Preface: communication studies, part 3.

4) Instagram

To understand how Instagram distributes content, start from one premise: it isn't chasing the strongest stimulus. It's trying to keep the user experience stable and predictable. The algorithm's goal is not to produce as many viral hits as possible, but to pick out, for each user, the accounts and content they're likely to want to see again and again. To do that, the system makes judgments on roughly four levels: relationship, identity, time, and only last, the content itself.

The first level is relationship signals. Following an account is only the minimum threshold. Instagram treats a follow as a permission, not as proof that a relationship exists. What sets distribution priority is whether low-friction, unforced exchanges of attention keep happening after that: DMs, mutual likes, mutual comments, repeat views, profile visits, saves, shares sent to a specific person. These behaviors tell the system that the attention channel between you and an account is still open.

These signals don't carry equal weight. Public likes and passing comments are weak, closer to social gesture. DMs, saves, targeted shares, and repeat viewing are stronger because they sit closer to real need and better predict whether someone will keep paying attention. So when you post, the system usually looks first at existing relationships and interaction history to judge who is most likely to stop, watch to the end, or engage again. If those channels have been "warm" recently, even ordinary content tends to reach people reliably. If they've gone cold, even polished work gets distributed more cautiously.

This explains a common pattern: the quality of the content hasn't noticeably dropped, but reach keeps shrinking. The cause usually isn't the content. It's that the relationship signals have cooled. Switching topics, changing rhythm, disappearing for a while, or cutting back on two-way interaction all tell the system that the old channels are becoming harder to reach. Distribution drops accordingly. This isn't punishment. It's resource conservation: the platform won't push content to people who have long since stopped responding.

The second level is recognizable identity. Instagram assumes that one account corresponds to one lasting person, not a series of unrelated content experiments. The system keeps asking the same question: when users scroll past you, can they quickly tell who you are, roughly what you do, and why they should keep watching?

Consistent identity doesn't mean the content has to be narrow. Growth, expansion, and change of direction are all allowed, as long as the change stays within a range people can follow. When subject, tone, values, and visual language all break at once, both the system and the audience have to recalibrate. The algorithm then has a harder time deciding who to recommend you to, and a harder time predicting whether you still fit what your existing audience expects. Usually it narrows distribution first and watches how your core audience responds, rather than pushing that uncertainty onto strangers too early. This is why Instagram isn't friendly to frequent "persona switching." Shrinking distribution isn't a verdict on the content. The platform is waiting for the account to settle into a stable signal again.

The third level is continuity over time. Instagram cares a great deal about whether an account looks like "one person who keeps showing up." It doesn't demand frequent posting. It demands a rhythm people can anticipate. One post a week, or even slower, is fine as long as it reads as steady output from a life being lived. Irregular disappearances, sudden bursts, and then another silence all read as uncertainty in how the relationship is being maintained. When an account returns after a long gap, the algorithm usually watches first: is this a brief appearance, or a real return?

Only the fourth level is content feedback. On Instagram it mostly reinforces relationships rather than deciding whether something takes off. Likes, comments, and saves matter less for their numbers than for whether they come from relationships that already exist. Saves in particular are read as a private-use signal, a sign that the post serves a practical purpose in someone's life. When the same group of people keeps interacting with you, the system doesn't read "this post is exciting." It reads "this account is useful in some people's daily lives." That matters more than a one-time spike in engagement, because it signals sustainability.

Discovery and spread do exist, but they're conservative. Explore and Reels work more like risk-controlled amplifiers than testing grounds. Usually the system becomes more willing to introduce an account to strangers only after its relationships, identity, and rhythm have stabilized. That spread works more like a reward than a test. For the same reason, Instagram is wary of gambler-style growth. Chasing trends, sudden pivots, abrupt persona changes may produce short-term numbers, but they rarely earn long-term credit. The platform would rather slowly amplify a predictable account than bet on a high-uncertainty hit.

Taken together, Instagram is closer to a display-oriented social network built on relationships and identity. Who you are and who you interact with often matter more than any single post. Content mainly serves to keep your self-image updated and your social presence alive. Recommendation keeps getting stronger in Explore and Reels, but in the core spaces, Feed and Stories, interaction history, follow relationships, and continuity of identity still dominate. What it measures is not explosive force but compatibility: whether an account can be encountered again and again without interrupting the user's rhythm. Here, "being remembered" doesn't mean being remembered by large numbers of strangers. It means the system has judged that you are already part of a small group of people's daily lives. For creators, exposure therefore tends to be fairly even. Change happens mostly between different entry points, not as sudden platform-wide amplification or wipeout.

5) Xiaohongshu (RED) vs. Instagram

If Instagram organizes distribution around relationships, Xiaohongshu organizes it around whether an experience can serve as a reference. It's less a content platform than a system for passing on experience, built around real-world decisions. People usually come here not to be stimulated or entertained, but to decide: should I do this, which one should I pick, is it worth it, will I regret it? Given that, whether a post amounts to an experience someone else can borrow often matters more than how intense or emotionally charged it is.

That's why treating Xiaohongshu as a pure self-expression platform tends to put effort in the wrong place. Content here needs, to some degree, to show a path of judgment. Even emotional posts that get shared and saved repeatedly usually contain a specific situation, preconditions, and the author's own reasoning. Readers care less about the emotion itself than about whether the experience lets them rehearse a decision in advance. Without that transferability, a post might get short-term exposure, but it rarely keeps being distributed.

User behavior reflects this. Time spent on a post, repeat views, saves, and later clicks from search often say more than likes about whether content is being "used." These signals don't necessarily bring fast amplification, but they point toward long-term value. The result is something rarely seen on Instagram: some posts show unremarkable numbers during the recommendation phase, yet keep being seen for a long time. Rather than spreading in a single burst, they gradually work their way into search and comparison, called up again and again as samples of experience. Their lifespan runs noticeably longer than content built on emotional stimulation.

Author credibility forms differently as a result. Instagram cares about whether an identity is recognizable. Xiaohongshu puts little weight on personal charisma or dramatic storytelling, and more on predictable judgment: when users run into you again on a similar question, do you still offer reasoning that is consistent, clear about its conditions, and free of contradiction? Credibility builds up across many similar situations. Once it's recognized, an account usually settles into steady, ongoing distribution, and doesn't get wiped out because one post fails.

By the same logic, high-stimulus content such as exaggerated headlines, emotional polarization, and manufactured conflict isn't useless on Xiaohongshu, but it struggles to enter the long-term paths where content gets called up again. The problem isn't that the platform rejects emotion. It's that such content is hard to use for real decisions. It lacks traceable premises, conditions, and consequences, and it's hard to match reliably to specific questions in search. Once it leaves its immediate context, its usefulness drops fast, and distribution drops with it.

The difference between the two platforms finally comes down to one question: what is content kept around for? Instagram rewards a steady presence in a few people's daily lives. Xiaohongshu's distribution revolves around being used. It doesn't reward performance for its own sake, nor does it reliably reward stance or volume. It favors content that offers a dependable reference point when someone else is hesitating, comparing, or deciding. The first makes people remember you. The second makes sure they can still find you when they need to decide. 📖