SocialHub and the Substrate of Decentralised NetworksSocialHub is one of the primary forum where fediverse developers can talk about ActivityPub, how to implement the protocol, and have conversations about how the technical interoperability can be improved with Fediverse Enhancement Proposals. The forum has been searching for new ownership, but making decisions on how to move forward has been challening. Most developers aren’t interested in taking responsibility of community management, while the current admin will only hand over control to a team of people who can not only do the technical administration but can also manage the community. There is also no shared vision for what SocialHub should become, and multiple developers openly wonder if it is even worth it to continue with the forum. Most crucially, nobody has clear authority to make final decisions, making it incredible hard to move past the phase of ‘making a forum post with some ideas and suggestions’.
One of the core challenges with building a decentralised network is that decentralisation is about building alternative power structures, where no single actor has control over the entire network. But power is hard to diffuse: when you build a system that spreads out power, from one control point to many nodes, often this means that new places of gatekeeping and centralisation pop up. The result is often a kind of governance vacuum where important decisions get stuck in endless discussion loops, or where informal power structures emerge that aren’t accountable to the broader community.
Building a decentralised network like the fediverse thus means not only building a social network that spreads out over many different nodes, but also building an infrastructure for the network to run on that is itself decentralised. What’s happening to SocialHub is symptomatic of this broader tension, where these decentralised systems promise to distribute power, but they still need coordination mechanisms to function.
Hobart and decentralised substrates
In an essay titled The Promise and Paradox of Decentralization, tech writer Byrne Hobart wrote about decentralised networks, and how one of their paradoxes is that they require centralised substrates. One quote from the article regularly pops up, where Hobart writes: “Any decentralized order requires a centralized substrate, and the more decentralized the approach is the more important it is that you can count on the underlying system.”
With this, Hobart means that decentralised systems require a shared agreement on how to communicate with the system, usually via a set of agreed-upon protocols. For a decentralised system to work well, people have to agree to a single method of interaction. The internet cannot function if every website implements their own incompatible version of HTTPS, for example.
This leads Hobart to the observation that open networks are prone to being captured by companies that figure out an onramp to the network, writing: “these onramps are built on an open system, but part of their function is to close off some of it. And the better they do that, the more value they can capture.” Twitter and Facebook, but also crypto companies like Coinbase are examples for Hobart of this dynamic.
He writes: “This pattern raises a question: is centralization just a natural tendency of all networks? Are we destined to have a ‘decentralization sandwich,’ where there’s a hard-to-change set of protocols, something open built on top of that, and a series of closed systems built on top of that, which are the only ones the average person interacts with?”
On a surface-level reading, it feels straightforward enough: the fediverse is a decentralised network, and its technical function depends on the ActivityPub protocol. You can view the ActivityPub protocol as the centralised substrate to the decentralised network.
But when you start looking more closely, the picture that emerges is significantly more complicated.
The technological substrate
When you start looking more closely at how the fediverse operates in practice, the picture that emerges is significantly more complicated than Hobart’s centralised substrate theory suggests. Rather than a single protocol that serves as the foundation for a decentralised network, there is fragmentation at multiple levels. Moreover, the more this network pushes towards decentralisation, the more fragmented it becomes.
On a protocol level, there is no singular ActivityPub. The ActivityPub protocol as maintained by the W3C is the official canon version of the protocol, but most platforms don’t implement the full ActivityPub spec, instead opting for a combination of ActivityPub’s Server to Server protocol in combination with the Mastodon API. This means that the ‘centralised’ substrate is already fragmented in practice. While it is possible to make a case that developer adoption would go smoother if ActivityPub implementations were more standardised, the current fragmentation is a result of the network consisting of independent actors that coordinate with each other only to a limited extend.
Quote posts provide a concrete example of how this fragmentation plays out in practice. There are multiple different ways to implement quote posts. Misskey notably has a different method than the method that Mastodon is now using to implement quote posts. When Threads decided to implement quote posts, they decided on supporting both implementation methods for quote posts. This would seem like a good example of the value of a centralised substrate to a decentralised network: things would go smoother if everyone had agreed upon a singular implementation method of quote posts. So when a new fediverse platform that wants to be fully interoperable with other platforms would only have to implement one method, and know exactly in advance which one to use. But the reality shows that even basic features resist standardisation.
