What Happens When the Moderators Leave?

Not that long ago, we were arguing about fairness.

Was Meta being too aggressive in Nigeria? Too soft on India? Was Twitter targeting state-linked propaganda networks based on democracy scores, or just flagging enemies of the West?

Researchers built models. Journalists filed FOIAs. Activists demanded transparency. We wanted to know who got silenced, and why.

That debate now feels quaint.

Because here in 2025, the question isn’t how platforms moderate disinformation. It’s whether they’re bothering to do it at all.

Since Elon Musk’s takeover, Twitter—now rebranded as X—has gutted its moderation teams, reinstated formerly banned accounts, and walked back policies on everything from hate speech to vaccine misinformation.

Meta followed suit, dissolving its third-party fact-checking program, letting violent rhetoric quietly bleed through its filters, and shifting its operations to more politically sympathetic turf in Texas.

The era of platform responsibility, if it ever really existed, is over.

And yet, one year before this slow-motion collapse, researchers Nihat Mugurtay, Özenç Kök, and Niklas Helwig published a study that now reads like a postcard from a fading empire.

Their paper, Politics and Propaganda on Social Media, examined takedown patterns on Twitter and Meta from 2018 to 2022.

What they found was eye-opening: content moderation wasn’t neutral. It was deeply political.

Platforms enforced the rules unevenly—targeting authoritarian states more than democracies, penalizing geopolitical rivals more than allies, and pulling punches where markets were too lucrative to offend.

In other words, they were running foreign policy under the cover of “community guidelines.”

Back then, this selective enforcement felt like the real scandal. Today, it feels almost like nostalgia. At least it was a system. Today, as moderation collapses, it’s not the rules we debate, but the absence of rules at all.

This article revisits Mugurtay et al.’s work not to critique it, but to contextualize it—as a last look at the architecture of platform power before the whole thing went off the rails.

We’ll walk through what the study revealed, how the ecosystem has unraveled since, and what happens when platforms stop even pretending to guard the public square.

Because when the moderators leave, the content doesn’t just get louder.

It gets meaner, riskier, and harder to look away from.

The Lost Age of Strategic Moderation

Before X was a lawless meme pit and Meta decided truth-checking was bad for business, there was a window—a narrow, uneasy window—when the tech giants were still trying to look like responsible stewards of the public sphere.

From roughly 2018 to 2023, moderation was not exactly fair, but it was at least framed as a moral imperative.

Platforms published transparency reports. They made high-profile announcements about dismantling "state-linked information operations." They floated terms like “coordinated inauthentic behavior” (CIB) with the precision of a PR team trying to sound like the CIA.

It was during this era that Mugurtay et al. dropped their quietly radical paper.

In Politics and Propaganda on Social Media, the researchers analyze takedown activity by Meta and Twitter over a five-year span.

What they discover is not some rogue intern with a delete button, but a pattern—a systematic bias in what gets taken down, and where.

Here’s what they found:

  • Countries with low democracy scores (think authoritarian regimes, flawed elections, restricted media) were more likely to be targeted for content removals.
  • Nations with high political instability—those mired in violence, unrest, or contested leadership—were also frequent takedown targets.
  • But perhaps most tellingly, geopolitical alignment mattered: countries that routinely voted against U.S. interests at the UN faced stricter scrutiny than those aligned with American foreign policy.
  • And finally, money talked: the bigger and more economically important the market, the more softly platforms stepped.

This wasn’t content moderation. This was platform diplomacy.

Meta and Twitter weren’t just enforcing rules. They were managing relationships.

They punished small, unstable, or adversarial countries more aggressively than powerful allies, even when the nature of the content was similar.

They weren’t arbiters of truth—they were strategists, shielding their bottom lines and avoiding state retaliation.

But here’s the paradox: for all its inconsistency, this system functioned. There were takedown teams. There were criteria—however flexible.

There was a vague sense that the platforms were trying, however imperfectly, to hold the line against digital manipulation.

And that’s what makes this era—the one Mugurtay et al. document—so eerie in hindsight. It wasn’t a golden age. But it had guardrails.

What followed, as we’ll see next, wasn’t a reform of the system.

It was a decision to burn the whole thing down.

