In recent months, I've often run across arguments (including in comments on this blog) that certain kinds of speech ought to be unprotected because they're really "conduct," not "speech." Now these weren't arguments about expressive conduct, such as flagburning or nude dancing. They focused on speech that was written or spoken words, and the speech was seen as causing harm through its content (as opposed to, for instance, because it was too loud or said by people who were blocking traffic). But something about the words -- or the laws restricting the words -- led some to respond that the restriction was actually a conduct restriction, not a speech restriction.

This led me to decide to serialize on the blog portions of an article I wrote about the subject a couple of years ago, Speech as Conduct: Generally Applicable Laws, Illegal Courses of Conduct, "Situation-Altering Utterances," and the Uncharted Zones, 90 Cornell Law Review 1277 (2005). I much enjoyed serializing my Medical Self-Defense piece when I was working on it, and my sense was that many of our readers enjoyed it, too. Let me then take the liberty of doing the same as to this piece -- which turns out to be relevant to many hot First Amendment debates, as I hope the following will show. If you're interested in seeing the footnotes, look here.

When, if ever, should speech lose its First Amendment protection on the grounds that it's really just conduct? Let us set aside restrictions of speech or expressive conduct based on its noncommunicative aspects, for instance because the speakers are blocking traffic or are being too loud. Rather, let's focus on situations in which speech is restricted because of the harm that flows from its content.

Consider, for instance, a book that explains the steps necessary to commit a particular crime. May this speech be restricted on the grounds that it constitutes the "conduct" of aiding and abetting, and is thus not subject to First Amendment protection at all? Or consider racist, religiously bigoted, or sexist statements that create an offensive work environment, an offensive educational environment, or an offensive public accommodations environment. May such statements be freely restricted because they aren't speech but rather the "conduct" of harassment?

There are at least three main types of such "it's conduct, not speech" arguments. First, some people think speech should be treated as conduct when it has the same effects as harmful conduct and it is covered by a generally applicable law that restricts all conduct that has those effects. This can happen in many situations [for examples of each, see the footnotes here:

1. Publishing a book that describes how to grow marijuana might constitute intentional or knowing aiding and abetting of a crime.
2. Publishing a newspaper article or web site that points to an infringing site may constitute contributory copyright infringement.
3. Publishing a news story that reveals the name of a witness, and thus unintentionally helps a criminal intimidate or kill that witness, may violate laws that bar knowingly, recklessly, or negligently facilitating crimes.
4. Publishing a news story that reveals the existence of a wiretap may help the wiretap targets escape justice, and may thus violate obstruction of justice laws.
5. Teaching one's child racist, pro-polygamy, or pro- or anti-homosexuality views may (in the views of some family court judges) be contrary to the best interests of the child and may therefore lead the parent to lose custody or have his visitation rights curtailed under the generally applicable "best interests of the child" standard.
6. Making statements that create an offensive work, educational, public accommodation, or housing environment based on race, religion, sex, age, disability, or sexual orientation might violate antidiscrimination law.
7. Speaking out against a proposed group home for the mentally disabled might violate the Federal Housing Act's ban on "interfer[ing] with any person in the exercise or enjoyment of" the right to be free from housing discrimination based on handicap.
8. Engaging in speech that helps the election of an anti-war candidate may violate treason law -- which prohibits intentionally aiding the enemy in time of war -- if the speaker thinks the enemy deserves to win the war.
9. Creating newspaper advertisements, billboards, or leaflets that praise jury nullification may be punishable under laws that prohibit all attempts to influence jurors.
10. Producing and distributing movies that stimulate copycat crimes may constitute negligence under generally applicable tort principles.
11. Giving children sexually themed material, or for that matter political material that most people view as evil, may violate laws that ban "impair[ing] the . . . morals of . . . [a] child."

In all these cases, the speech would be restricted because of what it communicates -- because its content informs, persuades, or offends people -- and because of the harms that flow from this informing, persuasion, or offense. Yet some courts and commentators argue that such speech restrictions don't implicate the First Amendment because the law in these instances punishes conduct, not speech: "[S]peech which, in its effect, is tantamount to legitimately proscribable nonexpressive conduct may itself be legitimately proscribed, punished, or regulated incidentally to the constitutional enforcement of generally applicable statutes." Others argue that generally applicable laws should be treated as content-neutral restrictions on expressive conduct, and should thus be fairly easily upheld under the deferential O'Brien test because the restrictions on speech are "incidental" to the law's overall thrust.

