Re-Centralizing the Fediverse

Ever since the takeover of Twitter by Elon Musk and the subsequent exodus from that platform, the Fediverse has been in the news more and more. (Really, Mastodon, the most Twitter-like of the many applications to use the ActivityPub standard, has been in the news; to the extent that most people hear the term “Fediverse” at all, they probably assume it’s synonymous with Mastodon.) And, over and over again, to the point of becoming a cliché, one hears the same analogy:

It’s just like email! There are lots of different email servers, but they all use an open protocol. So if you use Gmail and someone else uses Yahoo, you can still send and receive email with them. And since all Mastodon (and MissKey, Pleroma, PeerTube, etc.) servers speak ActivityPub, you just set up an account with one of them and you can engage with content on all of them. You can even run your own server! It’s easy!

One glaring omission from this analogy is that, while it is relatively easy and common for one to run one’s own Mastodon, MissKey, or Pleroma server, pretty much nobody runs their own email server these days. In fact, email could be the poster child for an open standard that has ceded control to a Big Tech oligopoly. Most people’s email addresses end in,,, or a small handfull of other dot-com domains; even if they have an institutional email account with a company or school, there’s a good chance that Google or Microsoft is actually running the server their email domains point to. And yet most people seem to be in denial about the risk that ActivityPub and the Fediverse could suffer the same fate.

Back in the 1990s and 2000s, most people’s email accounts were with their internet service providers, and it was still common for ISPs to be locally run. And while it wasn’t exactly common, it also wasn’t all that unusual for people to run their own email domains. So what happened? Well, the Internet got popular. Email got popular. And everyone who ran their own email servers soon discovered that the open standard that was email just couldn’t scale effectively. Spam became a huge problem. If your email address was published anywhere on the Web, you could bet it would soon be inundated with unsolicited emails from questionable commercial entities who would scrape the Web for new addresses and pummel them with automated emails, or sell them to other people who would do the same. Scams also became a problem: it was trivially easy to fake the origin of an email, making it seem like it came from someone else’s address. People flocked to email providers who ran better spam filters, and the best spam filters were at the bigger email providers, since they had the most training data. More infrastructure and standards grew up for verifying that an email came from a reputable source, with the not-quite-unintended consequence being that emails from smaller servers wound up getting sent to spam folders. What had been an exceedingly simple set of protocols became so complex and expensive to maintain that nobody wants to do it anymore.

Already, the Fediverse is experiencing similar growing pains. Servers are having to deal with an influx of users from the sinking ship that is Twitter, and many of the smaller ones are unable to deal with the added traffic. The Fediverse also found that it was rapidly diversifying away from its original user base of mostly affluent mostly white tech workers, and those who didn’t fit that description often found themselves the target of harrassment and intimidation. It’s easy to create fake accounts that look like they belong to some other person, and the hacky way Mastodon currently “verifies” users (asking them to put a link in their profile to their personal web page, which itself contains a link back to that Mastodon profile) is more of a hack than a security measure. These problems are only going to get worse as more users and services join the Fediverse.

We are fortunate that ActivityPub was designed much later than SMTP and can learn from email’s mistakes. There’s more security built into the protocol. There’s not an implicit assumption that everybody has a right to talk to everybody else, and Fediverse servers can and do block entire domains that don’t do a good job of moderation. There’s a grass-roots #fediblock hashtag that allows server admins to find such questionable instances and block them. But when it comes right down to it, safeguarding the Fediverse is left to the individual server admins (or users), and many, many of them are going to find they’re not up to the challenge. As Mekka Okereke points out, this situation, much like the earlier problems with email, is likely to be “fixed” by corporations and their walled-garden approaches if something isn’t done to stop that. Okereke suggests a twofold solution: make it harder for a server to be listed on, and make it easier for a Mastodon admin to check a box opting in to an allow-listed federation.

The problem with both of these solutions is the “Gargron bottleneck”: For all intents and purposes, Mastodon is, if not the whole Fediverse, the part that most people are familiar with, and the part that’s seeing the most growth in the post-Twitter era. And is Mastodon’s front door. Both are run by the maybe-not-so-benevolent dictator Eugen “Gargron” Rochko. Which means that the Fediverse is already somewhat centralized: Gargron exerts an outsized influence over what form the Fediverse takes. If he doesn’t want to make it harder to discover poorly moderated instances, it won’t happen, and if he has a different understanding of what “poorly moderated” means, his vision will win out. Sure, it’s possible one of the many competing services to Mastodon will become popular and Gargron’s influence will wane, but making this happen is an uphill battle when all the major media outlets only talk about Mastodon and only want to interview Gargron.

