Big Tech is Working Just Fine. That's Why We Need the Fediverse

I am not trying to be cheeky or provocative, but I do want to revise title of this session. Big Tech was never not broken, or maybe it's working exactly as intended. It's accruing power and influence.

As Nick writes in _The Space of the World_, the development of social media over the past twenty years has been marked by four errors (155-156).

The first error is that social media have scaled up without any regard to the effect on sociality.

The second is that global-scale sociality has been mediated by for-profit corporations.

The third error is allowing those for-profit corporations to continually design and refine social space.

Finally, with almost no exception, we have allowed social media corporations to operate without any consideration of public benefit.

These errors weren't on the part of Big Tech -- they largely have been the fault of regulators. Considering the lack of regulation of social media over the past two decades -- particularly from my home country of the USA -- what we now call "Big Tech" has operated precisely as it intended to. It is not broken -- it is massively profitable and has an outside influence on nearly every sphere of technological, political, and civil life.

This is not new. In 2003, we were in the world of "Web 2.0", a more participatory web. Everywhere, people were talking about wikis and blogs and folksonomies.

By 2006, _TIME_ magazine famously declared "you" to be the Person of the Year.

This was presented as a democratization of the media, but what was happening in the 2000s is that for-profit companies -- Google and Facebook in particular -- realized that they would not have to make media content. Instead, the user would generate it, and other users would react. Over time, users of web applications did all the work of building social media.

And while users did so, their every move was tracked. Web 2.0 or corporate social media relies on ordinary people to do the work: of socializing, of declaring interests, and of (as it is often put) 'creating content.' For-profit social media has gained power by capturing sociality and labor of everyday people.

The situation may only now be shifting as Big Tech develops generative AI systems. Chatbots have been built by analyzing people's interactions, and now companies like Meta see AI friends and content as the future of social media, a means to keep us engaged. This exploitation of the ordinary in service of media power has become the dominant media form of the world -- the Space of the World. Again, this is exactly what was intended when Big Tech started this mission 20 years ago. That's why we need ways out.

We have many ways out of Big Tech. Activists and academics have been mapping such ways out for years. Nick has consistently conluded his books with a call for alternatives -- from _Media Rituals_ (a book from 2003 that had a major influence on my work) to _The Space of the World._

Since 2010, throughout my career, I've considered my work aligned with people like Nick, Clemencia Rodriguez, Chris Atton, Alessandra Renzi, Gordon Gow, and others who have been critically exploring alternative media as a way out of big media power.

Specifically I have been studying alternative social media -- media built as a response to the concentration of power found in corporate social media (what we are calling "Big Tech" today). I have been doing this work since the early 2010s, following the activists and developers who build alternatives.

The work continues with a group of scholars found at socialmediaalternatives.org -- the Network of Alternative Social Media Researchers.

Based on my years of study, I want to make the case that noncentralized, noncapitalist, federated social media -- as found on a significant part of the fediverse -- is a path forward away from the problems of Big Tech. This is a major linkage between my work and Nick's -- he concludes _The Space of the World_ with a call to take federated, alternative social media seriously.

My forthcoming book, _Move Slowly and Build Bridges: Mastodon, the Fediverse, and the Struggle for Democratic Social Media_, draws on interviews with developers, moderators, admins, artists, and activists who build and run federated social media. I argue that the activists building and running the fediverse are charting the most viable path away from Big Tech.

So, what's the fediverse? To understand the fediverse, let's start with its underlying technology. A way to do that is to consider something familiar:

...email. Email is a federated system.

Email resources appear on multiple computer servers, which can communicate with one another. For example, let's say a friend of mine uses Gmail, another friend uses Microsoft Outlook, and I use Protonmail. Despite the fact that we're each using the services of very different companies, we can still communicate. This is because all email servers agree to abide by a set of technical rules, a protocol called SMTP (Send Mail Transfer Protocol.) Today, there is a means for social media data to be treated like email.

