Twitter is Fragmenting

1

Imagine a neighborhood bar. It’s kind of like the one in Cheers, where everybody knows your name, and they’re always glad you came.

Now imagine the bar got bought by a rich asshole. He makes crude jokes, and when people don’t laugh he just gets louder. Worse, his lack of organizational skills are starting to take their toll on the basic functioning of service. Everything’s mostly fine—you’re still a regular—but it feels like something that was special about the place died. Meanwhile, the financial math, which was always slightly precarious, is now definitely not working. One wonders how close the place is to Chapter 11.

So what happens? Do people scatter? Do they re-convene at a new location? Or do they stick around?

My belief is that they scatter.

2

What holds a social app together? Ask any tech enthusiast and they will quickly tell you: network effects. Put simply, social apps allow us to communicate with familiar faces in a familiar way, and they make it easy to meet new people. Starting a new social app is kind of like starting a new town: it’s just a bunch of empty space at first. This is why most social apps fail.

New networks usually are formed when a new way of communicating is enabled by a technology shift. Broadband internet and flash video enabled YouTube. SMS enabled Twitter. Mobile photography enabled Instagram.

But sometimes networks form for other reasons. Facebook, for example, was technically possible to build nearly 10 years before it was founded. All you need is personal computers, the internet, .edu email addresses, the web, and the <img> tag—all of which were available in 1995. The reason Facebook succeeded is it was the first to understand a cultural shift (the web was becoming mainstream amongst college students) and build a network structure that suited it (private networks within universities).

Today it feels like we’re living through another important cultural shift—the decline of Twitter—and a new generation of networks are experimenting with new structures to try and capitalize on it. If we want to understand what’s next, it helps to ask: What caused the culture shift? What new structures would be better? And what will it feel like to use the internet in the future?

These are big questions. They’re impossible to answer with data or certainty. But they’re important questions to consider, especially now, because if we’ve learned anything from the history of the internet so far, it’s that the structure of the networks we communicate through influences the way we live.

Churchill once said, “We shape our buildings; thereafter they shape us.” He was right.

3

First, let’s consider what went wrong with Twitter and how it relates to the culture shift away from large networks.

To be clear: Twitter is not going to die, but it is in structural decline. The proximate cause is Elon’s (mis)management, but there is a deeper issue: throwing the entire world into one chat room with one set of rules was never going to last. As Noah Smith put it, the internet wants to be fragmented.

Back when Jack Dorsey ran Twitter, the platform was consistently spawning right-wing alternatives (Gab, Parler, Truth Social). Now that Elon is in charge and angering a wider variety of people, a wider variety of alternatives are cropping up. To name a few: Mastodon, Substack Notes, Bluesky, Farcaster, and more. But no matter who is running Twitter, their decisions are going to anger some subset of users, and those users will seek alternatives.

Until recently I thought those alternatives were all doomed. But now I see that I was judging them by the wrong yardstick. They aren’t meant to become “the next Twitter.” There will be no “next Twitter.” But there will be a shift toward an internet that looks more like it did before Twitter and Facebook became dominant. The big networks will still exist, but their share of attention will shrink and there will be an increase in smaller communities that have the power to enforce their own rules.

Some of these communities will have their own tech stack. Others might use existing platforms that are structurally more suited to their purposes. But generally I believe power is shifting towards the edges.

Noah Smith already did an excellent job making the case for why the internet is fragmenting, so I won’t dwell on this point at length, but the quick version is:

• People get angrier and more polarized when confronted with others who have irreconcilable points of view.
• People are happier and more open to thinking critically about new ideas when they’re in a space that makes them feel safe.
• Groups can only achieve harmony when they have the power to arbitrarily exclude people who don’t fit.
• Being excluded from a group isn’t as big a deal when there are many groups to choose from.
• People benefit from being a part of multiple semi-overlapping groups, where you can explore different interests and identities in different spaces.

Putting all these facts together, it’s easy to see why Facebook and Twitter struggled to keep so many different types of people all together under one roof. For a long time, they’ve functioned kind of like the “downtown of the internet” where people come together to meet but retreat back to their neighborhoods for more meaningful interactions. But now, thanks to Elon, it feels increasingly precarious to rely on Twitter as a public space. The rules seem to change every day and the service feels increasingly buggy and broken. This throws everything off balance.

The question is, what could come next?

4

The list of Twitter alternatives grows each week, but to me the most interesting players are:

Substack Notes

(Note: I am biased! I used to work at Substack.)

