Center for Technology and Innovation

Amer⁠i⁠can Conserva⁠t⁠⁠i⁠ve: Rou⁠t⁠e Around Depla⁠t⁠form⁠i⁠ng W⁠i⁠⁠t⁠h Decen⁠t⁠ral⁠i⁠zed Tech

By: Andrea O’Sullivan / 2021

Andrea O’Sullivan


Center for Technology and Innovation


As centralized platforms censor more, it’s time for dissidents to turn to peer-to-peer options.

Conservatives have been told that the solution to most big business problems is to offer or to patronize a better alternative in the marketplace. As we have seen with digital technology, this is simplistic. In many cases, it’s not merely a matter of building a superior product that attracts enough users and funding. To survive online, targeted users and alternative technology platforms must also find ways around the central points of control that other services may exploit or cut off at their discretion.

Call it the Parler Paradox. The conservative-friendly social networking platform heeded the free market wisdom. Its creators were unhappy with heavy-handed social media moderation, so they built a competitor more to their users’ likings in 2018. Millions flocked to the upstart app, whose culture and conversational tone could be surmised from top users such as Sean Hannity and Rudy Giuliani. It grew steadily in popularity as conservatives sought more hospitable conversational climates before rocketing up the most-downloaded app store lists in the wake of the chaos at the Capitol.

Then came the shutdowns. Apple and Google removed Parler from their app stores, limiting new users’ abilities to install the software on their phones. Then Amazon Web Services—the largest cloud hosting provider—made the unprecedented step of refusing service for Parler, knocking the site offline. The social network remains inaccessible even on standard web browsers as their techies scramble to find or build alternative hosting. It may fold if it finds few enough sympathetic providers.

It doesn’t matter whether these actions were biased or hypocritical. By volume, legacy social media platforms necessarily host more violent and hateful speech because of their much larger user bases. Even from the establishment’s moral framework, there are more calls for “right wing insurrection” (let alone apolitical horrors like child exploitation) on big tech platforms with their billions of users despite earnestly tough takedowns. Indeed, the chief complaint from the left is that platforms do not censor enough.

Parler was deplatformed because it could be. It was a centralized application requiring centralized identity verification—user data was later hacked and posted publicly—that relied on other centralized services. There is a failure point at each centralized node. After Parler had flouted enough establishment conventions—namely, not replicating or even enhancing the incumbents’ moderation policies and thereby defeating its purpose—it got the boot.

It’s not just a question of “Silicon Valley leftism,” either. Yes, many of these companies are overwhelmingly left-leaning. Yes, many of their employees were surely quite happy to take part in these deplatformings. But such controls are tantalizing to all power centers, public and private. Dominant groups can and do lean on central platforms to effectuate their own objectives. Service providers may be more moved to deplatform clients due to pressure campaigns than personal malice.

The deplatforming problem is properly understood as an inevitable consequence of centralized computing. Minimizing the risk of being deplatformed means technologically minimizing the number and power of trusted central nodes that can exert control over any user. In other words, it means moving to a networking environment that is more decentralized by design.

This is a technical problem, although it has taken on an overtly political dimension in our times. We are accustomed to an internet environment that is largely operated by what we can call “trusted third parties”—centralized bodies that manage things like user identity, security, hosting, and communications. But for many functions, there is no technological reason that this should be the case.

Consider email. We do not all rely on the same email provider in order to send asynchronous messages online. There is a core set of rules or “protocol,” like IMAP and POP3, that any service provider or individual may use to connect to the “email network.” We can choose to use email provided by Google or by our ISP or we can choose to run our own email server. It is a fairly decentralized system, although centralized providers operate atop it.

There is no such protocol for “Facebook”—you either have an account with them or you don’t. You can’t go to an alternative provider and interact with Facebook users or operate your own “Facebook server” that you can use to interact with the social network. Facebook is the trusted third party for the “Facebook network” and determines the rules and protections of the game (such as there are). Facebook is computationally sovereign, and users are their serfs.

