not land in that thread. Not a deboost.

Ranking uses relationship and behavior because that is how a reply list stays readable. An account that replies dozens of times an hour to non-followers, always with a link, will lose that ranking even if no human reviewer looked at it. The same thing can happen after reports on a reply that people marked as spam.

• Reply volume that spikes, especially to accounts that do not follow you.
• Repeated text. Ranking and spam systems both treat duplicates as low value.
• Links in replies. A link is not forbidden. A link as the whole reply, repeated, is a classic spam shape.
• Low reply reputation: new accounts, or accounts whose recent replies were hidden or reported.
• The conversation itself. A post with thousands of replies will bury you no matter how healthy the account is.

None of these requires a conspiracy or a personal decision by X staff. They also do not require you to have broken a written rule in an obvious way. A promotion habit that felt normal, replying to every large account in a niche with the same pitch, is exactly the shape these systems are built to push down.

Content labels can downrank too. If the reply contains media that X has marked sensitive, the warning itself changes who bothers to open it, and the label can carry a visibility effect. Check the media before you blame reply ranking alone. Details are in the sensitive media guide.

Choose parent posts that are public, not yours, and not exploding. A few dozen replies is a better lab than a post on the explore page. Reply with a unique phrase and no link. Wait a few minutes. Open the parent URL logged out. Note whether you are on the first screen. Expand more replies and note whether you appear.

Do this three times with three different authors. Keep a one-line record: parent link, first screen or not, visible after expand or not. A deboost pattern is 'not first, visible after expand' on more than one of those tests. A single 'not first' on a busy post is the product working as designed.

While you are there, glance at search so you do not mix up two limits. A logged-out from:handle search can be healthy while replies are deboosted. If search is also empty, you have a second problem, covered in the search ban guide. The checker separates these probes so a reply warning does not get misread as a search ban. An error on the reply card means the probe did not complete.

Reduce the behavior that looks like reply spam, then give ranking something normal to score. For several days, reply less often, only when you have something specific to say, and not with a copied line. Prefer conversations where people already interact with you. Leave links out of replies until a retest looks normal. Verify the account's phone and email if that is still unfinished.

Do not buy replies, likes, or followers to 'push the reply up.' Purchased engagement is its own spam signal and can make the next ranking decision worse. Do not create a fresh account solely to escape a deboost. New accounts start with less reply reputation, which is the resource you are trying to rebuild.

Retest with the same three-post method, not with a screenshot of one viral thread. You are done when quiet threads show your reply without requiring an extra click, at least often enough that the earlier pattern is gone. You are not done when you outrank the poster's friends. There is no official number of days for this. How long shadowbans last is about that gap. The practical sequence sits inside how to recover.

Reply rank is per viewer and per post. A logged-out test is a fair public sample. It is not your follower's app, and it is not the author's app. Two checkers will disagree if one of them never expands Show more and calls every lower reply a ghost. Ask that question before you trust a red badge.

X's Under the Hood report, for eligible accounts, can show that labels affected visibility in the prior month. It does not show the rank position of a reply you wrote today, and it does not use the phrase reply deboost. Live placement in the thread is still the test that matches this symptom. Read Under the Hood vs checkers before you treat a monthly label count as a live ranking score.