The Statute That Punished One Thing and Protected Another

A statute named for decency is remembered for a provision that protects platforms from liability. The Communications Decency Act, enacted as Title V of the Telecommunications Act of 1996 (Public Law 104-104, signed February 8, 1996), set out to criminalize the transmission of indecent material to minors on the Internet. What endured was something else entirely: 47 U.S.C. 230, the immunity provision that shields online services from liability for content created by third parties. The law’s title describes what Congress tried to ban; its legacy describes what Congress allowed to live.

The short answer to how this happened is that Congress never chose between the two approaches. The Senate wrote a criminal ban, the House wrote a competing immunity, and the conference committee that reconciled the two chambers’ telecommunications bills kept both halves in the final text. Eighteen months later, in Reno v. American Civil Liberties Union, 521 U.S. 844 (decided June 26, 1997), the Supreme Court struck down the criminal half as unconstitutional. The immunity half survived. Congress did not choose between criminalizing and immunizing; the Court chose.

The Senate approach began in 1995, when Senator J. James Exon of Nebraska offered an amendment making it a federal crime to transmit obscene or indecent material to a person known to be under eighteen, carrying fines up to $100,000 and prison terms of up to two years. The Senate adopted it on June 14, 1995, by 84 to 16, and the criminal provisions traveled with the Senate bill into conference.

The House approach was nearly the mirror image. Representatives Christopher Cox and Ron Wyden offered an amendment titled the Internet Freedom and Family Empowerment Act: no provider or user of an interactive computer service would be treated as the publisher of content supplied by someone else, and no provider could be held liable for good-faith screening of offensive material. The logic traced to the Stratton Oakmont decision, in which a New York court treated the online service Prodigy as a publisher, and therefore liable for user postings, precisely because Prodigy moderated its bulletin boards. The House passed the amendment 420 to 4.

The conference kept both halves. The conferees placed the Senate’s criminal indecency provisions and the House’s immunity provision into the same title of the same act. The conference report, filed January 31, 1996, stated that Section 230 was added to overturn decisions exposing services to liability because they had restricted access to objectionable material, yet the same report preserved the Exon provisions exposing speakers to prosecution. Both chambers agreed to the report on February 1, 1996, and the President signed the bill a week later. The contradiction became law.

The Court’s choice came quickly. The criminal provisions were challenged the day they were signed, and in Reno v. American Civil Liberties Union the Court struck them down as unconstitutional under the First Amendment. Justice John Paul Stevens, writing for seven Justices, held the provisions substantially overbroad and ruled that the Internet was entitled to the full First Amendment protection given to the print press. Two Justices concurred in part and dissented in part. Section 230 survived untouched.

This article traces the Senate’s criminal approach, the House’s immunity alternative, the conference decision to enact both, the Supreme Court’s invalidation of the criminal half, and the afterlife of what remained. The procedural lesson is the namable claim at the article’s center: when Congress could not decide between punishment and protection, it enacted both, and the conference kept both.

What penalties did the Exon amendment attach to indecency?

The Exon amendment made it a crime to transmit obscene or indecent material to a person known to be under eighteen. Conviction carried fines of up to 100,000 dollars and prison terms of up to two years. The Senate adopted the amendment on June 14, 1995, by 84 to 16, and the penalties entered conference with the Senate bill.

What was the House alternative called?

Cox and Wyden titled their amendment the Internet Freedom and Family Empowerment Act. It provided that no provider or user of an interactive computer service would be treated as the publisher of content supplied by someone else, and that no provider could be held liable for good-faith screening of offensive material. The House passed it 420 to 4.

How was each approach supposed to be enforced?

The Senate approach worked through the criminal courts: federal prosecution after the fact, with fines and prison terms for people who transmitted indecency to minors. The House approach worked through private hands: parents and services filtering content for themselves, with the legal risk of screening removed and the FCC barred from regulating online content.

Who co-sponsored the Senate amendment, and why did the pairing matter?

Senator Dan Coats, a Republican from Indiana, co-sponsored the amendment with Senator James Exon, a Democrat from Nebraska then in his third term. The pairing was bipartisan by design, and the sponsors invoked it throughout the debate as evidence the measure answered a shared concern rather than a partisan one.

When did the Supreme Court decide Reno, and how did the Justices divide?

The Court decided Reno v. American Civil Liberties Union on June 26, 1997, some eighteen months after enactment. Justice John Paul Stevens wrote for seven Justices, holding the indecency provisions substantially overbroad. Two Justices concurred in part and dissented in part. Section 230 was not before the Court and survived untouched.

The two-approach table

	Mechanism	Enforcement	Constituency it served	Fate
Senate measure (Exon-Coats amendment, S. Amdt. 1268, to S. 652)	Criminal ban on transmitting obscene or indecent material to minors	Federal prosecution after the fact, through fines and prison terms	Senators and voters who wanted the state to punish distributors of indecency to children	Struck down as unconstitutional in Reno v. ACLU, 521 U.S. 844 (1997)
House measure (Cox-Wyden amendment, H.Amdt.744, from H.R. 1978, to H.R. 1555)	Immunity: services not treated as publishers of third-party content; Good Samaritan screening shield; FCC barred from content regulation	Private user control: parents and services filtering without legal risk	Online services needing legal certainty; parents wanting household-level control over content	Survived as 47 U.S.C. 230, the Act’s enduring core

The Senate Approach: Crime, Not Categories

The Senate did not approach online indecency as a classification problem. It treated it as a crime. Where the House would later build its case around making content easier to sort and filter, the Senate reached for the tool it knew best: federal criminal penalties applied to people who put indecent material where children could find it. The result was the Exon-Coats amendment, offered as S. Amdt. 1268 to the telecommunications reform bill S. 652, and it became the criminal-indecency core of what Congress would eventually enact as Title V of the Telecommunications Act of 1996, the Communications Decency Act.

The amendment’s chief sponsor was Senator James Exon, a Democrat from Nebraska serving his third term. His co-sponsor was Senator Dan Coats, a Republican from Indiana. The pairing was bipartisan by design, and the sponsors invoked that bipartisanship throughout the debate as evidence that the measure answered a shared concern rather than a partisan one. Exon had pushed similar legislation in the preceding Congress, where it had stalled. In 1995 the political weather changed.

The Political Moment

The change began with a magazine cover. In July 1995, Time published a cover story on what it called cyberporn, illustrated with a child silhouetted against a screen. Inside, the magazine reported what it described as an exhaustive study of online pornography conducted by a research team at Carnegie Mellon University, which had supposedly measured for the first time what people actually downloaded. The reported findings were startling: 83.5 percent of the images stored in Usenet newsgroups, the magazine said, were pornographic. ABC’s Nightline cited the same research in a program about pornography online, with its host telling viewers the material was the kind of smut many pornography stores would not even carry. The coverage gave the public a single impression, that the internet abounded in filth and that a child could reach it with a few keystrokes.

The study’s authority did not survive scrutiny. Online discussion communities challenged the magazine’s claims quickly, and later accounts identified the work as the project of Marty Rimm, an undergraduate in Carnegie Mellon’s electrical and computer engineering program, rather than a formal university research team. Critics also pointed out that much of the underlying data came from private online bulletin boards that sold pornography to adults by credit card, which meant the material was not freely available to children. These objections circulated widely among people who used the internet, but they did not undo the political effect. By the time Exon rose to offer his amendment, the image of a porn-saturated network sat firmly in the public mind, and the Senate’s debate record is full of senators treating the Time cover as the factual premise of the discussion.

Exon had a personal stake in that premise. He asked a friend to download some of the rawer images available online, and the experience converted him. He later said he had known the material was bad, but that seeing it made Playboy and Hustler look like Sunday-school material by comparison. That sentence, repeated in the press, captured the distance between what lawmakers imagined the internet contained and what they confronted when they looked. Exon decided his colleagues needed the same confrontation.

The Blue Folder

Commerce Committee chairman Larry Pressler, a Republican from South Dakota, had earlier told Time that he intended to move to table the Exon bill. Tabling would have killed it quietly, without a recorded vote, and the telecommunications bill to which it was attached was the committee’s signature project. Pressler’s calculation changed the week Exon arrived on the Senate floor carrying a blue folder.

Exon had the downloaded images printed out and stuffed them into the folder, which colleagues and reporters began calling his blue book. He invited senators to stop by his desk on the Senate floor and view them. According to later accounts in the Congressional Record, he also invited members into the Democratic cloakroom, Republicans as well as Democrats, to view images copied from the internet and asked whether they had realized the material was a click away. The senators who accepted the invitation were shocked. Coats later recalled that what they saw was not a magazine centerfold but material he described as brutal, sadistic, and sexually explicit, including the sexual exploitation of young children, bestiality, and the exploitation of women. Whatever one’s view of the amendment’s merits, the blue folder functioned exactly as Exon intended: it replaced an abstract policy question with a stack of photographs and dared his colleagues to look away.

The floor debate was carried live on C-SPAN, and the cameras mattered. Senators who might have been willing to kill the amendment on a procedural motion found themselves weighing a nationally televised vote that could be described, accurately or not, as a vote for pornography. That framing did not come only from the sponsors. The blue folder made it inescapable, because any senator voting no would have to do so with the knowledge that opponents in the next campaign could say he had seen the pictures and voted to keep them a click away from children. The amendment that had been written off for dead weeks earlier suddenly commanded a floor, an audience, and a momentum no committee chairman could table. Pressler’s planned tabling motion, once a routine piece of committee defense, became politically unusable on camera, since moving to kill the amendment would have read as protecting the images in the blue folder rather than defending the telecommunications bill.

What the Amendment Said

The Exon-Coats amendment made it a federal crime to use an interactive computer service to transmit indecent material to a person known to be under eighteen, and to display patently offensive material in a manner available to a person under eighteen. The operative verbs were transmission and display, and the operative mental state was knowledge. The debate record described penalties of fines up to $100,000 and prison terms of up to two years for each offense. The amendment thus operated on two prongs. The first reached private, point-to-point communications: a message sent to a recipient the sender knew to be a minor. The second reached public display: material made available where minors could encounter it without any sender knowing their ages. The first prong depended on the sender’s knowledge of the recipient; the second did not, because on a network open to anyone, availability to minors was the default condition. Critics seized on this asymmetry. The display prong, they argued, effectively required every provider of patently offensive material to assume a child might be watching, which meant every provider had to either block all access, verify every user’s age, or abandon the material. On a medium where anonymous browsing was normal and age verification was barely a concept, the choice was not really a choice. The provision functioned, in the critics’ telling, as a general ban wearing the language of a targeted one.

An earlier draft had threatened computer-system operators with liability for material their subscribers sent, a provision that would have placed colleges and other large network hosts, whose computer systems were the workhorses of the internet, in legal jeopardy. The version the Senate took up generally shielded system operators and aimed instead at the individuals who distributed material deemed offensive. The shield was a concession to the universities, libraries, and service providers that had flooded Congress with petitions and phone calls, warning that criminal liability for subscribers’ conduct would make them police their networks or shut them down. The concession narrowed the amendment’s target but did not answer the deeper objection, which was that the distributors themselves often could not know who was reading.

The language of the indecency standard was not original to the amendment. The sponsors copied it in the exact wording of the federal dial-a-porn statute, the law governing commercial telephone messages that the Supreme Court had upheld in Sable Communications v. FCC. The sponsors’ reasoning was explicit and repeated: if the Court had approved the standard for telephone pornography, it would approve it for internet pornography. The medium changed; the constitutional question, in their telling, did not. Dialing a 900 number and logging on to a network were both, in Coats’s formulation, invasive services that entered the home at the user’s request. The government had already been permitted to keep children off one of them, and the same permission should cover the other.

