In January 2011, somewhere in the crush of bodies in Cairo’s Tahrir Square, a man held up a handwritten sign. It read: “Facebook: against every unjust” [sic]. He probably didn’t think of himself as making a statement about technology policy. He was making a statement about Hosni Mubarak—the dictator who, within weeks, would be forced from power, in part because tools like Facebook had given Egyptians a way to coordinate, organize, and find each other when state media was telling them they were alone.
Fourteen years later, almost to the day, the man who founded and still runs Facebook sat in a place of honor at the inauguration of an American president openly contemptuous of democratic norms. He was seated ahead of that president’s own cabinet nominees, alongside nearly every other major tech CEO. The same platform. A radically different relationship to power.
Get Liberalism.org in your inbox.
It would be easy to look at these two images and conclude that the techno-optimism of the early internet era was a fraud, and that the tools we thought would liberate us were always going to be captured by the powerful. Easy, and wrong.
But not entirely wrong. Something real did change between those two photos, and pretending otherwise gets us nowhere.
What changed wasn’t the technology, exactly. Facebook in 2011 was already a centralized platform owned by a single company. What changed was that the underlying incentives of that centralized architecture had time to work. Centralized systems create chokepoints. Chokepoints, once they exist, attract everyone with an interest in squeezing them: companies looking to extract more value from users, governments looking to extract compliance from companies, and political movements looking to extract influence from both. In 2011, Facebook hadn’t yet figured out how lucrative those chokepoints would be, or how much leverage they offered to the powerful.
By 2025, everyone had figured it out.
This is the part most debates about tech and democracy miss. The real question is whether the underlying architecture creates incentives that concentrate power or that distribute it. It’s not about whether technology is inherently good or bad, liberating or oppressive. Architecture shapes incentives; incentives shape outcomes. And once you’ve built a chokepoint, the attempts to capture it will be relentless, because the payoff for whoever controls it just keeps growing.
To understand why this change happened, and why it was arguably inevitable given the business models that emerged, you need to understand something about the economics of digital abundance.
The Incentives of the New Middlemen
Digital abundance reshaped what it means to be a cultural intermediary.
The early-internet conventional wisdom was that the network would kill the middleman. Yet the promised era of “disintermediation” never quite happened. What actually happened was that the middlemen changed character.
The pre-internet middlemen—record labels, book publishers, movie studios, magazine editors—were gatekeepers. Their entire job was rejecting 99% of what came through the door so that a curated few could reach an audience. The internet middlemen who replaced them did the opposite. Their approach depended on letting nearly everything through. They became enablers rather than gatekeepers, and an entire generation of cultural production—music, video, writing, podcasts—flowed through them in volumes the old gatekeepers couldn’t have imagined.
But the new middlemen still needed a business model. The successful early internet companies quickly ran into what appeared to be a fundamental economics problem: if anyone can create content, and all this content is available for free, how does anyone make money?
Economics has traditionally been the study of allocating scarce resources. Digital technology broke that frame. Suddenly information could be copied at near-zero marginal cost, and what do you charge for when what you’re providing is effectively free? Google and Facebook found the answer that defined the next two decades. The scarcity that wasn’t going away was human attention. Build a massive ad machine, sell that attention to advertisers, and the more data you collect, the more precisely you can target—which means the more you can charge. To build the data collection architecture meant taking increasing control over the flow of information itself.
So the same platforms that broke the old gatekeepers and let billions of people participate in cultural and political life also built the most concentrated attention-extraction machines in human history. The empowerment was real. So was the price tag: a business model that required centralization. The architecture that empowered users was the architecture that extracted from them. Building one required building the other.
From Enshittification to Despotification
That initial burst of user empowerment produced things like the Arab Spring. Distributed voices that the old gatekeepers had silenced were suddenly able to find each other, organize, and act. But abundance is loud. The same wave of user-generated content that let suppressed voices break through also produced a cacophony, plenty of it garbage, some of it deliberately weaponized.
