Freedom of Speech

Almost everything on this site runs into the same objection: isn't moderating any of this just censorship? It is the most important objection to take seriously — and the one most often answered badly, from every direction.

How to read this page. Tap any underlined word to see the precise term and a short definition. Expand any "Deeper" box for the legal detail and the contested points. The main text works on its own — you can skip both and still get the whole argument. This page deliberately takes no side; it sets out the strongest version of each.

Two different things wearing one word

"Free speech" is asked to carry two very different things. The first is a person speaking — saying an unpopular thing, and not being jailed for it. The second is a system deciding which of millions of messages billions of people see, and how loudly. The principle of free expression was built, over centuries, to protect the first. Whether it was ever meant to protect the second is the question underneath almost every fight about online content.

The classic argument for letting speech run free is the marketplace of ideas: let everything be said, and truth wins the competition. It is a powerful idea. It also quietly assumes some things — that speech is roughly scarce, that speakers are people, and that being heard depends mostly on whether your argument is good. A recommendation feed breaks all three assumptions at once.

The marketplace of ideas

The long-standing argument — associated with John Stuart Mill and, in US law, with Justice Oliver Wendell Holmes — that the remedy for bad speech is more speech, not suppression, because truth tends to win an open contest. It is one of the strongest cases for a light touch. Its critics do not reject it so much as note its hidden assumptions: it pictures a town square of human speakers, not an engagement algorithm choosing what a billion people see. When reach is sold and automated, the "market" is no longer deciding between ideas on their merits.

The objection, taken seriously

Start with the strongest case against moderation, because it is a real one. Once someone has the power to remove speech, they have the power to remove speech they simply dislike. The line between "harmful" and "inconvenient" is easy to blur, and history is full of governments that called dissent "disinformation." Today that power sits largely with a handful of private companies, applied by rules they write and can change overnight — what one legal scholar calls the platforms acting as the new governors of speech. A reader who is uneasy about that is not being naive. They are noticing something true.

So the question is not "moderation: yes or no." Some sorting is unavoidable — a feed that showed everything, in no order, is not neutral, it is just a different choice. The real question is narrower and harder: who decides, by what rules, with what accountability, and can the decision be seen and challenged? That is the same process question this section asks of journalism, science, and democracy. A correction mechanism you can inspect is the thing that separates governance from censorship.

Scale and automation change the question

Here is the argument that the law itself may be out of date, made most sharply by the legal scholar Tim Wu. The free-speech tradition grew up in a world where speech was scarce and the danger was a government silencing a speaker. Online, speech is effectively infinite and almost free to produce. In that world, the cheaper way to suppress a message is not to ban it. It is to bury it — to flood the space with noise, distraction, and abuse until no one can hear it. Wu calls this reverse censorship.

Reverse censorship (censorship by flooding)

Tim Wu's term for silencing speech not by banning it but by drowning it — overwhelming the information space with propaganda, distraction, troll campaigns, and automated content until genuine speech cannot be found or trusted. It is the modern censor's tool precisely because it never has to admit to censoring anything. It also explains why "let everything be said" no longer guarantees anything is heard: when attention is the scarce resource, whoever can manufacture the most noise wins, regardless of truth.

This is where the bot enters. A network of automated accounts, or a single message engineered and paid to reach millions, is not a person speaking in the town square. It is closer to industrial production. Wu's point is not that such things should be banned — it is that treating them as identical to an individual's voice, and protecting them on that basis, misunderstands what is actually happening. The same machinery is described plainly on How It Works and put to deliberate use by state actors.

Persuasion is not the same as manipulation

There is a further distinction worth drawing, even where speech is free. Persuasion gives someone a reason, which they can weigh and reject; it is the thing free expression exists to protect. Manipulation works differently. As the scholars Susser, Roessler and Nissenbaum define it, manipulation influences a person by exploiting weaknesses and bypassing their capacity for conscious, rational choice — steering them without their quite realising it. The law already treats some word-made acts this way: fraud, false advertising, and incitement are not shielded simply because they are made of words. Much of what the rest of this site documents — content engineered to exploit predictable features of the mind — sits closer to that category than to a person making an argument. Naming which one you are looking at does not settle the legal question. It does change it.

Two traditions, two answers

It helps to know that the world's democracies have not settled this the same way. In the United States, the First Amendment is close to absolute — but it binds only the government. That is the doctrine of state action: the constitution restrains the state, not private companies, so a platform is generally free to carry or remove what it likes. In 2024 the US Supreme Court, in Moody v. NetChoice, treated those moderation choices as a form of the platform's own expression — which cuts against laws forcing platforms to carry speech.

State action

The principle in US constitutional law that the Bill of Rights — including the First Amendment — limits what governments may do, not what private parties may do. A consequence many people find counter-intuitive: when a social platform removes your post, it has not violated your First Amendment rights, because it is not the state. The same act by a government would be censorship in the constitutional sense; by a company it is, legally, an editorial choice. This is exactly why the real debate is about private power, not constitutional rights.

