In Part 1 in this series, I explained why there are no easy answers to social media censorship. The problem is not a “Big Tech” conspiracy to silence particular voices. The problem is that too much of our public discourse is mediated by private platforms — centralized, proprietary, walled gardens. We’ve traded in the public square and the open internet for proprietary social media.
It doesn’t have to be this way.
To understand the alternatives, we need to look to (1) programs, (2) protocols, and (3) policy.
Unfortunately, there’s no simple “just use X” solution. We need to understand more about technology and policy to protect free speech online. That’s my goal in this post. In Part 3, I’ll offer some more practical suggestions.
Let me summarize up front, but don’t worry if you don’t understand any of these concepts — they’re explained through the rest of this post.
- First, the software programs that we rely on for communication make a big difference. Software freedom matters. Although it’s not sufficient, software freedom is the cornerstone of internet freedom and provides a stronger foundation for protecting free speech online than proprietary software.
- Second, software protocols can provide technological solutions for protecting free speech and defending against deplatforming. Decentralized, federated social networks (like Mastodon) are an alternative to proprietary, centralized, walled gardens, and a focus on protocols over platforms can shift more control from the services to the users.
- Finally, technological solutions can only get us so far. The issue of free speech online has a social and political dimension, though many of the legislative or regulatory proposals today are much more likely to make the problems worse. It’s important to understand the problems correctly to determine which policy solutions can actually be helpful — those that encourage interoperability rather than entrench incumbents.
Through programs, protocols and policy we can shift power from a few private platforms back to the public, and return to the promise of free speech on an open internet.
1. Programs: Software Freedom Matters
The first problem is that users are tenants on social media, not owners. Tenants can be evicted. Too much of the current technopanic is seeking a more benevolent or ideologically-aligned landlord — like conservatives looking to Parler or Gab for refuge. This is short-sighted. Twitter, for example, used to describe itself as the “free speech wing of the free speech party.” Platforms can change, or fail.
We don’t need a better landlord, we need more owners — and we need tenants’ rights when ownership is not possible.
That is, we need software freedom.
What is software freedom?
One key part of the problem is that so much of the software we rely on to participate in public conversations is proprietary, owned by private companies. In the software freedom community, we call this software “non-free” because it doesn’t respect your freedom. It doesn’t have to be that way.
Free (libre) / open source software (FLOSS), on the other hand, respects user freedom. Free software — which is about freedom, not price — allows anyone to run, edit, share or modify the program. By ensuring these freedoms, FLOSS enables shared community control of software, rather than one company having all the power and the users having none. Popular examples of FLOSS are Firefox, Chromium, WordPress (which powers this website), Wikipedia, GNU/Linux, VLC, Android (the core), or Signal. (Here’s a list of free software I rely on.)
Software freedom is not a complete solution to the problem of free speech or privacy on the internet, but it sure makes it harder for abuses to happen. Source code is auditable by anyone, and if the community is unhappy, they have the freedom to “fork” the code and do something different with it (e.g. see Brave, a privacy-focused web browser based off Chromium). I’m just scratching the surface here, and I’ll write more about software freedom in the coming months, but the point is that FLOSS benefits the users by design by protecting freedom, whereas proprietary software tends to benefits its private owners over the interests of the users.
Software Freedom is a Foundation for Free Speech
Software freedom alone can’t solve every problem, but provides a much more solid footing for protecting free speech and privacy online. It gives ownership and rights to the users.
Whether a program is “liberal” or “conservative” doesn’t matter as much as if it’s free (libre) or non-free. Thus, the first part of the solution is technological: We should prefer FLOSS over proprietary software.
Don’t like WhatsApp because its owned by Facebook? There’s a big difference between Signal, an free software app run by the non-profit Signal Foundation, and Telegram, another proprietary service owned by a single company.
Don’t like Twitter’s content moderation policies? There’s a big difference between Parler, another proprietary walled garden owned by a single company with a single point of failure, versus Mastodon — a free software, decentralized social networking platform.
What is a decentralized social network, you might ask? That leads us to the next point: protocols.
2. Protocols, not Platforms: Decentralization and APIs
Protocols can provide technological solutions for protecting free speech online, both at the server level through decentralization and the client level through APIs, rather than focusing on which “platform” is best (e.g. Twitter vs Parler vs Facebook).
