The Twitter Xodus: Time to stay or time to go

“We’re here, because we’re here.” Until recently, leaving a social media platform seemed likely to be an act of self harm, a way to lose friends and influence. Yet people are now leaving Twitter/X, in greater numbers, to join a range of different platforms. It’s a good moment to take stock of what is happening and why.

Our toxic relationship with Twitter/X

The open nature of contact on Twitter/X has long made it the target of harassers and distributors of disinformation. People’s experience of the platform has been in long term decline, which accelerated under Elon Musk. Yet organisations and individuals have been reluctant to abandon the years of work they have put in building networks; and while journalists and decision makers remain on the platform, it is hard for many organisations to decide to depart.

The de facto monopoly position of social media platforms, albeit within specific niches, makes it difficult for people to decide that the game is up. Nevertheless, the naked abuse of personal power by Musk to promote a selection of abhorrent views which coincide with the political platform he has decided to associate himself with, will push matters over the line for many users, as it did for some 50-plus German NGOs and the Guardian.

Whoever is left will find themselves in a space that is increasingly unpleasant, and more like a replica of Truth Social than the Twitter of the past; potentially its own problem, but a different one. Twitter/X can probably be said to be in terminal decline.

Our toxic relationship with new wannabee social media monopolies

New platforms, backed by investor capital, are often joyful experiences. Audience reach is easy, there is no advertising and the bad actors are yet to arrive. As the platforms gain momentum, they need revenue, and turn to monetisation. Our investment of time and effort in their product is rewarded by data profiling, time spent watching adverts and being asked to pay to reach the audiences we have built up.

Platforms are likely to continue to want to give access to their product free to its audience, and to serve the audience with content that others provide for free. The market dynamics are likely to remain the same, where audiences are not paying: user profiling, combined with cheap content that provides entertainment and engagement, dominates the model. These dynamics play against matters like truth, insight and investigation, or accurate moderation of behaviour and abusive content. They create a space where reputation is difficult for users to judge. Bad actors can divorce their standing from adherence to ethics and truthfulness, and put in their stead an alternative ‘truthiness’; what feels like it ought to be true, for a particular audience.

The power dynamics of global monopolies controlling what is acceptable or desired speech should be very plain at this point. That is not to say that Twitter/X created the MAGA movement, far from it, but it has actively facilitated it, as the choice of a single billionaire, who stands to gain personal and regulatory advantage.

A healthy relationship with social media

If we want a healthy relationship with a telephone provider, or an email provider, then we need to be in control of that relationship. Ideally, we should:

1. Not suffer disadvantage because we move provider, eg be unable to contact people on the old service from the new;

2. Be able to take our contacts to a new provider;

3. Be able to take our identity, eg telephone number, to a new service, or at least be able to forward messages sent to the old identity;

4. Be able to move content to a new service.

All of these are true and obvious with your mobile phone today. Changing provider does not have a material consequence, you can take your number with you. Email is similar, but you need in practice an overlap, where the old service forwards mail to the new one.

Monopoly social media on the other hand does not meet any of these criteria. BlueSky and Mastodon try to meet them, more or less, while Threads makes promises to do so in the future.

This approach can be backed by regulation. Mobile phone switching exists because of the Communications Act 2003 and is governed by Ofcom. Open Banking, a tool which makes payment systems work better in the UK, is based on legislative requirements.

There are measures in the UK’s Digital Markets, Competition and Consumers Act 2024 that support obligations for digital interoperability (“open choices”) which can be applied to social media.

Steps towards a positive relationship with social media

Another world is possible: and is essential, if we want a sustainable, democratic information economy. For instance, decentralised communities, where actors know each other, find it easier to moderate content, remove bad actors, and for reputations to remain founded on shared values and ethics. Payment and editorial control need not be the factor that reduces circulation of poor information, but instead proximity of a platform to its community can play a similar role.

