Content Regulations Where the Message Causes Harm in Two Steps

A. The Nonregulable Second Step

Sometimes government wishes to prevent receipt of a message, not because re­ceipt will cause harm directly, but because receipt will likely cause the audience to act in ways that cause harm either to others or to the audience itself.

1058 (9^th Cir. 2002). The publishers may be intending to threaten the doctors with injury or death. On the other hand, they may be legitimately suggesting that those who oppose abortion boycott those doctors. Suppose the latter is their actual intent, but the doctors on the list infer the former intent. Is the website list a protectable exercise of freedom of expression? It is causing a one-step harm. And although its publishers are not intending to threaten, why should its status as protected or unprotected by freedom of expression - as opposed to the criminal liability of its publishers - turn on culpability rather than harm?

For a similar question on the context of incitement, see text following note 59 infra.

It should be noted that, just as it has required intent to incite in order to punish an inciter - see Brandenburg v. Ohio, 395 U.S. 444 (1969) - the United States Supreme Court has required an intent to threaten in order to punish threats. See Virginia v. Black, 123S. Ct. 1536, 1548 (2003). Interestingly, the Court has not clarified when the less speech-protective doctrine regarding threats trumps the more speech-protective Brandenburg rule regarding advocacy and incitement. If a speaker intends both to incite one segment of his audience and at the same time to intimidate another segment, who might fear the incited conduct, does the speaker’s message fall under the Brandenburg rule or under Virginia v. Black?

cases, the government can attempt to prevent the harm only by preventing the message from being received. In others, the government has two chances to prevent the harm: It can attempt to prevent the audience from receiving the message, and, if it fails, it can attempt to prevent the audience from taking harmful action.

Here are some examples of the first type of message-induced harm. Bob spreads false and vicious rumors about Brenda that if believed will cause the audience to shun her. Bertha tells Bill, falsely, that her potion will cure his cancer and obviate the need for chemotherapy. Brian prays aloud in the presence of Brad, who is a deranged atheist and prone to go on murderous rampages whenever he hears someone proclaiming the existence of God.

In each of these cases the government may want to prevent the audience from receiving the message because it fears that the audience, in response to the message, will act in ways harmful to others or to the audience itself. Moreover, the only tenable way for the government to prevent the harm is to prevent the audience from getting the message. Once the audience has heard the message, whether it then acts in a harmful way is beyond government’s ability to control. Those who hear credible lies about Brenda cannot be punished for shunning her. Nor can Bill be reliably prevented from going off his chemotherapy. And Brad, who is deranged, is totally beyond the law’s ability to control.

Thus, to prevent harm, government must prevent Bob, Bertha, and Brian from getting their messages out. Such attempts fall squarely within principle (5), but do they violate freedom of expression? Should government be precluded from penalizing defamatory speech, dangerously deceptive speech, or speech that causes irresponsible actors to commit harms? What would justify a right that so restricts government?

B. The Regulable Second Step

Before answering the questions just raised, let us look at the final set of laws regulating messages, those premised on preventing harms that government may also attempt to prevent by punishing the acts that directly cause them. For exam­ple, when government punishes the incitement or solicitation of crime, it seeks to prevent the harmful act - the crime - by preventing the criminal from receiv­ing the message encouraging the crime.

Because government can punish the audience for any crimes it commits, should government be able to prevent the audience from hearing the messages encouraging those crimes? Or does so preventing the audience contravene the right of freedom of expression? On the one hand, because the acts that the messages encourage are legitimately forbidden, it may appear that any messages that encourage them must be valueless as messages and equally condemnable. On the other hand, if freedom of expression does not protect the audience’s right to assess for itself the truth and value of a message that the government deems false and valueless, then it is hard to see what freedom of expression could mean.