(A “server” is just another computer that provides a service – like a website or email or messaging or video calling – and a “client” is the software you use to connect to a server. For example, the proprietary Zoom app is a client that connects you to Zoom’s proprietary servers for video calling, or your web browser right now is a client that’s connecting you to my web server so that you can view my website. A “protocol” is a defined set of rules that allows one piece of software to communicate with another in a defined way, usually behind the scenes out of sight from the user — for example, HTTP is the protocol that allows a web browser to “speak” to a web server.)
What is Federation?
The best way to illustrate the difference between centralized and decentralized systems is with a few examples.
In the early days of email, you could only send messages to other people on the same server. If you were using the University of Toronto’s systems, you could only communicate with other people from the university using that same system. But then a protocol was developed (eventually standardized as SMTP – Simple Mail Transfer Protocol) that works behind the scenes to allow users on one email server to communicate with users on another email server. That is, I can use my email address @blaise.ca to communicate with users @gmail.com or @protonmail.com. This is a federated system — different servers (blaise.ca, gmail.com, protonmail.com) can communicate with each other. It’s decentralized — there’s no one central “email” server, but a decentralized network of many email servers that can communicate with each other.
Another example would be cellular text messaging. At first, you could only send text messages to other users with the same cell phone provider, but a standard protocol was developed (SMS – Short Message Service) and eventually cell phone providers federated, so Freedom Mobile customers can send SMS text messages to customers of Rogers, Bell, or TELUS. There are many less popular examples: SIP is a protocol for federated VoIP, XMPP for federated instant messaging, etc.
Since email is federated, you can pick whatever email service you want and it doesn’t limit who else you can communicate with. You can pick whatever cell phone provider you want, and it doesn’t limit who you can call or text. The software protocols enable communication between servers to create a federated network.
Mainstream social media is clearly not federated. It’s a centralized silo, a walled garden. Facebook users can only communicate with other Facebook users. Twitter users can only communicate with other Twitter users. If you choose a different platform — or if you’re deplatformed — you’re cut off from communicating with people on Twitter or Facebook/Instagram.
It doesn’t have to be this way.
Mastodon is a FLOSS Twitter-like application that connects to a federated social network (using a protocol called ActivityPub behind the scenes). There is no one Mastodon server — it’s like a federated FLOSS version of Twitter. I have an account @balleyne@floss.social — @balleyne on the server floss.social — but I can follow and communicate with people on any other Mastodon server. You or your organization or community could even run your own Mastodon server (like I used to for an older predecessor federated social network, called StatusNet) and still communicate with everyone else. You don’t have to be on the same server as the people you’re communicating with, because the servers are federated to form one decentralized network.
Federation is a Defence Against Deplatforming
If you’re following along so far, it should be obvious why decentralization is a key defence against deplatforming. How can you deplatform someone on a decentralized network? It’s not possible.
Sure, a user could be blocked from any given server — and that’s the right of whoever is running it to boot off people they think are causing problems — but no one server controls everything. If your Gmail account is cancelled, that doesn’t prevent you from ever emailing anyone again using another email account. If I lose my floss.social Mastodon account, I can just create another Mastodon account on another server (or run my own) and still connect to other people on the network. The switching costs are dramatically lower, and no one party is the sole arbiter of whether or not you can access the network.
Also, Amazon’s deplatforming of Parler (however justified or unjustified) could never happen to Mastodon — there’s no one single website to take down.
Think about what this means for content moderation too — there’s no one set of rules for a decentralized network. Mastodon.xyz might have a different set of rules than mastodon.social or floss.social. If you don’t like the rules on one Mastodon instance, you can pick another or run your own — and still communicate with everyone else on Mastodon. Content moderation at scale is impossible, but with decentralized networks, content moderation doesn’t have to scale nearly as far — power is distributed, and so is responsibility for content moderation. Parler created a new walled garden to compete with the content moderation policies of Twitter’s walled garden, but a decentralized network allows different servers to compete on different content moderation policies within the same ecosystem.
There are a lot of silly things about the way Parler has been run, but the most fundamental criticism I have is that they’ve tried to solve the problem of free speech online by creating yet another proprietary, centralized walled garden — when decentralized alternatives to Twitter already exist (and have since 2008…).
