Concept
Overbreadth in Later Cases
Overbreadth
Overbreadth in Later Cases
Overbreadth in the later cases is one test applied to different texts. Under the standard from Virginia v. Hicks, a law regulating speech is facially invalid only if it prohibits a substantial amount of protected speech, judged in relation to its plainly legitimate sweep, and the challenger bears the burden of showing that. A law that falls short of the showing remains enforceable against unprotected conduct, even if some hypothetical applications would reach protected speech. Think of a ratio, with the law's protected applications on top and its legitimate reach below. The earlier sections of this chapter showed the two poles. In Board of Airport Commissioners v. Jews for Jesus, a resolution declared the airport's Central Terminal Area closed to First Amendment activities by any individual or entity.1 It prohibited virtually every expressive activity, so the legitimate side of the ratio was nearly empty and the resolution fell on its face. In Hicks, the trespass policy applied to everyone entering a housing development, so protected speakers were a small remainder against a large legitimate sweep, and the challenge failed.
The cases in this section add no new doctrine. They run the same ratio on texts that exam questions favor, and from one text to the next only two things change: what the court counts, and how hard the counting is. A statute may give a court real applications to tally on each side, as in the animal-cruelty case. An ordinance may use words so open-ended that it reaches speech far beyond the narrow category the state may punish, which raises the question who may supply a limiting reading. A law may regulate a new medium in terms the legislature never defined, so that overbreadth and the vagueness analysis from the previous section apply to the same text.
Four questions organize every case below. What does the law actually prohibit? What is its plainly legitimate sweep? Are the protected applications realistic? And are they numerous enough, in absolute terms and against that sweep, to justify suspending enforcement for everyone?
Counting Applications in Stevens
United States v. Stevens shows the ratio worked out with real numbers. The Court held, 8–1, that a federal statute banning certain depictions of animal cruelty for commercial purposes was substantially overbroad.2 The statute, 18 U.S.C. § 48, reached any depiction in which a living animal is intentionally maimed, mutilated, tortured, wounded, or killed, if that conduct violates federal or state law where the depiction is created, sold, or possessed.3 Robert Stevens ran a business and website that sold videos of pit bulls fighting and attacking other animals. After his indictment he moved to dismiss, arguing that § 48 was facially invalid, and the District Court denied the motion.4 Overbreadth is invoked only when a litigant has engaged in some type of expressive activity, and a seller of videos had.5 The question was therefore not whether Stevens's own videos could be punished but what else the text reached.
Because the standard compares two sets of applications, the Court first had to learn what the statute covered. A law may be invalidated as overbroad if a substantial number of its applications are unconstitutional, judged in relation to its plainly legitimate sweep, and the first step is to construe the challenged statute, since a court cannot tell whether a statute reaches too far without knowing what it covers.6 The step carried unusual weight here. The Government did not defend a broad ban and rested its defense on reading the statute as limited to extreme material, so the statute's constitutionality hinged on how broadly it was construed.7
Two features of the text made the protected side of the ratio large. The definition reached any depiction in which an animal is “wounded” or “killed,” words that in their ordinary meaning require no cruelty, and the Court found little ambiguity in them despite the Government's proposal to read in an element of accompanying cruelty.8 The illegality requirement did little to narrow the statute. Many laws governing animals have nothing to do with cruelty, and the conduct need be illegal only in the single jurisdiction where the depiction is created, sold, or possessed, wherever the wounding or killing took place.9 A depiction of entirely lawful conduct therefore becomes criminal wherever the same conduct is forbidden. All hunting is unlawful in the District of Columbia, other jurisdictions permit or encourage it, and the national market for hunting magazines, television programs, videos, and websites is enormous, with demand exceeding that for crush videos and animal-fighting depictions by several orders of magnitude.10
The Government then tried three ways to shrink that count, and each failed for a reason that carries over to other overbreadth problems. The first was the exceptions clause, which exempts any depiction with serious religious, political, scientific, educational, journalistic, historical, or artistic value.11 The Court said that “serious” should be taken seriously and that excepted speech must also fall within one of the listed categories, which much speech does not, since most hunting videos are not obviously instructional.12 The clause borrowed its wording from the obscenity test in Miller v. California, but Miller excluded works of serious value from the definition of obscenity and did not make serious value a precondition for protecting speech in the first place.13 The defect is structural. Most of what people say to one another lacks serious value in that sense and is still sheltered from regulation, so a clause that spares only valuable speech leaves a large body of presumptively protected expression in the count.14
The second attempt was an assurance. The Executive Branch said that it read the statute to reach only extreme cruelty and that it neither had brought nor would bring a prosecution for anything less.15 The Court responded that the First Amendment protects against the Government and does not leave speakers at the mercy of noblesse oblige, and that it would not uphold an unconstitutional statute merely because the Government promised to use it responsibly.16 The reason follows from the chilling-effect rationale taught earlier. A seller deciding whether to publish a hunting video can read the statute but cannot read a prosecutor's intentions, so the deterrence comes from what the text permits.
