Facebook censorship isn't just about free speech. It's about who controls what you see and the information you will never see.
Ryan Hartwig did not go to work intending to become a whistleblower.
In 2018, he took a job with Cognizant, a contractor performing content moderation for Facebook and Instagram. His job was straightforward in principle: review what users posted and determine what stayed online and what came down.
Some of that work is difficult to argue with. Employees were warned during the hiring process that they would encounter disturbing material and could not simply skip past it. Hartwig described moderators reviewing pornography, violent cartel videos, terrorist material and other graphic content.
Someone has to make those decisions.
The question is what happens when the decisions go beyond asking if content is pornography, violence or terrorism, and a moderator is asked to consider, for example, which political speech people should be allowed to see. That became the center of our conversation.
When Moderation Becomes Something Else
Hartwig said he began questioning the system when he saw Facebook making exceptions to its policies. He gave examples of where he believed that line had been crossed. One was during Pride Month. Moderators were told that Facebook was making an exception to its normal hate-speech rules that would allow certain attacks against straight white males.
Another was being instructed to remove a viral video that he believed would generate sympathy for Donald Trump, despite saying he could find no justification for its removal in Facebook's policies.
While moderating content outside the United States, he recalled instructions to closely monitor posts involving right-wing protesters in Spain, flag the content, and escalate it to Facebook.
Hartwig said Cognizant supervisors communicated regularly with Facebook. Moderators received policy updates, and presentations told them what kinds of content to delete or allow. He recalled one sensitive policy that was not presented during an in-person PowerPoint. The relevant slide was skipped or not in the presentation deck. When the material was emailed to moderators later, the slide was included.
That raises the question: How and when does moderation become curation?
Removing a cartel execution video is one thing. Deciding which political arguments, viewpoints, or groups receive exceptions is different. Once a platform begins making those decisions, what users see is no longer simply what other users posted. It's which user posts someone else decides you should see.
Facebook Is a Private Company. So What's the Problem?
A newspaper has an editorial perspective. Nobody expects every newspaper to be perfectly neutral. If I own a newspaper, I can decide which letters to the editor I publish. Should Facebook be any different?
Hartwig's concern went beyond ordinary editorial bias. He described American moderators reviewing content originating in Latin America, Spain and Canada. He said politically significant content could be flagged, escalated or removed, and argued that such decisions potentially influence political discourse and elections.
That makes the scale of the problem different. A local newspaper can influence its readers. A global social platform can influence what billions are able to find, share and discuss. That doesn't automatically answer the legal question. Facebook is not the government.
That led us into Section 230.
The Section 230 Question
Hartwig suggested that the way Section 230 of the Communications Decency Act has been interpreted gives social-media companies protections they should not receive when they move beyond hosting other people's speech and begin curating it. He distinguished between an interactive computer service and an information content provider, and argued that Facebook increasingly behaves like the latter, while retaining protections associated with the former.
One of the things I appreciated about the conversation was that we didn't pretend the legal issue was simpler than it is. Hartwig acknowledged that the common "platform versus publisher" explanation is useful and incomplete. Underneath the legal complexity is a simpler problem:
A social-media company needs moderation.
Few people want Facebook filled with child exploitation, terrorist videos, graphic violence or doxing. The question is where moderation should stop. Hartwig's answer is fairly radical in its simplicity: when adults are discussing politics, leave it alone, unless the content violates the law or falls into narrow categories requiring protection. He would make stronger exceptions for minors, and enforce laws against conduct such as doxing.
You can disagree with his solution.
It requires us to clearly define the problem.
The More Important Line May Be Government Involvement
A private company deciding what it wants on its platform is one question.
A government telling a private company what speech to remove is another.
Hartwig suggested this is where censorship becomes troubling. His analogy was a private restaurant. The restaurant can establish rules for customers, but when someone commits a crime, government authority enters through a legal process. His concern is what happens when government officials instead begin using a private company's moderation machinery to accomplish things the government itself may not legally be able to do.
Whether a particular government interaction amounts to persuasion, pressure, coercion or unconstitutional action has to be established from the evidence.
The question isn't, can Facebook remove your post.
The question is, who made Facebook remove it?
What You Don't See May Matter More Than What You See
We talked about Google, search engines, AI and the systems increasingly standing between us and information. This is where the problem becomes relevant to someone with no interest in debates about Facebook censorship.
Most of us cannot personally examine everything happening in the world.
We search.
We scroll.
We ask AI.
An algorithm decides what appears first. A recommendation system decides what appears next. A moderation system decides what disappears entirely.
That doesn't require a sinister conspiracy. It can be done by ordinary people who sincerely believe they're making the system better. That may be the problem. If someone believes that certain information is harmful, misleading or politically dangerous, removing it may feel less like censorship and more like responsible stewardship.
The result is the same for the user:
Someone has changed the information available to you before you've had an opportunity to evaluate it for yourself.
I compared a search engine to a library card catalog. I want the librarian to help me find the book. I don't want the librarian quietly removing index cards just because he or she thinks I shouldn't read certain books. The books may exist. If the system you depend upon to find them makes them effectively invisible, your perception of what information exists has been altered.
Censorship Is Not the Only Way to Influence
Instead of asking only what am I not being allowed to see, ask, Why am I seeing this?
Suppose a story about Donald Trump, Joe Biden, Israel, Iran, climate change or vaccines suddenly appears in your feed. The story might be accurate. The question worth asking is:
Why did this particular story reach me?
Who selected it?
What caused the algorithm to recommend it? What information would give me a different understanding of the same event? Who benefits from my seeing this version of the story? Selection influences perception.
With an enormous amount of information available online, people can find something that supports whatever they already believe. It's confirmation bias. More information does not produce better understanding - it gives us more material with which to confirm ourselves.
The Solution Isn't to Stop Using Technology
The answer isn't to abandon a useful technology just because the systems behind it are imperfect. Hartwig favors greater decentralization and personal control. His proposed approach includes reducing dependence on a small number of companies and taking privacy more seriously.
I came away with a different, complementary, solution: Keep thinking for yourself, as you use a technology. When Google gives an answer that matters, ask where it came from. When AI gives an explanation, ask what assumptions went into the question. When a story appears in your feed, ask why that story was selected.
When everyone around me seems to agree about something, I can ask whether I'm observing a genuine or filtered consensus consensus. When something disappears, I can ask who decided to remove it.
The danger isn't that Facebook, Google or AI may tell us what to think. The greater danger is that we stop noticing that someone or a system is helping decide what we have available to think about in the first place.
That is why the answer to biased information isn't finding a source that agrees with you, it's learning to recognize filtering. Once someone controls the information you see, asking better questions may be the only part of the process that still belongs to you.
Editor's Note
This article is based on my podcast conversation with Ryan Hartwig. The ideas discussed here originated in that conversation, while the structure, emphasis, and commentary are my own. Any errors or interpretations are mine, not Ryan Hartwig's
About Ryan Hartwig
Website: https://ryanhartwig.org
Behind the Mask of Facebook: A Whistleblower's Shocking Story of Big Tech Bias and Censorship with Kent Heckenlively.
Listen or Watch
Apple Podcasts • YouTube • Spotify
About the Author
Daniel Stih (danielstih.com) is an aerospace engineer, software engineer, indoor environmental consultant, and author of 12 books. For more than 30 years, he has investigated complex problems spanning engineering, technology, the built environment, and human decision-making. His work explores how evidence, assumptions, and systems shape the conclusions we draw—and whether we're solving the right problem. Learn more about his approach in Why I Think This Way.