Social Media Has Not Brought Liberation

In the spring of 2011, I sat in my tiny undergrad campus apartment with friends watching Egyptians march in Tahrir Square against the Mubarak regime. I remember it distinctly because it was one of the first times as a young adult when I was really conscious of seeing capital-H History in real time. The only other time that I had been aware of that happening, in something approaching an adult way, with the significance of what was happening readily apparent to me without needing an adult to translate it into the pettier and more immediate concerns of a child, had been the election of Barack Obama.

Tahrir Square was more exciting to me than the Obama election had been, because it was not orderly, formal History but messy and uncertain History, with troops and tanks in the street, things that teenage boys, no matter how sophisticated and mature they may think they are, find exciting (provided it’s not them or people that they know in the firing line). And there was something even more exciting about this: I could participate. I could be part of it. 

Like many young men of my generation, I had been online since I was able to register a vBulletin forum account. And at the time you couldn’t hear enough about how Twitter, Facebook, and other social media had made Tahrir Square possible. There were columns about how the Information Age made it impossible for authoritarians to suppress the flow of information that wanted to be free. The signal could not be stopped. This wasn’t just something that was happening on TV. It was something that was happening due to spaces that I was in, spaces that I was part of. Without ever leaving my room, I could feel like—I was—part of something greater.

Though this is likely not news that my fellow millennials will welcome, children born the year Obama was elected are now the same age that I was when I was watching those crowds of hopeful, brave Egyptians walking up to the guns. So it’s worth reiterating just how widespread this democratic optimism was at the time. Pundits everywhere rhapsodized about how authoritarianism was on life support and how the nature of the internet made it impossible for states to repress the free flow of communication in the future. The future was freedom, enabled by 21st-century technologies of seamless connection.

To be fair, there have been mea culpas by some of the sunnier optimists, saying that the techno-utopian dream of digital democracy has not occurred due to effective repression, disinformation, etc. But it all is still talked about in terms of social media failing to live up to its supposed liberatory potential. The information can still set us free, we just need to figure out how to structure it so that it does—just need to figure out the right algorithms—just need to figure out the right moderation strategies.

Few seem willing to contemplate that the story of social media might not be that of still-unfulfilled democratic potential, but rather the opposite: that social media has been actively good to autocrats. That its effects are not the fault of imperfect implementation, but baked into the form itself. The evidence is getting increasingly hard to deny. It was fashionable to laugh at claims that Russia “hacked” the 2016 election through Facebook, and for a while you could confidently scoff at the ham-fisted images of “LGBT for Bernie” cartoons that were evidently part of such efforts.

But in 2026 to deny that such operations have been, to a greater or lesser extent, successful, requires you to be one of the people whose salary depends on not seeing it. Unfortunately, there is no shortage of such people, largely for the reason that such operations are more successful when they exploit pre-existing and often genuine fault lines within coalitions that authoritarians want to see fracture. So there are people with entirely genuine politics who are understandably upset by the implication that they are the useful idiots of foreign agents posing as friendly randos on X who egg them on, sowing distraction and discord and fighting. But then, that is the nature of a useful idiot. It is exactly this ambiguity that makes such efforts so successful.

It’s fair to ask what makes social media such a uniquely vulnerable attack surface for authoritarians. After all, foreign countries have always been able to print newspapers or operate radio stations in other nations, and frequently have. But the difference in the ease of social media as an attack vector means that a difference in degree becomes a difference in kind.

Consider a direct comparison with traditional media: The Soviet Union is estimated to have spent over three billion dollars a year on propaganda, the euphemistically titled “active measures.” That's a conservative figure, from an era when three billion dollars was real money. After 1982, participating in these operations was the official duty of every officer in the KGB's 15,000-strong foreign intelligence directorate. And what did all that money and manpower buy? The crown jewel of the entire enterprise, the AIDS-as-American-bioweapon hoax, required the KGB to fund an entire newspaper in New Delhi years in advance just to have somewhere to plant the story, which then took four more years of patient laundering through friendly outlets in the Third World before it finally surfaced in American media. Four years, a bespoke newspaper, and the resources of a superpower: one conspiracy theory. Compare the Internet Research Agency, which reached over a hundred million Americans on Facebook alone for about $1.25 million a month—about the cost of opening a McDonald's franchise. The Soviets weren't short on resources or ambition. They were short for not yet having an information environment where placement is free, laundering is instant, and distribution is handled, at scale and for pennies, by the target country's own most widely used consumer products. 