In 2018, five technologists working with the World Wide Web Consortium published an open standard called ActivityPub. ActivityPub provides a protocol for social media data, much like SMTP provides a protocol for interoperable email servers. ActivityPub standardizes typical social media activites: making a profile, posting something, liking or favoriting that post, boosting a post, and commenting on posts. Let me walk through this by considering three distinct fediverse systems.

The top logo is that of PeerTube, which is server software that allows for people to post videos, comment on them, and share them. The logo to the left is for Pixelfed, which allows people to post images. And the logo to the right is Mastodon, which is a microblogging system. Mastodon, incidentally, is the most popular fediverse system -- it grew exponentially after Elon Musk bought Twitter. All of these servers can communicate with one another via ActivityPub. That means my friend could post a video on a PeerTube server, another friend could see it in Pixelfed and comment on it, and I can like and boost that post in Mastodon.

It's hard to overstate how radical a departure this is. You CANNOT do that with corporate social media. You cannot post to Facebook and expect a friend on X to comment on it.

You cannot even post to Instagram and expect a response from Facebook, even though they are both owned by the same company, Meta.

The result of ActivityPub has been a global network of heterogenous social media servers. The average size of each server is around 500 accounts, but the network itself is comprised of 10s of thousands of servers with somewhere between 8 to 14 million users.

Ok, so the fediverse is kind of like email in that multiple services can talk to each other. But federated resources across a network -- like email -- do not necessarily chart a path out of Big Tech. There's more to the fediverse than its technical structure. Here I will discuss several problems with Big Tech and point to ways in which the fediverse mitigates those problems.

Specifically, I will talk about four things the fediverse is against: surveillance capitalism, atomization, repressive tolerance, and growth.

The most obvious problem of corporate social media is what we now call 'surveillance capitalism.' Shoshana Zuboff's 2019 book gave us this label but 'surveillance capitalism' has long been with us. As I noted earlier, the idea of the "users building the business for you" dates back at least to the early 2000s and Web 2.0.

Surveillance capitalism functions through a very complex infrastructure. It involves gathering increasingly granular data on users across multiple platforms and channels in order to auction off user attention to the marketers of products and political positions. It tends to work quite well in centralized systems that can suspend a connection between two users for just long enough for an auction for their attention to occur.

The vast majority of the fediverse has been built as a direct repudiation of surveillance capitalism. If you look at the code, you will see most federated systems do not implement these tools -- at all. There is no data-gathering, no auctioning of attention. Moreover, for such gathering to work, the system would likely have to be either much more centralized, or multiple instances would have to collaborate. Simply put, the fediverse is not implementing the infrastructure of surveillance capitalism.

In addition, I would point to something deeper than the implementations. A series of accidents led to the protocol underlying the fediverse to be developed outside of the needs of surveillance capitalist firms. I have a chapter in my book about this. ActivityPub, the underlying protocol of the fediverse, was developed without the direct involvement of Facebook, Google, Twitter, or Sina Weibo. Instead of having to design a protocol with the needs of corporate social media in mind, its developers produced a system they thought would work well for their imagined users.

We can see this in their use of "user stories" in the development process -- the imagined users of ActivityPub were people looking to socialize. There are no user stories of marketers looking to advertise, or of data analysts who want to make consumer profiles. I believe that such use cases would have been considered had corporate social media been involved, but because ActivityPub developed outside of the needs of corporate social media, its authors did not have to consider the needs of marketers.

As a result of both ActivityPub development as well as the critical response to surveillance capitialism, there is now a sociotechnical prohibition against the political economy of behavioral advertising on the fediverse. Mastodon, for example, organized as a not-for-profit from its earliest days in 2017. Indeed, much of the fediverse runs as informal or formal not-for-profits, and anarchist practices of mutual aid and care economies are quite strong in this space.

Perhaps the most ringing endorsement of the fediverse being opposed to surveillance capitalism comes from articles like this one in Ad Age, which states that the fediverse is simply not good for brands.