Thomas Scanlon, in 1972,3¹ and David Strauss, in 1991,³² put forward, as the core principle of freedom of expression, one that forbids government from interdicting messages for the reason that those messages might persuade the au­dience to act in ways the government deems harmful. Scanlon calls his principle the “Millian Principle.”³³ Strauss calls his the “Persuasion Principle.”³⁴ They elaborate and ground their principles slightly differently, but for my purposes it is what these principles have in common that is significant.³⁵

The first thing to note about the Scanlon/Strauss position is that it can be given both a very broad and a very narrow reading. On the broadest of read­ings, the position forbids government from penalizing solicitations and other encouragements to crime, and it forbids government as well from penalizing defamations and harmful misrepresentations. For if we take the idea that an

³¹ Thomas Scanlon, A Theory of Freedom of Expression, 1 Phil. & Pub. Aff. 204 (1972).

³² David A. Strauss, Persuasion, Autonomy, and Freedom of Expression, 91 Colum. L. Rev. 334 (1991).

33

There are certain harms which, although they would not occur but for certain acts of expression, nonetheless cannot be taken as part of a justification for legal restrictions on these acts. These harms are (a) harms to certain individuals which consist in their coming to have false beliefs as a result of those acts of expression; (b) harmful consequences of acts performed as a result of those acts of expression, where the connection between the acts of expression and the subsequent harmful acts consists merely in the fact that the act of expression led the agents to believe (or increased their tendency to believe) these acts to be worth performing.

Scanlon, supra note 31, at 213. See also Charles Fried, Saying What the Law Is (2004) 89-93 (proposing as a free speech principle one very similar to Scanlon’s and Strauss’s); Thomas Nagel, Concealment and Exposure (2002), 43-5 (arguing for a general moral principle very similar to Scanlon’s “Millian Principle” for legitimate governmental action); James Weinstein, Hate Speech, Viewpoint Neutrality, and the American Concept of Democracy, in The Boundaries of Freedom of Expression & Order in American Democracy (T. Hensley, ed., 2001), 146, 164-6 (arguing that the fundamental precepts of American democracy require allowing attempts to persuade citizens to violate fundamental rights and even to repudiate democracy).

³³ The persuasion principle... holds that the government may not suppress speech on the ground that the speech is likely to persuade people to do something that the government considers harmful. Put another way, harmful consequences resulting from persuasive effects of speech may not be any part of the justification for restricting speech.

Strauss, supra note 32, at 335.

³⁴ Scanlon justifies his principle on the ground that equal and autonomous citizens would not cede authority to government to violate it. In other words, Scanlon’s principle is one of the legitimacy of governmental acts.

Strauss, on the other hand, views his principle as resting on the right not to have one’s autonomy, in the sense of control over one’s reasoning processes, violated; it is thus a principle that extends beyond governmental action. Strauss, supra note 32, at 354-6.

Scanlon later repudiated his principle -T.M. Scanlon, Freedom of Expression and Categories of Expression, 40 U. Pitt. L. Rev. 519,532-4 (1979)-arepudiation Strauss attributes to its narrow focus on governmental acts. See Strauss, supra note 32, at 356 n.62. audience has a right to make its own assessments of messages - the right that grounds the ScanlonZStrauss restriction on the government - then laws against solicitation of crime, defamation, and misrepresentation all appear to violate the right by attempting to preempt the audience’s assessment of messages.³⁶

³⁵ In the United States, the common law of defamation has been modified by the Supreme Court in order to conform to the Court’s view of freedom of expression. See, e.g., New York Times v. Sullivan, 376 U.S. 254 (1964) (holding that speech libeling public officials is protected unless defendant knows of or is reckless regarding its falsity); Curtis Pub. Co. v. Butts, 388 U.S. 130 (1967) (extending the New York Times rule to cover public figures); Gertz v. Robert Welch, Inc., 418 U.S. 323 (1974) (requiring negligence as a condition of liability when private figures are defamed in discussions of matters of public concern). The United States Supreme Court protects some false and defamatory speech in order not to “chill,” through fear of liability, negative speech that is true. The Court has never explained how it knows that the true speech it has thawed is more valuable than the harms caused by the false speech it has thawed. In other countries that recognize a right of freedom of expression, the courts have found that ordinary defamation law is not violative of thatright. See, e.g., Barfod v.