Opponents attacked every element of that logic. The amendment’s scienter requirement, the word knowingly, struck them as hollow on a network where a sender could not reliably know the age of a recipient. The vagueness of indecency, defined by community standards, struck them as a grant of power to the most conservative community in the country, since a sender could be judged by the standards of any locality where the material was received. And the criminal penalties struck them as a mechanism that would make adults self-censor. The Electronic Frontier Foundation’s Mike Godwin argued that the indecency provisions would transform the internet’s vast library into a children’s reading room, a place where only subjects suitable for children could be discussed. The Electronic Privacy Information Center’s Marc Rotenberg called the amendment government censorship and warned that it would put computer networks under the jurisdiction of the Federal Communications Commission, the agency that enforced the broadcast indecency rules, thereby importing a broadcast regulatory model onto a medium that functioned more like print.

The Vote

The Senate adopted the Exon-Coats amendment on June 14, 1995, by a vote of 84 to 16. The margin was lopsided, and the reason was the politics of the blue folder and the cameras. Few senators wanted to cast a nationally televised vote that might later be characterized as pro-pornography, and the amendment’s framing gave every voter the shelter of the blue folder’s photographs. The 16 votes against it belonged mostly to senators who argued that the measure was unconstitutional on its face, but even some of them paired their opposition with declarations of sympathy for the goal of protecting children. The recorded vote told the story the sponsors wanted: overwhelming, bipartisan, and decisive. It also told a second story that the sponsors did not emphasize: the Senate had voted on the amendment as a referendum on child protection rather than as a statute about speech, because the blue folder had fused the two questions. A vote against the criminal penalties could not be separated, on camera, from a vote against protecting children from the images Exon had shown. That fusion was the amendment’s greatest political asset and its greatest legal vulnerability, since the courts would eventually insist on separating the questions the Senate had deliberately combined.

The lopsidedness also reflected the Senate’s institutional habits. A crime aimed at pornographers is a difficult thing to oppose in public, whatever a senator’s private doubts about its drafting. The debate record shows sponsors returning again and again to the same moves: the indecency standard had survived the Supreme Court in the dial-a-porn context, the penalty targeted distributors rather than carriers, and the alternative was to do nothing while Time’s cover story continued to shock parents. Against those moves, constitutional objections sounded abstract. They were not wrong as law, as the Supreme Court would demonstrate two years later, but as politics they lost decisively on June 14.

The Leahy Alternative

Senator Patrick Leahy, a Democrat from Vermont, offered the Senate the only organized alternative. His proposal would have directed the Justice Department to study the applicability of existing obscenity statutes to computer networks before Congress legislated, and to report back on the least restrictive means of protecting children from exposure to indecent material online. Leahy’s argument was that indecent speech, unlike obscenity, was protected under the First Amendment, and that the Senate was about to ban it across an entire medium without knowing whether narrower tools existed. He waved petitions bearing the names of tens of thousands of opponents and told his colleagues that the internet was something that finally worked in America, and that Congress was about to break it.

The alternative received barely any discussion. The Senate’s attention was fixed on the amendment that had the blue folder behind it, and the study-first approach could not compete with the urgency Exon had manufactured. Leahy’s proposal was scrapped once the Exon-Coats amendment was adopted. Its supporters objected to the indecency provisions on several grounds that would later reappear in court: that an outright ban was not the least restrictive means of protecting children, since parents already had tools to filter material; that the ban would force adults to censor themselves; that community standards would reduce adult speech to what was decent in the most conservative community or appropriate for very young children; that the law created different standards for the same material in print and on a screen, so that works legal in a bookstore or on a library shelf would become criminal when transmitted over a network; and that the measure ignored the interactive character of the medium, in which users exercised far more control over what they encountered than television viewers did. These arguments formed the intellectual core of the opposition, but they could not slow the amendment’s momentum.

The Sable Precedent and Its Critics

The sponsors’ constitutional confidence rested almost entirely on Sable. In Sable Communications v. FCC, decided in 1989, the Supreme Court had upheld a federal ban on indecent commercial telephone messages while striking down a broader ban on noncommercial ones. Exon and Coats lifted the indecency language from the dial-a-porn legislation word for word and built their case on a syllogism: the Court approved it for telephone pornography, so it would approve it for internet pornography. The services were alike in the respects that mattered, both invasive, both entering the home, both requiring the user’s affirmative act to engage. Coats put it in the Record as a principle extending across every medium: the responsibility for restricting minors’ access fell on the provider of the material. The Ferber principle was the amendment’s constitutional foundation, and the sponsors cited it as settled law: the state’s interest in the well-being of minors was compelling, and the indecency standard was the tool the Court had already validated. In their view, the Senate was not breaking new constitutional ground at all. It was applying an approved rule to a new set of facts, the way legislatures do whenever technology moves into territory the courts have already mapped. The internet’s novelty was, to the sponsors, a distraction. The transaction was the same one the Court had judged in the telephone context: a provider offering indecent material to whoever called or clicked, and a child on the other end.

The critics’ answer was that the medium was not the same, and that the Supreme Court had said so itself. The degree to which content could be regulated, they argued, depended on the characteristics of the medium, and the internet’s characteristics cut against the broadcast analogy. Broadcast indecency rules rested on the scarcity of spectrum and the pervasiveness of television, which entered homes uninvited. The internet was neither scarce nor uninvited; its users dialed in, navigated menus, and chose what to download, a degree of user control that resembled print more than broadcasting. Godwin’s FCC-jurisdiction warning followed from this: if Congress treated the network as a broadcast medium, the Commission’s indecency machinery would extend to computer networks, and the famous prohibitions that governed the broadcast airwaves would migrate onto a medium the Commission had never been given authority over. The Senate’s vote settled the policy question for the moment; the medium question remained open, and it would be answered not by the Senate but by the courts.

Dissent From the House Side

The Exon approach did not command uniform support even within Congress. House Speaker Newt Gingrich publicly opposed it, calling it a violation of free speech and of the rights of adults to communicate with one another. The Speaker’s objection carried weight because the amendment still had to survive the House and the conference process, and because it came from the leader of the chamber that had the most say over what the final telecommunications bill would contain. Gingrich’s criticism foreshadowed the House’s different path: where the Senate criminalized indecency, the House would take up the question of how to empower users and parents to filter content for themselves. The two chambers were solving the same problem from opposite directions, and the tension between the Senate’s criminal statute and the House’s filtering provisions would have to be reconciled before anything reached the President’s desk.

Gingrich’s opposition also underlined a fault line inside the Republican coalition. The Exon-Coats amendment was bipartisan, and its co-sponsor was a prominent conservative, but the party’s House leadership was moving toward a free-market, technology-empowerment position that treated government censorship as the greater threat. The Speaker was not defending pornography; he was defending the adults’ right to communicate and the principle that the government should not police the content of a medium whose defining feature was user choice. That argument would find its strongest institutional expression in the House’s alternative provisions, and its strongest legal expression in the Supreme Court’s eventual ruling. Gingrich’s position also carried a practical political implication: the amendment could not become law on Senate votes alone, and the House leadership’s skepticism meant the criminal-indecency approach would have to survive a second chamber whose instincts ran toward technological self-help rather than federal prosecution.

After the Vote

The 84-to-16 vote sent the Exon-Coats amendment into the Senate-passed telecommunications bill and, from there, into the conference negotiations that would produce the Telecommunications Act of 1996. The amendment’s sponsors left the floor confident that they had done the constitutionally safe thing: they had borrowed language the Court had already blessed, aimed it at distributors rather than carriers, and grounded it in the compelling interest in safeguarding the physical and psychological well-being of minors that the Court had recognized in New York v. Ferber. They believed the Court would see dialing and clicking as equivalent acts.

What the Senate had done was choose a criminal model of governance for a communications problem. A crime is investigated after the fact, punished by fines and imprisonment, and aimed at deterring conduct through the threat of prosecution. The internet the amendment governed was a medium where conduct was hard to attribute, where senders could not verify recipients’ ages, and where a single act of transmission could reach thousands of screens across dozens of jurisdictions at once. The two-year prison term was the measure’s answer to that difficulty: if the wrongdoing could not be monitored in advance, it could be punished severely enough after the fact to make providers stop. The critics’ objection was that this severity would fall not on pornographers alone but on everyone who discussed sex, health, literature, or art online in language that any community might call indecent. The Senate weighed those arguments for a debate and chose the blue folder’s photographs over them.

The House Approach: The Cox-Wyden Bill

While the Senate moved toward criminal bans for objectionable material on the new computer networks, the House produced a rival theory of the same problem. Its authors argued that the way to make the networks safer for families was not to punish the companies that carried traffic but to free them, legally and financially, to help parents control what reached their children. That argument became H.R. 1978, the Internet Freedom and Family Empowerment Act, offered on the House floor as Amendment 744 to the House telecommunications bill and adopted on August 4, 1995, by a recorded vote of 420 to 4.

The House’s answer deserves attention on its own terms, not only as the source of the immunity provision that survived. It represented a fully formed alternative theory of online safety, with named authors, a public rationale, and a near-unanimous vote. Understanding that theory explains why the provision looked the way it did, and why its sponsors believed that immunity would produce more protection for families rather than less.

The sponsors and the trigger

The bill belonged to two men whose politics overlapped almost nowhere else. Christopher Cox was a Republican congressman from California, serving the 47th district. Ron Wyden was a Democratic congressman from Oregon, serving the 3rd district. Their partnership on this measure was deliberate. The bill needed to read as a bipartisan rejection of the Senate’s criminal approach, and its sponsors’ parties, from opposite ends of the ideological spectrum, made that rejection explicit before a word of the text was read.

Timing made the purpose plain. On June 14, 1995, the Senate had voted 84 to 16 to adopt Senator James Exon’s Communications Decency amendment, which treated the transmission of indecent material over computer networks as a criminal offense. Sixteen days later, on June 30, 1995, Cox and Wyden introduced H.R. 1978. The press understood the sequence as a response. A Chronicle of Higher Education report from that month described the Cox-Wyden bill as aimed directly at the Senate’s legislation, a House counterproposal rather than a companion measure.

The second trigger came from a courtroom. On May 24, 1995, the New York Supreme Court ruled in Stratton Oakmont, Inc. v. Prodigy Services Co. that Prodigy, an online service, qualified as the publisher of defamatory statements posted on its bulletin boards by users. The court pointed to three facts: Prodigy posted content guidelines for users, enforced them through board leaders, and used screening software to remove offensive language. Because Prodigy had exercised editorial control, the court held, it had assumed the responsibilities of a publisher, including legal liability for content it had not written. An earlier federal decision, Cubby, Inc. v. CompuServe Inc. in 1991, had treated a service that did not screen its content as a distributor, with far lower liability exposure. The two rulings together created what industry lawyers called a moderator’s dilemma. A service that ignored objectionable material enjoyed protection. A service that tried to remove it became legally responsible for everything it missed. For any company building an online service in 1995, the rational move was to look the other way.

Cox and Wyden treated both triggers as symptoms of one error. The Senate’s answer to online indecency was to threaten punishment, and the courts’ answer to online screening was to punish the screeners. The House bill reversed the second logic and offered an alternative to the first.

The House leadership shared the sponsors’ skepticism of the Senate approach. Speaker Newt Gingrich said publicly that the Exon amendment violated free speech and the rights of adults to communicate with one another, and he predicted it would not survive. That posture mattered. In a Republican-controlled House whose leadership doubted the Senate’s criminal approach, the Cox-Wyden bill did not arrive as a fringe alternative. It arrived as the preferred answer of the chamber’s leadership, carried by a bipartisan pair.

The bill and its declared purpose

The official title of H.R. 1978 read, in full, as an instruction: “to encourage and protect private sector initiatives that improve user control over computer information services.” The short title, the Internet Freedom and Family Empowerment Act, arranged the bill’s two selling points in one phrase. The official title, by contrast, stated the mechanism: private-sector initiatives, user control.

The Congressional Research Service summary, the standard plain-language account of an introduced bill, described the policy in three moves. The bill declared a national policy of promoting the continued development of the Internet and other interactive computer services while preserving the competitive free market that existed for them. It addressed concerns about objectionable material on those services, sexual material including child sexual exploitation and otherwise, through user control rather than criminal bans. And it shielded service providers and users in the ways described below.

