When Elon Musk took over Twitter in 2022, the whole idea of using open protocols to build social networks that are not owned by oligarchs took hold in the mainstream. For good reason, people do not like that the global communication infrastructure is sits under the control of few billionaires, and they thus started looking for alternatives. In 2022 that mainly started with Mastodon, and later came Bluesky as well. But these networks were obviously much smaller, so people had to start formulating ideas of change: how do we get the majority of people to stop using platforms like X, and start open networks like the fediverse or the atmosphere instead? There were roughly three theories of change:
- Convince people of how bad the Big Tech platforms are, and that alternative networks do not have these downsides.
- Build new apps and software on these open protocols that is so cool, new and interesting that people will want to join.
- Governments would adopt regulation that would force platforms to interoperate.
In 2022 and 2023 there was a strong conversation around the EU’s Digital Markets Act (DMA), which was in the process of adding interoperability requirements for messaging services to the regulation. There was also a clear understanding that social media interoperability requirements would be added later. So the assumption that government regulation, most notably the EU’s DMA, would be a pathway for open social networking protocols like the fediverse and the atmosphere to gain mass adoption, made sense.
A few weeks ago, the European Commission, in a first review of the DMA, decided to not expand the interoperability requirements of the DMA to include social media. This means that at least for the next three years, the pathway to adoption of open protocols via government regulation, can safely be assumed to be dead.
So I hope I can make you care slightly about European legislative processes, because it does explain quite a bit on how adoption of this new generation of open social networking protocols actually happens (or more sadly in this case, won’t happen, at least not via a hoped EU regulatory intervention).
First, some context, what actually happened, and how does this work.
The Digital Markets Act is an EU regulation that came into force in 2023. It targets a handful of designated “gatekeeper” companies (such as Apple, Meta, X, Microsoft and ByteDance) and imposes obligations on them that are meant to make digital markets fairer and more competitive. The provision most relevant here is Article 7, the interoperability requirement, which currently forces gatekeepers operating messaging services like WhatsApp and Facebook Messenger to let competing services connect to them. The DMA requires a review every three years, and this spring was the first review, with one of the main questions being whether that interoperability requirement should be extended from messaging to social media as well.
The Commission gives two main reasons for not extending Article 7. The first is that a study commissioned by the EC found “no clear demand” from European consumers and businesses for social network interoperability. The second is that the existing messaging interoperability framework under Article 7 has not yet seen wide adoption, and it is therefore too early to assess whether extending it to a new domain would work.
I don’t think either reason is particularly strong however.
The “no clear demand” finding has a methodological problem, namely that consumer-survey demand for an unfamiliar capability is a notoriously weak signal. Explaining to people who join these open networks what new sort affordances they offer, that you can follow a Pixelfed account from a Mastodon account, or that you can use your same atproto account for blogging on Leaflet, chatting on a live stream on Stream.place as well as publishing your git repo on Tangled, is already hard enough as it is, and that is for people who are interested and invested in open protocols. That a survey shows little demand is no surprise, because the affordance that open protocols are so new and without any clear previous referent, that people have a hard time imagining what is even possible.
The argument that there has not been a strong demand yet for interoperable messaging services is much worse however, because the EC refuses to engage with the question of why that is. The reason the messaging interoperability framework of the DMA is failing is not that users do not want it or that the technology does not work. It is that Meta’s compliance strategy, requiring per-user geolocation proof for any interoperating service, has rendered it practically unusable. Matrix has documented this in detail, and that is a story worth in itself for another time. The EC frames this as a problem of consumer interest, and that the regulation is still early. But the actual problem is that the EC tried to pick a fight with Meta via regulation, and that Meta outsmarted the EC, and that the EC has simply lost that fight.
Back in 2024, I held in my own hands one of the testing phones that had WhatsApp fully interoperable with Matrix. The technology for this totally works, and has existed for a while now. But because Meta requires the IP address of every user that wants to connect to WhatsApp via Matrix, to validate that they are located in Europe, and Matrix does not track user IP addresses, Matrix simply could not implement the bridge that connects Matrix with WhatsApp, and they decided against complying with Meta’s restrictions and thus did not roll out their bridge between Matrix and WhatsApp. So while the interoperability for messaging services exists in theory, it does not exist in practice. Meta implemented the system in such a manner that complies with the letter of the law, while ensuring that no real interoperability exists in practice.
This makes it both incredibly funny and frustrating that the EC opens the report by saying that the DMA exists because “a small number of very large platforms were able to restrict competition including by imposing unfair conditions both on their business users and on end users”, only to then claim that this is now being addressed. According to the report, “alternative messaging services can now interoperate with WhatsApp and Facebook Messenger, which has already led to new providers, including a European SME entering the market with innovative new functions”. The European SME that the EC proudly points to as evidence of a healthy interoperable market is a chatting app with a single rating on the Apple Store. Single rating, as in, one (1) individual person has bothered rating this app. This is the only app currently operating that allows for WhatsApp interoperability. Meanwhile, the report says nothing about why other apps like Matrix, with significantly larger userbases and real interoperability work behind them, are not actually interoperating with WhatsApp and Facebook Messenger.