In all of these countries, though to a varying extent, misleading speech that is likely to induce an audience to act to its detriment is subject to sanctions. Thus, commercial claims that are false are subject to civil and sometimes criminal punishment. See, e.g., Lillian Bevier, “The Invisible Hand of the Marketplace of Ideas,” in Externally Vigilant (L. Bollinger & G. Stone, eds., 2001), 233 et seq. So, too, are all sorts of misrepresentations, commercial, professional, and miscellaneous, right down to “that bridge is safe to drive over.” See, e.g., The Restatement ofTorts 2d, 310,311. See generally Susan M. Gilles, “‘Poisonous’ Publications and Other False Speech Harm Cases,” 37 Wake Forest L. Rev. 1073 (2002). Indeed, even accurate medical and legal advice can be sanctioned if the speaker has not been licensed by the government, presumably because of the danger that unlicensed persons may purvey dangerously misleading advice. See, e.g., National Ass’n for the Advancement of Psychoanalysts v. Cal. Bd. of Psychology, 228 F. 3d 1043 (9th Cir. 2000) (involving claim that regulation of medical advice through licensing violated freedom of speech). And in some countries, false and misleading political speech can be sanctioned. Compare Brown v. Hartlage, 456 U.S. 45 (1982) (holding that a political promise incapable of being performed did not constitute grounds for overturning an election), with Frieson v. Hammell, 57 B.C.L.R. (3d) 276 (1999) (stating that political speech that contains misrepresentations can be basis for setting aside election results); Bonneville v. Frazier, 2000 B.C.D. Civ. J. 3690 (2000) (same). Arguably, misleading political speech is the most dangerous of all misleading speech. But see Fried, supra note 33, at 117.

Of course, rendering liable for damages those who make false or misleading statements could “chill” speakers and result in the reduction of true statements. However, given that we do not know what truths will be suppressed through chilling, and given that we lack an evaluatively neutral framework for assessing the importance of those truths, we cannot conclude that the harm of imposing liability exceeds the harms caused by false and misleading speech.

Finally, every country that purports to honor the right of freedom of expression outlaws criminal solicitation. See, e.g., Brandenburg v. Ohio, supra note 30 (striking down mere advocacy of crime but stating that the state may legitimately punish incitements of imminent lawless acts); Faurisson v. France, 2 B.H.R.C. 1 (1996) (U.N. Human Rights Commission) (upholding conviction under French “holocaust denial” law enacted to prevent incitements to anti-Semitism); Express Newspapers v. Keys, [1980] I.R.L.R. 247 (deeming incitements to breach contracts as unprotected by freedom of expression); Islamic Unity Convention v. Independent Broadcasting Authority and Others, (2002) 5 B.C.L.R. 433 (CC) (stating that incitements to hatred likely to cause harm may be punished consistently with freedom of expression). And speech that

One might attempt to pare down the reach of the ScanlonZStrauss principle by exempting intentional (knowing or reckless) defamations and misrepresen­tations from its scope. For after all, if both the government and the speaker believe that X is false, it is difficult to see how we, as audience, have any inter­est in being told, with apparent sincerity, that X is true, especially when treating X as true will turn out to be harmful to others or to ourselves.³⁷

Eliminating intentional misrepresentations does go some distance in reduc­ing the reach of the ScanlonZStrauss principle. However, the principle still inval­idates lots of laws. If Barney really believes that his quack potion cures cancer, then the Scanlon/Strauss principle protects his right to convince Betty to take it, even if government is correctly convinced that the remedy is useless and dan­gerous. Moreover, ordinary criminal solicitations remain protected expression under the Scanlon/Strauss principle.

One might attempt further to pare down the Scanlon/Strauss principle by eliminating from its purview assertions of “facts.”³⁸ Facts in the free expres­sion context are frequently contrasted with opinions, on the one hand, and with values, on the other hand. But neither the fact-opinion nor the fact-value dis­tinction seems capable of producing a tenable and robust domain of freedom of expression.

The fact-opinion distinction, which surfaces occasionally in the Supreme Court’s defamation jurisprudence,³⁹ is particularly problematic.⁴⁰ Almost any statement of opinion can be viewed as an assertion of facts of various types. If Bob says, “In my opinion, Brenda is a drunk,” rather than merely “Brenda is a drunk,” his “opinion” can be interpreted as implying the existence of specific

advocates disobedience or disloyalty, if intended to give aid and comfort to an enemy, can constitute the crime of treason, a doctrine in some tension with Brandenburg.