The third attempt was the broadest. The Government's primary submission was that depictions of animal cruelty, as a class, are categorically unprotected by the First Amendment, as obscenity is.17 That argument goes straight to the count, because a statute that reaches only unprotected speech cannot be overbroad and the protected side of the ratio would be empty from the start. The Court disagreed and declined to recognize depictions of animal cruelty as a new unprotected category.18
With those rescues set aside, the count was lopsided. The Court observed that the Government did not seriously contest that the presumptively impermissible applications of the statute, properly construed, far outnumber any permissible ones, and that the markets for crush videos and dogfight depictions, however growing and lucrative, are dwarfed by the market for hunting magazines and videos within its text.19 The Court held only that the statute was substantially overbroad. It left undecided whether a statute limited to crush videos or other depictions of extreme animal cruelty would be constitutional.20
Set beside the other cases in this chapter, Stevens sits between the poles. The airport resolution in Jews for Jesus left no legitimate sweep to weigh, because a ban on all First Amendment activities reached virtually every expressive activity. The trespass policy in Hicks had an enormous legitimate sweep that dwarfed its few protected applications, so the challenge failed. The animal-cruelty statute had a real core of applications to crush videos and dogfight depictions, but the protected applications dwarfed that core. One standard produced three results, and the difference lay in how large each side of the ratio turned out to be.
Justice Alito, the lone dissenter, applied the same test and reached the opposite result by reading the statute differently. He stressed that a court looks to a statute's application to real-world conduct, not fanciful hypotheticals, and that the challenger must show substantial overbreadth from the text and actual fact.21 He invoked the canon that a federal court construing a federal statute challenged as overbroad should adopt a limiting construction if one is available. He would have read “animal cruelty” to cover only acts of cruelty as defined by applicable law and treated hunting depictions as having serious scientific, educational, or historical value within the exceptions clause.22 On that reading the statute reached almost none of the depictions the majority counted. A substantial core of crush videos and dogfight videos remained, and no substantial overbreadth.23 He also found nothing in the record suggesting that anyone had created or sold a depiction of food-animal slaughter that would not easily qualify under the exception.24 The majority had found little ambiguity in “wounded” and “killed,” and so no limiting reading available.25 Majority and dissent thus ran the same test. They disagreed about the first step, construction, and that disagreement decided the case.
An exam stem built on this pattern asks whether a statute aimed at a disfavored category of expression is facially valid. The answer starts with the text: what does each operative word cover, and does any requirement, such as illegality, actually limit the statute? It then states the legitimate core the government would defend, sets beside it the everyday protected expression the text also reaches, and tests each rescue the government offers. An exceptions clause measures value, and protection does not depend on value. A promise of narrow enforcement changes nothing the text says. An argument that the whole class is unprotected must itself succeed, and the Court rejected it for depictions of animal cruelty. If the statute is federal and readily susceptible to a limiting reading, the answer should say so, because that was the path by which the dissent's count came out the other way.