What the fediverse shows is that a decentralised network tends to split up into multiple different subnetworks. These networks themselves are also decentralised, and while technically part of the larger fediverse supernetwork, they are often quite separated. For example: The collection of Misskey servers are largely catering towards the Japanese audience. They are technically interoperable with the ‘Threadiverse’, a set of link-aggregator platforms (Reddit-likes, basically), but in practice interoperability and connections between these two sub-networks of the fediverse is negligible. Streaming software Owncast is seen as part of the fediverse, but the ActivityPub-enabled interactions between Owncast streamers and the Mastodon-verse are arguably even more limited.
What’s seen as ‘the fediverse’ turns out to contain more protocols that are interoperable with each other to a certain degree, such as Hubzilla’s Nomad protocol. And if we expand our perspective to look at the open social web as a set of decentralised social networks that are all interoperable with each other, we see even more protocols, such as ATProto and Nostr. At this level, the idea of a single centralised substrate becomes even more tenuous.
So what this means is that the more decentralised a network becomes, the network tends to split into subnetworks, where each cluster of this supernetwork becomes more distinct from each other. Interoperability and connections between these clusters is possible and happens occasionally, but for social and cultural reasons can be fairly limited.
From a technical perspective, Hobarts claim that “the more the decentralized the approach is the more important it is that you can count on the underlying system” turns out to be recursive: the more decentralised approach means that networks start to fragment into subnetworks, each with slightly different technological substrates, and it becomes more important that you can count of the underlying substrate of the subnetwork.
The social substrate
Hobart’s centralised substrate theory assumes that decentralised networks require centralised governance of their foundational protocols. But examining how the fediverse actually governs itself reveals multiple, overlapping authority structures that challenge this assumption. Rather than a single centralised point of control, there are competing forms of governance, spread out over multiple places and communities.
The W3C, the organisation that governs ActivityPub, usually focuses on protocol governance via W3C members, where these members are often required to be organisations. This represents the closest thing to Hobart’s “centralised substrate” – a formal institution with official authority over the protocol specification.
The SocialHub forum is one of the main places for structured long-form communications about ActivityPub. It is also the main place for conversations about Fediverse Enhancement Proposals (FEP). A FEP is a document that gives structured information about ActivityPub and the fediverse, with the goal of improving interoperability and well-being of fediverse applications. Anyone can submit a FEP, and conversations about them on places like SocialHub is how they get legitimacy and buy-in for other projects to implement the proposals.
The grassroots system of the FEPs, in which the SocialHub plays a major part, shows that a single protocol can be used in a manner that is highly decentralized: there is no central authority that can mandate implementation of FEPs, yet they gain legitimacy through community discussion and voluntary adoption.
Conversations about ActivityPub and the fediverse are spread out fairly wide, over a variety of places on the network. Some of the notable places for conversation are the SocialHub forum and the Fedidev matrix channel. The SocialCG of the W3C has various places for discussion, including an email list, GitHub discussion boards and regular meetings. Other places include discussions on microblogging feeds, various (semi)private chat groups and Lemmy communities. Notably, each of these places for conversation only has a small subset of fediverse developers that are participating, and developers are spread out over all these places. This indicates that the ‘social substrate’ of the fediverse development is decentralised as well, there is no single place that owns or controls the conversations about protocol development.
Decentralisation and political power
Hobart is not the only one who has thought and written about how decentralised networks relate to the (potentially centralised) governance of the protocols that powers them, as well as how they are vulnerable to capture. But Hobart’s alignment with the tech-right political wing makes his writing relevant to me, specifically because I strongly disagree with his political views, and the people he aligns himself with. Understanding why this thesis appeals to certain political actors helps makes it all the more important to challenge this way of thinking.