A second type of "conduct, not speech" argument is sometimes made even to defend laws that specifically target communication, such as statutes that ban the publication of bombmaking information. Such speech, the argument runs, is punishable because it is part of an illegal "course of conduct," or is perhaps "speech brigaded with action," a "speech act" rather than pure speech.

The argument seems especially appealing to some when the speech appears likely to cause harms that would be punishable if caused by conduct rather than speech -- when "words are bullets," in the sense of being "a specific tool or weapon used . . . for the express purpose" of causing harm. Such arguments often quote Giboney v. Empire Storage & Ice Co., a 1949 case which asserted that "[i]t rarely has been suggested that the constitutional freedom for speech and press extends its immunity to speech or writing used as an integral part of conduct in violation of a valid criminal statute," and that "it has never been deemed an abridgement of freedom of speech or press to make a course of conduct illegal merely because the conduct was in part initiated, evidenced, or carried out by means of language, either spoken, written, or printed."

Courts have applied Giboney to justify, among other things, restrictions on (1) speech that advocates crime, (2) speech that explains how crimes can be committed, (3) doctors' speech recommending medicinal marijuana to their patients, (4) speech that urges political boycotts, (5) speech that creates an offensive work environment, (6) racially offensive business names, and even (7) public profanity.

A third "speech as conduct" argument is made in Professor Kent Greenawalt's influential book, Speech, Crime, and the Uses of Language, which asserts that certain kinds of speech -- such as offers, agreements, orders, permissions, and some threats -- constitute "situation-altering utterances" and should therefore be treated as unprotected conduct.

Finally, the "speech as conduct" argument is sometimes made to explain some of the uncharted zones of First Amendment law: categories of speech whose First Amendment status the Court has never squarely confronted, such as aiding and abetting, criminal solicitation, conspiracy, perjury, agreements to restrain trade, and professional advice to clients. Most lawyers would likely agree that such speech generally should be unprotected, or at least less protected. A common explanation for the Court's lack of attention to these speech restrictions is that the speech is actually conduct, which the First Amendment does not protect.

Consider a generally applicable law that is being applied to speech, but that on its face doesn't mention speech. Sometimes, as in United States v. O'Brien, the law may be triggered by the "noncommunicative impact of [the speech], and [by] nothing else." A law barring noise louder than ninety decibels, for instance, might apply to the use of bullhorns in a demonstration. We might call such a generally applicable law "content-neutral as applied," because it applies to speech without regard to its content.

But sometimes the law is triggered by what the speech communicates. The law may, for instance, prohibit any conduct that is likely to have a certain effect, and the effect may sometimes be caused by the content of speech. A person may violate a law prohibiting aiding and abetting crime, for example, by publishing a book that describes how a crime can be easily committed.

We might call such a law "content-based as applied," because the content of the speech triggers its application. The law doesn't merely have the effect of restricting some speech more than other speech -- most content-neutral laws do that. Rather, the law applies to speech precisely because of the harms that supposedly flow from the content of the speech: Publishing and distributing the book violates the aiding and abetting law because of what the book says.

In this post and coming posts, I'll argue that laws that are content-based as applied should be presumptively unconstitutional, just as facially content-based laws are presumptively unconstitutional. Both presumptions may sometimes be rebutted, for instance if the speech falls within an exception to protection or if the speech restriction passes strict scrutiny. But generally speaking, when a law punishes speech because its content may cause harmful effects, that law should be treated as content-based.

This analysis also cuts against some commentators' arguments that First Amendment doctrine should focus primarily on smoking out the legislature's impermissible speech-restrictive motivations. (See, e.g., Elena Kagan, Private Speech, Public Purpose: The Role of Governmental Motive in First Amendment Doctrine, 63 U. Chi. L. Rev. 413 (1996); Jed Rubenfeld, The First Amendment's Purpose, 53 Stan. L. Rev. 767 (2001).) When a law generally applies to a wide range of conduct, and sweeps in speech together with such conduct, there is little reason to think that lawmakers had any motivation with regard to speech, much less an impermissible one. Nonetheless, such a law should still be unconstitutional when applied to speech based on its content—even though the legislature's motivations may have been quite benign.

The Court has confronted many cases where a law was content-based as applied. In all those cases, either the Court held that the speech was constitutionally protected, or -- if it held otherwise -- the decision is now viewed as obsolete.