Cold War information control was control over discrete activities: the paper you read in the morning, the broadcast you watched at night, the pamphlet someone pressed into your hand. You consumed media, and then you stopped consuming it and went about your day. But social media is not an activity—it is the substrate of daily life. It is where people find jobs, sell furniture, meet spouses, argue with their uncles, and store their baby photos. A state that controls a newspaper controls what you read. A state that controls the platform controls the space where your social existence happens, and it knows who you talked to, what you lingered on, and what you almost said. This penetration is why it is such an effective attack vector: Soviet propaganda had to break into an American's media diet from the outside, past editors and gatekeepers who decided what got printed. Modern influence operations do not break in at all. They are already inside, sitting in the feed between a cousin's wedding album and a meme about gas prices, dressed exactly like everything else there. Social media isn’t “the news.” Social media is ordinary life.

Even the Russian efforts of 2016 almost look quaint these days. In the age of LLMs, the process can be, and is, entirely automated. There are, right now, today, bots that can convincingly argue for or against any particular political party, candidate or topic their operators choose. You’ve probably encountered them, though you may not know it.

The most crucial thing here is that this attack surface is asymmetric. The playing field of the social media informational environment is uneven, and it systematically benefits autocracies. The reason that none of the promised liberatory effects of social media have materialized is simple: it actually is not that difficult for competent authoritarian states, unbound by any kind of commitment to rule of law or free expression, to control the medium. 

First, platforms not controlled by the state or aligned interests can simply be banned altogether. No ban is foolproof, and those really determined to circumvent them can turn to VPNs and other means, and some people do these things. But the average citizen is not going to get a VPN or figure out how Tor works. They aren’t even going to download anything that isn’t on the App Store.

Consider Russia: It blocked Facebook and Instagram in 2022 and declared their parent company to be an extremist organization. YouTube, it throttled into uselessness in 2024. Signal, Viber, Discord, and eventually WhatsApp, blocked or strangled, one by one. Yes, Russians turned to VPNs by the tens of millions, and for a while the pundit class treated this as proof that the signal could not be stopped—the 2011 dream in miniature. Then Roskomnadzor blocked 469 VPN services and started blocking the underlying protocols, every ISP in the country was required to install state-controlled interception hardware, searching for banned content became a fineable offense, and the state rolled out its own WhatsApp clone, preinstalled on new phones (and it reportedly checked whether the user had a VPN). None of this required any special technical genius. It required a state that was willing and a medium that centralizes chokepoints in a way no samizdat network ever did. All of those millions of posts flying through the air rely on the cloud, and ‘the cloud,’ while not a series of tubes, is in fact a set of physical things that the state can control.

Secondly, while it’s true that even a very repressive state will have a difficult time preventing someone who really wants to upload a certain image or post a certain viewpoint online from doing so, the era when the ability to do such things represented potentially existential challenges to illiberal governments has firmly come and gone. It flatters the sensibilities of Western journalists who fancy themselves Woodward or Bernstein to imagine that the greatest possible challenge to autocracy or corruption is the dissemination of information, that the right set of facts put into circulation represents the greatest possible tool against repression. That sentiment may on some level be admirable. It just isn’t actually true. 

In the 21st century it’s far easier and more effective to flood the zone with shit than it is to try and shut off some underground signal. It may be difficult to track down a single dissenting voice if it belongs to someone with the determination and technical competency to post anonymously. But it’s trivially easy to drown them out. This isn’t me speculating or trying to Say Something About Society: it’s what the Chinese government actually does. And we know that they were doing this systematically, with hundreds of thousands of personnel dedicated to these operations, in 2014. Now imagine what that probably looks like today, where, again, large parts of the process can be entirely automated. 

The reason that the dream of information liberation hasn’t come true is because it isn’t true. If you’re a sufficiently competent and unbound authoritarian state, you can in fact stop the signal.  