Now, decentralized social networks aren’t a panacea either. There are many problems with federation, and lots of pros and cons of different systems as people experiment. Some FLOSS messaging tools, like Matrix, have embraced federation, while others, like Signal, have deliberately eschewed federation.
When I wrote about free (libre) network services in 2011, federation was just one of four considerations. However, protocols that decentralize power are one important factor to consider: Do we need to use a centralized systems, or are decentralized alternatives available?
Third-Party Clients: A Diverse Ecosystem
It is also helpful to look at Twitter’s original approach for another example of how protocols can better protect free speech online. While Twitter was never decentralized or FLOSS, it used to have a powerful API.
An API — Application Programming Interface — is like a custom protocol to allow other software to communicate with a particular system. Twitter’s API allows for people to write third-party Twitter clients that can communicate with Twitter’s service. The Twitter API isn’t a standard protocol used by other services beyond Twitter, but it allows anyone to write software that can work with Twitter. For example, I don’t use Twitter’s Android client. I use a third-party FLOSS Twitter client called Twidere, which uses Twitter’s API to connect to Twitter.
Whereas federation protocols allow for multiple servers, an open API allows for multiple clients.
In the early days of Twitter, the proprietary service was quite open with its API. It was common for Twitter uses to pick their favourite client among many, many different options, all with different features and interfaces. Tweetdeck was a popular proprietary third-party client (before Twitter acquired it). I used Gwibber under GNU/Linux. Twitter initially didn’t have a search engine, but Summize built a third-party Twitter search engine using the API! (Twitter later acquired Summize to turn it into the official Twitter search engine.) There was a rich ecosystem of third-party applications for Twitter, when it used to have a strong API — the protocol that enabled it.
Think about email again, and the wide variety of email software you can use from Mozilla Thunderbird to Microsoft Outlook to Apple Mail, or K-9 or Gmail’s web interface or MailChimp, etc. Facebook is not like this, nor is Instagram or TikTok, and Twitter is no longer like this — but Mastodon and Matrix are.
(Android is another good example of a relatively open API — you can use the Google Play store, but you can also install alternative software repositories like F-Droid to manage your apps instead, or swap out any of the Google apps for alternatives.)
Protocols can enable decentralization at the server level (through federation) and a diverse and competitive ecosystem at the client or app level (through a combination of protocols and APIs). Another question we can ask is: Does a service have an API that allows people to write third-party software for it, so you can choose between more than one client? Or do you have to use software from the service provider to use the service?
Make the Internet Open Again
John Gilmore famously said that “the Internet perceives censorship as damage and routes around it.” This may have been true in the early days of the open web, but it is no longer true in the age of proprietary, centralized social media. However, it doesn’t have to be this way.
Mike Masnick presents a detailed argument in his essay Protocols, Not Platforms: A Technological Approach to Free Speech on how we might be able to return to a decentralized internet built around protocols. Here is the key argument (emphasis added):
This article proposes [an approach] that enables more free speech, while minimizing the impact of trolling, hateful speech, and large-scale disinformation efforts. As a bonus, it also might help the users of these platforms regain control of their privacy. And to top it all off, it could even provide an entirely new revenue stream for these platforms.
That approach: build protocols, not platforms.
To be clear, this is an approach that would bring us back to the way the internet used to be. The early internet involved many different protocols—instructions and standards that anyone could then use to build a compatible interface. Email used SMTP (Simple Mail Transfer Protocol). Chat was done over IRC (Internet Relay Chat). Usenet served as a distributed discussion system using NNTP (Network News Transfer Protocol). The World Wide Web itself was its own protocol: HyperText Transfer Protocol, or HTTP.
In the past few decades, however, rather than building new protocols, the internet has grown up around controlled platforms that are privately owned. These can function in ways that appear similar to the earlier protocols, but they are controlled by a single entity. This has happened for a variety of reasons. Obviously, a single entity controlling a platform can then profit off of it. In addition, having a single entity can often mean that new features, upgrades, bug fixes, and the like can be rolled out much more quickly, in ways that would increase the user base. […]
Moving to a world where protocols and not proprietary platforms dominate would solve many issues currently facing the internet today. Rather than relying on a few giant platforms to police speech online, there could be widespread competition, in which anyone could design their own interfaces, filters, and additional services, allowing whichever ones work best to succeed, without having to resort to outright censorship for certain voices. It would allow end users to determine their own tolerances for different types of speech but make it much easier for most people to avoid the most problematic speech, without silencing anyone entirely or having the platforms themselves make the decisions about who is allowed to speak.