More fundamentally, this shows the incoherence in the EC’s thinking. The DMA is about creating fair digital markets, and the goal of the legislation is ostensibly to provide a framework that constrains the biggest market powers, to create “contestable and fair markets in the digital sector”. But when push comes to shove, the EC firmly points to the consumer instead. It says there is no consumer interest and uptake, so there is no need to implement further regulation on social media. Consumer interest and a healthy market for businesses are two very different things, and the EC has quietly shifted from trying to regulate markets to trying to serve a vague version of “consumer interest”.
The reason this matters for open protocol adoption is that open protocols are not just another product category competing for consumer attention. They are a different way of organising the market itself, and the DMA’s stated purpose, contestability and fairness in digital markets, is exactly what they offer when they work. The shift from market regulation to consumer interest is hides the thing the regulation was supposed to be doing, which is creating the structural conditions for alternatives to gatekeepers to actually exist. Reframing this as a question of demonstrated consumer demand puts the burden of proof on the alternatives to show they are wanted before the regulation will create the conditions for them to be usable, which is exactly backwards from how the original DMA was supposed to work.
The DMA’s interoperability framework operates with a distinction between “horizontal” and “vertical” interoperability. Horizontal interoperability is what happens when systems with the same role connect as peers, like Mastodon servers federating with each other. Vertical interoperability is what happens when one system runs “on top of” another, like a third-party Twitter client using Twitter’s API. This distinction has become the operative analytical frame for thinking about interoperability in EU digital regulation, and it shows up throughout the Commission’s review.
Robin Berjon recently wrote a great piece on this, pointing out that this distinction is not actually a frame that the people who design and build interoperable systems use. The horizontal/vertical distinction lives in the heads of economists and regulators, not in the heads of practitioners. And worse, it has been actively used to lobby for the interests of incumbent platforms, with arguments that horizontal interoperability would somehow entrench gatekeepers by making them more central to user experience, and that the DMA should therefore not require it. This is the frame the EC is now relying on to decide whether to extend interoperability rules to social media.
But the bigger point Berjon makes is that “the purpose of interoperability is simply to make institutional arrangements possible”. When you design a social networking protocol, you are not just deciding how data moves between computers. You are deciding who gets to do what. A protocol creates roles, like server operators, labelers, relay providers, and gives each role specific powers and limits. The protocol is effectively a small constitution for the system it governs, distributing authority among the actors that participate in it.
This is a very different way of reading protocols as plumbing, pipes that move data from one place to another. Once you see this, the limits of the horizontal/vertical framework that the EC now follows become obvious, because it allows for only one kind of question: do the pipes connect? Either two systems exchange data as peers (horizontal), or one system runs on top of another (vertical). It cannot ask the questions that actually matter once you treat protocols as governments. Who has authority to refuse to federate with whom, and what does the process behind that refusal look like? When a Mastodon instance decides to defederate from Threads, that is an independent community exercising power that does not exist at all in a centralised platform. When atproto separates the role of labeler from the role of AppView, that is a constitutional choice about whether moderation and delivery are bundled into one centralised function or distributed across different actors that users can mix and match.
When Bluesky users launch their weekly scheduled discourse about Bluesky’s moderation decisions, that is a clear signal that the thing that people actually care about in an open interoperable system is the rules on how and when connections should be prevented, and who gets to set and enforce these rules. So when the EC now weakly commits to some vague ‘horizontal/vertical interoperability’ framework, it misses the actual interesting challenge: when should interoperability not happen, and who gets to set these rules?
With this decision, the European Commission has closed off one of the potential pathways for mass adoption of open social networking protocols for the foreseeable future. But it has done so using weak reasoning, while quietly shifting its own understanding from market regulation to measuring consumer demand, all using an analytical frame about interoperability that cannot see what open protocols actually do in the first place. At the same time, the social media team of the European Commission is proudly posting that they have added follow buttons for Bluesky and Mastodon to their website, and removed X, to showcase their support for open social networks. And to make matters even more complicated, the EU’s Court of Justice’s recent ruling makes it very unclear if the concept of federation itself is even compliant with GDPR.
So if you feel confused, bewildered or frustrated about how Europe’s talk about ‘digital sovereignty’ matches up in practice with what the EU is doing policy wise, join the club. Europe is certainly doing ‘things’ regarding digital sovereignty, as long as you don’t expect these things to be particularly coherent or sensical.