By contrast, democracies have a far more difficult time keeping adversaries out of their information environment. TikTok, an app that the Chinese government does not allow domestically, received a half-hearted attempt at a ban by the U.S. government after it repeatedly refused to be transparent about its governance structure. The attempted ban prompted U.S. teenagers to call Congressional offices threatening to kill themselves. (There were claims that the ban plan was due to content critical of Israel on the app. It’s unclear how much actual truth there was to this, but once again, a preexisting political fault line complicated efforts to do anything about the platform.) Then Jeff Yass, a major investor in TikTok, bought Trump like groceries, and Trump flipped on enforcing the ban, which was U.S. law at that point, and then put his cronies in charge of the platform. (And I do mean in charge: the deal Trump blessed by executive order handed TikTok within the U.S. to a consortium led by Oracle, whose founder Larry Ellison is a longtime Trump ally, with the algorithm itself, the thing that decides what 170 million Americans see, placed under Oracle's control to "retrain" as it sees fit. The family that now steers TikTok's feed also owns CBS.) Though not at all unique to the medium, this is another major vulnerability that social media presents: we see here how authoritarian regimes both foreign and domestic can simply co-opt the business elites that run them.

Is it necessary to go on? Just a few weeks ago Elon Musk’s father sat with English right-wing provocateur Tommy Robinson in a luxury hotel bar in Moscow, live-streaming about how Russia was the UK’s ally.  At the time, his son was on the platform that he owns and bought for the more-or-less explicit purpose of influencing politics, cheering on the burning of immigrants' homes in Northern Ireland.

The evidence continues to grow. Social media, as it is currently constructed, redounds to the net benefit of illiberal regimes and their allies in democracies. The question is what to do about it. Contra the points that often get raised whenever turning off or altering these social balkanization machines is proposed as a possibility, none of this is inevitable. None of it is baked into technology itself. None of these are things that we have to live with. All of them are the result of policy choices we have made.

Altering the attack surface that social media presents requires neither some all-encompassing revolution that reorders the foundations of society, nor any compromise on fundamental values such as freedom of speech. Despite what the inevitable rejoinder may be, there is no hypocrisy here. Democracies already regulate speech in all kinds of ways, from libel law to product safety to truth in advertising. Simply making platforms assume publisher liability for the content that gets served by “For You” pages and similar endless scroll mechanisms would be a massive step towards bringing these companies under control. 

I am not saying that Facebook should be liable for what your QAnon uncle posts. I am saying that when Facebook's recommendation engine takes your uncle's post and inserts it into the feeds of thousands of strangers who never asked to hear from him, the company should be liable for that, for the act of promotion itself. If the New York Times runs a column baselessly calling someone a pedophile, the Times is in legal trouble, and nobody considers this a crisis for the First Amendment, because the paper chose to run it and used its infrastructure to amplify it. The only thing the platforms have changed is laundering that same editorial choice through different machinery. Imagine a newspaper that selected its front page stories via tarot cards and dice rolls. If it printed something defamatory about you, would you accept "the dice did it" as a defense? Of course not. The paper chose to delegate its editorial judgment to that process and it owns the results. An engagement-maximizing algorithm is just those dice but with fancier math. Courts are already starting to move in this direction: TikTok recently was held to be liable for promoting content that got a young girl to hang herself. The precedent already exists. And it is not the end of the First Amendment. 

This change reintroduces friction in the form of human gatekeepers with some kind of skin in the game in terms of what gets pushed and what gets eyeballs in front of it. Social media firms would likely have to substantially staff up their compliance departments, and probably move to a whitelist model to avoid exponentially increasing liability risks when objectively false content goes viral. But it is not censorship to subject Mark Zuckerberg and Elon Musk to the same liability that traditional media sources with editorial standards  have operated under for over a century. And it is this added friction, the need to get content consciously signed off on by humans with something to lose, that will reduce the attack surface that these platforms represent compared to traditional media. 

Such reforms will not end the risk of illiberal regimes attempting to use these platforms for propaganda purposes. None is a cure-all. But they reintroduce some of the same friction that made the active measures of previous generations a more manageable threat. They put humans with something to lose into the loop. Such change would not be easy, it would not be tradeoff-free, but the tools to bring these platforms to heel already exist. We just need the political will to use them.

Featured Image is Tahrir Square on Nov 18th, by Lilian Wagdy