In short, it would push the power and decision making out to the ends of the network, rather than keeping it centralized among a small group of very powerful companies.
Moving back to a focus on protocols over platforms can solve many of these problems… potentially creating the best of all worlds: useful internet services, with competition driving innovation, not controlled solely by giant corporations, but financially sustainable, providing end users with more control over their own data and privacy—and providing mis- and disinformation far fewer opportunities to wreak havoc.
Masnick engages with a lot of the challenges with protocols that led to the recent pendulum swing towards platforms, but outlines the potential for solutions to many of these problems. And people are taking notice, like Twitter’s CEO Jack Dorsey:
Twitter’s Blue Sky is an exciting new initiative, though it is far from the first attempt to do this. However, maybe Twitter has the clout to make it happen.
There is, however, a limit to how far technological solutions can get us when we’re dealing with a fundamentally human problem. If companies don’t elect to adopt these technological solutions, how can we decentralize the internet? This is where we need to turn to policy.
3. Policy: Encourage Interoperability, Don’t Entrench Incumbents
The goal of any policy intervention should be decentralization. Unfortunately, too many policy proposals seek to regulate centralized walled gardens, rather than to decentralize them. This is more likely to entrench the incumbents than to fundamentally change the status quo.
Regulating Big Tech Makes Them Stronger
The problem with free speech online is that we’re far too dependent on a few private companies for public discourse. Why on earth would we want to further entrench their dominance? That’s not the intended goal of most popular policy proposals, however it would certainly be the effect.
Facebook’s ideal outcome would be to avoid regulation. However, their second best outcome is to be regulated in such a way that only they could comply.
Many of the policy proposals from across the political spectrum would place state-like duties or immense legal liability on companies like Facebook, making it so expensive to comply that only these big companies would be able to live up to the laws or survive the lawsuits. This is a great way to entrench the tech giants and stifle competition from new startups. Silly attempts to “punish” Big Tech are more likely to backfire than solve the problems of free speech online. Raising the cost of running a social media platform is a formula for entrenching the incumbents, not encouraging competition.
If the problem is that we’re too dependent on a few companies for online speech, increased competition should be the goal.
To encourage competition and bring about a more decentralized internet, we need interoperability.
Adversarial Interoperability
Interoperability is the technical ability to plug one product or service into another product or service. It’s precisely what protocols enable. Interoperability is the antidote to the centralized power of the incumbents.
Cory Doctorow and the Electronic Frontier Foundation have been advocating for interoperability as the answer to protecting free speech online. Doctorow argues we should use interoperability to fix the internet, not the tech companies, and that we should revive adversarial interoperability as a weapon against today’s tech monopolies, rather then entrenching their dominance through ill-conceived regulation. Mai Sutton makes a similar case on Techdirt in Defeating Tech Giants with Open Protocols, Interoperability, And Shared Stewardship.
The basic premise is that interoperability is sometimes “indifferent” (e.g. power adapters for car cigarette lighters), sometimes cooperative (e.g. defining USB as a standard for computer and device manufacturers to implement), but often it’s adversarial — a weapon used to challenge monopolies through competition without consent.
Think of the tech giants of the past. 20 years ago, Microsoft was the dominant player, and the subject of much anti-trust attention. They used their dominance with Microsoft Office to lock people into Windows. There was a time when you needed Microsoft Office to reliably open Microsoft Office documents. What happened? Apple and OpenOffice.org waged a war of adversarial interoperability through reverse engineering their file formats — despite Microsoft’s opposition. Today, you can use The Document Foundation’s LibreOffice, Apple’s iWork, or Google Docs to read and write Word, Excel and PowerPoint files. You no longer need Microsoft Office — and therefore you no longer need Windows.
