Writing about open & equitable product development

Fragmentations in the fediverse

Last week, the official Mastodon app rolled out a new change which presents mastodon.social as the default way to sign up for an account on the app.

This did not land well with many folks on the #fediverse.

From @Fedi.Tips, aka @FediThing:


The official Mastodon app is doing something new which is potentially very dangerous to the existence of Mastodon and the Fediverse.

..continued with an issue on GitHub requesting this feature to be reverted.

From Aral Balkan:


Dear @Gargron,

Please reevaluate your decision to incentivise centralisation on mastodon.social in the official app.

This is the sort of design a VC-funded startup would implement, not a non-profit acting in the interests of a healthy commons.

I’m sure you don’t want mastodon.social to become mini-Twitter and you don’t want to become mini-Musk.

That’s not how we win this.

More instances, not larger instances is the key.

Onboarding vs Resilience

I agree with the general sentiment here. Instance diversity is unequivocally a Good Thing. The friction comes from two opposing objectives:

Option 1: Improving the Mastodon/ActivityPub onboarding story


Option 2: Improving the ActivityPub/Fediverse resilience story

(2) has clearly been the priority since Mastodon's founding in 2016, as (1) was always an obvious option and has never stopped being promoted by user experience advocates.

The UX impact of a no-questions-asked signup method cannot be understated. I've noted an abundance of anecdotes detailing fedizens failing to bring their peers on to the network due to the cognitive load of having to choose a server.

What's especially worrisome about this type of usability barrier is that every unsuccessful onboarding goes unrecorded, failing silently. We've surely lost hundreds of thousands of prospective newcomers to the multi-server signup paradigm, possibly millions.

Perhaps I'm being naive, but I believe the fediverse has grown resilient enough to handle a temporary centralization event within its richly decentralized ecosystem. WordPress started as completely centralized and has grown into a 600bn market with no single actor bigger than ~5% of the total pie.

Growing pains of this kind can be good problems to have, as they force the ecosystem to evolve to new levels of scale. You’re gonna run into bottlenecks with whichever growth strategy you go for, it’s just a matter of choosing whichever seems most tractable.

Weighing the options

Eugen laid out the tradeoffs pretty clearly in his very recent interview with The Verge:

Nilay Patel:

The Gmail comparison is really interesting here. I cannot believe I’m about to explain Gmail to the Decoder audience, but I’m going to do it just so I’ve said it out loud. Email is an open protocol that is run by standards organizations. Gmail runs on those protocols — SMTP, IMAP, and the rest — and so does Outlook or whatever. The reality for most people is that there’s a collapse between the protocol and the application on their phone. If you have Outlook, you are almost certainly not using the Gmail app. If you have Gmail, you are almost certainly not using the Outlook app. You’re going to the service and putting that app on your phone. The only real exception to this rule is the Apple Mail app on the iPhone. With everything else, there’s a collapse between the protocol and how it’s expressed to the user.

mastodon.social is the one you run. You download the official Mastodon app, and it’s going to default you into it because defaults are important. I mean, literally, the number one criticism I hear from people is, “Well, you have to pick a server and it’ll never work,” because no one wants to pick a server. You’re going to solve that problem, but aren’t you now getting closer to that collapse between the open protocol and the user experience, where people download the Mastodon app and end up on the Mastodon server?

Eugen Rochko:

Well, for reference right now, mastodon.social isn’t actually the default in our app. It’s just one of the top ones that shows up. However, I think that possibly going forward, we might rework the onboarding user experience into presenting a default option as well as an advanced option, where all that stuff with choosing a server would basically be hidden away from the people who get intimidated by choice.

Yes, you are correct in that it gets us closer to the Gmail situation. But it’s kind of unavoidable with the constraints of the problem I’ve described, where the choice is too complicated. You need to convince people that this is better and that this is something they should invest some time into. Then they’ll realize both how it works and what they can do with it.

That is the idea. The idea is the funnel. They get started on mastodon.social, but afterwards they can move to an account on their own server that they create or on a different server provided by a different company or person. Historically, that has been the case. A lot of people who are currently running their own servers had their first account on mastodon.social. So it is working, and I imagine it will make mastodon.social somewhat disproportionately large in the future, but that’s just part of making it work, I think.

It's quite regrettable however that the Mastodon team didn't make more of an effort to restate their delicate position on this issue in a dedicated blog post and accompanying toot.


In the absence of any official response from Eugen & co., we've arrived at a precarious impasse:


I don't think people are realising the danger the Fediverse is in.

The only thing stopping corporations and VCs taking over this place is that the Fediverse is spread out on many different servers, which makes it very difficult to purchase.

If most of the Fediverse ends up on mastodon.social, which is now a strong possibility, there will be nothing to stop most of it being sold to Musk or Zuckerberg or whoever.

The bigger mastodon.social becomes, the more likely a buyout is to happen.


With all that in mind, here's a suggestion:

➡️ IF mastodon.social becomes more than 50% of the Fediverse, either by total users or monthly active users, the rest of us should defederate it.

Sticking with mastodon.social because “that's where the people are” is pointless. Centralised growth will simply cause the governance problems we've seen on Twitter and Facebook to be replicated on here.

Growth has to be decentralised in order to protect the independence of all Fedi servers.

I love everything else you've been doing as a community advocate @FediThing, but this latest line of thinking is not conducive to a healthy fediverse. In fact, it could very well be its downfall. Large scale separation is antithetical to the social fabric of the world wide web.

Can we take a breather here?

Consider for a second what a terrible business decision it would be for anyone to attempt another high profile social network buyout in this day and age. Add to that the massive risk of buying an AGPL codebase that cannot be relicensed but can be forked, and an un-locked-in userbase made up largely of people whose one commonality is that they heckin' love sticking it to the man.

Hostile VC takeovers are very far down the list of problems facing our nascent community of fedizens. Onboarding and retention on the other hand is easily in the top 3.

Mastodon ≠ Fediverse. I'm not interested in discussing what the Mastodon team should do with their app. That is ultimately their call to make, and it seems like they've made up their mind already.

So what should we do? Can we redirect all this energy in a manner that will be generative instead of divisive? I think so.

Nomadic Identity

This whole discussion would be a non-issue if Mastodon and other AP servers had more comprehensive support for:

  1. Account migrations (tenants can easily move all their essential data to a new instance)
  2. Account aliases (tenants may tie multiple separate accounts together)
  3. Account sovereignty (the user is the ultimate owner of their identity)

In a world where tenants can effortlessly move their account from one server to another, mastodon.social can innocently serve as a temporary staging ground for new entrants as they are encouraged to move elsewhere once they've found their footing.

Many thoughtful people have made this point, and I support it wholeheartedly. If I was the CEO of the fediverse (which will never be a thing), I'd make it my highest priority to see it through.

To get involved with the deeply complicated but ever more important subject of data portability on the fediverse, here are some good places to start:

This thing will never work if at the slightest sign of conflict the first solution we reach for is putting up walls between one another.

I'm going to bed. Please be nice. 💜