Many of the most controversial internet services use this trusted third-party model. We don’t hear calls for email protocols to moderate the content of user messages. Where there is centralization, there is the potential for control. This means power centers will fight over who gets to benefit from those controls.

Trusted third parties did not come to dominate networking for any sinister reason. It was simply easier to do at the time. Centralized computing can be more user friendly and accessible. It does not require delicate peer-to-peer systems that may be hard or expensive to build and maintain. And from an operator’s point of view, it provides an easy road to profitability—just look at the riches that digital advertising has brought to the top platforms. Yet we now see how this convenience comes at great cost to the user.

The challenge, then, is to develop working alternatives to centralized networks that minimize the potential for control by eliminating the need for trusted third parties as much as possible. This is difficult, but not impossible, and in fact has been something of a trend in computing.

Bitcoin is the best example, both as a technical illustration and a sociological case study of a decentralizing technology that challenges an existing computing arrangement.

Before cryptocurrency was developed, all digital transactions needed to be managed by a central party like a bank or credit card company. These trusted third parties were simple, user-friendly ways to transfer money. Yet they bring downsides. Users must trust that intermediaries do not get hacked or lose their money. They must trust that intermediaries don’t share financial data with governments or other corporations. And they must trust that intermediaries faithfully execute their transactions and do not make mistakes or block them due to public or private pressure. All of these worst-case scenarios occur quite frequently.

Cryptocurrency replaces trusted third-party computing with decentralized computing. Rather than being managed by a single third party, cryptocurrency transactions are facilitated by a voluntary global network of computers guided by what is called a “consensus mechanism,” which is a mathematical technique for individual agents to agree to validate new data.

You don’t need to understand the technical details to grasp the simple distinction that a decentralized or “peer-to-peer” network is not controlled by any single party and therefore is not a chokepoint for users or the platforms that build on it. (But as an “open source” project, interested parties may vet the publicly available code and network for security and authenticity at any time—it works, and it works well.) No wonder so many dissidents have turned to cryptocurrency after being deplatformed by central providers like Paypal or Visa.

The logic of Bitcoin can be and is being applied to other networking applications. Some of them literally borrow the cryptocurrency technique of a blockchain to decentralize other online activities.

Consider the LBRY protocol, a YouTube competitor that uses a blockchain to decentralize video hosting and payments to content creators. Because LBRY is a decentralized system, it is censorship-resistant: there is no one for governments or corporations to lean on to shut down content. Rather, users can decide whether or not to watch or engage with content they find distasteful, while service providers that build on the LBRY protocol can set their own rules for acceptable content. Crude deplatforming is not an option.

But there are many other ways to decentralize networking. Like our email example earlier, social networking can be achieved through a protocol that allows choice among hosts. This is the approach taken by the “Fediverse” suite of federated (get it?) servers running open communication protocols.

The most popular is Mastodon, which is an alternative to Twitter. Anyone can set up their own Mastodon instance, which can then “connect” with other servers running the Mastodon software. Moderation is voluntary; servers can decide which other groups to connect with or block.

The free speech alternative Gab is actually (and infamously) run as a Mastodon fork. This caused quite the brouhaha. Mastodon instances tend to be fairly left-leaning, and many servers and applications impose certain language requirements for connection regarding pronouns and so forth. Gab clearly disregards those rules. But even if 99 percent of Mastodon users hated Gab, they could not prevent anyone from applying the protocol. All they can do is cut others off from the instances and services under their control.

Incidentally, Gab’s iterative deplatformings over the years explain why it has proven so resistant in recent days when compared to the far more centralized Parler. Gab runs on Mastodon, largely hosts its own servers, and has accepted cryptocurrency payments for years. There are still chokepoints that may be exploited in the near future, but it serves as a case study in how decentralization promotes robustness.

Encryption is another key piece of the decentralization toolkit. These mathematical techniques to conceal data from unwanted parties are receiving new interest in this world of subversion and snitching. The encrypted communications app Signal is one of the most popular app store downloads this year.