Fighting Words Ordinances
Fighting words are a small unprotected core, and a breach-of-the-peace law survives overbreadth review only if it stays inside that core. In Chaplinsky v. New Hampshire the Court defined fighting words as those “which by their very utterance inflict injury or tend to incite an immediate breach of the peace.”27 Later decisions narrowed the category: unprotected fighting words occur only if the speech is directed to a specific person and likely to provoke a violent response.28 Everything outside that definition stays protected. The Court has said that offensive or disagreeable speech is protected even when it takes the form of expressive conduct, and offensive speech, unlike fighting words, cannot be prohibited.29
A narrow category with a blurry edge invites broad drafting. Legislatures write breach-of-the-peace and disorderly-conduct laws around words that are abusive, offensive, or annoying, and those words reach speech that provokes a reaction without meeting the fighting-words definition. Such a law has two sweeps: a legitimate one, the fighting words it covers, and an unconstitutional one, the merely offensive speech it also covers. Because the line between them is hard to draw, the Court has increasingly invalidated these statutes on their face for overbreadth and vagueness, a technique that spares it from defining fighting words with particularity.30 The technique works because it changes the question. A defendant limited to an as-applied argument would have to persuade the Court that the defendant's own words fell on the protected side of an uncertain line, whereas overbreadth asks what the statute, as the state courts have construed it, reaches, and that can be answered without classifying anyone's particular speech.31 These laws also sit between the poles described earlier. Unlike the airport resolution in Jews for Jesus, a breach-of-the-peace statute has a genuine legitimate sweep in the fighting words it covers, so the comparison required by Hicks must be run against that sweep.
Gooding v. Wilson is the model. The defendant was convicted on two counts of using opprobrious words and abusive language under a Georgia statute covering such language used to or of another without provocation.32 Justice Brennan's opinion for the Court held the statute facially overbroad. Unlike the New Hampshire courts in Chaplinsky, the Georgia courts had failed to construe the statute narrowly to limit it to fighting words, so the Court voided it on its face.33 As Georgia applied it, the statute made it a breach of the peace “merely to speak words offensive to some who hear them,” and so it swept too broadly.34 The Court set the defendant's abusive language aside and examined the statute as the state courts had construed it.35 Overbreadth works this way because the analysis does not reach whether the challenger's speech is protected. It strikes the statute down entirely, and a law invalidated for overbreadth is not narrowed but becomes wholly unenforceable until a legislature rewrites it or a properly authorized court construes it more narrowly.36 A defendant who directed abusive words at police officers could therefore prevail on a claim about other people's speech.
Houston v. Hill applied the same reasoning to a city ordinance making it unlawful to “assault, strike, or in any manner, oppose, molest, abuse or interrupt any policeman in the execution of his duty.”37 State law had preempted the assault and strike provisions, so the enforceable remainder dealt with speech, and the Court found it not narrowly tailored to disorderly conduct or fighting words.38 That reduction matters under the substantiality comparison. When the conduct clauses drop out of the enforceable text, little legitimate sweep remains to set against the protected speech the remaining words reach. The Court also stressed police discretion, observing that it had repeatedly invalidated laws giving the police unfettered discretion to arrest individuals for words or conduct that annoy or offend them, and that “The Constitution does not allow such speech to be made a crime.”39 The ordinance was held overbroad on its face.40 The same pairing of defects appears in an ordinance that required groups meeting on a sidewalk to avoid annoying any police officer or other passerby. It was vague because conduct that annoys some people does not annoy others, so the ordinance specified no standard of conduct at all, and it was broad because it authorized the punishment of constitutionally protected conduct.41
In R.A.V., the Justices divided over whether to reach overbreadth at all. A cross was burned in the yard of a black family, and the charge was brought under a St. Paul ordinance making it disorderly conduct to place a symbol, object, appellation, characterization, or graffiti that one knows or has reasonable grounds to know “arouses anger, alarm or resentment in others on the basis of race, color, creed, religion or gender.” The Minnesota Supreme Court construed the ordinance with the Chaplinsky definition so that it reached only fighting words.42 The majority accepted that narrowing and decided the case on a different ground, that the ordinance singled out some fighting words for punishment based on the content of what they say.43 The Court's holding therefore rests on content discrimination within a proscribable category. A student who writes that the majority held the ordinance overbroad has attributed to it a ground it left alone.