By the mid-2000s, the dominant complaint about the internet was “information overload.” We had everything; we couldn’t find anything. The fix that won was algorithmic curation, delivered by centralized platforms. RSS readers and personal sites could in principle have solved the same problem, but they required users to do the curating themselves. The centralized platforms said: don’t worry about it, we’ll figure out what to show you.
Most people happily took that deal. And at the time, the deal was fine.
But that trade-off resulted in several downstream impacts that we’re just beginning to reckon with today. The first, as colorfully noted by writer Cory Doctorow, is the enshittification curve:
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.
The companies that embraced centralized control over the user experience did so initially because they were, in fact, making things better for their users. Information overload was a real issue. Having a better system for managing it was a good thing. It’s why so many people flocked to these centralized social media platforms and became so enraptured by their algorithms.
The problem of centralized systems is that they create an irresistible temptation to control and exploit. Users who found value early on feel stuck: they can leave, but doing so means abandoning their community. That lack of easy exit creates lock-in, and lock-in enables enshittification.
And the same chokepoints that let companies extract value also let governments extract power. Those seeking control hunt up and down the network stack for leverage, and centralized providers concentrate it.
Call this despotification: the political analog of enshittification, where the same chokepoints get exploited to extract compliance from platforms—and ultimately to gain control over what people can say and hear.
The temptation of those in power to twist the knobs to their liking became irresistible. This took many forms: X downranking posts with links to external sites, Amazon choosing which products to show you as the promoted results, Instagram choosing which content deserves to be sent to you as a reminder notification, Substack choosing which newsletters to suggest to you. Each of these choices can be tweaked in ways that enable greater usage, engagement, and revenue, and not necessarily in the interests of the users.
They could often justify those tweaks. The incentives are just too great. For CEOs of companies, they were trying to justify their valuations, or they were responding to public pressure regarding what they allowed (or didn’t allow) on their platforms. Governments would float almost any argument that worked, even if they contradicted one another: some wanted control over these tools to prevent disinformation, others to “enable free speech,” while almost all political actors trotted out “protect the children” and “protect national security” at one point or another.
All this was always about gaining some sort of control over a centralized point of leverage, where one could influence what kinds of conversations were allowed, encouraged, or promoted.
Decentralization Brings Back Democratic Ideals
It doesn’t need to be that way.
Remember, the original promise of the web was that it would break down the gatekeepers and set up systems that enable more people to participate in the broader conversation space. It certainly changed the landscape, away from the earlier generation of media gatekeepers and toward more internet-powered enablers. But over the last decade and a half, many platforms have gradually taken the role of the centralized provider, which is tempting for those seeking either economic or political control.
There’s no reason we can’t build a new generation of services that restores that democratic, decentralized promise—an open internet that empowers users rather than funneling control to gatekeepers.
We are already seeing small examples of that happening. In 2019 I wrote “Protocols, Not Platforms,” which laid out how a more decentralized architecture for online services could push power to the edges of the network, away from the chokepoints that make both corporate extraction and political capture so tempting.
There have since been various experiments and attempts to make that a reality. Specific decentralized protocols for social media have found audiences and developers: ActivityPub, Nostr, and Farcaster have all seen experimentation and useful apps developed. Then there’s the ATprotocol, which was developed in direct response to my paper, and Bluesky, where I’m on the board, and which has seen over 44 million users sign up. But much more importantly, the underlying protocol is now inspiring a larger, decentralized reinvention of the way the entire web works (in some ways, Bluesky itself is the least interesting development in the ecosystem, and what’s happening among the many developers is where the excitement lies).
Unlike the rise of Google or Facebook, these more decentralized offerings are designed to be locked open, so that we can build systems that present the empowering, enabling aspects of an open web, but without the chokepoints that make both enshittification and despotification all too tempting.