The bill’s formal life was short. It was introduced in the House on June 30, 1995, referred to the House Commerce Committee the same day, and referred to the Subcommittee on Telecommunications and Finance on July 10, 1995. It attracted cosponsors across the aisle and across the country: Ron Wyden himself on introduction day, then Robert Matsui, Rick White, and Bart Stupak on July 25, and Dana Rohrabacher on July 31. The subcommittee never reported it. Its future lay not in committee but on the floor, as an amendment to the larger telecommunications bill.

The bill opened with a statement of national policy rather than a penalty, and the placement was deliberate. It declared that the policy of the United States was to promote the continued development of the Internet and other interactive computer services and to preserve the competitive free market then existing for them. Policy declarations perform a function beyond rhetoric. They tell courts how to read the provisions that follow, directing interpreters to construe the protections broadly and the limitations narrowly. By placing development and competition first, the bill framed its concern with objectionable material, including the sexual exploitation of children, as a problem to be addressed inside that policy rather than as a reason to set the policy aside.

The mechanics, clause by clause

The bill contained three operative protections. Each solved a distinct legal problem, and together they formed a coherent strategy: make screening safe, assign responsibility to the author, and keep the regulator out.

The first clause provided that a provider or user of an interactive computer service was not to be treated as the publisher or speaker of any information provided by another information content provider. In plain language, legal responsibility for a message belonged to the person who wrote it, not to the service that carried it. The clause answered the Prodigy ruling directly. Under that ruling, exercising editorial control had converted a service into a publisher. The bill restored the distinction and made it unconditional: carrying someone else’s words did not make a service the speaker of those words. The clause also covered users, not only companies, so that a person who forwarded or posted another’s material was not treated as its publisher either. The effect was to collapse the dangerous line the courts had drawn between services that screened and services that did not. After this clause, screening could not change a service’s legal category.

The second clause provided Good Samaritan protection: no provider or user of an interactive computer service would face liability for good-faith actions taken to restrict access to material the provider or user considered objectionable. In plain language, a service that filtered out pornography or blocked harassing messages would not be punished in court for the material its filters missed. This was the direct inversion of the moderator’s dilemma. Under the Prodigy logic, the act of filtering was the evidence of control that triggered liability. The Good Samaritan clause made the act of filtering the evidence of good faith that defeated liability. The standard was subjective in one respect and objective in another: the service decided for itself what counted as objectionable, but its actions had to be taken in good faith, meaning an honest effort to restrict access rather than a pretext for something else.

The third clause declared that nothing in the bill would be construed to authorize the Federal Communications Commission to regulate the content of interactive computer services. In plain language, the new legal protections were not a license for the broadcast regulator to extend its reach to the networks. The clause addressed a suspicion that ran through the entire telecommunications debate: that any federal action touching the Internet would become an opening for content regulation modeled on the broadcast rules. Cox and Wyden closed that opening in the text itself.

The three clauses worked as a system. The publisher clause removed the baseline risk of carrying third-party content. The Good Samaritan clause removed the added risk of screening it. The FCC clause removed the risk that the protection would be reinterpreted as an invitation to regulate. A service that did nothing was safe. A service that filtered was safer. No clause punished the service that tried and failed to catch everything, and no clause gave a regulator authority to punish the service that tried too little.

The August 4 vote

H.R. 1978 never advanced as a standalone bill, because it did not need to. The House was assembling its own telecommunications legislation, H.R. 1555, and the Cox-Wyden language was offered as Amendment 744 to that bill during floor consideration. The amendment carried the full substance of the original bill: the policy declaration, the publisher clause, the Good Samaritan clause, and the FCC limitation.

On August 4, 1995, the House adopted Amendment 744 by recorded vote, 420 to 4. The same day, the House passed H.R. 1555 itself by 305 to 117. The two votes told different stories. The telecommunications bill as a whole divided the chamber, drawing substantial opposition on its many other provisions. The Cox-Wyden amendment, by contrast, drew near unanimity.

That margin carried a specific meaning. A recorded vote puts every member’s name beside a choice, and members facing competitive elections do not vote 420 to 4 for a provision they consider obscure or risky. The amendment was offered as a named alternative to the Senate’s criminal approach, and the House treated it as such. The vote did not merely add a provision to a bill. It declared the House’s theory of the problem: user empowerment and private screening, not criminal prohibition, was the chamber’s answer to objectionable material online. A margin of 420 to 4 is the legislative equivalent of a signed confession of intent.

The contrast between the two August 4 votes also gave the House negotiators leverage. When the two chambers met to reconcile their telecommunications bills, the House side could point to a 420 to 4 vote for the immunity approach and note that the broader bill carrying it had passed with votes to spare. The Senate, meanwhile, had committed itself to criminal language that the House leadership considered unworkable. The final conference product reflected that imbalance. The Senate’s criminal provisions entered the final title, but so did the Cox-Wyden language, and the two sat side by side in the statute that became Title V of the Telecommunications Act of 1996.

Immunity as encouragement versus prohibition as deterrence

The Cox-Wyden philosophy stood opposite to Exon’s in method while claiming the same goal. Exon’s approach worked through deterrence. It defined criminal offenses around the transmission of obscene or indecent material and attached jail terms and fines, on the theory that punishment would keep the material off the networks. The mechanism was prohibition, and its enforcement arm was the prosecutor.

Cox and Wyden worked through encouragement. Their bill removed legal risk from the services themselves, on the theory that the risk of liability was what kept services from screening and what kept investment from flowing into user-control technologies. Where Exon threatened the people who carried traffic, Cox and Wyden protected them. Where Exon trusted criminal law to define acceptable content, Cox and Wyden trusted households, armed with private screening tools, to decide for themselves.

The two theories also made different assumptions about what was feasible. The criminal approach assumed that prosecutors could identify and punish the sources of objectionable material at the scale of a global network. The House approach assumed that screening technology was developing and that parents, given tools and information, would use them. Cox and Wyden did not argue that objectionable material was harmless. They argued that the government’s chosen weapon, criminal prohibition, would fail on constitutional and practical grounds while the private sector’s tools, filtering software and user-controlled blocking, would succeed if the law stopped punishing the companies that built and deployed them.

The Senate’s theory assumed that the threat of prosecution could scale to the size of the networks. The House’s theory assumed that it could not, and that the networks would grow faster than any enforcement apparatus. In 1995 the networks were small enough that both assumptions were still arguments rather than facts. Cox and Wyden bet on growth. They designed their bill for a network with millions of users and a volume of traffic no service could review message by message, where no prosecutor could pursue every source. Immunity made that scale survivable. Criminal liability would have made it uninsurable.

The Prodigy case made this philosophy concrete rather than abstract. A service that marketed itself as family-oriented and screened its boards had been told by a court that its screening made it a publisher. The House bill’s answer was to make that screening legally safe. Under Cox-Wyden logic, the service that screened most aggressively deserved the most protection, not the least, because screening was the behavior the bill wanted to encourage.

Why immunity was framed as pro-family rather than pro-industry

The bill’s name put the family before the industry. The short title led with Internet freedom and closed with family empowerment, and the official title spoke of improving user control. This ordering was not decorative. Cox and Wyden faced a political problem: a bill that shielded online companies from lawsuits could be read as a gift to industry. They solved it by defining the industry’s interest and the family’s interest as the same thing.

The family empowerment logic ran as follows. Parents, not prosecutors, were the proper decision makers about what entered the home. Criminal bans set a single national standard through the blunt instrument of prosecution, and they reached adults as well as children. User-control technologies, by contrast, let each household set its own standard. A family that wanted strict filtering could have it; a family that wanted none was not bound by the strictest one. Immunity served this logic because it gave services a reason to build, sell, and install the tools. Without immunity, a service that offered filtering invited lawsuits over everything its filters missed. With immunity, the service could compete on the quality of its screening without legal risk, and parents could shop among competing tools.

The Prodigy episode supplied the concrete example the sponsors needed. Here was a company that had tried to keep its service clean for families and had been punished for the effort. The Cox-Wyden argument was that the law should reward exactly that behavior. Immunity, on this account, was not a shield for irresponsibility but a reward for responsibility. The service that screened most would be the service families trusted most, and the law should make that competition possible rather than dangerous.

The FCC limitation fit the same frame. A regulator empowered to police content would substitute bureaucratic judgment for parental judgment. By barring the FCC from content regulation in the bill’s text, Cox and Wyden signaled that the family’s control was the endpoint of the policy, not a way station on the road to government control.

The phrase user control performed specific work in the bill’s design. It located authority at the edge of the network rather than at its center. A criminal ban locates authority with the government, which defines the forbidden category once for everyone. User control locates authority with the household, which chooses among competing tools. The sponsors preferred the second arrangement for a reason they stated openly: families differed in what they found objectionable, and a single federal standard would fit none of them well. Private screening tools could be calibrated to different households. Criminal statutes could not.

The vote record against the slipped-in story

Later commentary sometimes described the platform immunity provision as a measure that slipped into the final law with little notice, inserted at the eleventh hour during the House-Senate conference by its two sponsors and discovered only afterward by observers who had overlooked it. That account treats the provision as an accident of process.

The House record does not support it. H.R. 1978 was introduced as a standalone bill on June 30, 1995, with a short title that announced its purpose and an official title that stated its mechanism. Its policy was summarized in public by the Congressional Research Service in plain language: promote the Internet, preserve the free market, and address objectionable material through user control. It gathered cosponsors from both parties. It was then offered as a named amendment, H.Amdt.744, to the House telecommunications bill, and the House adopted it on a recorded vote of 420 to 4 on August 4, 1995, as the chamber’s explicit alternative to the Senate’s criminal approach. Every one of those steps is the opposite of quiet insertion. A recorded vote is a public act by hundreds of named individuals, and a 420 to 4 margin is a declaration, not a whisper.

The sponsors themselves described the provision’s purpose on the record. Representative David Dreier, in House remarks in January 2001, returned to the Cox-Wyden amendment and stated what it did: it ensured that Internet service providers, website hosts, portals, and others were not burdened by tort liability for content created by others. Dreier, who had supported the amendment, called it the chief reason for the Internet’s growth as the “vast democratic fora” described by the Supreme Court. His remarks treated the amendment as a deliberate policy choice with named authors and a named purpose, not as language that had escaped anyone’s attention.

The eleventh-hour account also misplaces the provision’s journey. The language did enter the final statute through the House-Senate conference, which is the normal mechanism by which the two chambers’ bills are reconciled, and it entered as the House’s voted position rather than as new material introduced by its sponsors. Conference committees resolve differences between the chambers’ bills; they do not invent them. The difference the conference resolved was real and recorded: the Senate had voted 84 to 16 for criminal prohibition on June 14, and the House had voted 420 to 4 for immunity and user empowerment on August 4. The conference kept both.

The provision that survived conference and became law is the surviving provision, explained elsewhere in this series. Its survival was not an accident of drafting. The House had voted for it, by name, by a margin of 420 to 4, before it ever reached a conference committee.

The Conference: Keeping Both

The Communications Decency Act arrived at the end of the legislative process carrying a contradiction that no one had voted on directly and that no drafter had resolved. The conference committee assigned to merge the Senate and House telecommunications bills faced two irreconcilable philosophies about the internet. The Senate believed the medium needed criminal indecency prohibitions. The House believed the medium needed liability protection for the services carrying it. A conference exists to dispose of the differences between two chambers, and the conferees on this bill disposed of the difference by refusing to choose. They kept both philosophies and placed them side by side in Title V of the conference report, the portion of the bill that carried the Communications Decency Act name. That decision, made mostly in December 1995, explains why the enacted law simultaneously treated online services as criminals to be punished and as platforms to be immunized. The conference kept both, and the statute never reconciled them.

Two bills, two philosophies

The structural reason a conference existed at all was that the two chambers had passed fundamentally different bills. In the Senate, S. 652, the Telecommunications Competition and Deregulation Act of 1995, carried the Exon-Coats amendment as its internet provision. Senators James Exon and Dan Coats had pushed criminal penalties for transmitting indecent material over computer networks, a provision the Senate adopted in June 1995. The philosophy behind it treated the internet like a broadcast medium whose content Congress could police: because children could encounter indecent material online, the adults who transmitted or made it available faced prosecution.

