Centralized social media platforms face an existential crisis. Almost every day brings news that underscores the folly of having so much of our digital lives flowing through a handful of corporations, as we surrender our personal information to companies like Facebook, Twitter, YouTube, and TikTok in order to connect and communicate with others.
These and other platforms shape public discourse by extracting massive amounts of data from the people that use them, applying them to drive dark patterns of addictive engagement with the goal of selling user attention to marketers and advertisers. Each successive breach of trust illustrates the disadvantages of this asymmetrical relationship. Misinformation, hate speech, algorithmic bias, behavioral manipulation, and mass surveillance are all byproducts of these networks.
What makes these social media so powerful is their organization around central rule-making authorities and data repositories, which leaves users ripe for exploitation. Corporations that monopolize social platforms monitor and control who can participate, how they communicate, and what information or data they can access. When something goes wrong, users have little recourse to hold them accountable.
But rather than simply critique these platforms, we should apply their lessons to new technologies. Critical reverse engineering of social media can produce decentralized alternatives and open network infrastructures that empower users to govern by consensus and own their identities and data, ultimately returning control to individuals and smaller collectives.
Decentralized, alternative social media projects suggest a future where our personal information isn’t auctioned off, our connections aren’t mapped for profit, and our social lives aren’t modulated by a CEO and a team of software engineers.
In computer networking, a network’s logical shape or layout is referred to as its topology. The topological characteristics of a network are revealed when we map how various nodes are connected. One of the simplest computer networks to build is a star topology, where a central node controls all of the network’s other nodes. If you use a wi-fi router in your home, you’re using a star topology: all of your devices are connected to that router, which allocates a portion of your internet connection to each of them. Everything you and your family does online is seen by that router.
Corporate social media like Facebook or Twitter are star topologies on a massive scale. When we use them, absolutely everything we do — our likes, comments, shares, and connections — flows to the logical center before being routed to people in our networks. Like our home routers, what the apps present to us is also entirely determined by their center. And like our home routers, everything we do while connected to centralized social media is recorded and monitored.
This model is extremely attractive to marketers, who can use a single source to get valuable data and insert their marketing messages into our lives. It’s also a huge problem — when a star topology becomes really big, whomever controls it exercises incredible power. A single source of private data on citizens is also a tempting and convenient target. And if the central node gets hacked, then whoever takes over the center can access that power.
Critical reverse engineers are experimenting with solutions to these problems, including exciting topological experiments with decentralization. One such approach is federation. Federated systems allow individuals to install software on their own servers, placing the software that handles social connections and data in the hands of local administrators rather than on distant server farms. The owners of these servers can then open them up for regular users to connect to them, and the servers can also connect to one another. Federated systems avoid a logical center by creating constellations of smaller star topologies that can be linked together into a massive network. If one of those stars gets hacked or is shut down, the network will survive.
While federation sounds like quite an innovation compared to centralized systems, it’s actually more in line with the internet’s original conception: a collection of small interconnected servers.
A wide range of decentralized social media projects, including diaspora*, GNU social, and Mastodon, use the federated model. Inspired by the ideas of Columbia law professor and decentralization advocate Eben Moglen, diaspora* appeared in 2010 as a challenge to Facebook. Around the same time, Free Software Foundation employees developed GNU social as an alternative to Twitter. Mastodon, which is a newer implementation of GNU social, is arguably the most popular among them, boasting roughly three million users and thousands of servers (called “instances” in Mastodon parlance) that function as individual micro-blogging nodes that can connect to one another.
Developed in 2016 by Eugen Rochko, Mastodon is a textbook example of critical reverse engineering. It provides features that are familiar to Twitter users: microblog posts, the follower/followed relationship, and a timeline of posts with the newest on top. But it also includes innovations that build on Twitter’s model: server administrators gain a great deal of power in how their instances are run and which other Mastodon instances they connect to. Users can shop around, finding an instance that they prefer to use. Paid advertising — and the whole infrastructure of user surveillance that drives it — is nonexistent on Mastodon.
Mastodon’s critical reverse engineering of Twitter is largely driven by the developers who contribute to the open-source project.
“It always seemed to me that Twitter’s core product is fairly simple, and could’ve just as easily been implemented with an open/free protocol similar to email, which would allow a larger group of people to work on improving or customizing it as well as finding new use-cases,” Patrick Figel, a former Mastodon administrator, told me. “This is, to me, the most exciting aspect of Mastodon.”
Along with coding new features, skilled Mastodon developers also take time to help new people set up servers to contribute to what they call the “fediverse.” Federation, however, isn’t fully decentralized. Individual servers can become large, giving their administrators a smaller-scale version of the power that Facebook enjoys.
Other critical reverse engineers are experimenting with fully distributed, peer-to-peer networks. One example is Twister, another Twitter alternative. Developed by the Brazilian software programmer Miguel Freitas, Twister was inspired by the “Brazilian Spring” of 2013, when activists protested high public transportation costs, government corruption, and police brutality. At that time, Freitas was getting his news from Twitter and feared that the Brazilian government would crack down on activists that were sharing news about the protest. He set out to solve that problem.
According to a technical article by Freitas in the International Journal of Parallel, Emergent and Distributed Systems, Twister relies on:
“three mostly independent overlay networks, two unstructured and one structured. The first unstructured overlay network is based on the Bitcoin protocol and provides distributed user registration and authentication. The second is a structured [distributed hash table] network, providing key/value storage for user resources and tracker location for the third network. The last, also unstructured, overlay network is a collection of possibly disjoint ‘swarms’ of followers, based on the BitTorrent protocol, which is used to provide efficient near-instant notification delivery to many users.”