In such a system, the incentive structure is flipped. If the users themselves retain control over their data and which tools they can use to intermediate that data, then attempts to gain control over more centralized chokepoints become self-defeating. The ability of users to easily and seamlessly “exit” from one provider’s control, into some other provider (or even a self-hosted solution) means that the incentives to grab greater control are limited. Each step towards enshittification makes it more likely users will click a button and remove the problematic provider from their personal ecosystem without losing access to any content, data, or services.
Danny O’Brien, a Senior Fellow at the Filecoin Foundation, has argued that decentralization itself should never be the “terminal value” we pursue—it’s a means, not an end. The actual terminal value is “cognitive liberty”: the ability to think, reflect, and form ideas without those processes being surveilled, manipulated, or controlled by whoever owns the infrastructure.
The original World Wide Web absolutely decentralized some aspects of our cognitive liberty, enabling more people to speak, to associate, and to be heard. And that created tremendous opportunities, but also massive challenges, and the solutions to those challenges planted the seeds of the more enshittified, despotified world that we see today, with economic and political powers focused on controlling the chokepoints and manipulating them to their own interests.
Truly decentralized tools push power to the ends. Users control their data and choose which intermediaries operate on it—and that arrangement is a poison pill to both enshittification and despotification. Any move in those directions only pushes users across the deliberately low barrier to exit, taking their content, data, and community with them.
To Build the Decentralized Future
None of this happens automatically. The forces seeking control already have their arguments ready: centralized systems are safer, more accountable, and better at stopping bad actors. We’re already seeing this logic embedded in regulations being pushed under the banner of “protecting children” or “national security”—rules that would, in practice, further concentrate control in ways that make both enshittification and despotification more likely, not less.
Decentralization, like democracy itself, is something we have to fight for. Absent deliberate effort, the default trajectory runs toward centralization, because centralization is convenient, and convenience wins in the short term.
Which means the decentralized alternatives have to be genuinely better, not just philosophically purer. The centralized platforms won the last round because they removed friction. They didn’t ask users to manage config files or understand network topology—they said “click here and it works,” and most people took that deal. Any decentralized successor that requires users to become their own sysadmins will lose the same way the last generation of open protocols lost.
What’s different now is that we’re closer than we’ve ever been to having decentralized systems that are actually more convenient and more empowering, where the user experience is competitive with the centralized incumbents, and the democratic benefits come built in rather than bolted on. The goal is to build systems where those two things point in the same direction.
That’s what’s actually at stake in the work being done on decentralized protocols right now. A different architecture—one where the chokepoints that made enshittification and despotification inevitable simply don’t exist, and where democratic participation isn’t a side effect of good technology but the whole point of it.
The Next Internet
The pattern this piece describes is happening again right now, in real time, with the technology that will likely define the next decade.
The leading frontier AI companies—OpenAI, Anthropic, and Google, with xAI, Meta, and Mistral close behind—have every incentive to run the same playbook that the last generation of internet giants ran (and yes, some are the very same companies): Build something useful, attract users, create lock-in, exploit the chokepoints. The enshittification curve doesn’t care what the underlying technology is.
But the same forces that make decentralized social protocols viable apply here, too. The models themselves are increasingly interchangeable—users of agentic AI tools are discovering that the underlying model is a small piece of the puzzle, and the real value lives in their own data, context, and accumulated knowledge, all of which can live in files and databases they control. And open-weight models are getting good fast. Models you can run on hardware you own are inherently not subject to centralized control. Every step toward AI centralization makes the decentralized alternative more attractive.
The choice in front of us is the same one that’s always been in front of us, just with higher stakes and less time: Do we let the next generation of tools get built around chokepoints, or do we insist on architecture that distributes power instead of concentrating it?
The internet’s first generation empowered billions of people who had no voice. The second traded that empowerment for convenience, but the convenience came with chokepoints that turned into weapons. The third generation—social and AI both—can have empowerment and convenience together. But that outcome isn’t automatic. It requires treating decentralization not as a technical preference but as the architecture democracy actually needs, and fighting for it with the same urgency we’d bring to any other democratic institution under threat.
The alternative is already visible. Two photos, fourteen years apart.