Decentralised communities can be large, at least in theory, if the values of the community are clear and shared. Interoperability allows multiple communities that trust each other to share content and discussions. This is the model that has been established by ActivityPub and Mastodon. It also explains the skepticism many Mastodon users have of allowing ‘federation’ with Meta’s 250 million Threads.net users. It is hard to believe that Threads won’t include the same kinds of bad actors as have been found on other large platforms. Scale means that abuse is easier to engage in and harder to stop.

Both BlueSky and Threads promise interoperability, but of different kinds. Threads promises to work with Mastodon and other ActivityPub services; this is today partially true (Mastodon users can follow some Threads users, but Threads users cannot yet follow Mastodon users). The kind of partial interoperability provided by Meta so far does not provide a competitive environment; you can’t leave or even join Threads and take your audience with you.

BlueSky says that it will allow people to move to other compatible services, but there are question marks. Federation does not really exist yet; and users cannot move services. It is a platform designed by a single commercial company whose motivations may change with scale. BlueSky’s architecture means that user content is picked up by relays, sometimes called a “firehose”, whereby decentralised services or apps can pick out posts from these central indexes. It gives the prospect of scale and the advantages Twitter had of creating reach for content. It makes data more searchable, and so on, but has the potential to create a central and controlling aspect (why should anyone try to replicate the expensive to run relay / firehose). The assumption is that everything is accessed from the firehose, and marked up with moderation decisions, etc. The relay / firehose is also available to everyone as a complete dataset; this could create opportunities for commercial profiling, data mining and AI abuses, from both commercial and state surveillance, which are harder (but not impossible) to achieve in more distributed models.

The relay / firehose model may make running services easier, and the model attempts to distribute decision making over moderation decisions. It allows users to choose their own algorithms for content prioritisation. Nevertheless, it could also replicate many of the current problems with centralised moderation struggling with bad actors, as all of the content is in theory accessible, and many content decisions will be accepted by third parties on trust, given the scale of the data being dealt with. On ActivityPub / Mastodon, in contrast, a server has selected or blocked other servers according to their trustworthiness, and external posts are pulled in from (those) other servers according to preference; an approach that seems instinctively simpler, but can suffer from problems around scale. Thus BlueSky emphasises centralised accessibility, while ActivityPub as implemented so far emphasises decentralised trust.

BlueSky may have good reasons to try to do what they are doing, but deciding not to support ActivityPub means their federation does not work with Threads or Mastodon. This has produced the absurd outcome that both Threads with 200 million users and BlueSky with 10 million or more may claim to provide interoperability and account porting, but so far intend to ensure that there is no significant interoperability between them, because of business decisions to back different protocols.

Of course, we don’t know the long term future of either kind of federation; things can change. BlueSky has promised to hand governance of its protocol to a standards body in the future; this would help address these kinds of concerns. Meta and Threads are relying on existing standards, but progress is slow.

We can safely say that the shape of our democracies depends on the way federated social media develops. Without working and genuine interoperability, efforts to regulate social media for trust, reputation and truth, as well as civility, risk and toxicity, cannot work; all the power and incentives will continue as present.

Whether competition powers are ever used to force interoperability in social media is a political and regulatory one. We need to see a shift to tackling the root power of monopolies, as well as through content and data regulation. Monopoly abuses can be reduced through interoperability, by empowering users to choose providers they trust, in a way that content regulation like the Online Safety Act cannot. The OSA attempts to ameliorate the worst symptoms of our toxic relationship with social media by leveraging the power of those monopolies to remove access to specific kinds of content. Where interoperability asks monopolies to yield power and control, the OSA requires them to exercise it for policy ‘success’.

Likewise, while data protection is a necessary component to balance the relationship of users with global data monopolies, enforcement has so far let down the promise of the law, especially in the UK.

Even so, the policy imperatives for the Online Safety Act – desires for greater safety, less online abuse, less toxic behaviour – can potentially be better served by interoperability, by shifting power to users. To that extent, the demise of Twitter/X and the rise of interoperability within the new social media services like Mastodon, BlueSky and Threads is to be welcomed. We need to grasp how important that is, and how to get it right; it may not happen through the promises of corporations alone.