Another problem is atomization. We could point to a variety of ways in which corporate social media atomizes people into individuals, but in the interest of time I would point to terms of service agreements. When people sign up for corporate social media, they must scroll through thousands of words of legalese that, essentially, state that the individual is on their own when it comes to relating to the corporate platform. The terms attempt to structure that relationship, attempting to prohibit collective actions like class-action lawsuits or demanding that legal actions happen only within specific jurisdictions. (Nick talks about this in _The Space of the World_ in similar ways in his chapter on political theory.) In contrast to corporate social media atomization, in my analysis of the fediverse, I have come to refer not to "users" but rather "members." I do this for two reasons.

One is that instances, or servers, are the key units of the network. Instead of signing up to join a central service -- X.com, Facebook.com -- fediverse members join specific instances.

The instance has become how people interact with the network, and instances function as communities.

The instance-as-community is reinforced through a cultural practice: codes of conduct. Instead of terms of service agreements, instances create their own codes of conduct. Codes of conduct came out of progressive tech activism in the mid-2010s, an effort to push back against meritocracy and hyperindividualism in the tech sector. They were adopted by Mastodon instances at that time and spread across the fediverse. On the fediverse, codes of conduct typically lay out ethical norms: no racism, no casteism, no queerphobia, no marketing. Unlike terms of service agreements, which atomize people, codes of conduct are community standards. One of the great challenges -- and virtues -- of the fediverse is that, to join it, one must select an instance, and to do that well, one must consider that instance's code of conduct.

However, there is no rule that a fediverse instance must have a code of conduct, or it might have one that says "we are a free speech instance." There has been a longstanding valorization of absolute free speech, which has become a demand to be heard. In its search for growth, corporate social media has also largely adopted this line. Nick refers to this in _The Space of the World_ as "expanding the scale of human interaction without limit" (155). When corporate social media executives speak of its value, one of the first justifications they offer is that they allow "anyone to be heard." They are open to all. This is a naive view of how speech works.

The problem with a naive free speech view, or a naive openness view, is that it leads to the recapitulation of the power of the dominant -- the loudest, the wealthiest, and those in power are advantaged in such a system. Conversely, the poor or marginalized are drowned out. This is what Marcuse recognized with his concept of "repressive tolerance."

Taken to the logical conclusion, "free speech" then becomes code for "let's keep listening to whomever is in power" -- and our history shows that this is a call to hear out people such as white supremacists, or people who believe trans folks should not exist. Elon Musk's free speech absolutism is only one way in which this has been a norm in corporate social media.

As Marcuse argued, “Tolerance toward that which is radically evil now appears as good because it serves the cohesion of the whole on the road to affluence or more affluence.”

This brings me to contradictory meanings of the word "federation." In computer networking, the term can mean openness and connection with no regard to the ethical content of those connections. Federation in computer science simply means allocating resources across a network. A good federated network here ought to be 'transparent' to the end user, meaning that the user doesn't need to worry about the internal details. The network's components reliably and openly connect to one another. The network ought to be open to all -- there is no judgement about content. Information simply flows.

Federation in political science, however, is quite different. It is about autonomous, like-minded polities banding together through shared ethical values. As I argue in the book, I think that this is the meaning we need to keep in mind as we consider the fediverse: federation as a political or ethical theory. There are many examples of this meaning of 'federation' -- one that has been generative for political philosophers is the Haudenosaunee Confederacy here on Turtle Island or North America.

Likewise, with the instance as the political unit of the fediverse, instances-as-communities band together when they see that they have shared values, usually apparent through codes of conduct. If my instance sees that your instance shares the same values, we are likely to make and keep a connection.

However, fediverse instances do not feel obliged to remain connected to instances that harbor transphobes, racists, or other trolls. Fediverse members do not wish to "hear them out." The network is not naively open.

Instead of remaining connected to a server full of hate speech, fediverse servers simply defederate. There is no requirement that this network expand to include all, to hear all voices. Federation also involves defederation.

There are many examples of defederation. One is the Fedipact, an agreement by hundreds of fediverse instances to block Meta, which set up an ActivityPub-enabled system called "Threads." Rather than federate with Threads, many on the fediverse block it, citing its transphobic and racist content moderation policies as well as how it engages in surveillance capitalism.