In the House, H.R. 1555 carried an entirely different answer. Representatives Christopher Cox and Ron Wyden had added an amendment immunizing interactive computer services from liability for third-party content and for good-faith efforts to block offensive material. The Cox-Wyden philosophy treated the internet as a communications infrastructure whose growth depended on shielding the carriers: if services faced publisher liability for everything their users posted, they would either censor aggressively or shut down, and the medium would collapse under legal risk.

These two approaches did not merely differ. They assumed opposite theories of who the online service was. Under the Senate approach, the service was a publisher-like actor whose transmission of indecent material to minors was a crime. Under the House approach, the service was a conduit whose protection from liability was the precondition of the medium’s existence. Both chambers passed their versions, the Senate on June 15, 1995, by a vote of 81 to 18, and the House on October 12, 1995, without objection. A conference committee was then required, as how conference committees resolve competing measures describes, to produce a single text from the disagreeing votes of the two Houses. That committee held sessions in October and December 1995. Its members had to decide, or avoid deciding, what the internet was.

The composition of the conference made the outcome predictable. Senators Exon and Coats, the cosponsors of the indecency provisions, were named to the committee, and every other senator on it had voted for those provisions. The lone opponent of the indecency language on the committee was Representative Rick White, who had cosponsored the House immunity amendment, known in that chamber as the Online Family Empowerment Amendment. The Senate side of the table was therefore unanimous for criminalization, while the House side was split, with most of its conferees representing a chamber that had passed its telecommunications bill without objection and without any recorded fight over the internet title. In that asymmetry, the House resistance to the indecency language had little institutional weight, and the immunity language survived mainly because it had already passed the House and no senator had a strong reason to strike it.

December 1995: the House conferees’ choice

The decisive moment came in December 1995, when the House conferees agreed to the Senate-style indecency language. The House had been the chamber more resistant to internet regulation, partly because of the Cox-Wyden amendment’s logic and partly because its members were, on average, younger and more familiar with the medium. In conference, that resistance collapsed. The committee met on October 25, December 6, and December 12, 1995, and it was during the December sessions that the House members accepted the criminal provisions, betting that the liability shield they had already won in their own chamber would survive alongside them.

Senator Patrick Leahy of Vermont, who had opposed the Exon-Coats provisions from the start, went to the Senate floor to register what had happened. Speaking in December 1995, he said the action of the House members on the conference should send tremors through the internet community and defenders of the First Amendment, because they had agreed to a provision that would effectively ban constitutionally protected speech on the internet. His warning turned on a practical mechanism. If the amendment became law, internet users could no longer engage in freewheeling discussion in newsgroups and other areas accessible to minors. They would have to limit all language used and topics discussed to that appropriate for kindergarteners, just in case a minor clicked onto the discussion, because any user who crossed the vague and undefined line of indecency would face two years in prison and fines. Leahy illustrated the chilling effect with a report that, only the week before, America Online had deleted the profile of a Vermont woman who communicated with fellow breast cancer survivors online, because she used the word breast. The anecdote made his point concrete: a vague federal indecency standard would push private services to over-censor, and the over-censorship would reach speech no one would describe as pornographic, from literary quotation to discussions of birth control and AIDS prevention.

Leahy’s warning was a minority view inside the process. The conference majority treated the indecency language as a settled moral commitment. No senator wanted to cast what could be construed as a pro-pornography vote, and the political cost of opposing the provision was understood by everyone in the room. The House conferees’ December agreement meant the Senate philosophy would be in the final text. The question was whether the House philosophy would be there too.

The mechanics of keeping both

It was, because nothing about conference procedure requires choosing between philosophies. A conference committee’s job is to dispose of all differences between the House and Senate bills by producing a single substitute text. Conferees may adopt provisions from either chamber, modify them, or drop them, but the rules do not compel them to harmonize the underlying theories. Each chamber then votes the entire package up or down. This structural fact made the contradiction possible: the conferees could satisfy the Senate by keeping the criminal provisions and satisfy the House by keeping the immunity provisions, and the resulting text would face only a yes-or-no vote in each chamber, with no member able to strike one half of the internet title without killing the entire telecommunications bill.

That packaging was decisive. The Communications Decency Act was Title V of a 128-page conference report rewriting the nation’s communications laws for the first time since 1934. Members who wanted telecommunications deregulation, which was the overwhelming majority, had to accept Title V as part of the bargain. The House sealed the arrangement procedurally when it adopted House Resolution 353, which waived all points of order against the conference report and its consideration and provided that the report would be considered as read. Members could not amend the conference text on the floor. They could not strip the indecency provisions while keeping the immunity provision, or keep the immunity while striking the criminal sections, because the conference report was an amendment in the nature of a substitute offered as a single proposition. A member who objected to half of Title V had only one lever, a vote against the entire telecommunications bill, and almost no one was willing to pull it. Members who wanted the indecency provisions could vote for the report confident they were included. Members who wanted the Cox-Wyden immunity could vote for the report confident it was included too. The conference did not resolve the tension between criminalizing indecent transmissions and immunizing the services that carried them. It made the tension a feature of the compromise, because each side’s provision was the price of the other’s support.

The drafters did insert one textual bridge between the two halves, and it revealed how little reconciliation they intended. The Cox-Wyden immunity section carried a savings clause stating that nothing in the section shall be construed to impair the enforcement of section 223 of title 47, the very indecency provision the title had just created. The clause acknowledged the contradiction and declined to settle it. It told courts that both provisions were meant to operate, and left the courts to figure out how a service could be simultaneously immunized from liability for third-party content and criminally liable for transmitting indecent content to minors.

Title V as enacted

As the conference reported it, Title V, sections 501 through 509 of the conference report, carried the short title Communications Decency Act of 1996. Its internal architecture was a map of the compromise. The criminal prohibitions rewrote section 223 of the Communications Act of 1934. Section 223(a)(1)(B) criminalized the knowing transmission of obscene or indecent communications to a person known to be under eighteen by means of a telecommunications device. Section 223(d), aimed squarely at the internet, went further: it criminalized the knowing use of an interactive computer service to send patently offensive material to a specific person under eighteen, or to display such material in a manner available to a person under eighteen, with penalties of fines under title 18 or imprisonment of not more than two years, or both.

The breadth of those prohibitions was qualified by two affirmative defenses at section 223(e)(5). One covered defendants who took good faith, reasonable, effective, and appropriate actions to restrict access by minors to the prohibited communications. The other covered defendants who restricted access to the material by requiring certain designated forms of age proof, such as a verified credit card or an adult identification number or code. These defenses were the conference’s nod to technological reality: Congress was criminalizing conduct in a medium where identifying the age of the recipient was, for most speakers and most services, practically impossible. The defenses assumed a future of age-verification infrastructure that did not yet exist for ordinary users, bulletin boards, newsgroups, or chat rooms.

Then, in the same title, section 509 of the conference report added a new section 230 to the Communications Act, the Cox-Wyden platform liability provision. Section 230 provided that no provider or user of an interactive computer service would be treated as the publisher or speaker of information provided by another information content provider, and it protected good-faith action to restrict access to objectionable material. Where section 223 threatened services with prison for indecent transmissions, section 230 assured them they were not publishers of their users’ speech. The same title that defined the service as a potential criminal defined it as an immune conduit. No conference report language explained how a provider was supposed to read these sections together, because the conferees had not written them to be read together. They had written them to be voted on together.

The operational conflict was immediate and concrete. Consider a service hosting a discussion forum where users posted material that was patently offensive as measured by contemporary community standards. Under section 223(d), the service faced fines and up to two years of imprisonment if that material was displayed in a manner available to a person under eighteen. Under section 230, the same service was not the publisher of its users’ posts and was protected when it acted in good faith to restrict access to objectionable material. The criminal half of the title punished the failure to keep indecent material from minors. The immunity half encouraged the service to filter and block, yet offered no protection against the criminal charge if the filtering was imperfect. The affirmative defense for good faith, reasonable, effective, and appropriate action only deepened the trap, because it invited courts to judge after the fact whether a service’s precautions had been effective enough to earn the defense. A service that screened aggressively risked missing something and facing prosecution. A service that did nothing forfeited the defense entirely. The title demanded perfect age-gated filtering from an industry that lacked the tools to build it, then immunized the industry from civil liability while exposing it to criminal liability for the same conduct.

Section 561 and the fast track

The conference anticipated that the indecency provisions would be challenged in court, and it built the challenge into the statute. Section 561 of the Act created a special judicial review procedure: an expedited hearing by a three-judge district court panel, with direct appeal to the Supreme Court. The purpose was speed. Ordinary constitutional litigation would wind through a single district judge, a court of appeals, and then a petition for certiorari that the Supreme Court could decline. Section 561 compressed that path so that the constitutionality of the indecency provisions would be settled quickly, by a full evidentiary hearing before three judges whose factual findings the Supreme Court could review directly.

The inclusion of section 561 was itself an admission. Congress does not write expedited constitutional review into every statute. It writes it into statutes whose authors know the courts will be asked to strike them down. The conferees who kept the criminal provisions also supplied the procedural machinery for their invalidation, a tacit acknowledgment that Leahy’s December warning might prove correct. The fast track served both sides of the congressional argument. Supporters of the indecency provisions could tell constituents they had voted to protect children, knowing the courts would render the final verdict. Opponents could count on the judiciary to test the law without years of uncertainty. The provision transformed the statute from a final legislative word into an opening move in a lawsuit. It also reflected a hard-headed calculation about the medium itself. The internet was changing faster than the legislative calendar, and a constitutional challenge that took five years to reach the Supreme Court would have regulated a version of the network that no longer existed by the time the Court ruled. Expedited review kept the legal question and the technology on something closer to the same schedule.

Adoption and the same-day lawsuits

The conference report was filed on January 31, 1996, as House Report 104-458, and both chambers adopted it on February 1. The House agreed to the conference report by a recorded vote of 414 to 16, Roll Number 25. The Senate agreed by a vote of 91 to 5, Record Vote Number 8. The margins were overwhelming, and they were votes for the entire telecommunications package, not for the Communications Decency Act standing alone. No member had the option to approve the deregulation of local telephone markets while rejecting the indecency provisions, or to keep the Cox-Wyden immunity while striking the criminal sections. The conference’s keep-both compromise reached the floor as a take-it-or-leave-it proposition, and both chambers took it.

President Bill Clinton signed the bill on February 8, 1996, as Public Law 104-104, the statute described in the 1996 telecommunications statute. The enrolled bill had been presented to the President on February 2, six days after the chambers adopted the conference report, a brief interval that left no time for second thoughts about Title V. The signing made the Communications Decency Act law. It also started a clock the conference itself had wound. On the same day the President signed, February 8, 1996, the American Civil Liberties Union, the Electronic Frontier Foundation, and eighteen other plaintiffs filed suit in federal court in Philadelphia, challenging the constitutionality of sections 223(a)(1) and 223(d). The speed was not a surprise to anyone who had watched the conference. The plaintiffs had been organizing during the months of conference deliberation, and the section 561 fast track existed precisely because everyone expected this lawsuit.

The early litigation followed the path the statute had laid. A week after the filing, District Judge Ronald Buckwalter entered a temporary restraining order against enforcement of the indecency provision, finding the term indecent too vague to support a criminal prosecution. A second suit by twenty-seven additional plaintiffs was then filed, the two cases were consolidated, and a three-judge district court was convened pursuant to section 561 of the Act. After an evidentiary hearing, that court entered a preliminary injunction against enforcement of both challenged provisions, with each of the three judges writing separately but reaching a unanimous judgment. The direct appeal to the Supreme Court that section 561 provided was already in motion within months of the signing. The conference had written the law and its own judicial reckoning in the same text.

Why the contradiction was possible

The internal contradiction of Title V was not an accident of drafting. It was the predictable product of a procedure that rewards inclusion over coherence. A conference committee disposing of differences between two bills has no institutional reason to reject a provision that one chamber demands, and every reason to keep both chambers’ provisions if keeping them secures the votes for the package. The Senate demanded criminal indecency provisions. The House demanded platform immunity. The conference supplied both, separated by a savings clause that preserved the conflict rather than resolving it. The full House and Senate then ratified the package by lopsided margins, because the package was a telecommunications bill first and an internet bill second.