³⁶ Strauss does in fact exclude lying because lying, not its suppression, is paradigmatically a violation of autonomy. Strauss, supra note 32, at 335, 366-7. See also Michael Robertson, “Principle, Pragmatism, and Paralysis: Stanley Fish on Free Speech,” 16 Canad. J. Law & Jurisp. 287, 305 (2003).

³⁷ Both Scanlon and Strauss make such a move. Scanlon excludes statements of fact that cause harm only by enabling the audience to carry out what it otherwise has decided to do on other grounds - statements such as giving out the combination of the bank’s vault or the formula for making nerve gas out of common household ingredients. Scanlon, supra note 31, at 211-13.

Strauss, on the other hand, at times appears to be excluding all - but only - false factual statements from protection by his principle, although his principle might appear to exclude only intentionally false statements of fact. See, e.g., Strauss, supra note 32, at 366, for an example of Strauss’s ambiguity on this point.

It should be noted that there are a number oflaws in the United States that are meant to deprive people of information solely because the government fears that they will use the information in harmful ways. For example, employers are routinely forbidden from gathering information that might enable them to discriminate against employees or prospective employees on various bases.

³⁸ See Milkovich v. Lorain Journal Co., 497 U.S. 1 (1990) (discussing the fact-opinion distinction and its relevance to the constitutionality of defamation liability).

³⁹ See, e.g., FrederickF. Schauer, Language, Truth, and the FirstAmendme nt: An Essay inMemory of Harry Canter, 64 Va. L. Rev. 263, 276-81 (1978). but unmentioned facts (what he has observed or heard, what inferences are warranted from the evidence he possesses, and so on). On the other hand, if Bob lays bare all of these facts and then gives his “opinion” in addition, his opinion adds nothing. It is surely harmless, but for that very reason, no government would care about suppressing it. Therefore, if government fears harm from expressions of “opinion,” it must be for the reason that those expressions contain implicit assertions of facts that the audience might believe to the detriment of the audience itself or others.

I realize that my debunking the fact-opinion distinction will be controversial, so I ask the reader to engage in a thought experiment. Imagine that someone reveals to you all the facts he knows about a third person. (All the facts includes such things as background assumptions about the world, reasoning methodolo­gies, and so forth.) Assume further that all the facts are true. Finally, the person states an opinion about the third person that is defamatory. However, because the facts and reasoning methodology - or at least that much of the reasoning methodology that you agree with - do not support the defamatory conclusion, you will not draw that conclusion yourself despite its assertion by the speaker. All of the persuasive power lies in the facts and reasoning methodology, and they do not lead you to the defamatory conclusion. Bare opinions are harmless because they are inert. If they are harmful, it is because they imply unstated facts.

The same problem infects the distinction between assertions of fact and assertions of value. To the extent that government fears the latter, it does so because it fears the audience may be persuaded to adopt the values asserted - again, to the audience’s or others’ detriment. But to be persuaded by an assertion of value requires assigning that expression the status of something capable of persuading. If expressions of value had the status of, say, burps, they could not persuade an audience to adopt the values so expressed. One does not have to be a realist regarding the ontology of values to regard expressions of value to be heavily larded with implicit factual assertions. And surely it is these implicit “is” assertions in the assertions of “ought” that the government fears will persuade and thereby result in harm.

It is just not true, as some freedom of expression proponents contend, that government has no legitimate interest in suppressing statements of value that it fears will lead to harms that it has legitimate reasons to prevent. If the govern­ment has legitimate interests in suppressing statements of fact that are incorrect and that if believed may induce harmful acts, then it has the same legitimate interests in suppressing statements of value that it regards as dangerous for the same reason. A statement asserting an incorrect value must at some level be mistaken factually or imply a fact that does not exist (such as that the speaker has some special moral insight). When all the relevant facts about the world and the speaker are disclosed to the audience, the speaker’s assertions of value will themselves be inert and thus of no interest to the government.