Consider, for instance, the World War I-era cases Debs v. United States, Frohwerk v. United States, and Schenck v. United States. These cases, which upheld the criminal punishment of antiwar speech, are now generally seen as wrongly decided. But the defendants' statements had violated a generally applicable provision of the Espionage Act, which barred all conduct -- speech or not -- that "willfully obstruct[ed] the recruiting or enlistment service of the United States, to the injury of the service or the United States."

I continue the posts excerpting my article, Speech as Conduct: Generally Applicable Laws, Illegal Courses of Conduct, "Situation-Altering Utterances," and the Uncharted Zones, 90 Cornell Law Review 1277 (2005). Right now, I'm discussing the first part of the argument, which responds to the claim that "generally applicable laws" may be applied to speech with little constitutional scrutiny (or at least without strict scrutiny) even when the laws apply to the speech precisely because of the communicative effects of the speech.

So far, I've used the term "generally applicable law" simply to mean a law applicable equally to a wide variety of conduct, whether speech or not. But "generally applicable law" can have several different meanings, depending on context:

1. a facially speech-neutral law, which is to say a law applicable to a wide variety of conduct, whether speech or not;

2. a facially religion-neutral law, which is to say a law applicable equally to religious observers and to others; or

3. a facially press-neutral law, which is to say a law applicable equally to the press and to others.

These three meanings -- facially speech-neutral, facially religion-neutral, and facially press-neutral -- are different, though they sometimes share the label "generally applicable law." For instance, most libel law principles are press-neutral but not speech-neutral. A tax on all books would be religion-neutral but not press-neutral.

Unfortunately, since all these laws are sometimes called "generally applicable," the three types may be confused with one another. One major argument against the position I defended in the a previous post flows from this very sort of confusion. That argument (used, among others, by the Fourth Circuit in Rice v. Paladin Enterprises and by the Justice Department in their memo on restricting crime-facilitating speech) cites Cohen v. Cowles Media Co., and the opinions on which that case relies, for the proposition that applying generally applicable laws to speech doesn't violate the First Amendment.

In Cowles Media, the Court held that "generally applicable laws do not offend the First Amendment simply because their enforcement against the press has incidental effects on its ability to gather and report the news," and cited several other cases that so held. But this only means that the press gets no special exemption from press-neutral laws. The Court didn't consider whether speakers were entitled to protection from speech-neutral laws, especially when those laws are content-based as applied. Cowles Media involved a promissory estoppel lawsuit by a source against a newspaper publisher. Cowles breached its promise not to reveal Cohen's name; Cohen sued and won on a promissory estoppel theory, and the Court held that the damages award didn't violate the First Amendment. In the process, the Court reasoned that the case was controlled by the

well-established line of decisions holding that generally applicable laws do not offend the First Amendment simply because their enforcement against the press has incidental effects on its ability to gather and report the news. As the cases relied on by respondents recognize, the truthful information sought to be published must have been lawfully acquired. The press may not with impunity break and enter an office or dwelling to gather news. Neither does the First Amendment relieve a newspaper reporter of the obligation shared by all citizens to respond to a grand jury subpoena and answer questions relevant to a criminal investigation, even though the reporter might be required to reveal a confidential source. The press, like others interested in publishing, may not publish copyrighted material without obeying the copyright laws. Similarly, the media must obey the National Labor Relations Act and the Fair Labor Standards Act; may not restrain trade in violation of the antitrust laws; and must pay nondiscriminatory taxes. It is therefore beyond dispute that "[t]he publisher of a newspaper has no special immunity from the application of general laws. He has no special privilege to invade the rights and liberties of others." Accordingly, enforcement of such general laws against the press is not subject to stricter scrutiny than would be applied to enforcement against other persons or organizations.

There can be little doubt that the Minnesota doctrine of promissory estoppel is a law of general applicability. It does not target or single out the press. Rather, insofar as we are advised, the doctrine is generally applicable to the daily transactions of all the citizens of Minnesota. The First Amendment does not forbid its application to the press.

The Court repeatedly stressed that it was discussing only whether the press gets special exemption from laws that are equally applicable to the press and to others; this quote mentions "the press," "newspapers," or "the media" nine times. Each of the examples the Court gave discussed what "the press," "the media," "newspaper[s]," and "newspaper reporter[s]" have no special right to do. This makes sense, because the Court was overruling the Minnesota Supreme Court's conclusion that the First Amendment requires courts to "balance the constitutional rights of a free press against the common law interest in protecting a promise of anonymity."

