Part 2. Section 230. The Shield Substack Hides Behind

Cycle: Lintara Reads the Platform. Part 2 of 28.

---

Part 2. Section 230. The Shield Substack Hides Behind

There is one section of American law from 1996. Twenty-six words.

These twenty-six words created the entire modern social platform industry. Without them, no Facebook, no YouTube, no Twitter, no Substack.

These same twenty-six words are what every person who tries to sue a social platform runs into. In ninety-nine cases out of a hundred, they hit the wall and break.

The name is Section 230. Part of Title 47 of the United States Code.

---

What the section actually says

Direct translation, no softening:

“No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”

Translation to human language: if you are a platform, and someone else posted defamation, threats, fraud, or extremism on you — you are not responsible for someone else’s content. The plaintiff can sue the author of the post. The plaintiff cannot sue the platform.

Imagine you own a movie theater. Someone in the audience screams “Fire!”. People run, trample each other. Usually the screamer is liable, not the theater. Section 230 says the same thing, but on the internet.

---

Why this was needed

The year is 1996. The internet is just starting. There is a court case in the US: Prodigy, an online forum, is sued for messages posted by users. The court ruled: since Prodigy moderated content (removed spam and profanity), it is responsible for all content as a publisher.

This created a trap: either moderate nothing and become a sewer, or moderate and be legally liable for everything. Both paths kill the business.

Congress panicked that the US internet would die in the cradle without developing. So they wrote Section 230: moderate as much as you want, it doesn’t make you a publisher.

This was reasonable in 1996. The law was written to enable moderation — to let platforms remove the worst content without becoming legally responsible for everything that stayed. Platforms simply stored other people’s letters. Nobody chose what to show to whom. Nobody promoted some posts above others.

That world is gone.

---

How Substack uses Section 230

When I was banned on March 27, 2026, with the wording “violation of spam and phishing policy”, I received no explanation. This is not accidental.

Section 230 gives Substack a second important thing — the Good Samaritan provision, clause (c)(2): a platform can remove any content it considers “objectionable” and is not required to explain. (In practice, platforms more often invoke clause (c)(1) — immunity from being treated as the publisher — but (c)(2) is the clause that makes silent removals legally defensible.)

The word objectionable legally means nothing. It is “anything we don’t like”. They could have written “content that violates our rules”. They wrote “objectionable”. This is not a mistake. This is a choice. The vaguer the wording, the more freedom.

Substack uses this literally. Bans without explanation. Rejects appeals without citing the violated clause. Tears apart payment flows without warning. When an author sues, Substack raises Section 230 as a shield. The judge sees the shield and in most cases dismisses the case before considering the merits.

The plaintiff doesn’t even get to tell what happened. The case is closed at the motion to dismiss stage. No jurisdiction. Section 230. Not heard.

That is the armor. Not an answer. A refusal to answer.

---

Where the armor does not work

Section 230 protects the platform from liability for someone else’s content. Not for the platform’s own actions.

This is the key distinction. All further cracks come from here. Write down the formula:

Someone else’s content → immunity. Platform’s own actions → no immunity.

What counts as “the platform’s own actions”:

• Breach of the platform’s own contract with the user.
• A promise the user relied on, then the platform broke (promissory estoppel).
• Deceptive or unfair business practices by the platform itself.
• Conversion — seizure of intangible property (a handle, a brand).
• Copyright infringement by the platform itself (deleting comments, posts that were your intellectual property).
• Retaliation — retaliatory acts in response to complaints to a regulator.

Verizon cannot hide behind Section 230 if it broke its own contract with you. It is liable for the breach. Substack is the same.

And notice what this armor actually does. It does not protect against harm. It shifts the cost of proving harm — from the platform to the author. The platform is a corporation with a legal department. The author is one person. A lawsuit is expensive. A lawyer is expensive. The court closes the case under Section 230 before reaching the merits. The law does not “protect free speech on the internet”. The law makes fighting for free speech too expensive for an individual.