By using these three networks, Twister eliminates the need for any servers at all. Processing power, routing, and connection algorithms are handled only by the end-user devices in the network — there is no center. Topologically speaking, Twister looks more like a fishing net than a constellation of stars.
Like Mastodon, Twister critically reverse engineers Twitter by implementing a decentralized approach while still providing recognizable microblogging to end users. Individuals can still follow other accounts, make short posts, and share images. The major difference is that no one, not even Freitas himself, can interfere with a Twister user. Remarkably, Freitas implemented this fully distributed microblogging system in a matter of months, which was a stunning achievement considering the technical challenges of decentralization.
Critical reverse engineering is not merely technical. Developing a new technology isn’t enough — we have to find ways to adapt that technology to meet social needs. So while federated or distributed technologies represent technical innovations, they also need to be accompanied by social innovations.
A key area is moderation. Rochko didn’t just develop Mastodon to offer people the chance to create their own smaller versions of Twitter. He did so because he was concerned about hate speech and harassment on Twitter, and he built several affordances into Mastodon to give users and instance administrators more control over the content that they see.
“As a user, one has few different options — one can mute and block, which can hide individual trolls,” Rochko explained to the Indian news site Scroll. “The next… [option is to] hide all content from a particular instance… You can also contact your admin [to] take action on behalf of your instance, which is the harshest step, as the admin can cut off access to and from a particular instance.”
Because instance administrators have complete control over their own servers, they can easily moderate conduct on their instances, posting “codes of conduct” for their users to abide by and federating with other servers that comport to similar values. A potent example of this happened when a group of alt-right trolls tried to bring their online community, Gab, into the Mastodon fediverse. Mastodon administrators simply blocked the Gab server from connecting to their servers, thus repudiating Gab’s hate speech.
Moreover, as some colleagues and I have shown in a recent article in New Media and Society, Mastodon administrators tend to be far more accessible to users. Unlike Facebook’s opaque “Supreme Court-like” oversight board, which the company created to rule on permissible and prohibited content, Mastodon administrators and users communicate openly about policies and codes of conduct, negotiating solutions rather than having them imposed by fiat.
Whether moderators are involved, as in the case of Mastodon, or whether the end user is completely in control, as in a fully distributed system like Twister, decentralized projects give us an idea of the possibilities that can exist outside of the centralized social media model.
As with any new product or service, decentralized social platforms need to overcome difficulties in attracting users and building strong network effects to rival those of corporate social media.
Depending on their governance systems and whether or not they seek to compensate users for producing and sharing content, they may also need to introduce users to tokenomics models that create and distribute value within a given platform. Steemit, for example, is an online community resembling Reddit that uses a digital token to incentivize contributions, while Minds and Voice are social networks like Facebook that rewards users with tokens for interacting and spending time on the platform. Steemit and Minds have struggled with influxes of racist users, however, further highlighting the need to add authentication, accountability, and social moderation mechanisms to such technical and economic innovations.
Decentralized infrastructures may also contend with friction if individuals are required to run their own servers in order to connect. New networking technologies and core protocols can help by creating secure environments that support decentralized software, interoperability, and data autonomy while giving access to everyday internet users. As Freitas argues, ease of use has to be foremost in developers’ minds, since end users often “don’t care about anti-surveillance technologies until they are personally exposed. So you should assume that by the default they will just choose the solution that is easier and more pleasing to use. The anti-surveillance [aspect] has to come as a bonus, not as a trade-off for usability.”
But even within decentralized systems, centralization can rear its head. Network effects and economies of scale can raise the prospect of consolidation, and not just from the inside out. Preexisting, centralized powers often watch decentralized alternatives and attempt to appropriate their success. Twitter CEO Jack Dorsey has announced Project BlueSky, a research project investigating ways to transform Twitter into a decentralized platform. This move would represent another episode of a centralized corporation attempting to take over a community-developed standard, much as Google has done with RSS or SMTP.
There’s also a potentially problematic reliance on the influence of central figures. Freitas recently announced that he would stop developing Twister and that its website would shut down. Because it’s an open source project, he noted, others can step into the developer role by taking over the code. But if no one does, Twister may dissipate over time. Even the most decentralized projects can be vulnerable if there is only one or a few developers behind them.
Nevertheless, these weaknesses also underscore that because these decentralized projects are free and open source, they are ultimately user-controlled. Even after Jack Dorsey hinted that Twitter would develop its own decentralized protocol — or latch onto an existing one, like ActivityPub — Mastodon founder Rochko expressed confidence that the community-driven, decentralized systems would ultimately be strong enough to avoid being dominated by Twitter.
After all, despite all of the criticisms we level at centralized social media, all of the congressional hearings and media revelations of abuse of data and invasions of privacy, Mark Zuckerberg and Jack Dorsey will never allow the public to access the source codes of Facebook and Twitter. But in the case of open-source projects like Mastodon and Twister, we already have the code and can modify it if we choose to.
We don’t have to accept the problems of centralized social media. Combining topological decentralization with social innovations can produce responsive moderation systems, increased user control of data, accountable network administration, a better balance between individual expression and community, equitable distributions of value, and code that is socially owned rather than centrally controlled. This alternative future is well within reach.