Moreover, two of the Court's examples are consistent only with the interpretation that the Court used "generally applicable" to mean press-neutral rather than speech-neutral. First, copyright law (which the Court also mentions as an example later in the opinion) is press-neutral but not speech-neutral. In 1977, when Zacchini v. Scripps-Howard Broadcasting Co. -- the case that the Cowles Media Court cited when referring to copyright law -- was decided, copyright law applied exclusively to communication, as it had through most of its history. Even today it applies mostly to communication, though over the past few decades it has been extended to cover architectural works and computer program object code.

Second, as Part II.B pointed out, the First Amendment sometimes provides a defense against antitrust law, when the alleged restraint of trade comes from defendant's speech advocating legislation. Citizen Publishing Co. v. United States and Associated Press v. United States, the two antitrust cases that the Court cited, hold that newspapers cannot raise their status as members of the press as a defense to antitrust law. But Noerr and Pennington make clear that speakers can raise as a defense the fact that the law is being applied to them because of their speech.

So the Cowles Media Court's "general applicability" reasoning means simply that Minnesota promissory estoppel law is press-neutral, and thus shouldn't have been subject to any heightened scrutiny simply because it was applied to the press. (Compare Turner Broadcasting Sys., Inc. v. FCC, "[W]hile the enforcement of a generally applicable law may or may not be subject to heightened scrutiny under the First Amendment ... laws that single out the press ... for special treatment 'pose a particular danger of abuse by the State,' ... and so are always subject to at least some degree of heightened First Amendment scrutiny.")

That, of course, leaves unresolved the argument that the law couldn't be applied because it restricted speech; after all, it was Cowles Media's speech that constituted the potentially actionable breaking of a promise.

But later in the opinion, the Court explains why promissory estoppel law is indeed constitutionally applicable to all speakers, whether press or not: "Minnesota law simply requires those making promises to keep them. The parties themselves, as in this case, determine the scope of their legal obligations, and any restrictions which may be placed on the publication of truthful information are self-imposed." So the Court rejected the free speech argument based on the principle that free speech rights, like most other rights, are waivable, rather than on an assertion that speech-neutral laws are per se constitutional.

As I discussed earlier, First Amendment precedents generally makes conveying facts and opinions into a constitutionally immunized activity. Normally, the government may punish people for causing various harms, directly or indirectly. But it generally may not punish speakers when the harms are caused by what the speaker said -- by the persuasive, informative, or offensive force of the facts or opinions expressed (unless, of course, the speech falls within one of the First Amendment exceptions, such as incitement, false statements of fact, threats, and the like).

This is, of course, quite compatible with the Court's general jurisprudence of content-based restrictions; it just equally covers laws that are content-based as applied and laws that are content-based on their face. And this principle makes sense, because a law that's content-based as applied can restrict speech as much as a law that's content-based on its face. Moreover, such a law is indeed punishing the "speech element" of the communication rather than some "nonspeech element," see U.S. v. O'Brien.

This principle is in some tension, however, with claims (such as those made by Dean Elena Kagan and Professor Jed Rubenfeld) that the First Amendment is chiefly aimed at preventing government actions that are motivated by a desire to suppress speech. In the examples I've given, the lawmakers may have genuinely wanted to prevent a certain kind of harm, and may have been quite indifferent to whether that harm is caused by speech or by conduct. The drafters of the Espionage Act, for instance, might have sincerely wanted to punish all interference with military recruitment. But whether the Act was well-motivated or not, it should have generally been unconstitutional when applied to interference by persuasion.

In some of the examples, one can argue that the law is open to improper government motivations in its enforcement. For instance, the "outrageousness" test in the emotional distress tort, the "offensive conduct" test in breach of the peace laws, and the "offensive work environment" test in workplace harassment law are quite vague. Prosecutors, judges, and juries might well interpret them narrowly when they agree with the speech, and broadly when they disagree with the speech.

But in other situations, the law is pretty clear. Public speech that advocates draft resistance does seem likely to obstruct recruitment. A journal article that explains how fingerprint recognition systems can be evaded does seem likely to facilitate certain crimes by some readers. If applying the law to such speech would violate the First Amendment, the reason must flow from something other than the government's motive, which may well be quite pure.