Substack is interesting because it already has relationships with so many influential writers and their audiences. This is a unique starting point for a social communication platform. Each writer is like the nucleus of their own community of readers, but now, with Notes, readers and writers can interact in a more fluid way than before. It’s basically like Twitter, except it exists on Substack and therefore is useful to writers to help grow their newsletters. Now anyone can share a note, even if they don’t have a newsletter.

If that were all there was to Notes, it would still be enough to make it an interesting feature. But the far more important thing is that Substack’s writers have an incentive to make it good. When I follow someone on Notes, I’m also subscribing to their newsletter, and they get my email address. Compared to other decentralized alternatives to Twitter, email is by far the most valuable and widely adopted decentralized communications protocol. Also, Substack is not just about email, it has a business model built around paid subscriptions. The writers that use it don’t just want Notes to be good because it’s fun, they actually depend on Substack for part or all of their livelihood.

It will be interesting to see how this evolves over the next few months. The big risk is whether writers will actually be able to catapult it into a large enough network to be a valuable source of subscriber growth for them. Will readers and casual writers meaningfully participate?

If I were Substack, I’d strongly consider integrating the comments and chat functionality into Notes in order to simplify the product and give the network every possible chance to succeed.

There’s also the question of moderation. How do you make writers and readers feel in control so you avoid the “one world group chat” problem? Under the old network structure, this was clear: each writer had their own fiefdom where they were in charge. Notes muddles it a bit by allowing anyone to post outside the context of a writer-centric community. This means either Substack is in charge of moderation (a tough position to be in where you can’t please everyone) or no one is.

Bluesky

Email addresses are great, but as many Substack Notes users have pointed out, people are somewhat reluctant to share them. Our inboxes are full enough already. What if there were a new protocol that was designed from the start for casual, short-form communication?

This is what Bluesky is all about. It is building a direct competitor to Mastodon with a different set of ideas about how decentralization should work. The three main principles are:

1. Federation: you can connect with anyone on any service that uses the AT protocol (the open standard that powers Bluesky).
2. Algorithmic choice: anyone can build their own algorithm and share it with the world to use.
3. Portability: you can switch from one hosting service to another without losing your followers or content. (This is the main thing that distinguishes AT from ActivityPub, the open standard that powers Mastodon.)

I’m honestly pretty excited about Bluesky and the AT protocol. I would love it if Substack Notes supported it, and I don’t see any reason why that wouldn’t be possible in the future.

I’ve written before about how cool algorithmic choice could be. Here’s a mockup I created almost a year ago of what a marketplace for algorithms could look like within Twitter:

Ironically, the title of the post I created this design for was “Elon is Right: Twitter Should Open Up The Algorithm.” In that post, I argued that merely open-sourcing the algorithm wouldn’t matter, and that the real benefit would come from being able to choose an algorithm. There were a few people within Twitter at the time who reached out and it seemed like something that actually might happen. Now, under Elon, it seems incredibly unlikely that the company can get itself together to ship a major feature like that.

Honorable mentions: Wavelength, Farcaster, Discord

Besides Substack Notes and Bluesky, I also wanted to mention a few other companies that I think are working on interesting versions of this problem.

Wavelength is a new app that John Gruber of Daring Fireball is advising. I installed it a few days ago but my first impression is very positive. It’s like a group chat app, except each group chat has threaded discussions, which makes it much easier to manage multiple topics and support larger groups than would be possible in a single-threaded discussion like on WhatsApp or Telegram. The design is intuitive, the app is lighting fast, and it also has a native Mac client. Oh, and you can chat with AI in any thread you want by tagging “@AI.” 

Farcaster is a lot like Bluesky, except you register your domain name on the Ethereum blockchain. I can understand why that would be important, but I think the relatively frictionless onboarding of Bluesky may prove decisive.

Also, I would be remiss if I did not mention Discord. It is an absolutely giant service and as far as I can tell it just keeps growing. I spend just as much time there as I do on Twitter these days.

5

Maybe I’m wrong. Maybe this is just a blip. It’s possible that Elon is just getting his sea legs, and will start making better decisions (or at least delegate them to someone who will), and in retrospect this whole thing will have been a mirage.

Only time will tell. As I said, I don’t think Twitter will go away, but I do think it will consume a smaller share of the world’s attention going forward. I think more space will be created for communities that can enforce their own values. It won’t feel like it did in the past, where there was nowhere else to go.

This will be a good thing. In safer, weirder spaces, people thrive.