Finally, let's turn questions of growth. This has much to do with the ecological impact of media. We already are seeing the energy and resource demands of generative AI, a sector we're told over and over again must grow.

Even before the extreme resource demands of generative AI, environmental scholars were pointing to the energy and water demands of server farms. Corporate social media increased the demand for data centers that are always on so that users can infinitely scroll. As a colleague of mine, Roel Roscam Abbing, notes, "always-on" is measured in uptime, with more 9s being better. 99.99% uptime is better than 99.9%, which is better than 99%. But every additional 9 presents a kind of Zeno's paradox of reliability: increasing reliability requires exponentially more resources, so much so that even green data centres rely on diesel-powered generators as a fallback to meet user demand.

The fediverse, by contrast, can be a degrowth medium. We already discussed its repudiation of surveillance capitalism: that alone reduces its energy demands, since it does not need to power data-gathering, consumer profiling, or the auctioning of attention.

In addition, many fediverse instances willfully remain small -- they close registrations rather than seek endless growth. For example, scholar.social, a Mastodon instance meant for academics, currently has its registration closed. Why don't they want to grow? It's because they want to make sure their ratio of content moderators to members is healthy. This is a major contrast with corporate social media, which always seeks to grow.

And there are experiments on the fediverse to reduce power demands and even rely on solar power to run infrastructure. This is because the network and its software are open to experimentation and modification. Anne Pasek of Trent University here in Canada is working on this as we speak, building a solar-powered fediverse instance.

Such an instance would require low-power hardware and attempt to be very efficient in terms of media storage and bandwidth. It might make social media running on reclaimed hardware more viable. In other words, it's the opposite of the massive server farm. This is possible because federated technologies are open to experimentation and modification.

Taken as a whole, my book is about what I call the covenantal fediverse: a network of thousands of autonomous communities who agree to band together through a thin set of global values. We see those values in noncapitalist practices and mutual aid, the use of codes of conduct, practices of defederation and blocking, and experiments to mitigate the climate disaster. What I find most exciting about the fediverse is that we see a widely-shared set of ethical norms in communication emerge as small autonomous communities choose to connect to one another. This is unique. Perhaps BBSes or Usenet anticipated it. But neither of those systems built a culture of shared values across multiple independent communities. The closest analog I can find is federated political structures, but of course when we speak of the fediverse, we do not speak of armies or currency. (This is not to say the state is withering away, or that there is no monetary exchange.)

So, the covenantal fediverse is the future of social media, right?

It might have felt so for a minute. 2023 was a giddy year. The purchase of X by Elon Musk brought about a massive wave of new people to the fediverse. New projects and instances were springing up seemingly every day.

However, thinking with Stuart Hall, there are no guarantees. The fediverse as I'm detailing here requires continual care and activism. It also requires energy from developers, and it requires resources.

It can be captured by corporate interests. I mentioned earlier the fedipact, the project to block Meta. What the fedipact is blocking is Meta's new microblogging app, Threads, which is implementing ActivityPub. In theory, that means that Threads users can connect to fediverse members. Given the size and power of Threads, that would mean Meta could come to dominate the fediverse.

Alternatively, much of the energy that was going into the fediverse has been sapped away by a competing project, Bluesky. Bluesky began as a project within Twitter and then later spun off. That corporation is developing its own social media protocol called ATProto. I know Nathalie will talk about Bluesky. I myself am increasingly critical of Bluesky, but I will hold back for now in the interest of time. Regardless, there is no guarantee the ActivityPub-based, covenantal fediverse I describe in my book is the future. As always, a better communicative future requires struggle.

However, what I've learned in writing my book is that even if the covenantal fediverse fades away, the struggle – the care work – to build better methods of communication and community will no doubt continue on. To put it in terms from Nick's book, even if they fail, the activists who have built the fediverse will keep trying to build a better space of the world.

Thank you!