Four concurring Justices would have decided the case on overbreadth. Justice White wrote that it could easily be decided within the contours of established First Amendment law by holding the ordinance fatally overbroad, because it criminalizes not only unprotected expression but expression the First Amendment protects.44 Even as limited to fighting words, he reasoned, the ordinance reaches conduct that is unprotected, but it also makes criminal expressive conduct that causes only hurt feelings, offense, or resentment.45 The reason follows from the trigger. Much protected speech, including sharp social and political criticism, leaves some listeners angry or resentful, so a prohibition keyed to those reactions cannot separate the unprotected core from the protected speech around it. A state court's narrowing does not end the inquiry. The mere presence of a state court interpretation does not insulate a statute from overbreadth review, and the Court has stricken legislation when the construction supplied by the state court failed to cure the overbreadth.46
The two lines of reasoning leave a legislature little room. A fighting-words law narrow enough to escape overbreadth risks singling out some fighting words for punishment because of their content, while a law broad enough to avoid that problem probably reaches protected speech and falls to overbreadth or vagueness.47 The picture changes when the state's own courts have confined the law to the category, as the New Hampshire courts had in Chaplinsky, which is the contrast Gooding drew with Georgia.48 A statute so confined, and keyed to a tendency to provoke immediate violence rather than to a listener's offense, reaches no substantial protected speech. Its challenger loses the overbreadth route and must argue that the challenger's own words were protected.
An essay stem that describes a defendant who cursed and threatened an officer invites a long debate about whether those words were fighting words, and that debate is usually the wrong place to begin. The efficient answer starts with the statute.
The third step is where answers go wrong. A narrowing to fighting words can still leave a protected trigger underneath the label, which is what the concurring Justices found in the St. Paul ordinance. If the law survives that step, the overbreadth argument fails, although the separate ground the R.A.V. majority applied, content discrimination among fighting words, remains available against a law that singles out some fighting words by subject.
Internet and Online Speech
The Court's leading application of overbreadth reasoning to online speech is Reno v. American Civil Liberties Union, which reviewed two provisions of the Communications Decency Act of 1996 (CDA). One barred the knowing transmission of obscene or indecent messages to any recipient under 18, and the other barred knowingly sending or displaying patently offensive messages in a way available to a person under 18.50 The Court held that both provisions abridge the freedom of speech, because the Act is a content-based blanket restriction that effectively suppresses a large amount of speech adults have a constitutional right to receive and to address to one another.51 The ratio from the earlier sections still organizes the opinion, but the medium changes what must be counted.
A law that protects minors by restricting speakers works only if speakers can sort their audiences, and the trial court found that online they could not. At the time of trial, existing technology included no effective way for a sender to keep minors from its Internet communications without also denying access to adults, and verifying that users are adults would have been prohibitively expensive for noncommercial speakers and some commercial ones.52 Those limits, the Court reasoned, must inevitably curtail a significant amount of adult communication.53 A speaker who cannot screen an audience has one safe course, which is to say nothing a minor might encounter, and that course silences adults in every forum the statute reaches. This is the chilling effect described earlier in the chapter, operating across a whole network at once. The Court assumes that an overbroad law's very existence may cause others not before it to refrain from protected speech, and an overbreadth ruling is designed to remove that deterrent effect.54
The text of the Act widened the problem. The undefined terms “indecent” and “patently offensive” covered large amounts of nonpornographic material with serious educational or other value, and the Act reached noncommercial speakers as well as commercial ones.55 Because liability turned on community standards, a message available to a nationwide audience would be judged by the standards of the community most likely to be offended. The Court's examples included discussions of prison rape or safe sexual practices, artistic images with nude subjects, and arguably the card catalog of the Carnegie Library.56 The government's interest in protecting minors also weakened across that coverage. A parent who e-mailed birth control information to a 17-year-old college freshman could be incarcerated if the college town's community found the material indecent, even though neither the parent, the child, nor anyone in their home community did.57 In ratio terms the legitimate sweep was real, but the protected side was enormous, and much of it lay beyond the interest that justified the law.