This is why the Communications Decency Act read as two statutes wearing one title. The criminal sections assumed services could identify minors, verify ages, and restrict access, and punished them when they failed. The immunity section assumed services could not be treated as publishers of their users’ speech at all. The affirmative defenses assumed a technological infrastructure for age verification that the medium did not possess. And section 561 assumed the courts would have the last word, which they did. The conference kept both philosophies because keeping both was the only way to get the bill to the President’s desk.

The episode also revealed something durable about how Congress legislates new technology under political pressure. When a subject is poorly understood and the political cost of restraint is high, the conference becomes a place where contradictory provisions accumulate rather than cancel out, because each provision has its own constituency and its own floor votes behind it. The drafters did not need the provisions to agree with each other. They needed them to agree with the vote count. The statute that resulted told online services, in the same breath, that they were criminals and that they were immune, and it left the resolution of that command to the judges.

The decision: Reno v. ACLU

President Clinton signed the Telecommunications Act of 1996 on February 8, 1996. The same day, the first lawsuits challenging the Communications Decency Act landed in federal court. The American Civil Liberties Union, joined by a coalition of publishers, online service providers, writers, librarians, and civil liberties organizations, filed suit in the Eastern District of Pennsylvania in a case styled ACLU v. Reno. A second coalition, led by the American Library Association, filed a parallel challenge, ALA v. DOJ, on the same day. Congress had anticipated immediate constitutional attack: the Act contained a special review provision, section 561, that funneled challenges to the indecency provisions directly to a three-judge panel of the district court, with appeal lying straight to the Supreme Court. The two cases were consolidated before that panel, and the litigation moved with unusual speed because the criminal prohibitions Congress had written applied to speech already flowing across the network every hour of every day.

The plaintiffs’ theory was straightforward, and it made no concession to obscenity or child pornography. They did not defend material that met the Miller test for obscenity, nor did they contest Congress’s power to criminalize the sexual exploitation of children. Their argument targeted the Act’s two broader categories: “indecent” communications under section 223(a)(1)(B), and “patently offensive” communications under sections 223(d)(1) and (2). Indecent speech, the plaintiffs emphasized, was constitutionally protected speech for adults. Under existing First Amendment doctrine, the government could not ban adults from reading, viewing, or hearing indecent expression simply because the same expression would be inappropriate for children. A 1989 decision, Sable Communications of California, Inc. v. FCC, 492 U.S. 115, had made the point explicitly for a neighboring technology: sexual expression that was indecent but not obscene was protected by the First Amendment, and the government could not suppress it for adults in the course of shielding children. The CDA, the plaintiffs argued, repeated Sable’s error on a vastly larger scale. It criminalized constitutionally protected communication between willing adults whenever a minor might possibly encounter it. Because the statute attached criminal penalties to the transmission or display of indecent material without requiring any intent to reach a child, the only safe course for any speaker was silence. A publisher, a librarian, a health educator, or a private citizen discussing sexuality online faced the same criminal exposure as a purveyor of pornography, because the statute drew no distinction among speakers and imposed no requirement that the speaker target, know, or even suspect the presence of a minor. The plaintiffs also raised a vagueness claim under the Fifth Amendment, arguing that terms like “indecent” and “patently offensive” gave no speaker fair notice of what the law prohibited. The district court ultimately rested its holding on overbreadth under the First Amendment, treating vagueness as a supporting consideration rather than an independent ground.

What made the district court’s ruling distinctive was the foundation it built before reaching any legal conclusion. The three-judge panel held evidentiary hearings and made extensive factual findings about the architecture of the internet as it existed in 1996, findings the Supreme Court later adopted almost wholesale into its own opinion. The district court found that the internet was a decentralized network of networks, linked through packet switching: messages were broken into small packets, each routed independently across whatever paths were available, and reassembled at the destination. No single entity controlled the network, and no central choke point existed through which all communications could be filtered or supervised. The court found that the internet’s reach was global, that a message posted from one computer could be received anywhere in the world, and that content once published could be replicated instantly and indefinitely.

The panel documented the variety of communication the network supported. Electronic mail allowed one-to-one and one-to-many messaging. Newsgroups and mailing lists gathered participants around shared interests, from technical subjects to frank discussions of health and sexuality. Real-time chat rooms let strangers converse instantly. And the World Wide Web, then in explosive growth, allowed anyone with a computer to publish documents retrievable by anyone else on the network. The court found that the internet had grown with extraordinary speed, that tens of millions of users had come online, and that the medium supported an unprecedented diversity of speakers, from major publishers to individual citizens typing on home computers. It also found that most speakers on the network were noncommercial: individuals, nonprofit organizations, and small publishers who could not afford elaborate screening systems.

One set of findings proved decisive. Congress had written affirmative defenses into the statute: a speaker could avoid liability by requiring age verification, such as a verified credit card or an adult identification code, before granting access to questionable material. The district court found that these defenses were illusory for nearly everyone they were supposed to protect. Credit card verification was used by some commercial providers of sexually explicit material, but the court found it was not economically feasible for most noncommercial speakers, who gave their content away and had no billing relationship with their readers. Adult identification codes existed, but obtaining and administering them imposed costs and burdens far beyond the reach of the typical website operator. The findings meant that the statute’s supposed safety valve protected almost no one, and that the criminal prohibition therefore operated as a near-total ban on indecent speech by ordinary speakers.

Those findings mattered because the government was asking the courts to treat the internet like broadcast television. The factual record gave the courts a basis for refusing. A broadcast licensee operated on a scarce public resource, the electromagnetic spectrum, under a government license, and pushed programming into homes where children might encounter it without any deliberate act. The internet, as the district court documented it, inverted every element of that picture. There was no scarcity: anyone with a computer and a modem could publish. There was no licensee exercising editorial gatekeeping over the whole medium. And access was not passive: a user had to take affirmative steps, typing an address or following a link, to retrieve particular content. The panel’s findings established that the architecture of the medium placed the burden of access on the receiver, not the sender, and that this structural fact distinguished the internet from any medium the Court had previously regulated.

On the strength of that record, the three-judge panel entered a preliminary injunction in June 1996 against enforcement of sections 223(a)(1)(B) and 223(d), except as applied to obscenity and child pornography. The scope of the injunction was carefully drawn. It restrained the government from prosecuting speakers for transmitting or displaying indecent or patently offensive material, but it expressly preserved the government’s right to investigate and prosecute obscenity and child pornography under the same statute. The panel found that the plaintiffs were likely to succeed on the merits of their First Amendment claim and that enforcement of the provisions would cause irreparable injury to protected speech. The government, invoking section 561, appealed directly to the Supreme Court, bypassing the court of appeals entirely. The Justices noted probable jurisdiction, heard argument on March 19, 1997, with Deputy Solicitor General Seth Waxman defending the statute and Bruce Ennis arguing for the appellees, and handed down their decision on June 26, 1997, in Reno v. American Civil Liberties Union, 521 U.S. 844.

The government’s case at the Supreme Court rested on three precedents, and Justice Stevens’s opinion for the Court dismantled each in turn after what the opinion called a close look.

The first was Ginsberg v. New York, 390 U.S. 629 (1968), in which the Court had upheld a state law barring the sale of sexually explicit material to minors under a “variable obscenity” standard that defined obscenity with reference to its appeal to children. The government argued that the CDA did the same thing: it merely adjusted the definition of prohibited material to protect minors. The Court rejected the analogy because the statutes operated differently in a decisive respect. The New York law in Ginsberg restricted only sales to minors; it left adults free to purchase the same material, and it required knowledge that the buyer was a minor. The CDA contained no such limitation. Its prohibitions applied to communications that adults had a constitutional right to send and receive, and the risk that a minor might access the material was enough to make the speaker a criminal. Ginsberg protected children without confining adults to reading only what was fit for children; the CDA did the opposite.

The second precedent was FCC v. Pacifica Foundation, 438 U.S. 726 (1978), which had upheld the Federal Communications Commission’s authority to sanction a broadcaster for airing George Carlin’s “seven dirty words” monologue in the afternoon. Pacifica rested on two features of broadcast: the scarcity of the spectrum, which justified licensing and public-interest regulation, and the pervasiveness of radio, which the Court described as uniquely intrusive, entering the home uninvited and confronting listeners, including children, without warning. Neither feature applied to the internet. The spectrum rationale had no purchase on a medium with no spectrum and no licensing regime; the internet’s capacity for speakers was effectively unlimited, and the government had never allocated access to it. The invasiveness rationale collapsed against the district court’s factual findings: internet users did not encounter content passively. They sought it out through affirmative acts, typing addresses, selecting links, and choosing which newsgroups or sites to visit. A child who encountered indecent material online had, in the Court’s analysis, navigated to it through a series of deliberate steps, a pattern with no parallel in the radio broadcast that had simply played in a car. The Court concluded that the internet was far less like broadcast radio than like a vast library, and that Pacifica’s reduced protection for broadcast indecency could not be transplanted to a medium whose architecture was the opposite of broadcasting.

The third precedent was Renton v. Playtime Theatres, Inc., 475 U.S. 41 (1986), which had upheld a zoning ordinance restricting the location of adult theaters. The government argued that the CDA was analogous to a zoning law: it did not ban indecent speech but merely confined it to areas where minors would not encounter it. The Court rejected this argument because the Renton ordinance was treated as a content-neutral regulation aimed at the secondary effects of adult theaters, such as crime and declining property values, and because it left ample alternative channels of communication for the restricted speech. The CDA was neither content-neutral nor limited in its alternative channels. It targeted speech precisely because of its sexual or excretory content, which made it a content-based restriction subject to the most exacting scrutiny. And it left speakers no practical alternative: the statute’s affirmative defenses, such as requiring credit card verification or age screening, were, as the district court had found, unavailable or economically infeasible for the vast majority of noncommercial speakers. A zoning law that closed every theater in town was not a zoning law at all. The government could not recharacterize a criminal ban on disfavored content as a mere regulation of place and manner, and the Court refused to let the zoning label do that work.

The government offered one further argument, independent of the three precedents: that its significant interest in fostering the internet’s growth supplied its own justification for the statute. The theory was that the unregulated availability of indecent and patently offensive material was driving citizens, and especially parents, away from the medium, and that the CDA would therefore encourage broader participation. The Court found the argument singularly unpersuasive. The record before the district court showed the opposite of the government’s premise: the internet’s growth had been and continued to be phenomenal, with no evidence that indecent material was deterring users. The Court stated its presumption plainly: as a matter of constitutional tradition, in the absence of evidence to the contrary, governmental regulation of the content of speech was more likely to interfere with the free exchange of ideas than to encourage it, and the interest in encouraging freedom of expression in a democratic society outweighed any theoretical but unproven benefit of censorship. The doctrine of overbreadth permits a court to strike down a law on its face when the law, though it may have some legitimate applications, sweeps in a substantial amount of constitutionally protected speech. Justice Stevens held that the CDA’s indecency and patently-offensive provisions were overbroad because they burdened adult speech far beyond what the protection of minors required. The statute criminalized communications that adults had every right to exchange, and it did so without regard to whether the speaker knew a minor was among the audience. The chilling effect was not speculative. A speaker facing criminal penalties, unable to know with certainty whether a minor might access the material and unable to afford the verification technology the statute offered as a defense, would simply refrain from speaking. The Court noted that the CDA would suppress a large amount of speech that adults had a constitutional right to receive and address, from frank discussions of birth control and safe sex to the online edition of a major newspaper.

Stevens illustrated the statute’s defect with a comparison that captured the constitutional problem in a single image: the CDA, as written, would reduce the adult population to reading only what was fit for a children’s reading room. The government could not pursue its concededly compelling interest in protecting minors by means that burned down the library for everyone else. The interest was real, the Court acknowledged; shielding children from sexually explicit material was among the most legitimate ends a legislature could pursue. But the means had to be narrowly tailored, and the CDA was not. Less restrictive alternatives existed, and the district court had documented them. Filtering and blocking software, installed by parents on the home computer, could screen out categories of content the parents deemed inappropriate, and such software was becoming increasingly available and effective. A parent who wanted to shield a child could do so without asking the government to criminalize the speech of strangers. The existence of those alternatives, combined with the statute’s sweeping criminal penalties, made the overbreadth conclusion inescapable. The government had chosen the bluntest instrument, a criminal ban on the speaker, when narrower instruments aimed at the receiving end were at hand.

The opinion’s holding on the level of protection was the sentence that defined the decision’s place in First Amendment history. The Court held that speech on the internet was entitled to the full First Amendment protection accorded to print, not the reduced protection applied to broadcast. Quoting the district court’s characterization, the Court described the internet as “the most participatory form of mass speech yet developed,” entitled to “the highest protection from governmental intrusion.” That holding established the constitutional baseline for every subsequent dispute about online speech. The internet was not a suspect medium requiring special supervision, nor a scarce resource requiring government allocation. It was a forum for expression as protected as the printing press, and regulations of its content would face the same strict scrutiny that protected newspapers, books, and pamphlets. The medium-specific approach the Court had applied in earlier cases, examining the actual characteristics of each communications technology, led here to maximum protection rather than minimum. The decision became the foundation of the Supreme Court’s internet-law decisions, the first case in which the Court confronted the network directly and set the terms on which it would be judged.

The vote was 7 to 2. Justice Stevens wrote the opinion of the Court, joined by Justices Scalia, Kennedy, Souter, Thomas, Ginsburg, and Breyer. Justice O’Connor, joined by Chief Justice Rehnquist, concurred in the judgment in part and dissented in part. The partial concurrence accepted the core of the majority’s holding: it agreed that the provisions, as applied to displays of indecent material generally available on the internet, were unconstitutional, because the statute failed to confine itself to communications directed at minors. O’Connor’s disagreement concerned the transmission provision, section 223(a)(1)(B), insofar as it criminalized knowingly sending indecent material to a specific person the sender knew to be under eighteen. She viewed the CDA as, at its heart, an attempt to create “adult zones” on the internet, segregating indecent material away from minors while leaving adults free, and she argued that precedent supported such zoning when properly drawn. The provisions strayed from that blueprint, in her view, because they reached too far beyond communications actually directed at children.

Her reasoning turned on the statute’s scienter requirement. Section 223(a)(1)(B) punished only a speaker who knowingly transmitted an indecent communication to a person the speaker knew to be under eighteen. That knowledge requirement, O’Connor argued, made the transmission provision closely analogous to the New York statute upheld in Ginsberg, which had barred sales only to buyers the seller knew were minors. A speaker who addressed an adult, or who did not know the recipient’s age, faced no liability under her reading; only the deliberate transmission of indecent material to a known child was criminal. The display provision of section 223(d)(1)(B), by contrast, criminalized making patently offensive material available to minors generally, without any requirement that the speaker know or target a particular child, and it was this provision that carried the statute’s constitutional defect. On her view, the Court should have sustained the knowing-transmission prohibition as a valid exercise of the zoning power while striking the broader display ban. Her opinion thus accepted the invalidation of the broad display provisions while arguing that a narrower prohibition on transmissions knowingly directed at known minors could survive. The majority did not adopt that narrowing construction, holding that the statute as written could not be salvaged by judicial rewriting.

The judgment itself was precise about what fell and what survived. The Court affirmed the district court’s injunction against enforcement of section 223(a)(1)(B) insofar as it related to “indecent” communications, severing the words “or indecent” from the provision while leaving the prohibition on “obscene” communications intact. The government had invoked the Act’s severability clause, section 608, and asked the Court to construe doubtful terms narrowly rather than strike them down. The Court accepted that invitation in only one respect. Because obscene speech could be banned entirely, and because the word “obscene” in section 223(a) stood on its own textual footing apart from “indecent,” the two terms could be separated, and the obscenity prohibition survived on its own. The injunction against enforcement of sections 223(d)(1) and (2) was unqualified, because those provisions contained no separate reference to obscenity or child pornography and offered no textual seam along which a court could cut. The Court refused to rewrite those provisions from the bench, holding that fashioning new limitations to save the statute would have amounted to legislating. The Court expressly preserved the government’s right to investigate and prosecute obscenity and child pornography, noting that obscene speech could be banned entirely under Miller v. California and that the child pornography prohibitions were untouched by the holding. What the decision destroyed was the middle category Congress had tried to create: a criminal prohibition on indecent but non-obscene speech, applied to a medium used by adults and minors alike, with penalties that fell on the speaker regardless of intent toward any particular child.

The Court reached this result without deciding the Fifth Amendment vagueness claim. The opinion discussed the vagueness of terms like “indecent” and “patently offensive” because that uncertainty amplified the overbreadth problem: a speaker who could not tell what the statute prohibited would steer even further clear of the boundary, suppressing more protected speech. But the judgment was affirmed on First Amendment overbreadth alone, and the Court expressly declined to reach the vagueness issue as an independent ground. That restraint left the holding clean. The CDA’s indecency provisions fell not because their language was unclear, though it was, but because even a clear version of the same prohibition would have criminalized too much protected speech. Congress could protect children online, the Court held, but it could not do so by making criminals of adults who spoke to other adults in language the First Amendment protected.

What Survived the Decision

The Supreme Court decided Reno v. ACLU, 521 U.S. 844, on June 26, 1997, sixteen months after the Telecommunications Act of 1996 became law. The Court struck down the Communications Decency Act’s criminal prohibitions on the knowing transmission of “indecent” messages to minors and on the display of “patently offensive” sexual or excretory material where minors could see it, holding that both provisions abridged the freedom of speech. The decision left the obscenity and child pornography prohibitions untouched, severing the word “indecent” from the statute and leaving the word “obscene” standing. What remained on the books after that severance was a statute its authors would not have recognized: the criminal crusade against indecency had failed, and the provision that endured as the Act’s working core was the one that protected online speech rather than punishing it.

The surgical character of the decision mattered as much as its outcome. Because the Court severed the unconstitutional applications rather than invalidating the title wholesale, the conference’s gamble, that it could enact contradictory policies and let each stand or fall on its own, paid off in a form the conferees never intended. The statute survived its own amputation, and the surviving parts bore almost no resemblance to the political bargain that had produced them.

The sections that follow take inventory of that residue, explain why the surviving provision captured the statute’s name, rebut the story that the survivor was smuggled into law, state the claim this history supports, and draw the series’ thesis through the whole sequence.

The Residue

The provisions that fell were the heart of Title V as Congress had debated it. Section 223(a)(1)(B) of Title 47, as amended by the Act, made it a crime punishable by up to two years in prison to transmit obscene or indecent communications to a person known to be under eighteen, and Section 223(d) criminalized the knowing display of patently offensive material depicting sexual or excretory activities or organs in a manner accessible to minors. The Court held the indecency application of the first provision and the entirety of the second unconstitutional, on the ground that they burdened protected communication between adults in the name of shielding children. The prohibitions that had supplied the Act’s political energy, the reason its sponsors had fought for it and its opponents had mobilized against it, ceased to be enforceable law.

What survived fell into three categories, and each one deserves precise description because the residue is routinely misremembered as a single immunity clause. The first category was 47 U.S.C. 230, the platform provision, which the conference report had carried as Section 509 under the heading “Online Family Empowerment.” Its operative paragraph directs that no provider or user of an interactive computer service be treated as the publisher or speaker of information supplied by another content provider, which means that liability for user-generated material does not attach to the intermediary merely for carrying it. Its second paragraph shields providers and users from civil liability for good-faith voluntary efforts to block or screen material they consider objectionable. The two paragraphs work as a pair: the first removes the penalty for carrying others’ speech, and the second removes the penalty for filtering it. Together they answered the precise dilemma that had motivated the House’s alternative to the Senate’s criminal approach, namely that a provider which screened its service for offensive material risked being treated as a publisher of everything it carried, while a provider which screened nothing risked being condemned for indifference. The conference report’s own heading for the provision, “Online Family Empowerment,” captured the theory behind it: families, rather than federal prosecutors, would decide what entered the home, and the law’s job was to clear away the legal obstacles, publisher liability and the threat of lawsuits over screening decisions, that stood in the way of private filtering. The shield was thus conceived not as an afterthought to the criminal prohibitions but as their rival theory of how to protect children online.

The second category was the obscenity law. The Court’s judgment expressly preserved the government’s power to investigate and prosecute obscenity and child pornography under Section 223, and the Department of Justice treated those prohibitions as fully enforceable after the decision. The severance was surgical: the term “indecent” was cut away from Section 223(a)(1)(B), and the “obscene” prohibition in the same paragraph continued to operate. Whatever else the decision destroyed, it left the preexisting criminal law of obscenity exactly where it had been.

The third category was the set of affirmative defenses in Section 223(e)(5). A defendant could avoid conviction by showing good-faith, reasonable, effective, and appropriate actions to restrict minors’ access to the prohibited communications, or by showing that access had been restricted through a verified credit card, a debit account, an adult access code, or an adult personal identification number. These defenses had been drafted as safety valves for the criminal provisions, a way to soften the prohibitions’ reach by rewarding private screening and age verification. When the prohibitions they were built to soften fell, the defenses remained in the Code as defenses to the surviving prohibitions, a set of tools without the machine they had been designed to regulate. That mismatch illustrates the strange mechanics of severance. Defenses are drafted against specific offenses, and when the offenses fall, the defenses persist as free-floating permissions applicable to whatever prohibitions remain in the section. The Section 223(e)(5) defenses therefore outlived the criminal scheme they were meant to temper, a reminder that statutes are machines with interchangeable parts and that removing one component does not automatically remove the components built around it.

The residue, then, was a decency statute with no criminal indecency law in it. It contained a liability shield for intermediaries, a ban on obscenity and child pornography, and a pair of defenses built for offenses that no longer existed in the form Congress had written them. The title still promised decency. The enforceable text delivered immunity, an obscenity ban, and safe harbors.

Why the Survivor Got the Name

A statute enacted to restrict online speech came to be known almost exclusively for the provision that protects it. The mechanism behind that reversal is straightforward, and it reveals something general about how statutes acquire their popular names. Those names do not come from the titles Congress selects. They come from the provisions that continue to generate cases, disputes, and arguments after the political moment of enactment has passed. The indecency prohibitions produced exactly one landmark decision and then ceased to be enforceable law; there was nothing left to litigate, and so there was nothing left to name. Section 230 produced a continuing stream of litigation over who bears responsibility for words written by someone else, and each new dispute reattached the statute’s public identity to the provision that decided it. The name followed the live wire.

The irony has a second layer that makes the first one sharper. The provision that defined the Act’s public identity was not the work of the legislators who gave the Act its name. The Senate’s authors pursued criminal penalties aimed at the speakers and displayers of indecent material; the immunity provision arrived as the House’s alternative, offered precisely because its sponsors believed that criminal penalties were the wrong instrument for the problem. When the Court removed the Senate’s half of the compromise, the House’s alternative was all that remained with operative force, and the public name of the statute migrated to it. The legislators who wanted a decency law got a statute remembered for immunity; the legislators who wanted immunity got their provision remembered under a title about decency.

A title is a claim about purpose; a popular name is a record of which provisions survived contact with the courts. The Communications Decency Act claimed decency and delivered immunity, and the name followed the delivery rather than the claim. That pattern repeats often enough to count as a rule of legislative afterlife: the provisions that keep working keep the name, and the provisions that die in court take their purposes with them.

The reversal also exposes a habit of legislative branding. Titles like the Communications Decency Act are chosen for their political resonance at the moment of enactment, when the goal is to signal concern and claim credit. Popular names are assigned later, by lawyers citing the provision that decides their cases and by commentators reaching for the section number that explains an outcome. The first naming is an act of politics; the second is an act of use. When the two diverge, as they did here, the divergence itself becomes part of the history, because it shows which of the statute’s ambitions proved durable and which proved rhetorical.

The Story That It Was Slipped In

One persistent account holds that Section 230 was slipped into the conference report late and unnoticed, a minor provision that became consequential by accident and that no one in Congress understood at the time. The record does not support it, and the brief for this article requires the rebuttal to be stated plainly. In August 1995, the House took up the Cox-Wyden amendment, the Internet Freedom and Family Empowerment Act, as a deliberate and named alternative to the Senate’s criminal approach to the same problem of offensive material online. The amendment was debated on the floor, offered as a substitute policy rather than a technical correction, and passed on a recorded vote of 420 to 4.

Its sponsor, Representative Christopher Cox of California, described its purpose explicitly on the record. Rather than criminalizing users or conscripting service providers into the role of censors, the amendment would empower parents, users, and providers with filtering tools and remove the threat of publisher liability so that providers could screen offensive material in good faith without fear of being treated as its speaker. The official title of the measure stated the same purpose in legislative language, to encourage and protect private-sector initiatives that improve user control over computer information services, and the committee summary described it as shielding providers from treatment as publishers and from liability for good-faith blocking. None of this was hidden. The purpose was announced, the mechanism was described, and the vote was recorded.

The conference committee then carried the provision into the final bill as Section 509, titled “Online Family Empowerment,” alongside the Senate’s criminal provisions, and the completed Telecommunications Act passed both chambers and was signed into law. A provision adopted 420 to 4 after floor debate, with its purpose stated on the record by its sponsor and its function described in the committee summary, was not smuggled into law in the dark. It was the House’s chosen answer to the problem the Senate had answered with criminal penalties, and the conference kept both answers in the same title. That version of events belongs with the other common claims about this history that collapse once the legislative record is consulted.

The slipped-in story likely persists because the provision’s quiet, technical character contrasted so sharply with the noisy moral drama of the indecency fight. The criminal prohibitions dominated the debate, the headlines, and the litigation, while the immunity provision did its work in the background, attracting little attention until the courts began applying it. Inattention after enactment is not the same as inattention during enactment, and the record of August 1995 shows a House that knew exactly what it was voting for. The conference’s subsequent decision to retain the provision confirms the point. Conference committees exist to reconcile differences between the chambers, and they discard provisions routinely; the Cox-Wyden language survived that winnowing and emerged as Section 509 of the enrolled bill. A provision that passes the House 420 to 4, survives conference, and enters the statute under its own descriptive heading has cleared every procedural hurdle the legislative process offers. To call that sequence an accident is to misunderstand what the legislative record is for.

The Conference Kept Both

The namable claim of this history is direct: the conference kept both.

Congress did not choose between criminalizing online indecency and immunizing the intermediaries that carried it. The House voted overwhelmingly for the immunity approach, the Senate pressed the criminal approach, and the conference committee reconciled the two chambers’ bills by retaining both, stacking the Cox-Wyden shield on top of the Exon prohibitions inside a single title. The selection between the two approaches was therefore not made in the conference room in the winter of 1995 to 1996. It was made in the courtroom on June 26, 1997, when the Supreme Court struck the criminal half of the compromise and left the immunity half standing. The most important provision in American internet law survived not because Congress preferred it to the alternative but because the alternative was unconstitutional. The conference declined to decide, and judicial review decided in its place.

The evidence for the claim sits in the structure of the enacted text itself. A Congress that had chosen the immunity approach would have discarded the criminal prohibitions; a Congress that had chosen the criminal approach would have discarded the immunity. The conference did neither. It enacted a title that simultaneously threatened speakers with prison for indecency and shielded intermediaries from liability for carrying speech, leaving the contradiction for someone else to resolve. The courts were that someone else.

The claim also reframes the usual moral of the story. The familiar telling presents Section 230 as a fortunate accident, a good provision that survived the wreck of a bad statute. The record supports a colder version: Congress enacted a restrictive statute and a protective provision together, refused to choose between them, and was rescued from its own indecision by a court. The survival of the immunity was not luck but procedure, the predictable result of stacking two policies in one title and submitting both to judicial review.

Procedure Determining Text

This is the series’ thesis in its purest form: procedure determines text. The conference committee is a procedural device for reconciling the two chambers’ bills, and in this case that device produced a statute containing two incompatible policies toward a single problem, punish the speaker or protect the intermediary. Rather than resolving the incompatibility, the conference preserved it. The severability built into the legislative design then allowed each half of the compromise to live or die on its own, so that the statute could survive the removal of either policy without collapsing.

Judicial review completed the selection the conference had declined to make. The Court’s invalidation of the indecency and patently-offensive provisions operated as a second, post-enactment round of reconciliation, striking one of the two policies the conference had stacked together and leaving the other as the enforceable law. The final shape of the statute was therefore set by two procedural acts rather than by one substantive legislative judgment: the conference’s decision to keep both approaches, and the Court’s decision to eliminate one of them. The floor debates supplied the arguments, the hearings supplied the anecdotes, and the procedures supplied the outcome.

That sequence is worth stating in full because it reverses the usual civics account of how a law gets its content. In the usual account, legislators debate competing policies, choose among them, and write the winner into the statute. Here the legislators debated competing policies, refused to choose among them, wrote both into the statute, and left the choice to a court applying the First Amendment sixteen months later. The law that governed online speech in the years that followed was written first by a committee compromise and finished by a judicial opinion, and neither act alone explains the text that survived. The conference explains why both policies were present; the Court explains why only one remained. Procedure, twice over, determined the text.

Notice what drops out of this account. The substantive merits of criminalizing indecency versus immunizing intermediaries, the arguments that filled the floor debates and the committee reports, did not determine the outcome. Both sets of arguments were heard, both were written into law, and the one that prevailed was the one the Constitution permitted. Procedure determined which arguments received a second hearing, in court, and the Constitution determined which survived it. The series’ thesis does not claim that substance is irrelevant; it claims that substance reaches the statute book only through procedural channels, and that those channels can select, combine, and discard substantive commitments in ways the debaters never anticipated.

Closing

Taken together, the elements give the story its shape. The residue of the Communications Decency Act was a liability shield, an obscenity ban, and a pair of affirmative defenses, with the criminal indecency law that had defined the legislative fight excised from the Code. The name of the statute migrated to the shield, because popular names follow the provisions that keep generating cases rather than the titles Congress bestows. The shield itself was no accident of conference drafting: the House adopted it 420 to 4 as a named alternative to criminal penalties, its sponsor stated its purpose on the record, and the conference deliberately kept it alongside the Senate’s prohibitions. The claim that organizes these facts is that the conference kept both, criminalization and immunity together, and left the choice between them to the courts. The thesis that organizes the claim is that procedure determined the text, first through a conference compromise that refused to decide and then through judicial review that decided in its place.

Legislative history is often written as a contest of ideas, but the Communications Decency Act rewards a different reading, one that follows the procedures that assembled the statute and the procedures that disassembled it. The ideas were all present in the debate; the procedures determined which of them became law. That reading also explains why the statute’s afterlife surprised its authors. The Senate’s sponsors expected to be remembered for protecting children from indecency; the House’s sponsors expected their provision to operate quietly in the background of a criminal statute. Neither expectation survived the procedures that followed enactment. The statute that emerged from severance belonged to neither camp’s plan, which is precisely what makes it an artifact worth studying on its own terms. For readers who want to trace that sequence through the statute’s structure, the conference mechanics, and the Court’s reasoning in their own notes, the VaultBook legislation study notebook offers a structured place to work through it.

Frequently Asked Questions

Q: How did the Communications Decency Act become law?

Congress enacted it as Title V of the Telecommunications Act of 1996. The Senate had adopted the Exon-Coats criminal indecency amendment to S. 652 on June 14, 1995, by 84 to 16; the House had adopted the Cox-Wyden immunity amendment to H.R. 1555 on August 4, 1995, by 420 to 4. A conference committee reconciling the two chambers’ telecommunications bills met in the fall of 1995 and kept both halves in a single title: the Senate’s criminal prohibitions rewriting 47 U.S.C. 223, and the House’s immunity added as a new 47 U.S.C. 230. The conference report, filed January 31, 1996, was adopted by the House 414 to 16 and the Senate 91 to 5 on February 1, 1996. President Bill Clinton signed the bill on February 8, 1996, as Public Law 104-104. Because the conference text was a single unamendable package, no member could accept one half of Title V and reject the other.

Q: Who wrote the Communications Decency Act indecency provisions?

Senator James Exon, a Democrat from Nebraska, was the chief sponsor, and Senator Dan Coats, a Republican from Indiana, co-sponsored. They offered the provisions as S. Amdt. 1268, the Exon-Coats amendment, to S. 652, the Senate telecommunications bill, in June 1995. Exon had pushed similar legislation in the preceding Congress without success; in 1995 the political weather changed, driven by the Time “cyberporn” cover story and by Exon’s own blue folder of downloaded images, which he showed colleagues on the Senate floor. The amendment copied the indecency language of the federal dial-a-porn statute, which the Supreme Court had upheld in Sable Communications v. FCC in 1989, on the theory that the Court’s approval for telephone pornography would extend to the internet. The Senate adopted the amendment on June 14, 1995, by 84 to 16, and it traveled with S. 652 into conference, where it became the criminal core of Title V.

Q: Was section 230 an alternative to the Communications Decency Act?

Section 230 was an alternative to the Act’s indecency provisions, not to the Act itself, and it ended up inside the Act. Representatives Christopher Cox, a Republican from California, and Ron Wyden, a Democrat from Oregon, wrote H.R. 1978, the Internet Freedom and Family Empowerment Act, introduced June 30, 1995, as the House’s answer to the Senate’s criminal approach. The bill declared that no provider or user of an interactive computer service would be treated as the publisher of third-party content, added Good Samaritan protection for voluntary screening, and barred the FCC from regulating online content. Offered as H.Amdt.744 to the House telecommunications bill, it passed 420 to 4 on August 4, 1995. The conference kept the immunity language alongside the Senate’s criminal provisions, so the alternative and the original became co-tenants of the same title, and the alternative outlived the original.

Q: What did Reno v. ACLU do to the Communications Decency Act?

Reno v. American Civil Liberties Union, 521 U.S. 844, decided June 26, 1997, struck down the Act’s criminal indecency and patently-offensive provisions as violations of the First Amendment. Justice John Paul Stevens, writing for seven Justices, held that the internet deserved the full First Amendment protection given to print, rejected the government’s analogies to broadcast regulation in Pacifica and to adult-theater zoning in Renton, and ruled the provisions substantially overbroad because they criminalized protected communication between adults whenever a minor might encounter it. The Court severed the word “indecent” from 47 U.S.C. 223(a)(1)(B) while preserving the obscenity ban, and struck sections 223(d)(1) and (2) entirely. Justice O’Connor, joined by Chief Justice Rehnquist, concurred in part and dissented in part. Section 230 survived untouched. Sixteen months after the Act became law, its criminal half was gone and its immunity half stood.

Q: How did both approaches end up in the Communications Decency Act?

They arrived through the House-Senate conference that reconciled the two chambers’ telecommunications bills. The Senate’s bill, S. 652, carried the Exon-Coats criminal indecency amendment, adopted 84 to 16 on June 14, 1995. The House’s bill, H.R. 1555, carried the Cox-Wyden immunity amendment, adopted 420 to 4 on August 4, 1995. The conference committee, meeting in October and December 1995, had no procedural duty to harmonize the chambers’ philosophies, only to produce one text each chamber could vote up or down. The House conferees accepted the Senate’s criminal language in December, and the Cox-Wyden language survived because no senator had a strong reason to strike it. A savings clause in the immunity section said it did not impair enforcement of the new indecency provisions, acknowledging the contradiction without resolving it. Both chambers adopted the package on February 1, 1996, and the two philosophies entered the statute together.

Q: What survived after the Communications Decency Act was struck down?