Hobart is a techno-optimist, and his mode of thinking is illustrative of a wider thinking on technology and culture in Silicon Valley. His latest book, on why bubbles are actually good, got a foreword by Peter Thiel. This connection is not incidental, as Hobart represents a particular worldview about how technology, power, and governance should intersect.
Thiel fits well with the line of thinking of Hobart, both on the wider points of techno-optimism, as well as on the aformentioned quote, that decentralised networks require a centralised substrate. Thiel’s beliefs can be understood as techno-feudalism, where he wants to move power away from the political domain to domain of corporate tech, where power is held by a few corporate elites, not by a democracy. Decentralised networks in itself are an antithesis to the worldview of Thiel’s authoritarianism. The decentralisation of a network means divesting power away from the few corporate elites, and spreading it out over many individuals instead.
The line of thinking that decentralised networks often have a centralised substrate, and are vulnerable to being captured by building closed systems on top of the open systems, can be read as either a warning or as an instruction manual. And for noted democracy-hater Peter Thiel, whom Hobart seems to align himself with, it is much more likely that Thiel views this as an instruction manual on how to deal with open and decentralised systems.
The idea that a decentralised network still can have a single central point, namely the technological substrate that powers the network, is thus an attractive idea to an authoritarian figure. You might not be able to control a decentralised network directly, but by controlling or influencing the protocol that powers it, a chokepoint arises that the authoritarian feudalist overlord can leverage to extract rent.
Meta’s approach to the fediverse demonstrates the substrate capture strategy in action. By joining ActivityPub governance discussions while simultaneously building Threads as a massive onramp to the network, Meta places itself into a position to influence both the protocol, as well as to function as a primary gateway to the network. This follows the format of the “decentralization sandwich” that Hobart describes. Their sponsorship of the Social Web Foundation further embeds them in the governance substrate of the fediverse network.
In this context, Hobart’s quote takes on a new meaning. Hobart’s message resonates with the people and organisations who are building today’s social networks of extraction. They have built social networks where they are the gatekeepers, and with their gatekeeping power they have become richer than god. While decentralised networks might pose a threat to centralised networks, promising to take their gatekeeping power away, Hobart’s description points to a new place where they can extract rent. This is why it matters to understand how decentralised networks function matters: it also indicates that the substrates of decentralised network can be decentralised, and points to ways how corporate capture can be resisted.
Reframing decentralisation
Hobart’s statement that decentralised systems depend on centralised substrate makes it appealing to authoritarians, since it provides a guidebook on how to gain forms of centralised control over decentralised systems. But while the idea seems to fit well with a surface-level analysis, a closer look at how the fediverse operates in practice also shows that the substrate of the network is, and has the potential to be, a lot more decentralised than first might be assumed.
From a technological side, the assumption of ‘the fediverse is the decentralised network’, with ‘ActivityPub being the centralised substrate’ turns out to be a whole lot more complicated in practice. What’s seen as ‘the fediverse’ turns out to contain more protocols that are interoperable with each other to a certain degree. The ActivityPub protocol also turns out to contain multiple sub-protocols: most platforms don’t implement the full ActivityPub spec, instead opting for a combination of ActivityPub’s Server to Server protocol in combination with the Mastodon API.
On the social side, ‘decentralisation’ is both a technical description of a network architecture, as well as a more general description of the distribution of authority in a network. The grassroots system of the FEPs shows that a single protocol can be worked on in a manner that is highly decentralised.
This intertwining of technical and social decentralisation reveals why Hobart’s thinking on decentralisation and substrate s fails to capture the reality of how these networks actually operate in practice. At the same time, Hobart’s thinking does provide a good way of understanding how authoritarian-minded people and organisations might approach decentralised systems, and how they think about capturing and controlling such networks. It is this dual combination that makes Hobart’s thinking interesting to me, specifically because I disagree with it on multiple levels.
As for the SocialHub: after a period of uncertainty, Pavilion, the organisation that also build the Discourse plugin which connects the forum software to the fediverse over ActivityPub, will become the new admins of the community.
#nlnet
https://connectedplaces.online/socialhub-and-the-substrate-of-decentralised-networks/