The United States Supreme Court has at times made a distinction that seems more accurately to capture what lies behind the false distinctions of fact-opinion and fact-value. Thus, in Milkovich v. Lorain Journal Co.,^[121] the Court referred to provaby false factual assertions and implications as opposed to factual asser­tions and implications the falsity of which could not be proven.^[122] The Court gave as an example of the latter the statement “In my opinion Mayor Jones shows his abysmal ignorance by accepting the teachings of Marx and Engels.”^[123] Pre­sumably, the implication the truth of which is unprovable is that the teachings of Marx and Engels are false. On the other hand, “Jones is a liar,” if false, is provably so in the Court’s view.

Again, however, I think this distinction will not do the work intended. Some of the facts that are the premises of ordinary defamation and misrepresentation suits are routinely established to a lesser degree of certainty than are the facts on which assertions about the falsity of Marxism or the benefits of free trade rest. It is true that certainty about political and philosophical claims is almost never achieved, while certainty about mundane facts such as whether an airplane crashed or water is composed of hydrogen and oxygen is much more common. But many factual assertions that might be subject to governmental proscription - take for instance medical claims - are much less certain than assertions about, say, the desirability of a centrally-planned economy. No one’s liberty should be curtailed unless the possible harms to others, discounted by the probability of their occurrence - minus the benefits that can justify such harms, discounted by the probability of their occurrence - justify the restriction. Probabilities are probabilities, however, and the probability that laetrile will cure cancer is in principle no different from the probability that Marxism has got things right.

It might be useful at this point to mention David Brink’s recent interpretation of John Stuart Mill’s arguments for freedom of expression.^[124] Mill gives four arguments against message censorship:^[125] (1) censored opinions might be true;^[126] (2) even if literally false, they might be partially true;^[127] (3) even if false, they may prevent true opinions from becoming dogma;^[128] (4) true opinions held dogmatically lose their meaning.^[129] Brink is skeptical about the strength of the first two claims.^[130] Whether censorship will impede the search for truth or further it depends on the ability and motivation of the censor and myriad other factors; it is not preordained that without censorship, truth will vanquish falsehood, even in the long term. And in the short term in which we dwell, false opinions may cause worse harms than the loss of some truths.

Brink, however, finds Mill’s arguments against censorship based on prevent­ing true opinions from becoming dogmatic more powerful.^[131] Censorship, even if it targets only false ideas, preempts the exercise of our deliberative capacities. And development of deliberative capacities is important to human happiness.

Brink does not believe that Mill’s argument supports a total ban on censor­ship. Some ideas are not worth deliberating, and some harms caused by false beliefs outweigh the setback to development of deliberative capacities.^[132]

Mill’s arguments, as elaborated by Brink, might provide a rationale for distin­guishing among suppression of false messages regarding values. For example, when one states that the existing great inequality of income justifies the over­throw of capitalism by force, the statement contains many “factual” claims, some explicit, most implicit. Such claims include: (1) there exists great in­equality of income; (2) capitalism causes it; (3) inequality is a greater injustice than the use of force required to eliminate it; (4) when a greater injustice can be eliminated by means of a lesser injustice, it is right to eliminate the greater in­justice. The first two claims could be considered straightforwardly factual. The last two are evaluative. If they contain implicit factual claims, perhaps those are the type about which citizens must be able to deliberate. Such implicit factual claims may be that Scripture supports the claim, that Scripture is an inerrant moral guide, and so forth.

Still, I doubt that it will be possible to justify protecting some factual claims - those implicit in evaluative claims - without protecting others. Are we better able to deliberate about the inerrancy of Scripture than the causes of income inequality? I see little hope in any wholesale distinction between false factual and false value claims.^[133]

Moreover, suppose government suppresses the doomsday speech of, say, those who oppose genetically-engineered foods on the ground that the speech consists of (unintentionally) false and misleading statements of fact that if believed by the public will impair the public’s ability to deliberate rationally and thus its autonomy, and lead it to choose courses of action that cause harms (more expensive, less pest-resistant foods; violations of trade treaties; starvation in the Third World). Would we not consider such suppression a paradigmatic violation of the right of freedom of expression? If we say that our verdict depends on the truth about genetically-engineered foods, have we not given away the game?