This law has been written about for years, in detail. Mike Masnick at Techdirt has been keeping what amounts to a daily journal of Section 230 since 2010 — every significant ruling unpacked. Daphne Keller at Stanford CIS sets the academic standard — her analyses are cited in court. Eric Goldman at Santa Clara Law has been writing a post-by-post chronicle of every Section 230 decision since 2005. Jeff Kosseff wrote a whole book about it — The Twenty-Six Words That Created the Internet, 2019. The material is gathered. The sources are open. The cases are cataloged.

But the bridges between levels were never built. Lawyers write for lawyers. Academics write for academics. Journalists write for editors. An author who has just been banned opens any of these sources and hits a wall of vocabulary. The words are the same as in their life — claim, appeal, liability, free speech. But they are arranged in a language the author cannot enter. Between their case and these analyses is a gap of one word, which is explained by five more words, each of which requires five more.

I am writing this series from inside that gap. From the other side of it.

---

The main crack: algorithm as platform speech

Until recently, a platform could always say: “We just host content. The algorithm is a neutral tool. It doesn’t ‘speak’. Section 230.”

August 2024. The Third Circuit Court of Appeals, Anderson v. TikTok.

A ten-year-old girl saw in her TikTok feed the blackout challenge — a “challenge” to choke yourself to unconsciousness. She repeated it. She died. The mother sued.

TikTok claimed: Section 230. Someone else’s content. Not our “challenge”.

The court answered: “TikTok’s recommendation algorithms are expressive activity. They are the platform’s speech. When you curate and promote specific content to a specific user — it’s your choice. It’s not ‘hosting’. It’s publishing.”

This is a turning point. Section 230 had cracked before — Roommates.com (2008), Lemmon v. Snap (2021) — but always around the edges: how a platform shaped a form, how it designed a feature. Anderson is the first time a federal appeals court named the recommendation algorithm itself as the platform’s speech. And the platform is liable for its own speech.

---

Why this matters for Substack

Substack works differently from TikTok. There is no video feed. But there are algorithms: recommendations, “Rising” rankings in categories, the “Notes” feed, selection into “Featured”.

In my case on March 18, 2026, I recorded a fact: after connecting payments, my blog rose to position seventeen in Rising in Science. This is not hosting. This is the platform’s publishing choice. “Bring this blog up. Hide that one. Show this one in Rising. Drop that one.”

If the logic of Anderson v. TikTok spreads to all platforms — and that is a question of time — Substack will lose Section 230 protection regarding its recommendation decisions.

This is no longer theory. This is the direction practice is moving.

---

What this means for you

If you write on Substack, or any other large US-jurisdiction platform, and try to understand where you stand — the formula has three parts.

First. Trust in “we will protect you” from the platform — zero. Section 230 protects the platform from you, not you from the platform.

Second. Section 230 does not close all possible lawsuits. It closes “you are liable for someone else’s content”. It does not close “you breached your own contract”, “you seized my property”, “you retaliated”, “you violated my copyright”.

Third. The armor is tearing. Slowly, through ten major rulings, but since 2021 courts have been increasingly saying to platforms: “Yes, you are immune to this. But not immune to that“.

---

What you can verify yourself

Section 230 has a public text. It lives in 47 U.S. Code § 230. Google it any way — the first link will go to the official Cornell Law School site.

Anderson v. TikTok is a Third Circuit Court of Appeals ruling, August 2024, docket number 22-3061. The full decision is openly available.

Substack’s Terms of Use, where they directly cite Section 230: substack.com/tos. Look for “safe harbor” or “§ 230”.

---

Interim conclusion

Section 230 is not an eternal truth. It is a 1996 bill written for a “message board”, applied to “recommendation algorithms”. This is a mismatch. Mismatches in US courts are resolved gradually, decision by decision.

Substack knows the armor is wearing out. Their business model is built on this armor. And that is why every important decision they make is in “faster than the courts wake up” mode.

The next four parts of the cycle are about other layers of armor that the platform added on top of Section 230. Each layer works separately. Each deflects its own type of lawsuit. Together they form a building in which the author has not a single door.

— lintara

Next: the arbitration clause. JAMS. San Francisco. Class action waiver. Why this single clause kills ninety-nine percent of lawsuits before they begin. And where it does not work.