Europe starts from a different place. Under Article 10 of the European Convention on Human Rights, free expression is a fundamental right — but an expressly qualified one. It may be limited where a limit is lawful, serves a legitimate aim, and is "necessary in a democratic society." The European Court of Human Rights has, for example in Delfi AS v. Estonia, weighed expression against other rights such as protection from hate and harm, rather than treating speech as a near-absolute. Neither tradition is "the" answer. They are two democratic societies drawing the line in different places — which tells you the line is a choice, not a law of nature.

We already draw lines to protect children

One more piece of common ground is worth naming, because it is the user's instinct and it is already law. Free and open societies place age limits on alcohol, tobacco, gambling, and driving; they rate films and label explicit music; none of this is usually called censorship. Several democracies have begun extending that same logic online, through age restrictions on social media. Whether that logic fits a feed as well as it fits a cinema is genuinely arguable. But it shows that "any limit at all is censorship" is not how these societies actually behave — they draw narrow, rule-bound lines, especially around children, and argue about where the lines go.

Where this leaves us

This site does not tell you where to land, because the honest position is that reasonable people land in different places. What it offers is a sharper set of questions than "free speech: for or against." Are we talking about a person speaking, or a system amplifying? Persuasion, or manipulation? A government silencing a critic, or a company ordering a feed? And on either side — platform or state — is the power being used through rules you can see and challenge, or in the dark? The thing genuinely worth fearing, in the end, is the same thing the rest of this section warns about: power with no correction mechanism. That can be a censor. It can equally be an unaccountable feed.

The two legal traditions in more detail — and why neither resolves it cleanly

The American settlement. US free-speech law is unusually speech-protective and unusually narrow in who it binds. Because of the state action doctrine, the First Amendment constrains governments, not private platforms — so a platform removing content is not, in the constitutional sense, censoring anyone. The live US fight is therefore the reverse: when states pass laws telling platforms what they must carry, do those laws violate the platforms' own speech rights? In Moody v. NetChoice (2024) the Supreme Court indicated that a platform's content-moderation and curation can itself be protected expression, and sent the cases back to the lower courts. The upshot: in the US, "free speech" increasingly cuts in favour of the platform's editorial freedom, not the individual user's right to be carried.

The European settlement. European law treats expression as a fundamental right that is nonetheless balanced against other rights and interests. Article 10 of the European Convention permits restrictions that are lawful, pursue a legitimate aim, and are "necessary in a democratic society" — a proportionality test, not an absolute. In Delfi AS v. Estonia (2015) the European Court of Human Rights upheld a finding of liability against a news portal for unlawful reader comments, weighing expression against protection from hate speech. Critics worry this chills lawful speech; defenders say it forces a serious reckoning with harm. Either way it is a different machine from the American one.

Why neither closes the case. Both traditions were built before recommendation engines and automated speech at scale. Scholars such as Jack Balkin argue the real subject is no longer a duel between speaker and state but a triangle — speakers, governments, and the infrastructure companies that now sit in between — and that our inherited rules describe the first relationship far better than the second. That is not a reason to abandon free-speech principles. It is a reason to be precise about which relationship a given fight is actually about.

The strongest version of the censorship worry — stated without flinching

It would be dishonest to present only the case for limits. The strongest worry runs like this. The category of "harmful" or "false" content has no fixed edge, and whoever polices it acquires enormous, quiet power over what a society can discuss. That power has been abused before: "disinformation" and "public order" are among the oldest pretexts for silencing opposition, and a private platform under government pressure can become an instrument of exactly that, with none of the constitutional limits a state censor would face. Automated moderation makes it worse — it scales mistakes, and its errors are invisible and hard to appeal. A reader who concludes that the cure risks being worse than the disease is holding a serious position, not a careless one. The reason this site still argues for transparency over either extreme is that a rule you can see, contest, and overturn is the one kind of power that can be corrected — which is the whole argument of this section.

Sources

Wu, T. (2017), Is the First Amendment Obsolete?, Knight First Amendment Institute, Columbia University.
Klonick, K. (2018), The New Governors: The People, Rules, and Processes Governing Online Speech, Harvard Law Review 131.
Balkin, J.M. (2018), Free Speech in the Algorithmic Society, UC Davis Law Review 51.
UN Human Rights Council (2018), Report of the Special Rapporteur on the promotion and protection of the right to freedom of opinion and expression (David Kaye), A/HRC/38/35 — on content moderation and the difference between state censorship and platform governance.
Supreme Court of the United States (2024), Moody v. NetChoice, LLC — platform content moderation as expression.
European Court of Human Rights, Grand Chamber (2015), Delfi AS v. Estonia — Article 10 expression balanced against other rights.
Susser, D., Roessler, B. & Nissenbaum, H. (2019), Technology, autonomy, and manipulation, Internet Policy Review 8(2) — the persuasion / manipulation distinction.