Or take Microsoft leveraging their Windows market share to lock the web into Internet Explorer and its proprietary extensions. What happened to Internet Explorer? Firefox (a FLOSS browser), open web standards, and other browsers like Apple’s Safari and Google Chrome broke apart Microsoft’s grip of the browser market.
(Interestingly, Doctorow’s elementary school friend from Toronto, Tim Wu, writes about this dynamic of the cycle between “open” and “closed” information systems in his 2010 book, The Master Switch.)
The reason why interoperability is so important is that switching costs are a major barrier to change, and thus a major force of locking us into the incumbents. Sure, you can leave Facebook, but how will you talk to your friends or get invited to events (when those become a thing again)? You can pick your favourite chat program, but how will you talk to your friends on another system? Degooglifying is not easy either.
Interoperability makes switching easier, because it breaks down silos.
What does this have to do with policy? The goal of policy interventions should not be to “punish” (and ultimately entrench) the tech giants, but to remove the barriers to adversarial interoperability.
As Doctorow explains:
Adversarial interoperability is the consumer’s bargaining chip in these coercive “negotiations.” More than a quarter of Internet users have installed ad-blockers, making it the biggest consumer revolt in human history. These users are making counteroffers: the platforms say, “We want all of your data in exchange for this service,” and their users say, “How about none?” Now we have a negotiation!
Or think of the iPhone owners who patronize independent service centers instead of using Apple’s service: Apple’s opening bid is “You only ever get your stuff fixed from us, at a price we set,” and the owners of Apple devices say, “Hard pass.” Now it’s up to Apple to make a counteroffer. We’ll know it’s a fair one if iPhone owners decide to patronize Apple’s service centers.
This is what a competitive market looks like. In the absence of competitive offerings from rival firms, consumers make counteroffers by other means.
There is good reason to want to see a reinvigorated approach to competition in America, but it’s important to remember that competition is enabled or constrained not just by mergers and acquisitions. Companies can use a whole package of laws to attain and maintain dominance, to the detriment of the public interest.
Today, consumers and toolsmiths confront a thicket of laws and rules that stand between them and technological self-determination. To change that, we need to reform the Computer Fraud and Abuse Act, Section 1201 of the Digital Millennium Copyright Act, patent law, and other rules and laws. Adversarial interoperability is in the history of every tech giant that rules today, and if it was good enough for them in the past, it’s good enough for the companies that will topple them in the future.
There is no silver bullet. Trying to craft laws aimed at big tech companies will likely make it so that only companies that big can survive. Rather, we need to look at the thicket of laws and rules that create barriers for adversarial interoperability and competition, and prevent interoperablity from being used as a weapon to restore competition to the market — from “right to repair” laws to copyright and patent reform and reform of other laws that punish competitors who seek to offer interoperable alternatives.
We need policy that protects adversarial interoperability and promotes competition.
(Anti-trust is another important mechanism to seriously consider to encourage competition. However, the challenge with anti-trust is trying to define “the market” in an ever-changing technological environment. Are WhatsApp, Facebook, and TikTok in the same market? What about YouTube? Or FaceTime? Is Facebook Messenger in the same market as Zoom? Did the “Zoom” market exist 5 years ago? Anti-trust seeks the right goal — breaking up large monopolies to encourage competition — but I’m concerned it’s not an effective tool in this context because it’s very difficult to define what a monopoly is in the rapidly changing technology space.)
Conclusion: Asking the Right Questions
When we start from the premise that the problem of free speech online is simply a culture war issue — simply a matter of some evil tech giants that are either too intolerant of conservatives or too tolerant of “misinformation” — we end up with “solutions” that will only make the problem worse. The problem is that the public square has become a shopping mall. You don’t solve that problem by giving shopping malls state-like duties, or by moving to some sketchy plaza down the street.
Our public discourse is too dependent on a few private companies. The solution is more distributed ownership.
The way we can get there is through programs that respect user freedom, protocols that decentralize the internet, and policies that promote interoperability and competition.
The questions to ask are: Is software system FLOSS or proprietary? Does it interoperate or is it a walled garden? Is it federated or centralized? (Not “is it liberal or conservative?”)
Slogans won’t protect free speech online, nor will new proprietary walled gardens, or trying to simply punish the big guys. We need to redecentralize the internet and pursue software freedom, interoperability and competition to ensure that public discourse can thrive online.