2024–2025 – The Year the Guardrails Fell Off

If Mugurtay et al. gave us a map of platform enforcement shaped by political interest, then 2024 was the year someone took a lighter to it.

Let’s start with the most dramatic transformation: Twitter, now X, under the command of Elon Musk. What began as a chaotic acquisition turned into a full ideological rebrand.

Moderation teams? Dissolved. Trust and Safety Council? Disbanded. Disinformation policies on elections, vaccines, or climate change? Rolled back, reworded, or quietly deleted.

Musk promoted “free speech absolutism” while personally interacting with and amplifying white nationalist accounts, conspiracy theorists, and trolls once deemed too toxic to be verified.

And in 2025, it got worse. During the Trump re-inauguration, Musk stood onstage and gave what many interpreted as a Nazi salute.

Whether performance or provocation, the signal was clear: X was no longer pretending to play by old rules.

Then there’s Meta, once the poster child for over-engineered content policies. Under pressure from multiple political fronts and desperate to avoid regulation, Mark Zuckerberg made his own pivot.

The company ended its third-party fact-checking program and replaced it with “community notes”—a vaguely democratic gesture that, in practice, meant moderation by mob.

Meanwhile, moderation teams relocated to Texas, closer to a political climate more tolerant of regulatory rollbacks and, let’s be honest, more sympathetic to the idea that “free speech” means letting disinformation run free.

The tone shifted. It wasn’t just that platforms were doing less enforcement. It was that they no longer felt they needed to explain themselves.

  • Hate speech rates on X climbed and stayed high.
  • LGBTQ advocacy groups tracked a surge in targeted harassment with fewer takedowns.
  • Extremist networks reconstituted themselves with little resistance.

The rules, it turned out, weren’t rewritten. They were abandoned.

And what we learned—painfully, predictably—is that in the absence of moderation, the void fills fast. What used to be managed through clumsy enforcement is now shaped by algorithmic indifference, political spectacle, and CEO ego.

2024–2025 wasn’t the end of disinformation. It was the end of platforms pretending they cared about stopping it.

As we move to the next section, we’ll look at the role these platforms still play—not as referees, but as active political actors, shaping the battlefield even as they claim to have stepped off it.

Platforms as Political Actors—Then and Now

Let’s be clear: platforms were always political actors. They just used to wear suits and carry clipboards.

Back in the Mugurtay era, Twitter and Meta presented themselves as reluctant arbiters of the digital world. They used words like “neutral,” “consistent,” and “policy-driven.”

Behind the scenes, as the research showed, they were already choosing whose disinformation to target based on regime type, alliance, and profit potential.

Enforcement wasn’t fair, but it was deliberate. Strategic.

It was the kind of power that hides behind process.

Fast-forward to now, and the performance is gone. X isn’t hiding its ideology—it’s tweeting it.

Elon Musk didn’t just remove moderation infrastructure; he replaced it with a personal brand of chaos-as-principle.

When a platform CEO publicly mocks a disabled critic, boosts neo-Nazis, or interacts warmly with accounts pushing antisemitic conspiracies, we are no longer talking about bias.

We are watching a media ecosystem radicalize in real time, from the top down.

Meta, for its part, still clings to a kind of procedural aesthetic. But don’t let the interfaces fool you. What was once strategic enforcement has become selective abandonment.

Mark Zuckerberg doesn’t need to tweet swastikas. All he has to do is do nothing while hate speech metastasizes and “community notes” become a tool for coordinated trolling.

The platforms’ politics have gone from veiled discretion to techno-libertarian spectacle.

And yet, the power hasn’t disappeared—it’s just more opaque. When moderation was selective, we could at least track the pattern.

Now, the lack of enforcement is the policy. Musk and Zuckerberg have effectively said: “We won’t decide what’s true. We’ll let the people sort it out.” But the people sorting it out are often bots, billionaires, or rage-driven mobs.

This shift doesn’t depoliticize the platforms. It repoliticizes them through abdication. It says: "We’re not responsible for what happens here."

And yet, the consequences—the election outcomes, the hate campaigns, the social fragmentation—still radiate outward, as if the platforms were curating it all.

Which, let’s be honest, they still are. Just less cleanly. Less accountably.

