In November 2022, just after Elon Musk bought Twitter, I did a Science Moment arguing that almost everyone was missing what was actually valuable about the thing. People were talking about revenue, verification, moderation, features, whether Musk wildly overpaid, all that.
My point was that Twitter’s real value was already baked in.
By then, hundreds of millions of people had spent more than a decade deciding, one by one, whom to follow. That had produced tens of billions of connections, arranged into this insanely complicated hierarchy of little clusters inside bigger clusters — friends, acquaintances, experts, journalists, scientists, celebrities, weirdos, and thousands of people you couldn’t even name but had somehow found and decided were worth hearing from.
The analogy I used was moving the whole United States to a new, supposedly better United States. Maybe the houses are nicer. Maybe everything is cheaper. Fine. But everybody gets scrambled. Your neighbors aren’t your neighbors anymore. Your friends aren’t nearby. The familiar guy down the street is gone. Sure, you can go hunt down Fred and Doug and try to rebuild some of it. But good luck recreating the whole country.
That was basically the problem every Twitter competitor faced.
And now we’ve run the experiment over and over: Threads, Bluesky, Mastodon, Truth Social, Gab, Parler, GETTR, MeWe, Minds, CounterSocial, Post, Spoutible, Hive, Cohost, Pebble. Some survived. Some became little ideological islands. Some are basically dead. Most just don’t matter much.
Threads is the interesting one because it actually got huge. It now claims around half a billion monthly users, and by some measures its daily mobile audience is right there with X. Yet X still gets roughly ten times the web traffic. So Threads has managed to get something close to X-sized population without becoming anything close to the same kind of place.
Spend time there and you can feel it. Threads has been around more than three years now, and I still feel like I know almost nobody there. Twitter is full of people I “know” — social-media know. I know their personalities, their obsessions, their blind spots, their humor, what they’ll probably say when some new thing happens. A lot of them I’ve never met. Some I know almost nothing about outside Twitter. But they’re familiar characters in my world.
Threads still feels much more like an algorithm feeding me strangers. And this isn’t just some weird personal reaction. From the beginning, people have complained that Threads shows them loads of recommended accounts and posts instead of mainly the people they deliberately chose to follow.
Meta had basically every advantage imaginable. Billions of users. Instagram’s existing graph. Huge engineering resources. Instant distribution. It could populate Threads almost overnight. But Instagram’s graph was built for Instagram. The people whose vacation pictures you want to see are not necessarily the people whose thoughts you want passing through your head all day.
And the really hard part is that you can’t realistically reproduce Twitter’s follower graph. Those tens of billions of connections were built one by one over years. You might remember to re-follow Joe Rogan or a few dozen obvious people. But you don’t remember the hundreds or thousands of obscure accounts that, together, make your feed feel like your feed. And everyone else has the same problem at the same time. Scramble the graph and you scramble the place.
Plus, those connections had acquired character. You knew which scientist was careful, which journalist actually knew his beat, which stranger was hilarious, which person you disagreed with but still wanted to hear from. The graph wasn’t just huge and complicated. It had become familiar.
And think about the amount of decentralized computation buried in all this. Hundreds of millions of people making billions and billions of little judgments for more than a decade — follow this person, unfollow that one, trust her a bit more, ignore him, remember that prediction, notice that this stranger keeps saying interesting things. Behind all that are astronomical numbers of neural firings across hundreds of millions of brains. Out the other end comes this giant social structure, with little clusters joined to bigger clusters joined to still bigger ones.
Nobody designed that. Nobody could.
We measure social networks mostly by users, posts, likes, replies, minutes spent. But the thing that matters here is something more like social depth. How old are the connections? How often have the same people interacted? How densely are all the little neighborhoods connected? How much familiarity and reputation has built up?
A network can have half a billion users and still be socially shallow.
That was the thing I thought people were missing in 2022. Musk wasn’t just buying software, servers, employees, revenue, or a brand. He was buying a gigantic piece of already-grown human social structure.
Back then I called everything else the icing. Musk could change verification, screw around with the algorithm, try this feature, dump that feature, make a mess for a while.
The cake was already there.
And four years and a small graveyard of Twitter replacements later, it sure looks like the cake was the hard part.
The real value of owning Twitter is THIS. Moment 317, November 2022



