“Protocols, Not Platforms”: A Mantra For The (We Can Only Hope) Post-Twitter Internet

CW – briefly mentions misgendering & deadnaming

Sooo, Twitter’s a hot mess. I think we all kinda know the modern internet sucks in a lot of ways; privacy is dead, objectivity is dead, customizability and user-friendly design are dead, competition is dead, democracy is dead (not that it was ever really alive). If you’re a Twitter user fed up with Twitter and with Elon Musk doing asshole billionaire things, you might be thinking about switching social networks. You might also be finding it hard to leave your friends on the site. It probably feels like there’s nowhere to go. What if I told you the internet didn’t have to be like this? There is a better way, a better ethos for a more open, collaborative, secure, and user-focused internet: protocols, not platforms.

Let’s start at the beginning. In 1990, Tim Berners-Lee invented three new technologies which would become the basis of the World Wide Web. First was Hypertext Markup Language (HTML), an easy-to-use language for representing textual content with hyperlinks and formatting (markup). Next was the Hypertext Transfer Protocol (HTTP), which described how to transfer HTML documents from a server which stored them to a client who wanted to view them. Finally, there were Uniform Resource Locators (URLs), which described how to find the right server storing the specific HTML page the client wanted.

Here’s the key: these technologies were open protocols. Berners-Lee could have created a program which read opaque files that only he and his program understood and communicated with servers that he controlled. Instead, he carefully described what HTML files were and how to host them, send and receive them, and author them. This meant that anyone who wanted to create and publish content for this new thing called the internet could do so, easily. It also meant that anyone who wanted to read content could do so in any manner they chose. Think about Firefox, Chrome, and Safari. You can choose which browser to use precisely because HTML and HTTP are open standards, so multiple interfaces know how to read them; a browser is just a fancy name for an interface to these protocols. Their openness means you can consume web content in a huge variety of ways. For example, you can choose not to load content from advertiser or content-tracking URLs, you can override websites’ color-schemes, and you can choose to load no images from any page ever. All of this is only possible because Berners-Lee wrote open protocols.

That’s all a protocol is: a standard that describes how to do something, so that it can be implemented by multiple different interfaces, each with different features, business models, etc. We’re used to that with browsers, but think about how different it is from modern social media. Imagine if there was a standard Social Media Protocol which described how to represent data like posts, likes, and friends. Then, different companies (or open-source developers) could implement various user interfaces for interacting with that protocol. So if I want to just see a simple list of my friends’ posts, but you want to see an algorithmically curated set of posts based on your interaction history, we could both do so, by selecting different interfaces for the same protocol. If I wanted a lot of fancy features, I could pay for an interface with a professional development team, modern look, integration with a hypothetical open calendar protocol, whatever; whereas you might pay less or even nothing for a basic implementation with none of those bells and whistles.

You might have heard of people leaving Twitter to something called Mastodon. Mastodon is exactly what I just described: an interface which implements an open social networking protocol called ActivityPub. If someone doesn’t want to use Mastodon, they can use other implementers of this protocol, and still communicate with their friends’ posts from Mastodon. Compare that to Twitter, where you have to be on Twitter to talk to others on Twitter, and therefore you are always subject to their algorithm, spam, questionable moderation, data collection, targeted advertising, terrible UI choices, etc. etc. etc. If Mastodon makes a change to their UI — or their privacy policy — which you dislike, you can leave it for another interface. If Twitter makes such a change, you’re kind of stuck, unless you want to lose touch with a bunch of friends.

Email is built on a number of open protocols, including the Simple Mail Transfer Protocol (SMTP). I use a text-based email client from 1995 called Mutt. That’s weird, but I love its customizability, its keyboard shortcuts, its 90s feel, the fact that I can read its source code, the fact that it doesn’t steal all my data. And even though I make this weird choice, I can still email people who are using modern gmail.com, because email is built on open protocols. I have a filter that runs some code to scan emails for my deadname and send them to a separate folder, where I don’t look at them until I know I have the energy. The emails still get sent along the protocol — SMTP knows nothing at all about filtering — I just change how the interface interprets them. I can do that because the interface I use happens to support custom filtering. You can imagine the applications to social media, for example. Content filters could be implemented without needing to wait for a monolith like Twitter to add the feature, and if a smaller interface added the feature before larger ones, it would be much easier to pull users from the larger interface. Users could also add filters for triggers in private, without even the company making the interface knowing. This is not a silver bullet, but it would certainly help with the content moderation issue on social media today — trans people could much more easily filter out misgendering and deadnaming, for instance.

You know how texting between Androids and iPhones is horrible? Group chats randomly get created and recreated, people sometimes just don’t get messages, etc? Texting was originally based on a protocol called the Short Messaging Service (SMS), which is very outdated and not up to modern standards for group chats, security, reliability, embedded content, and more. Modern Android texts use a newer open protocol called Rich Communication Services (RCS). Modern iPhone texts use a closed platform called iMessage. Because Apple refuses to use the modern open protocol, instead preferring a walled-off platform for only iPhone users, the old SMS protocol becomes the fallback when messaging between platforms. And SMS is terrible, so messaging between iPhones and Androids is terrible. If Apple implemented the modern, open protocol, or if they made their proprietary platform open, the mess would go away.

So maybe you’re convinced this model is at least interesting. But who would take on the expensive task of data storage, without the financial benefits of a closed shop? Since data would also be stored in open formats like HTML, there’d be no need for it to be stored on servers owned by the platform you happen to use. Instead, there would be dedicated data store companies, who could (presumably for a cost) store your data completely encrypted, so even they cannot access it, but you still can. (I currently use such a service, called Tarsnap, for backups of my computer, my non-Gmail email data, etc.) Privacy and encryption could be segmented and sandboxed; these cryptographic problems are pretty well-studied. For example, your social network data could be open to anyone who’s online identity (based on something called a private key) is on your friends list, so they can see your posts, but completely private to anyone else. Those who want cheaper services and algorithmically selected content could opt in to data collection by the interfaces they use, while still communicating with their more privacy-focused friends. Again, the cryptography here is pretty solid, unless you think the NSA has broken the RSA encryption protocol; the problem is finding user-friendly ways to implement it.

The examples I’ve discussed here are mostly related to communication and social media, but I hope you can see a little bit about how they generalize. There are also a lot more details to work out that I didn’t have space (or expertise) for here. If you want more, check out a 2019 article by Mike Masnick called “Protocols, Not Platforms: A Technological Approach to Free Speech.” I don’t agree with everything he says (especially the cryptocurrency nonsense), but it’s mostly very well explained, and covers issues like competition, free speech, moderation, harassment, partisan bubbles, and processes for updating these protocols.

The key point, though, is that closed-shop platforms prevent user choice by forcing users who want the benefits of collaboration and communication to participate in whichever platform happens to be dominant, regardless of its subjective features like web design and of its more serious issues like privacy. And yeah, I don’t really have a great plan to get from here to there. But we should remember that there is a better way, the philosophy that made the internet the open sandbox that let Twitter and Gmail overtake MySpace and AOL in the first place: open protocols, not platforms.

Notify of

Inline Feedbacks
View all comments

Related Stories


We would love your thoughts, please comment!x
%d bloggers like this: