The Technological Poison Pill: How ATProtocol Encourages Competition, Resists Evil Billionaires, Lock-In & Enshittification
Disclosure: I’m on the board of Bluesky, so feel free to take as many grains of salt as you want in reading it, even though part of this is cheering on a new entrant looking to build an alternative to Bluesky.
There’s been some debate over the last year or so regarding Bluesky and how decentralized it really is. There has also been a growing fear that “enshittification is inevitable.” Or, worse, that an “evil billionaire” might take it over and ruin it the way other platforms have been ruined.
But I think it’s important to understand that Bluesky has, effectively, created a technological poison pill: by building on an open protocol, ATprotocol, the system itself can be rebuilt outside of Bluesky, but in a way where everyone can continue to communicate, and that creates incredible incentives that undermine any evil billionaires, and would actually punish Bluesky (or anyone else!) should they try to enshittify.
Last week, a group called Free Our Feeds announced itself to the world and kicked off a crowdfunding process to effectively build a Bluesky competitor, built on the same ATprotocol and fully interoperable with Bluesky, but wholly separate from the app.
This is both exciting and fantastic, in part because it’s cool, and in part because it demonstrates the real-world impact and importance of ATprotocol’s open design, showing how it enables the creation of alternative infrastructure that can prevent lock-in and empower users.
The Enshittification Fear:
For a few years now, Cory Doctorow’s concept of “enshittification” has been a hot topic in tech circles, with many worried that even the most well-intentioned platforms are doomed to become terrible over time.
Indeed, just a few months ago, Cory wrote a thoughtful piece about why he was not joining Bluesky and why he feared it was on the path to enshittification. He and I had actually discussed all of this much earlier (very early in Bluesky’s history) and I suggested to him that Bluesky had some tricks up its sleeve to be enshittification-resistant. In the piece, Cory says some very nice things about me before (correctly!) saying that even though he trusts me deeply, he doesn’t think that his trust of me (and me being on Bluesky’s board) means Bluesky is immune to the enshittification curve:
Bluesky has many federated features that I find technically admirable. I only know the CEO there slightly, but I have nothing but good opinions of her. At least one of the board members there, Mike Masnick, is one of my oldest friends and comrades in the fights for user rights. We don’t agree on everything, but I trust him implicitly and would happily give him the keys to my house if he needed a place to stay or even the password for my computer before I had major surgery.
But even the best boards can make bad calls.
And he’s correct. The best boards can make bad calls. And I can certainly make bad calls.
But the secret behind Bluesky was not that it has an amazing CEO or a non-evil board. It’s that it was built from the ground up with a focus on forced openness and a protocol on which anyone could build. I discussed this in great detail a few months ago on Ed Zitron’s Better Offline podcast.
The key points:
- Nothing can be enshittification-proof, but you can make things enshittification resistant.
- The key to doing so is building on a forced-open protocol, such that if people hate what you’re doing, they’re able to rebuild any part of the infrastructure and cut out the entity engaged in enshittification.
- Even if the alternative competing services don’t exist, the simple fact that the option is there for people to keep their content, keep their relationships, keep their ability to communicate while avoiding any particular platform, is a very strong incentive to resist enshittification.
- This is because even the looming possibility that someone can come in, piggyback on the existing network, but with their own infrastructure, creates incredibly strong incentives for any player in the space to avoid giving reasons to people to leave.
Going back to Cory’s original formulation of “enshittification,” I can explain this further:
Here is how platforms die: first, they are good to their users; then they abuse their users to make things better for their business customers; finally, they abuse those business customers to claw back all the value for themselves. Then, they die.
In traditional, centralized systems that shift away from being good to users to being good to their business customers is where the slide begins. For traditional companies, though, they can get away with it, because users are “stuck.” As Cory well knows from his work on adversarial interoperability, the real problem is lock-in.
Once a platform has you, it can start to squeeze you if you have nowhere else to go. And, in the case of social media, that’s particularly tricky, because you want to be where your social graph is, so once you’ve really built up connections, a platform has you.
But if the network is based on an open protocol, in which alternative infrastructure can be built, then any player in the system has a greatly diminished incentive to start being bad to users in favor of other constituents, because the worse you get for users, the more opportunity there is for someone else to jump in and offer something better.
I think that many people, though not necessarily Cory, have zeroed in on the idea of “VC funding” as the root cause of enshittification, rather than the lock-in. And it is true that some VCs might be looking to invest only in centralized platforms that have built-in lock-in, but it’s possible to recognize an alternative approach to building a sustainable business: by treating users well, allowing everyone to build on the same open network, and recognizing that this makes the whole system more valuable to everyone.
That’s what Bluesky is trying to do with the ATprotocol.
The company has said from the beginning that “Bluesky” itself is just a reference app, and the point is for others to build. Indeed, part of the company’s own mission is that “the company is a future adversary.”
As Cory notes, no CEO or board can protect against that. But building an open network that enables third parties to build every bit of the stack as alternatives does help protect against that.
Free Our Feeds!
And now it’s happening. With the launch of Free Our Feeds, which is running a crowdfunding project and looking to raise $30 million over the next three years, we’re seeing that fully independent infrastructure on the path to being built. And hopefully they won’t be the only ones.
The plan is to build entirely separate infrastructure, but all using the ATprotocol, so that anyone on Bluesky (or other ATproto services) can interoperate with the new service.
Bluesky is an opportunity to shake up the status quo. They have built scaffolding for a new kind of social web. One where we all have more say, choice and control.
But it will take independent funding and governance to turn Bluesky’s underlying tech—the AT Protocol—into something more powerful than a single app. We want to create an entire ecosystem of interconnected apps and different companies that have people’s interests at heart.
Free Our Feeds will build a new, independent foundation to help make that happen.
This isn’t just about bolstering one new social media platform. Our vision offers a pathway to an open and healthy social media ecosystem that cannot be controlled by any company or billionaire.
And, notably, among the names signed on to support it is Cory Doctorow, which is exciting to see.
This ability to build alternative infrastructure is possible (despite Cory’s fears in his piece), it just takes resources. Cory talked about the lack of “federation” in his piece, suggesting that Bluesky had somehow failed to “federate”:
Bluesky lacks the one federated feature that is absolutely necessary for me to trust it: the ability to leave Bluesky and go to another host and continue to talk to the people I’ve entered into community with there. While there are many independently maintained servers that provide services to Bluesky and its users, there is only one Bluesky server. A federation of multiple servers, each a peer to the other, has been on Bluesky’s roadmap for as long as I’ve been following it, but they haven’t (yet) delivered it.
And while it is true that Bluesky is, currently, the only source for some aspects of the ATprotocol stack, it has been built so that the other parts can be replicated elsewhere. Admittedly, some of it is more complicated than other parts, but it is possible.
And that’s what the new Free Our Feeds effort is trying to do.
So while Cory was worried that this was something Bluesky had refused to do, the reality is that the possibility of doing this has been there for a while. The problem is that it’s not simple. And it needed someone else to come along and build what was open for them to build.
Because if Bluesky built it itself, then it’s not a third party that is independent. It’s still Bluesky. And that’s why Free Our Feeds is so exciting. They’re proving that a third party can build out a system entirely independent of Bluesky, the company. And, admittedly, building the full stack is not cheap, which explains why Free Our Feeds is working towards an ambitious funding goal.
The ability for a third party like Free Our Feeds to build an entirely separate system while still allowing users to communicate across apps is the key to ATprotocol’s “technological poison pill” effect. Even if Bluesky or another provider tries to act against users’ interests, people can seamlessly shift to an alternative without losing access to their social graph and data. This creates a powerful counterweight to the usual lock-in and network effects that enable enshittification.
The Company is a Future Adversary
And this is where it’s important to understand some of the fundamental differences in how something like Bluesky/ATproto works and something like Mastodon/ActivityPub work. They have a slightly different approach to trying to tackle the same problem. Each are trying to create decentralized, protocol-based social media tools, but they take a fundamentally different approach.
ActivityPub works on the theory that almost anyone can effectively build and host “their own” mini-Twitter-like service. And then that mini-Twitter can speak to many of the other mini-Twitters, with the ability of any of them to “defederate” (or block all communications) with other mini-Twitters, as needed.
That defederation aspect is a unique (and fascinating!) kind of incentivizing tool, as platforms that want to be good neighbors have incentives to police their own mini-Twitters. But it also creates some challenges. It’s tough to run your own mini-Twitter, especially if you allow more than just yourself to use it. There’s a fair bit of work involved. And then managing users, managing which servers you defederate from, etc., is a chore.
Thankfully (yay, open systems) some people have been building tools and services to make that better and easier, but it has remained a challenge.
The ATProtocol approach is somewhat different. You can federate some aspects of things, such as hosting all your own data on your own PDS or Personal Data Server (which is great, as it means you have full control over your data, not Bluesky or anyone else), but it’s not designed for a random individual to spin up an entire mini-Twitter.
The philosophy is more that different parts of the stack may require different players to be involved, and some of them may require more resources than others. Running your own PDS is relatively inexpensive and easy. Running your own relay is more challenging and expensive but wouldn’t necessarily need a corporation. Other pieces require more, and that’s what Free Our Feeds appears to be building.
And, again, the most important bit is that this is always possible on this network. Because “the company is a future adversary.”
Bluesky’s approach directly addresses the enshittification fear by ensuring that no single entity, not even Bluesky itself, can gain too much control over the network. The open protocol acts as a check on any potential abuse of power.
The fact that Free Our Feeds can do this in the first place is almost more important than whether or not they actually succeed (which I hope they do!), because it creates strong incentives for Bluesky, the corporate entity, not to enshittify.
Indeed, if you look back at the history of Twitter, in the early days, it encouraged open development and building, but it wasn’t a protocol where the entire stack could be recreated. And, when one entity started buying up many of the independent developers with a pretty explicit plan to “steal away” all of Twitter’s users, Twitter started locking stuff up and blocking that ability, because there wasn’t an open protocol and there wasn’t any possibility of rebuilding certain parts of the infrastructure.
In contrast, Bluesky was built from the ground up for this very thing. And you can see that difference in how the Bluesky team has reacted to Free Our Feeds: many employees, including top management, are cheering on the Free Our Feeds team even as, ostensibly, they’re building “a competitor.”
This sincere welcoming of potential “competitors” is practically unheard of in the tech world. But it reflects a fundamentally different mindset enforced by the open protocol, one focused on growing the pie for everyone rather than jealously guarding a slice. It’s a recognition that, in a world of open protocols, a rising tide can lift all boats.
Of course, some may argue that the technical complexity and costs involved in building out alternative infrastructure will still limit how many can truly compete with dominant players, even with an open protocol. And that’s a fair point. Spinning up a full social media stack is no trivial task, as the $30 million Free Our Feeds campaign underscores.
But the key is that it’s possible, and that possibility acts as a check on bad behavior. Moreover, as a robust ecosystem emerges around the protocol, we can expect to see more tools and services that lower the barriers to entry. Already, the rapid pace of development and the ease of building new user experiences on top of ATProtocol hint at a future where a vibrant alternative tech scene, with empowered users, can thrive.
It’s a recognition that, in the networked world, this can be a non-zero-sum situation, and having more players building makes it better for everyone. It also allows for different kinds of experiments, which will create more features that more people are interested in. It’s an approach that is focused on making sure the whole ecosystem grows, rather than one company’s fiefdom.
Get Busy Building
Along those lines, there’s been a lot more development going on elsewhere as well, which is equally exciting. In just the last week, there’s been talk of independent developers building an Instagram competitor and a TikTok competitor on ATprotocol. That last one, by the way, was built in just a few hours. That’s what can happen when you have an open system. Over the weekend, Bluesky itself added to this by soft launching a new view that gives the service a TikTok-like feel. But, again, in an open way such that others can build algorithms and feeds for a similar video-only feed.
Similarly, Flipboard recently released an amazingly slick brand new app, called Surf, that works with both ATprotocol and ActivityPub (and RSS!!) to create a very cool tool for browsing, consuming, sharing, and creating posts across all of these networks. And last week, right after the Supreme Court upheld the TikTok ban, Flipboard’s CEO/co-founder Mike McCue showed off SkyTok, a quickly created feed (using Surf) that creates a simple TikTok-like experience. And, over the weekend, they tested SkyTok with Bluesky’s new video rendering setup as well.
And people are taking notice. Famed entrepreneur/investor Mark Cuban put out a call for proposals, saying he’d like to fund a TikTok competitor built on ATProtocol, so I imagine the links above won’t be the only examples of people building cool stuff.
Most of these alternative apps are really building different looks at Bluesky’s implementation of ATprotocol, rather than a fully independent stack. Think of it as services that build on-top of Gmail. But that’s also why the Free Our Feeds effort is so cool. It’s like someone is coming along and building an Outlook to compete with Gmail. And, assuming they’re successful, these alternative apps (like the TikTok-style apps) should be able to easily work with it as well. Or any other third party that builds out the infrastructure.