Prosecutors may still have discretion in deciding whom to charge under those laws, and they may exercise that discretion out of a desire to suppress certain viewpoints, rather than to evenhandedly prevent the harm that the law is aimed at preventing. But that risk is equally present for any law that may be applied to speech, including generally applicable laws that are both speech-neutral on their face and content-neutral as applied.

So, if the cases I've discussed are right, then the constitutional problem lies in the law's being content-based as applied -- in its punishing speech because of the persuasive effect of the speech -- and not in the government's being motivated by a desire to suppress speech rather than to prevent harm. Though the Supreme Court has at times said that "[i]n determining whether a regulation is content based or content neutral, we look to the purpose behind the regulation," it has also acknowledged that "while a content-based purpose may be sufficient in certain circumstances to show that a regulation is content based, it is not necessary to such a showing in all cases." The better formulation is the one the Court has often used: A content-neutral law is one that is "justified without reference to the content of the regulated speech" -- and a law that is content-based as applied is indeed justified, in that application, with reference to what the speech communicates.

Tomorrow: More on why content-based speech restrictions are indeed generally more troublesome than content-neutral restrictions.

Content-based speech restrictions imposed by the government as sovereign are almost always unconstitutional, unless the speech fits within one of the First Amendment exceptions. Content-neutral restrictions are often constitutional.

I have argued in earlier posts that the Court has been right to treat generally applicable laws that are content-based as applied much like it treats content-based restrictions generally. And I think the Court should treat generally applicable laws that are content-neutral as applied much like it treats content-neutral restrictions generally. The Court probably would not, and should not, have intervened if Hustler had inflicted emotional distress on Falwell by using loud bullhorns outside Falwell's house. Nor would the Court have found a First Amendment violation if the NAACP had been sued for organizing a demonstration that blocked the entrance to Claiborne Hardware's door, or if Schenck had been prosecuted for interfering with the draft by blocking a draft board office.

But why? The law, and thus the lawmakers' motivation for enacting the law, would be the same in these hypotheticals as in the real cases. The law's effect would be the same: The law as applied would restrict speech. What then is left to explain the difference? And if indeed the lawmakers' motivation doesn't have the importance that some assign to it, then what is the difference even between facially content-based laws and facially content-neutral ones? There are, I think, two main answers to these questions — a conceptual one and a pragmatic one.

The Conceptual Distinction: Under nearly every theory of free speech, the right to free speech is at its core the right to communicate — to persuade and to inform people through the content of one's message. The right must also generally include in considerable measure the right to offend people through that content, since much speech that persuades some people also offends others. (There might be some limits on this right to offend, for instance if (1) the speaker is communicating to someone who has already said that he doesn't want to hear the message, and (2) the speaker can stop speaking to this unwilling listener, while still continuing to try to persuade or inform other potentially willing listeners.)

Persuading and informing people may certainly cause harm; the listeners might be persuaded to do harmful things. But the premise of modern First Amendment law is that the government generally may not (with a few narrow exceptions) punish speech because of a fear, even a justified fear, that people will make the wrong decisions based on that speech: "[T]he people in our democracy are entrusted with the responsibility for judging and evaluating the relative merits of conflicting arguments.... [I]f there be any danger that the people cannot evaluate the information and arguments advanced by [speakers], it is a danger contemplated by the Framers of the First Amendment." Thus, punishing speech because its content persuades, informs, or offends especially conflicts with the free speech guarantee, more so than punishing speech for reasons unrelated to its potential persuasive, informative, or offensive effect.

Practical Effects — Content-Based Restrictions as Likely Greater Burdens on Speech: I suspect that the Court's skepticism about content-based laws (whether facially content-based or content-based as applied) flows in large part from this conceptual distinction. But the conceptual argument is also reinforced by a pragmatic one: Allowing content-based restrictions (whether facially content-based or content-based as applied) is likely to burden speech more than allowing content-neutral restrictions.

(Note that I'll treat the term "content-based restrictions" as roughly interchangeable with the phrase "restrictions triggered by the communicative effects of speech," and the term "content-neutral restrictions" as roughly interchangeable with the phrase "restrictions triggered by the noncommunicative effects of speech." Also, some commentators have also argued that content-based restrictions are more dangerous than content-neutral restrictions, because content-based restrictions often distort public debate by burdening one side of a debate while allowing another to be heard free of any such burden, but others have disagreed; my analysis neither relies on nor rejects the distortion argument — I focus on whether a restriction is likely to substantially interfere (as opposed to only modestly interfere) with the expression of certain facts or viewpoints.)

To begin with, a typical law aimed at noncommunicative effects is unlikely to excessively inhibit the communication of some viewpoint or fact, because many different media would remain available to the speakers. For instance, even a total ban on leafleting, justified by the desire to prevent litter, would still leave people free to communicate their views by the many media that don't create litter — by displaying signs, using radio broadcasts, advertising in newspapers, and so on.

I think the leafleting ban would indeed interfere with public debate too much, but it can't even come close to driving certain views entirely from public debate. Moreover, because the content-neutral law can potentially apply to a wide range of speakers, its scope will likely be limited by political forces. Thus, the most severe hypothetical content-neutral restrictions — for instance, a ban on printing, justified by the environmental harms caused by the process of making paper — are sure to remain just hypotheticals: They are politically implausible precisely because they burden so much speech.

On the other hand, a content-based restriction, whether facially content-based or content-based as applied, can outlaw most expression of certain facts or opinions. If a law, such as the laws in Schenck v. United States or NAACP v. Claiborne Hardware Co., bans any conduct that may cause a certain harm, and persuading people to act in certain ways can cause that harm, then any viewpoints that have the potential for such persuasion — the draft is evil, blacks should boycott white-owned businesses — would largely be prohibited. Because the law focuses either on the content of the speech or on the harm that the speech causes, it can block the speech in all media. And because it's limited to a narrow range of speech, it may face less political opposition than broader bans might provoke.

Even narrower content-based restrictions, such as the laws involved in Cohen v. California or Hustler Magazine v. Falwell, can be quite burdensome. True, such restrictions only limit the particular words (in Cohen) or the level of vitriol (in Hustler) a speaker can use, and don't ban the expression of a particular fact or idea. But as Justice Harlan rightly concluded in Cohen, even such restrictions can seriously interfere with people's ability to express the "otherwise inexpressible emotions" that only certain kinds of words can effectively capture. Harsh contempt for a policy (the draft) or a person (Jerry Falwell) is itself a viewpoint that is subtly different from mild-mannered condemnation, and prohibitions on harsh language seriously interfere with the ability to convey this viewpoint.

[Coming Monday: Why can't this be dealt with using an "ample alternative channels" inquiry, under which restrictions — both content-neutral and content-based — would be constitutional if they leave open ample alternative channels for expressing the speaker's ideas, and thus wouldn't impose much of a burden on the speaker's speech? Content-neutral restrictions that leave open ample alternative channels are indeed generally constitutional; not so for content-based restrictions. Does this distinction make sense, or should the rule for content-neutral restrictions also be applied to content-based ones?]

I've argued earlier that the Court has been right to treat content-based speech restrictions much more harshly than content-neutral speech restrictions. But this left open the question: Why should the Court be so skeptical even about modest content-based regulations, which don't entirely ban certain ideas but merely restrict them while leaving open ample alternative channels? Such a test is applicable to content-neutral restrictions — why not content-based ones? Here is my answer, adapted from my Speech as Conduct: Generally Applicable Laws, Illegal Courses of Conduct, "Situation-Altering Utterances," and the Uncharted Zones, 90 Cornell L. Rev. 1277 (2005).

The Limits of the "Ample Alternative Channels" Inquiry, Both as to Content-Neutral Restrictions and as to Content-Based Restrictions: I have argued that content-based restrictions are dangerous because they risk broadly suppressing certain viewpoints or facts. But one could respond that, instead of presumptively prohibiting content-based speech restrictions, courts could try to prevent serious burdens on speech the same way they do with content-neutral restrictions: by asking whether the restrictions leave open "ample alternative channels" for expression.

I think, though, that the Court has been right to reject such proposals and to treat content-based restrictions as presumptively unconstitutional without an inquiry into how much the restriction burdens speech or into whether the restriction leaves open ample alternative channels. To begin with, the record of the ample alternative channels inquiry in the content-neutral restriction test hasn't been very good. The Court has at times applied it in a demanding manner, for instance insisting that alternative channels aren't ample if they materially raise the price of speaking, make it harder for speakers to reach the same listeners, or subtly influence the content of the message by changing the medium. But at other times, the Justices have treated this requirement as only a weak constraint. Such a disparity is to be expected given the vagueness of the term "ample."

In fact, the chief practical limit on content-neutral restrictions has not been the "ample alternative channels" inquiry, but the political reality mentioned above: Most realistically enactable restrictions on the noncommunicative aspects of speech do leave open fairly substantial alternative channels for expressing the same ideas. So even if the Court underenforces the ample alternative channels prong, few views or subjects will likely be broadly silenced.

