People You May Know, Places You Will Never Go: The Cruelest Algorithm on the Internet

Somewhere inside a server farm the size of a Costco, a machine is working very hard to fix your loneliness. It has studied your behavior, cross-referenced your mutual connections, analyzed your browsing history, and arrived at a bold conclusion: you should probably be friends with a wedding photographer named Brent who lives in Scottsdale and once liked the same Reddit post as you in 2019.

Congratulations. You've been algorithmically matched.

The Flattery Trap

The "People You May Know" sidebar — or its Instagram equivalent, the suggested follows — is a masterpiece of false intimacy. It implies that your next great friendship is just one click away, that the social fabric of your life has a loose thread and the platform has kindly located the person to tie it off. What it doesn't mention is that this person is 2,300 miles away, has 47 followers, posts exclusively about CrossFit, and shares exactly one mutual connection with you: your dentist.

This is not a coincidence. It is a feature.

The cruel genius of the recommendation engine is that it makes you feel simultaneously seen and unworthy. Seen, because the algorithm has clearly done some homework. Unworthy, because when you look at the suggested profiles and realize you have nothing to say to any of them, the logical conclusion your brain arrives at isn't "this algorithm is broken." It's "I must be the problem."

And that, friends, is some genuinely impressive psychological judo.

The Ex-Adjacent Rabbit Hole

Nothing the internet has ever produced is quite as disorienting as opening Facebook to find your ex's college roommate's boyfriend staring back at you under the header "Someone You Should Meet." You've never spoken. You've never been in the same ZIP code. You share no discernible interests. But the algorithm, in its infinite wisdom, has decided that what your Tuesday morning really needed was a reminder that your ex has moved on so completely that even his peripheral social orbit is now being served to you like a cold dish.

The technical explanation is mutual connections. The emotional experience is something closer to haunting.

Instagram does a version of this that's arguably worse. It will surface someone you briefly stalked three months ago — maybe an old coworker, maybe someone you met at a wedding, maybe a person whose pasta photo you accidentally double-tapped at 1 a.m. — and present them as a suggested friend, as if the platform is nudging you with its elbow and whispering, "You know you want to." It is the digital equivalent of a mutual friend saying, "Oh, you two would get along so well," except the mutual friend is a trillion-dollar corporation that profits from your engagement, not your happiness.

The Spectacular Failure Nobody Admits To

Here's what's remarkable: these recommendation systems are, by virtually every meaningful human metric, spectacularly bad at their job. A 2022 study from the University of Michigan found that algorithmically suggested connections on major platforms resulted in sustained friendships at a rate that researchers diplomatically described as "not statistically significant." In plain English: almost never.

And yet the feature persists. It grows. It gets more aggressive. LinkedIn will now send you a notification — a push notification, on your phone, which vibrates and interrupts your lunch — to tell you that someone viewed your profile and maybe you should connect. That someone is almost always a recruiter trying to sell you a job you didn't apply for, or a second-degree connection who accidentally clicked on your name while trying to find someone else entirely.

But you checked, didn't you? Of course you did. Because the notification implied possibility, and possibility is the most addictive substance the internet has ever synthesized.

You Are the A/B Test

The uncomfortable truth hiding under all of this is that "People You May Know" was never really designed to make you friends. It was designed to make you click. Every time you visit someone's profile — whether out of genuine interest, morbid curiosity, or the specific horror of recognizing your ex's new partner — you generate data. That data teaches the algorithm who else to show you. You are not a lonely person being helped. You are an unpaid research subject in a matching experiment that has been running for fifteen years and has yet to produce a peer-reviewed result.

The platforms know this, of course. They know that the aspiration of connection is more valuable to them than actual connection, because aspiration keeps you scrolling and actual friendship gets you off the app. Your loneliness is, functionally speaking, a retention strategy.

So What Now, Brent?

If you've ever accepted a "People You May Know" suggestion and then watched that connection immediately go silent — no messages, no likes, just a number added to your follower count and then nothing — you've experienced the full lifecycle of what these platforms are actually selling. Not friendship. Not community. A simulation of both, rendered just convincingly enough to keep you coming back to check.

Brent from Scottsdale isn't your person. The algorithm knew that when it showed him to you. It showed him anyway, because your three-second hesitation before clicking away was worth something to someone, even if it wasn't worth anything to you.

The loneliness algorithm doesn't want you lonely enough to leave. It wants you just lonely enough to stay.

And honestly? For a piece of software, that's a pretty sophisticated read of the room.