The Moderation Problem We Keep Avoiding
There is a useful distinction that rarely makes it into the debate over Big Tech content moderation: the difference between having a problem and admitting you have a problem. For the better part of a decade, technology companies insisted they did not have a problem at all. Content moderation was simply applied neutrally, they said. The algorithms were objective. The humans reviewing flagged posts were trained to follow clear, consistent guidelines. Bias? That was what your political opponents believed in, not us.
The trouble is that this position has become increasingly difficult to defend. It is not that any single leaked internal memo or lawsuit proves systematic bias. Rather, it is that the accumulated weight of evidence—internal documents, leaked communications, employee testimony, and documented patterns of inconsistent enforcement—suggests that moderation at scale inevitably involves judgment calls, and judgment calls inevitably reflect the values of the people making them. The question is no longer whether platforms have some relationship to viewpoint. The question is whether they will be honest about it.
What the Record Actually Shows
Start with what we actually know, not what we suspect. Content moderation at the scale of billions of posts per day requires either automation or enormous teams of human reviewers. Automation relies on patterns learned from training data. Training data reflects the decisions of humans. This is not a conspiracy. It is how the technology works.
Several independent studies have found that content flagged as "misinformation" or removed for terms-of-service violations shows patterns that correlate with political viewpoint. A 2022 analysis found that posts from accounts identified as left-leaning sources faced lower removal rates for identical policy violations compared to right-leaning sources, though the difference was modest. Other research has documented that fact-checking labels appear more frequently on posts from conservative accounts than progressive ones making comparable claims—a finding that cannot be explained by accuracy differences alone.
None of this proves intentional conspiracy. It suggests something simpler and more troubling: that moderators and algorithms trained by humans working in heavily progressive tech hubs may, unconsciously and at scale, apply different standards to different viewpoints.
The Transparency Question
Here is where the real problem lies. When these patterns emerge, platforms resist transparency in ways that would seem unthinkable in other sectors. A bank hiding its lending patterns would face regulatory scrutiny. A hospital concealing its mortality rates would lose accreditation. But a technology company removing millions of posts based on guidelines it will not fully disclose, with reasoning it will not explain, answers to almost no one.
This is a conservative concern, but it should not be. Institutional conservatives should be among the first to demand transparency and accountability from concentrated private power. The conservative tradition holds that power without accountability is dangerous, whether that power wears a government uniform or a Silicon Valley badge. Edmund Burke did not write only about the state.
Several proposals have emerged to address this. Section 230 of the Communications Decency Act, which shields platforms from liability for user-generated content, could be modified to require that platforms disclose their moderation criteria clearly and apply them consistently. Appeals processes could be strengthened. External auditing of algorithmic decisions could be mandated. These are not radical ideas. They are the kind of structural accountability we expect from other industries.
The Harder Question About Power
But underneath the moderation question lies something larger and more unsettling: the concentration of communicative power itself. Even if platforms were perfectly neutral—even if every moderation decision were made by an impartial algorithm with no human judgment involved—we would still face the reality that a handful of companies control the digital infrastructure through which billions of people speak, listen, and think.
This is not a problem that transparency alone solves. A transparent monopoly is still a monopoly. A clearly explained bias is still a bias. The conservative insight here is an old one: that power tends toward abuse not because people are evil, but because people are human. When a single platform can silence a voice to billions, the incentives to use that power—whether for profit, ideology, or the simple conviction that you know better—become almost irresistible.
This does not mean government should regulate the internet into mediocrity. Federal content regulation would be worse than the problem it claims to solve. But it does mean that when tech companies claim they are merely neutral platforms applying objective rules, skepticism is warranted. They are not neutral. They are not objective. They are powerful institutions making consequential judgment calls about what billions of people may say and hear.
What Comes Next
The path forward requires clarity about what we are actually asking for. Conservatives should not demand that platforms become politically balanced. That is a wish for different bias, not the absence of bias. What conservatives should demand is: clear, published moderation standards; consistent application of those standards; meaningful appeal processes; and transparency about how decisions are made.
We should also consider whether the current legal structure, which treats platforms as publishers for some purposes and common carriers for others, makes sense anymore. Perhaps it does not. Perhaps new categories of private intermediary power require new kinds of accountability.
The deepest conservative concern should be this: a society in which the rules of public discourse are written by unaccountable private companies, applied inconsistently, and defended with false claims of neutrality is a society that has abandoned the conditions for genuine democratic deliberation. We can disagree violently about policy. But we cannot deliberate at all if we cannot trust the platforms on which we deliberate to tell us the truth about how they work.
That is not left or right. It is a simple requirement of intellectual honesty. And it matters more than any single content moderation decision ever could.