As in Stevens, the Government tried to shrink the protected side of the comparison with features of the statute, and each attempt failed because it left the text's coverage unchanged. The first was an argument that adults could still speak through the World Wide Web. The Court answered that the CDA is a content-based blanket restriction, so time, place, and manner analysis does not apply.58 It is therefore immaterial whether the speech would be feasible elsewhere, and the Government's position was equivalent to arguing that a statute could ban leaflets on certain subjects as long as individuals are free to publish books.59
The second attempt relied on the knowledge requirement. Because both provisions reach only messages sent to persons known to be under 18, the Government said, transmitters need not refrain from sending indecent material to adults.60 The Court found that argument untenable. Most Internet forums, including chat rooms, newsgroups, mail exploders, and the Web, are open to all comers, and even the strongest reading of the specific-person requirement would confer a “heckler's veto” on any opponent of indecent speech who logged on and announced that a 17-year-old child was present.61 The requirement sounds like a narrowing device, but it fails because the knowledge attaches to the identity of the audience. In an open forum a third party can change that fact at will, so the speaker cannot control the condition that triggers liability.
The third argument was that material with scientific, educational, or other redeeming value would necessarily fall outside the prohibitions, and the Court found no textual support for it.62 The Act's affirmative defenses fared no better. Defenses for good-faith, effective efforts to restrict minors' access, or for requiring age proof such as a verified credit card, burden speech by forcing content providers to forgo the anonymity the Internet provides and to bear significant costs of developing age verification technology.63 A defense that makes a speaker pay to reach adults does not remove adult speech from the count.
The breadth of this content-based restriction also placed an especially heavy burden on the Government to explain why a less restrictive provision would not be as effective, and the Government did not carry it.64 The alternatives discussed in argument included tagging indecent material so that parents can control it, exceptions for messages with artistic or educational value, some tolerance for parental choice, and regulating commercial Web sites differently from chat rooms. Congress had made no detailed findings and held no hearings on the special problems of the Act.65 That burden language shows that Reno is not a pure Hicks-style challenge. The Court's reasoning ran through narrow tailoring as well as overbreadth, and the content-based character of the law is what put the duty to justify its breadth on the Government. In a challenge to an ordinary conduct law, the allocation taught earlier in this chapter, under which the challenger proves substantial overbreadth, remains the starting point.
The vagueness holding reinforced the overbreadth holding. The Government argued that “patently offensive” cannot be vague because the phrase appears in the three-part Miller test for obscenity. The Court answered that Miller's second prong requires the proscribed material to be specifically defined by applicable state law, a requirement the CDA omitted, and that Miller's prurient-interest and serious-value prongs critically limit the definition in ways the CDA lacks.66 A single limb lifted from a three-part definition thus carries the vagueness without the limits that make the whole definition workable. The result was a statute that unquestionably silences some speakers whose messages are constitutionally protected, and the Court held that such a burden cannot be justified if a more carefully drafted statute could avoid it.67 The overlap between overbreadth and vagueness taught in the previous section appears here in a single text, where undefined terms spread the statute's reach and leave speakers unable to tell where it stops.