This is a case where the more people building on this open protocol and open network, the more it helps everyone.
And, it does so in a way that is still easy for people to use. Most users don’t need to know any of this is happening, or about ATprotocol in the background. It’s just creating the kind of more open web that we all need, without the lock-in.
Again, it’s that lock-in that creates the eventual enshittification. Without lock-in, any app could still enshittify, but the risks for that app would be much bigger, because it’s so easy for users to exit. It won’t be like leaving Twitter for Bluesky where you are effectively “starting over,” it will basically be “Oh, I don’t like how Bluesky is acting, so I’m just switching over to the Free Our Feeds system” where… you don’t lose any of your posts (they’re in your own PDS), you don’t lose any of your connections (your social graph is really yours), and you remain in full control.
This is what the early internet promised us, but it got lost in the early 2000s when big companies came along and effectively colonized open protocols (or recreated them as closed silos with nicer user interfaces). This time around, though, people are learning to create user-friendly interfaces with open protocols.
From Ulysses Pacts to Technological Poison Pills
In Cory’s piece, he talks about the concept of the Ulysses Pact, which is what he requires of any new service:
There’s a name for this dynamic, from the world of behavioral economics. It’s called a “Ulysses Pact.” It’s named for the ancient hacker Ulysses, who ignored the normal protocol for sailing through the sirens’ sea. While normie sailors resisted the sirens’ song by filling their ears with wax, Ulysses instead had himself lashed to the mast, so that he could hear the sirens’ song, but could not be tempted into leaping into the sea, to be drowned by the sirens.
Whenever you take a measure during a moment of strength that guards against your own future self’s weakness, you enter into a Ulysses Pact.
He argued that Bluesky didn’t have that because it hadn’t “federated.” But it had. It had locked the protocol open so that anyone could build. And now they are.
I think a better way of thinking about this isn’t the “Ulysses Pact,” but rather a technological poison pill. I had seen some people saying on Bluesky that the company needed to create some sort of “poison pill” in its financial setup to ward off evil potential buyers who might “make an offer they can’t refuse.”
But what Bluesky has done with ATproto is even better: it’s not relying on some financial contract. It’s created a technological poison pill, such that even if Bluesky (the company) was offered a deal it couldn’t refuse, others could just rebuild the stack… outside of Bluesky’s control (but where users could continue to communicate with each other), and Bluesky could do nothing to stop them.
Beyond enabling the “easy exit” Cory wants, it also acts to ward off “evil billionaires,” because as soon as they act evil, the poison pill is there to give everyone an escape route, thereby effectively destroying any evil billionaire’s plans. An evil billionaire has less reason to be evil in the first place since alternatives can spring up and users can exit without cost.
The “Ulysses Pact” here is in the setup. Evil billionaires and enshittification become self-defeating, thanks to the poison pill. That’s not to say it’s impossible. Because you never know what bad decisions some future version (future adversary) might make. But the nature of the locked open protocol means that it’s much easier to deal with that, and that simple fact should hopefully disincentivize any attempts in the first place.
If this approach succeeds, it won’t just protect individual users; it has the potential to reshape the fundamental dynamics of the social web. By reducing the power of walled gardens and returning control to users, an ecosystem of open protocols could realign the incentives of technology companies, ensuring that they prioritize serving their users’ interests to remain competitive. It would mark a major shift back towards the original decentralized vision of the internet.
The rapid pace of development and the ease of building new user experiences on top of ATprotocol are not just exciting for their own sake. They hint at a future where a vibrant ecosystem of interoperable “small tech” can thrive, with a diverse range of user-centric services emerging to meet different needs. Rather than being limited to a handful of monolithic platforms, internet users could enjoy a rich marketplace of apps and services, all built on shared open standards.
That’s the vision I had with my Protocols, not Platforms paper, and now it’s on its way to being truly real. Having the Free Our Feeds folks jump in is not just proof that this is possible, it’s a vote of confidence for the overall setup, and shows how we can actually build enshittification-resistant systems by locking them open as a technological poison pill against lock-in and against the threats of evil billionaires.
If this approach succeeds, it won’t just protect users; it will fundamentally reshape the dynamics of the social web. It will bring us back towards the original promise of the open web where users are in control, rather than giant companies. Companies will still have a place, but the job of platforms will be to serve the users’ best interests first and foremost.