The Feed That Slowly Stops Feeling Personal
You're scrolling. It's been three minutes and you haven't seen a single person you actually know. There's a sponsored post for luggage, a viral clip from an account you followed during a phase you'd rather forget, and a meme that was probably funny six months ago. Your college roommate is definitely still posting her very strong opinions about pasta. You're just not seeing them.
This isn't a glitch. It's a structural outcome baked into how large platforms are built, and once you see the mechanism, you can't unsee it.
More Connections Mean More Competition for the Same Space
Every platform has a fixed amount of screen real estate to sell you per session. Not literally fixed, you can scroll forever, but engagement data shows that most users make meaningful decisions about a platform within the first fifteen to twenty posts they see. That's the premium slot. Everything else is background noise.
When you had forty connections, forty candidates competed for those fifteen slots. Your friend-to-slot ratio was almost one-to-one. Now imagine six hundred connections. Six hundred posts generated daily are competing for the same fifteen premium positions, plus paid advertising inventory, plus algorithmically surfaced posts from accounts you don't follow at all.
The math alone dilutes your friends before the algorithm even enters the room.
Then the algorithm enters the room.
How the Ranking Engine Sees a Post From Your Friend
Platforms don't show you posts in chronological order. They score every candidate post using a ranking model optimizing for one thing above everything else: predicted engagement. Specifically, the kind that keeps you on the platform long enough to see more ads.
Here's the concrete mechanism. Say your friend Maria posts a photo of her new apartment. The model looks at signals: how often you've liked Maria's posts in the past six months, whether you've clicked her profile, whether you share any mutual interactions. If you've been drifting apart digitally, those signals are weak. The model assigns her post a low predicted engagement score.
Now a video from a creator you watched once gets scored. That creator posts four times a day, has millions of followers, and a historical click-through rate of twelve percent. The model has rich data on it. High confidence, high predicted engagement. It jumps Maria in the queue.
This happens thousands of times a day. Weak-signal friend content loses to high-signal viral content almost every time, not because the platform hates your friends, but because it knows less about your relationship with them than it knows about content engineered to be watched by strangers. It's a bit like asking someone who's never met your family to arrange the seating chart at a reunion. They'll put the loudest person at the front every time.
Take two people who joined the same platform on the same day. One kept their network tight at around eighty close contacts and rarely accepted requests from acquaintances. The other connected with everyone, every colleague, every person from every event, and hit six hundred connections inside a year. Five years later, the first person still recognises almost every post in their feed. The second person barely sees anyone they'd call a real friend. Same platform, same algorithm, completely different experience. Network size did that.
The Engagement Loop That Makes It Worse Over Time
Here's where it compounds. The ranking model learns from your behavior. Every time you scroll past Maria's apartment photo without tapping it, you're teaching the model that Maria's posts aren't worth your attention. The model down-ranks her further. You see her even less. You engage even less. The signal degrades to near zero.
Meanwhile, you do watch that viral video. And the next one. The model learns that viral accounts are exactly what you want, and serves you more.
Platforms sometimes call this "relevance." It is relevance, technically. It's just relevance to your most reflexive browsing behavior, not to your actual relationships. The social graph (who you know) and the interest graph (what you click) start as the same thing on these platforms, then slowly divorce. You end up with a feed that reflects your worst habits, not your real life.
Two Assumptions That Miss the Point
The most common assumption is that platforms deliberately suppress friends to force you onto their paid promotion tools. Satisfying theory. Mostly backwards. The suppression is an emergent property of engagement optimization, not a deliberate social engineering campaign. The outcome is the same either way, your friend's posts disappear, but the cause matters if you want to do anything about it.
The second assumption that trips people up: that unfollowing or muting accounts will fix it. It helps at the margins. But if you have six hundred connections and prune aggressively down to two hundred, you've still got two hundred candidates fighting for fifteen slots, with the algorithm still preferring accounts that have thick engagement histories over the ones you actually care about.
The honest fix is earlier and more boring: keep your network intentionally small from the start, or use platform features (close friends lists, favorites, see-first settings) to manually override the ranking model for specific people. You're essentially telling the algorithm to treat a low-signal friend as a high-priority signal. It works, but it requires maintenance, and most people never bother.
Found those settings yet? If you have fewer than a hundred people on your see-first or close friends list, you've probably still got the problem.
The Feed Isn't Neutral Ground
A social platform is not a window onto your social life. It's a ranked list of bets the platform is making about what will keep you scrolling. Those bets get more accurate and more impersonal the larger and looser your network becomes.
Maria's apartment photo wasn't buried because it was unimportant. It was buried because it was unimportant to an algorithm that has never met her, doesn't know your history with her, and is trying to hit an engagement target set by people who will never appear in your feed at all.
The network you built is the network you get ranked by. Worth remembering the next time you're about to hit accept on someone you met once near the canapés.