Congress responded to Reno in 1998 with the Child Online Protection Act (COPA), which differs from the CDA in two ways. It applies only to commercial communications on the World Wide Web, and it defines the offending material by a variation of the Miller obscenity test measured with respect to minors.68 In Ashcroft v. ACLU the Court reversed an injunction against COPA on narrow grounds. It held only that reliance on community standards to identify material harmful to minors does not by itself render the statute substantially overbroad, and it expressed no view on overbreadth for other reasons, on vagueness, or on whether the statute would survive strict scrutiny.69 The narrowness of that holding is the lesson. Narrowing a statute's target shrinks the protected side of the ratio, but a court deciding one feature of a facial challenge decides only that feature, and the broader question whether a local community standard applied to the Internet offends the First Amendment stayed open.70
Virtual imaging raised the same problem from another direction. In Ashcroft v. Free Speech Coalition the Court held the Child Pornography Prevention Act substantially overbroad because it extended the federal ban on child pornography to computer-generated images that appear to depict minors. It thus reached a significant universe of speech that is neither obscene under Miller nor child pornography under Ferber.71 Ferber rested on how such material is made, because production is intrinsically related to the sexual abuse of children, and virtual images involve no actual children.72 The Government argued that virtual images are hard to tell from real ones and so frustrate prosecution of those who use real children. The Court's answer, repeated in later cases, was that government may not “suppress lawful speech as the means to suppress unlawful speech.”73 The Act's separate pandering provision, covering sexually explicit images advertised in a way that gives the impression that minors are depicted, also failed. The pandering rationale of Ginzburg did not save it, because it banned substantial speech outside that rationale and did not require that the context be part of an effort at commercial exploitation.74
The Court has been more receptive to child pornography laws that a limiting construction can bring within the unprotected category. An Ohio statute criminalizing the possession and viewing of child pornography might have succumbed to an overbreadth challenge, since it punished “simple nudity, without more.” The Ohio Supreme Court, however, had narrowed it to reach only nudity that constitutes a lewd exhibition or involves a graphic focus on the genitals, and as narrowed the statute was not impermissibly overbroad.75
After Free Speech Coalition, Congress enacted the PROTECT Act, which criminalized the pandering or solicitation of child pornography in certain circumstances, and the Court rejected overbreadth and vagueness challenges to it in United States v. Williams. The law prohibited offers to provide and requests to obtain child pornography and did not require that the material actually exist, because its target was the collateral speech that feeds the marketing and distribution of such material.76 The Court distinguished Free Speech Coalition on several grounds. The law had a scienter requirement, and the defendant had to objectively believe the material to be child pornography. Its definition of the proscribed “sexually explicit conduct” closely tracked one already upheld against overbreadth challenge in Ferber.77 The law was also not confined to commercial transactions, since a person could distribute child pornography without expecting money in return, yet it stayed focused on offers and requests.78
Each of those features shrinks the protected side of the ratio, and the scienter requirement does different work from the knowledge requirement rejected in Reno. The CDA's requirement attached to who might be listening, a fact outside the speaker's control in an open forum. The pandering law's requirement attached to the speaker's own belief about the material, a fact within the speaker's control. A speaker who holds no such belief is outside the law and has no reason to fall silent.
A different line of online cases finds a law too broad without running the overbreadth ratio. In Packingham v. North Carolina the Court began from the principle that all persons have access to places where they can speak and listen, and that a street or a park is a quintessential forum for the exercise of First Amendment rights.79 It held that a state statute making it a felony for a registered sex offender to access a commercial social networking site, where the offender knows that the site permits minors to become members or to create or maintain personal pages, was too broad and violated the First Amendment.80 Packingham, who had earlier been convicted of taking indecent liberties with a child, was prosecuted after he posted on Facebook that a court had dismissed a traffic ticket against him.81 The Court assumed the statute was content neutral and applied intermediate scrutiny, under which a law must be narrowly tailored to serve a significant government interest and may not burden substantially more speech than necessary.82 The statute's broad wording could deny access to sites as varied as Amazon.com, Washington Post.com, and WebMD, and the state had not shown that so sweeping a law was necessary to keep convicted offenders away from vulnerable victims.83
The label matters on an exam. Packingham did not use the Hicks comparison of protected applications to a plainly legitimate sweep, and it did not use the term overbreadth. The two inquiries ask a similar question, whether a law reaches more protected speech than its purpose justifies, but they place the work differently. Overbreadth requires the challenger to show substantial protected reach relative to the law's legitimate sweep. The tailoring inquiry in Packingham faulted the state for failing to show that the ban was necessary. The challengers also differ. Packingham's post about a dismissed traffic ticket was ordinary protected speech, so his claim did not depend on the rights of third parties. An answer that calls a social-media access ban substantially overbroad and runs the ratio has misdescribed the case, and an answer that asks whether the law burdens substantially more speech than necessary under intermediate scrutiny has described it correctly.