But let’s say both of Okereke’s suggestions do come to pass: Gargron introduces stricter rules for which instances get listed on, and Gargron adds a feature to the Mastodon app allowing administrators, upon creating a new instance, to elect to federate with a list of “good” instances (curated by Gargron). This would result in the creation of the Federal Republic of Gargron, a collection of semi-independent instances led by an autocrat (named Gargron). The maintainers of each instance in Gargron’s federation would be free to run their servers as they see fit, so long as they don’t violate any policies (set by Gargron) that would result in their being dropped from the federation list and/or removed from So they have a choice: follow Gargron’s (possibly arbitrary) rules, or run the risk that fewer people will federate with them or even be able to find them. And while I’ve been repeating Gargron’s name to the point of absurdity in order to stress how much this system hinges on a single individual (and because Gargron’s just a funny name), it of course won’t be that way forever. Gargron could die, leaving ownership of the repo and website in limbo. Or Gargron could hand over maintenance duties to someone else – perhaps even someone who pays him a lot of money. Perhaps even Elon Musk. Is this really something we want to risk?

There are, of course, other ways a federation could form. For example, a number of instances could decide to form a cooperative, the People’s Federal Republic of Mastodonia. A governing council is formed, elected by the maintainers of the member instances. This council is empowered to draft and maintain a set of rules that instances will need to follow in order to federate with Mastodonia. They maintain a website that lists the member instances, and members are required to check it at regular intervals and update their list of allowed and blocked instances accordingly. If any user from any of these instances feels that their instance is being lax in applying the moderation rules, they can bring a complaint before the governing council. The council will also have the responsibility of promoting the federation, making sure those who are new to the Fediverse, or who are already in the Fediverse but are dissatisfied with their current instances, hear about Mastodonia and know what it offers.

Ideally, what I would like is to put to bed, once and for all, the myth that choosing an instance is all about what hobbies or interests it caters to. I made that mistake when I first joined the Fediverse, and I wound up on an instance whose moderation and blocking policies left much to be desired. The fact is, Gargron is right about one thing: local timelines don’t really matter much. If you really want to find people who are into anime or Rust game development or Byzantine numismatics, you’re far better off searching for hashtags than trying to find the instance with the perfect local timeline. The most important thing about an instance isn’t its users’ interests; it’s its maintainers’ values. The important questions to ask are whether the maintainer has both the will and the ability to prevent harrassment, whether they block instances that are likely to surface objectionable or illegal content, and whether they take users’ complaints seriously. A cooperative of federated instances that makes its values, and the ways in which it enforces them, known would be the best way to enable users to make an informed choice of instances.

Some will no doubt note that I’ve sidestepped one important topic: money. It costs money to run a server, and it costs money to maintain client software. When you donate to “Mastodon”, you’re not donating to your instance or anyone else’s, nor are you donating to any of the many unofficial Mastodon clients. You’re only donating to Gargron for maintaining the official Mastodon app and official Mastodon website. If you want to support your local instance, you have to donate directly to them, and if you want to support your “federation,” you’d have to donate individually to each instance with which your own instance federates. A cooperative-run federation could also facilitate funding; you’d just donate to your cooperative, and the board would allocate the funds to instances and infrastructure as needed. Some cooperatives (or instances within a cooperative) might even charge fees to open an account, though I suspect most would not, since most people would probably not see enough benefit in such a federation to pay the fees.

This means that most federations would be donation-driven, and here there is a good deal of risk. Because, much as happens with large and important open-source projects, the bigger federations will begin to attract corporate donors, whose contributions could easily outstrip those of individuals. Such donations would no doubt be billed as “giving back to the community” but would in reality give said corporations an outsized voice in how the federation and its instances are to be run. The end game would be wholly corporate-owned and corporate-run federations, whose member instances have an incentive to follow whatever rules their corporate owners set – including, perhaps, allowing ads or handing over users’ data. We could easily end up with Twitter 2.0, with the same problems as the original except allegedly better somehow because it’s “decentralized”.

With the Fediverse, as with many things, my instinct is that centralization is inevitable. Economies of scale are a real thing, and whatever benefits there are in remaining completely decentralized won’t do much good when individual instances are swamped with spam or just have too many users to moderate effectively. So I think that too much emphasis on remaining decentralized will just result in centralization happening on terms other than our own. Perhaps instead we should be looking into ways of centralizing better: forming democratically run governing bodies, legally incorporating as nonprofits, and banning donations from for-profit corporations in the charter. We were unable to avoid the corporatization of the World Wide Web, but if we act quickly, we just might be able to stave off the corporatization of the Fediverse.

Last modified on 2022-11-25