Yet even Signal is not ideal. Although it is an open-source application, users must trust the Signal project to responsibly run their servers and not gather and analyze metadata on communications. Furthermore, user identities are tied to real-world phone numbers. And the Signal project has raised eyebrows for its funding and support by the State Department’s Open Technology Fund. (Telegram, another popular choice, is not even end-to-end encrypted for all communications and uses a closed-source, proprietary encryption standard at that. It is far more centralized than Signal and may soon receive the Parler treatment as it has become a strange haven for Parler expats.)

One strong candidate is the Element/Matrix project, an encrypted chat application that allows users to either run their own servers or select from their choice of servers, similar to the Fediverse. Further, identities are not tied to cell phone numbers, which grants more privacy. Because Matrix chats are more decentralized, they are more censorship-resistant.

On the bleeding edge, there are efforts to bring decentralized computing to even the OS level. This is the ambitious goal of the Urbit project, a built-from-scratch open-source computer, network, and identity system that places users in control of their data along most steps of the “computing stack.” In other words, it integrates the various self-sovereign computing techniques discussed in this article into an entire system. It is a little harder to use, but it works, and it is an attractive option to those seeking more control of their data and networking.

Conservatives now find themselves in the same camp as those deplatformed for flouting other establishment conventions over the years, a ragtag group including intellectual property pirates and transparency activists and encryption pioneers and contraband dealers and foreign dissidents and, yes, good old-fashioned evildoers. If conservatives wish to continue their online communities unimpeded, they will need to quickly learn, adapt, and build upon the techniques of decentralized networking that have served other targeted groups.

To technological determinists, the development of decentralized computing was a foregone conclusion. In the words of privacy activist John Gilmore, “the ‘Net interprets censorship as damage and routes around it.” By shielding third parties from user-submitted content liabilities, policies such as Section 230 of the Communications Decency Act made centralized computing more profitable in the short term. Now that the costs of such arrangements are becoming clearer to coders, robust alternatives are finally being built.

This is not to say that centralized computing is going anywhere anytime soon. Although decentralized applications are indeed in development, this does not mean most people will actually use them. It’s not just a question of accessibility: it is about as easy to set up a Mastodon account as it is to sign up for Twitter. Rather, the problem is one of network effects. Centralized platforms are simply where most of the users are, which makes them sticky. It will take a lot to induce an average person to switch to a decentralized alternative.

The future of networking may therefore be multi-tiered, with the bulk of non-offensive, advertiser-friendly content corralled on controlled central platforms. Meanwhile, a more inscrutable “freeweb” will operate in tandem where users can freely connect beyond the control of any one party. The freeweb will be at the same time more open and more closed: respectively both censorship-resistant and protected by encryption to be digitally accessible by only the intended parties. Less mass broadcasting, more cloistered communities. Targeted groups will be able to communicate, but their potential audiences will be much smaller.

There is an irony for the political establishment now wielding the levers of centralized network control. Their incentives to crush political enemies are obvious. Their “deradicalization thesis” holds that deplatforming will splinter enemy groups and limit communicative reach as they are pushed to smaller and remoter platforms—starved of messaging, weaker acolytes will eventually tune out. Yet by catalyzing a move to censorship-resistant technologies, the establishment is seeding a more robust (yet perhaps smaller-scale) alternative internet infrastructure that undermines the project of “deradicalization.” It’s a long-term strategic blunder.

It’s also a bit of an awkward position for conservatives. Distributed systems facilitate a lot of filth. Beyond moral scruples, there are tactical risks. It would be trivially easy for agitators to post violent rhetoric to a freeweb network that could either manipulate users or poison the well and cast suspicion on anyone who uses these technologies. Bitcoin advocates spend a lot of time explaining why the good uses of decentralized technologies far outweigh any potential misdeeds. Conservatives may soon find themselves doing the same.

Regardless, any activists who wish to continue online activities with less fear of deplatforming will turn to decentralized technologies by necessity. Those who simply try to create a “better Parler” will be subject to the same paradox. It’s a lesson better learned sooner than later.

Andrea O’Sullivan is the Director of the Center for Technology and Innovation at the James Madison Institute in Tallahassee, Florida.

To read the full article in the American Conservative, click here.