So if Mugurtay and company helped us see how platforms once played subtle foreign policy games through takedowns, today we see a new mode of governance:

Not diplomacy, but spectacle.
Not strategy, but signaling.
Not moderation, but momentum.

And momentum, we’re learning, often runs in the direction of power—not truth.

The Vacuum – What Rushes In When Moderation Leaves

Nature abhors a vacuum, and so does the internet. The moment moderation steps back, something—someone—rushes in.

On X, the power vacuum left by moderation teams has been eagerly filled by bots, trolls, and coordinated hate campaigns.

Hate speech, especially targeting trans people, immigrants, and journalists, has surged to levels we haven’t seen since the early aughts, before platforms started hiring content moderation staff in the first place.

And this time, it’s not just fringe users; it’s algorithmically boosted and ideologically validated from the top.

Meta’s story is less theatrical but just as troubling. With the removal of third-party fact-checking, users are now asked to self-regulate truth via “community notes”—a feature easily gamed by organized misinformation networks.

In some countries, this has translated into state-aligned disinformation being upvoted into legitimacy, while independent media posts get swarmed, flagged, and buried.

The results are predictable:

  • Disinformation spreads faster than ever, with fewer friction points.
  • Marginalized groups are disproportionately targeted, facing harassment campaigns with little platform intervention.
  • Election interference is no longer covert; it’s out in the open, often using the very tools meant for civic engagement—like trending hashtags or viral quote-tweets.

And all of this is happening not in the shadows, but in full view of platform executives, who now claim that intervention is a kind of bias. The line between free speech and algorithmic sabotage has vanished, replaced by a shrug: “We don’t police ideas.”

But we know this isn’t neutrality. It’s a policy of non-policy, a decision to let chaos govern.

This vacuum doesn't only harm the targets of abuse or the readers of falsehoods. It erodes the very idea that the internet can function as a civic space. If no one is curating the garden, we get weeds—and eventually, the weeds win.

And maybe that’s the plan. Because the less users expect from platforms, the more platforms can justify doing less.

We’ve gone from selective enforcement to selective neglect. And now, with the guardrails off, the burden of defending truth, dignity, and public safety has shifted entirely onto the users—most of whom never signed up to play platform cop.

Which leaves us with one final question: what now?

Conclusion – What Do We Build Now?

We began with a study that tried to pull back the curtain on how platforms moderate disinformation. Back in 2024, the scandal was selectivity—platforms enforcing rules unevenly, guided by market logic and geopolitical alliances.

But at least there were rules. At least the curtain existed.

Now, we’re living in the after. The post-rules phase. The shrug emoji era of content governance.

X has traded in moderation for provocation. Meta has outsourced truth to whoever yells the loudest. The idea that tech platforms are neutral arbiters of public discourse?

That illusion has not only shattered—it’s been monetized.

And yet, this is not the end of the story. It’s just a very dangerous middle.

Because even in a deregulated, disinformation-rich, chaos-fueled feed, people are still trying to organize, inform, dissent, and connect. The stakes haven’t disappeared. They’ve escalated.

So what do we build now?

We build accountability, but not by waiting for platform CEOs to find their conscience.

We push for regulatory frameworks with real teeth—ones that treat platforms not like quirky startups but like the infrastructural behemoths they are.

We support independent research and civic tech tools that monitor the manipulation behind our timelines.

We educate—not with “media literacy” buzzwords but with frank conversations about power, visibility, and algorithmic harm.

Most of all, we reject the idea that nothing can be done. That the flood is inevitable. That the only options are censorship or chaos.

Because if we’ve learned anything from Mugurtay’s research—and from what’s happened since—it’s this: moderation was never neutral. But that doesn't mean we can't demand something better than this.

Better than arbitrary takedowns. Better than algorithmic neglect. Better than platforms that shrug at abuse while profiting from outrage.

The wool has long since fallen from our eyes.

Now comes the harder work: building the guardrails back—not for the platforms’ sake, but for ours.

Sources & Links

Academic Source

  • Mugurtay, Nihat; Kök, Özenç; Helwig, Niklas.
    Politics and Propaganda on Social Media: How Twitter and Meta Moderate State-Linked Information Operations.
    arXiv preprint, 2024.
    https://arxiv.org/abs/2401.02095

Platform Policy Changes & Moderation Retreat

Human Rights & Watchdog Reporting