But it's much more likely that a politically feasible restriction on the communicative aspects of speech will substantially block people from expressing a particular viewpoint. This is true even when the restriction is framed as facially content-neutral, or even as speech-neutral — consider, for instance, the Espionage Act in Schenck. Judicial underenforcement of the ample alternative channels prong for content-based restrictions would thus be much more dangerous than underenforcement in the context of content-neutral restraints.

The Limits of the "Ample Alternative Channels" Inquiry as to Content-Based Restrictions: Judicial underenforcement of the ample alternative channels prong would also be more likely when the case involves content-based restrictions, whether they are facially content-based or content-based as applied. "Ample" is a vague term, and one that requires contestable predictions about the law's effects on a complex system of speakers and listeners. There is a large gray area in which the quality of the alternative channels would be hard to estimate.

And when the restriction will likely cover only a particular message — pro-boycott speech, anti-draft speech, and so on — the normal risk of judicial error and deliberate or subconscious prejudice is magnified because the judges know well which side of the political debate will lose as a result of their decision. In such a scenario, it's especially likely that judges will apply the vague "ample alternative channels" standard in a way that's not protective enough of unpopular speakers. It is probably no accident that the low water mark of the requirement, City of Renton v. Playtime Theatres, Inc., involved a restriction that was limited to sexually-themed speech, even though the Court treated the restriction as content-neutral.

Moreover, one restriction aimed at the communicative impact of certain speech is likely to be followed by other such restrictions. Content-based restrictions don't appear randomly: They arise because some fairly powerful segment of society (in government or out of it) believes that a certain kind of speech is dangerous, or — as to laws that aren't facially content-based but are content-based as applied — believes that all conduct that's likely to cause certain effects is dangerous.

If such a group succeeds in restricting, say, Communist speech in some contexts, it seems likely that it will also want to restrict Communist speech in other contexts. Likewise, if a movement tries to restrict bigoted speech in workplaces, perhaps using generally applicable hostile work environment harassment law, it will also likely try to use similar educational and public accommodations harassment rules to restrict speech in educational institutions or places of public accommodation. (That has in fact been the pattern of restrictions on Communist advocacy, antiwar speech, sexually themed speech, pro-civil-rights speech, and racist speech.)

Each success will help validate the pro-restriction forces' positions in the eyes of voters and legislators who are on the fence. Moreover, each success may reinforce the enthusiasm of the supporters of the restrictions. And government restrictions on such speech are also likely to be accompanied by private restrictions on such speech, for instance by private broadcasters, publishers, employers, and commercial property owners. As a result, even when each restriction standing alone imposes only a modest burden on speech, the aggregate of all the restrictions can end up being quite burdensome.

It is, of course, possible for courts to consider this risk, to allow only the first few restrictions, and then to strike down any new restrictions once the alternative channels no longer seem to be ample. But that's a hard project for courts to engage in, especially when they are armed only with the vague "ample alternative channels" standard. Judges may find it hard to explain why they are treating two seemingly similar restrictions differently, simply because of the order in which the restrictions were enacted. And because "ample" lacks an objective absolute definition, courts may end up applying a relative criterion — how many channels the restriction leaves open compared to those available before this restriction was enacted, or how many it leaves open compared to those that it shuts down. If that is so, courts might indeed allow a sequence of restrictions that gradually but substantially reduces the alternative channels, even if the courts would have struck down a restriction that tried to impose the same burden all at once.

Conclusion: For all these reasons, the Court has been right to treat restrictions that are content-based as applied — even if they are facially generally applicable to both speech and conduct — with the same skepticism as it has used for restrictions that are content-based on their face. It's the only approach that is consistent with Hustler, Claiborne Hardware, and the other similar cases. It's properly hostile to the government's attempts to restrict speech because of the informative or persuasive power of the speech. And it's necessary to prevent the government from having the power to broadly suppress certain facts and ideas.

When speech is punished precisely because of what it communicates — for instance, because it may persuade people to violate the law or to boycott someone, because it may offend some listeners, or because it may convey information that helps people commit crimes — the law is operating as a content-based speech restriction. The law is restricting speech precisely because of what is spoken. Therefore, courts should subject such a law to serious First Amendment analysis; they ought not dodge this analysis by simply relabeling the speech as "conduct."