Three things survived. First, 47 U.S.C. 230, the Cox-Wyden immunity provision: no provider or user of an interactive computer service is treated as the publisher of third-party content, and good-faith voluntary screening receives liability protection; it became the Act’s working core. Second, the obscenity and child pornography prohibitions: the Court severed “indecent” from section 223(a)(1)(B) but left “obscene” standing, and it expressly preserved the government’s power to investigate and prosecute obscenity and child pornography. Third, the affirmative defenses of section 223(e)(5), which let defendants show good-faith, reasonable, effective, and appropriate steps to restrict minors’ access, or age verification through credit cards or adult codes; drafted as safety valves for the criminal scheme, they outlived the offenses they were built to temper. The title still promised decency; the enforceable text delivered immunity, an obscenity ban, and safe harbors.

Q: What was the Exon amendment in the Communications Decency Act?

The Exon amendment, formally the Exon-Coats amendment (S. Amdt. 1268), was the Senate’s criminal indecency provision, offered by Senators James Exon and Dan Coats to S. 652 in June 1995. It made it a federal crime, punishable by fines up to $100,000 and up to two years in prison, to transmit obscene or indecent material to a person known to be under eighteen over an interactive computer service, and to display patently offensive material in a manner available to minors. Its scienter requirement, “knowingly,” and its community-standards definition of indecency were borrowed word for word from the dial-a-porn statute the Supreme Court had upheld in Sable v. FCC in 1989, on the theory that an approved rule for telephone pornography would hold for the internet. The Senate adopted it on June 14, 1995, by 84 to 16. In conference it became the criminal core of Title V, the portion of the Act that Reno v. ACLU later struck down.

Q: How did the House vote on the Communications Decency Act alternative?

On August 4, 1995, the House adopted the Cox-Wyden amendment, H.Amdt.744, to H.R. 1555 by a recorded vote of 420 to 4. The amendment carried the full substance of H.R. 1978, the Internet Freedom and Family Empowerment Act: the publisher-immunity clause, the Good Samaritan screening shield, the declaration of national policy for internet development and the competitive free market, and the bar on FCC content regulation. The same day, the House passed H.R. 1555 itself by 305 to 117, so the amendment drew near-unanimous support in a chamber that divided over the broader bill. The 420-to-4 margin was a declaration of the House’s theory: user empowerment and private screening, not criminal prohibition, was the chamber’s answer to objectionable material online. That recorded vote gave the House conferees the leverage to keep the immunity language in the final title alongside the Senate’s criminal provisions.

Q: Why was the Senate’s vote on the Exon amendment so lopsided?

The 84-to-16 margin reflected political pressure more than constitutional consensus. Exon’s blue folder had turned the vote into a nationally televised referendum on child protection: any senator voting no risked being portrayed in the next campaign as having voted for pornography. The C-SPAN cameras made that framing inescapable, because the debate was carried live and the images Exon had shown colleagues were fresh in every senator’s mind. The bipartisan sponsorship, a Nebraska Democrat and an Indiana Republican, let members treat the measure as a shared moral commitment rather than a partisan one. And the sponsors’ constitutional argument sounded reassuring: the indecency language had been upheld for dial-a-porn in Sable, so the Court would approve it for the internet. Constitutional objections, which the Court vindicated two years later, sounded abstract against those pressures and lost decisively on June 14, 1995.

Q: What alternative did Senator Patrick Leahy offer to the Exon amendment?

Senator Patrick Leahy, a Democrat from Vermont, proposed that Congress study the problem before criminalizing it. His alternative would have directed the Justice Department to examine how existing obscenity statutes applied to computer networks and to report back on the least restrictive means of protecting children from indecent material online. Leahy’s argument was doctrinal: indecent speech, unlike obscenity, was constitutionally protected, and the Senate was about to ban it across an entire medium without knowing whether narrower tools existed. He waved petitions signed by tens of thousands of opponents and warned that Congress was about to break an internet that finally worked. The proposal received barely any discussion; the Senate’s attention was fixed on the Exon amendment, and the study-first approach could not compete with the urgency the blue folder had manufactured. It was scrapped when the amendment was adopted.

Q: What was Exon’s “blue folder” during the Senate debate?

The blue folder was Exon’s instrument for turning abstract policy into political momentum. He had asked a friend to download raw images from the internet, had them printed, and stuffed them into a folder that colleagues and reporters began calling his blue book. He invited senators to his desk on the Senate floor, and into the cloakrooms, to view the images, asking whether they had realized such material was a click away. The senators who looked were shocked; Coats later described material including the exploitation of young children, bestiality, and brutal sexual explicitness. The folder fused the child-protection question with the speech-regulation question, so that a vote against the amendment read, on camera, as a vote against protecting children. Commerce Committee chairman Larry Pressler had planned to table the amendment quietly; after the blue folder, a tabling motion would have read as protecting the images, and it became politically unusable.

Q: How did the Stratton Oakmont v. Prodigy decision lead to the Cox-Wyden bill?

On May 24, 1995, the New York Supreme Court ruled in Stratton Oakmont, Inc. v. Prodigy Services Co. that Prodigy was the publisher of defamatory statements posted by users on its bulletin boards, precisely because Prodigy posted content guidelines, enforced them through board leaders, and used screening software to remove offensive language. An earlier federal ruling, Cubby v. CompuServe in 1991, had treated a service that did not screen its content as a mere distributor with far lower liability. Together the decisions created a moderator’s dilemma: screening made a service liable, while ignoring objectionable material kept it safe. Cox and Wyden treated that incentive as perverse, since it punished the companies trying to keep their services clean for families. Their bill answered it clause by clause, declaring that no provider would be treated as the publisher of another’s content and adding Good Samaritan protection for good-faith screening, so that filtering could never again be the evidence of liability.

Q: Why did the Cox-Wyden bill bar the FCC from regulating online content?

The bill declared that nothing in it would be construed to authorize the Federal Communications Commission to regulate the content of interactive computer services. The clause closed an opening the sponsors feared: that any federal action touching the internet would become a pretext for extending broadcast-style content regulation to the networks. Critics of the Exon amendment had warned that criminalizing indecency would put computer networks under FCC jurisdiction, importing the broadcast indecency machinery onto a medium the Commission had never been given authority over. The bar fit the bill’s family-empowerment theory: a regulator empowered to police content would substitute bureaucratic judgment for parental judgment. By writing the limitation into the text itself, the sponsors signaled that the family’s control was the endpoint of the policy, not a way station toward government control, and that the new protections were not a license for the broadcast regulator.

Q: How did the Senate’s vote on S. 652 itself differ from the Exon amendment vote?

The two votes are often confused, and the confusion is the source of a persistent error. The Exon-Coats amendment, S. Amdt. 1268, was adopted on June 14, 1995, by 84 to 16. The underlying bill, S. 652, the Telecommunications Competition and Deregulation Act, passed the Senate the next day, June 15, 1995, by 81 to 18. The amendment vote measured support for the criminal indecency provision standing alone; the bill vote measured support for the whole telecommunications package. The one-day gap and the similar margins make the mix-up easy, and published accounts have repeatedly attached the bill’s 81-to-18 tally to the amendment. The distinction matters because the amendment, not the bill vote, is what carried the criminal provisions into the conference that produced Title V of the Telecommunications Act of 1996.

Q: Why did the conference committee keep both approaches instead of choosing one?

Choosing was not required, and inclusion bought votes. A conference committee’s job is to produce one substitute text that each chamber votes up or down; nothing in the procedure compels conferees to harmonize the chambers’ philosophies. The Senate’s conferees were unanimous for criminalization, including Exon and Coats themselves, while the House conferees were split, with most facing a chamber that had accepted the Cox-Wyden immunity 420 to 4. The conferees satisfied the Senate by keeping the criminal provisions and the House by keeping the immunity, separated only by a savings clause that preserved the conflict rather than resolving it. Each side’s provision was the price of the other’s support, and the resulting package was a telecommunications bill first: members who wanted deregulation had to swallow Title V whole, since House Resolution 353 barred any floor amendment to the conference report.

Q: What was the §561 expedited-review provision?

Section 561 of the Act created a special judicial review procedure for challenges to the indecency provisions: an expedited hearing before a three-judge district court panel, with direct appeal to the Supreme Court, bypassing the court of appeals and the certiorari process. The purpose was speed, so that the constitutionality of the criminal provisions would be settled quickly rather than after years of ordinary litigation. Its inclusion was itself an admission: Congress writes expedited review into statutes whose authors expect the courts to strike them down. It also reflected the medium’s pace, since a five-year case would have regulated a version of the network that no longer existed by the time the Court ruled. The procedure worked exactly as designed: the three-judge panel in Philadelphia enjoined the provisions in June 1996, the government appealed directly, and the Supreme Court decided Reno v. ACLU on June 26, 1997.

Q: What happened on February 8, 1996, the day the Act was signed?

President Bill Clinton signed the Telecommunications Act of 1996 into law that morning as Public Law 104-104, and the same day the lawsuits the conference had anticipated arrived. The American Civil Liberties Union, the Electronic Frontier Foundation, and eighteen other plaintiffs filed ACLU v. Reno in the Eastern District of Pennsylvania, challenging sections 223(a)(1) and 223(d). A second coalition led by the American Library Association filed ALA v. DOJ the same day. A week later, District Judge Ronald Buckwalter entered a temporary restraining order against enforcement, finding “indecent” too vague to support a criminal prosecution; twenty-seven more plaintiffs filed a second suit, the cases were consolidated before a three-judge panel under section 561, and after an evidentiary hearing the panel entered a preliminary injunction. The statute’s authors had written its judicial reckoning into the same text that created it.

Q: How did Reno v. ACLU treat the Pacifica broadcast precedent?

The Court refused to extend FCC v. Pacifica Foundation, 438 U.S. 726 (1978), to the internet. Pacifica had upheld sanctions against a broadcaster for airing George Carlin’s “seven dirty words” monologue, resting on two features of broadcast: the scarcity of the electromagnetic spectrum, which justified licensing and public-interest regulation, and the pervasiveness of radio, which entered homes uninvited and confronted listeners, including children, without warning. Justice Stevens’s opinion found neither feature present online. The internet had no spectrum, no licensing regime, and effectively unlimited capacity for speakers. Nor was it intrusive: the district court’s findings showed users took affirmative steps, typing addresses and choosing links, to retrieve content, so a child who encountered indecent material had navigated to it deliberately. The Court concluded the internet resembled a vast library rather than broadcast radio, and held that online speech was entitled to the full First Amendment protection accorded to print.

Q: What did Justice O’Connor argue in her “adult zones” partial concurrence?

Justice O’Connor, joined by Chief Justice Rehnquist, concurred in the judgment in part and dissented in part, accepting the core of the majority’s holding while defending a narrower prohibition. She agreed that the display provisions were unconstitutional because they reached far beyond communications directed at minors. But she viewed the statute, at its heart, as an attempt to create “adult zones” on the internet, segregating indecent material from minors while leaving adults free, and she argued the transmission provision, section 223(a)(1)(B), could serve that zoning purpose. That provision punished only a speaker who knowingly sent indecent material to a person the speaker knew to be under eighteen, which she treated as closely analogous to the statute upheld in Ginsberg v. New York. On her reading, a speaker addressing an adult, or unaware of the recipient’s age, faced no liability; only deliberate transmission to a known child was criminal. The majority declined to salvage the statute by such narrowing, holding it could not be rewritten from the bench.

Q: What affirmative defenses did §223(e)(5) give defendants?

Section 223(e)(5) provided two affirmative defenses, Congress’s nod to the technological reality that identifying a recipient’s age was nearly impossible for most speakers. The first covered defendants who took good faith, reasonable, effective, and appropriate actions to restrict access by minors to the prohibited communications. The second covered defendants who restricted access by requiring designated forms of age proof, such as a verified credit card, a debit account, an adult access code, or an adult personal identification number. The three-judge district court found both defenses illusory for nearly everyone they were meant to protect: credit card verification was economically infeasible for noncommercial speakers who gave their content away, and adult codes imposed costs far beyond the reach of typical website operators. When the criminal provisions fell in Reno v. ACLU, the defenses remained in the Code as defenses to the surviving obscenity prohibitions, tools without the machine they had been designed to regulate.