Indeed, is not the prohibition of libeling the government - making false claims about its policies and personnel - the paradigm of a violation of free­dom of expression?⁵⁴ And does not the libel consist of factual assertions, not assertions of value? On the other hand, if the claims are false or misleading and capable of causing great harm, how are they to be distinguished from other false or misleading and dangerous factual assertions?⁵⁵

justify restraint regarding potentially harmful messages than they justify restraint regarding potentially harmful conduct. See Robertson, supra note 37, at 311.

⁵⁴ Is that not the reason the Alien and Sedition Act is regarded as a central violation of the First Amendment despite their almost contemporaneous enactments? TheSedition Act of 1798, which made libeling the government a crime, also provided that truth was a defense. Supporters of the Act’s constitutionality expressed the belief that such a defense was sufficient to make the Act constitutional. Opponents believed that the Act would chill valuable speech because the truth of “opinions” could not be proved. See Leonard W. Levy, Legacy of Suppression (1960), 258-63. The latter view emerged victorious in 1964 when the Supreme Court found ordinary defamation law violative of free speech when applied to protect public officials. New York Times Co. V. Sullivan, supra note 36. See generally Harry Kalven, Jr., The New York Times Case: A Note on “the Central Meaning of the First Amendment,” 1964 Sup. Ct. Rev. 191, 204-7.

⁵⁵ Indeed, countless factual assertions, though literally true, can create misimpressions in the minds of particular audiences that will lead to harm to others or to themselves. A truthful statement that a celebrity constantly switches lovers or uses drugs may suggest falsely that promiscuity or drug use is good even if it does not literally make such claims. Indeed, a true assertion that the government has mismanaged some affair may lead people to believe that replacing the government would be beneficial when such a belief is unwarranted - which is why even good governments fear truthful but bad publicity. An omniscient governor, whose primary concern were that people possess true beliefs, might well censor lots of literally true factual assertions.

Consider in this regard the goal motivating various campaign finance laws - and some people’s pet policies regarding broadcast regulation - of equalizing the influence of various positions on issues. Putting aside such metaphysical imponderables as how “issues” and “po­sitions on issues” are identified and individuated, the core question is what justifies or pur­ports to justify such regulatory proposals. And the answer to that question appears to be pre­venting people from being led to hold mistaken views that they might ultimately act upon to their detriment. The idea is that not hearing the correct balance of views will create the misleading impression that those views heard more often because better financed are the cor­rect views. In other words, inequality in the number of times opposing views are presented is tantamount to making false or misleading assertions. (Notably, no one proposes equalizing rhetorical skill and other intangible facts that make assertions of views more or less persuasive; nor does anyone propose measures to make sure that each member of the audience hears both sides.)

If neither the fact-opinion nor the fact-value distinction carves out a robust, but not too robust, realm of freedom of expression in cases where government fears the expression will cause harm in two steps, are there any other distinctions that might do the trick? For example, can and should we not distinguish speech that will cause harm before it can be counteracted by opposing speech from speech that can be effectively so countered? One occasionally sees this asserted to justify the Supreme Court’s distinction between “mere advocacy” of harmful acts and “incitement” to such acts.⁵⁶

This distinction, however, also fails to withstand close analysis. In ordinary, everyday examples of criminal solicitation - which no one seems to regard as protected by a right of freedom of expression - the speaker is soliciting an undercover police officer who needs no counterspeech to be persuaded not to carry out the crime.⁵⁷ On the other hand, abstract advocacy of, say, violent overthrow of all capitalist governments may reach audiences who are beyond the reach of counterspeech, or who are impervious to it, and so forth. The advocacy may be open and notorious, or it may be transmitted to only a few people. Effectively countering such advocacy might require large resource allocations - to the detriment of other rights - which government may understandably not wish to make, especially given that it regards what is advocated as wrong and worthless. And considering every argument pro and con, of which there may be an indefinite number, will outstrip the always finite resources of time and attention.⁵⁸

So if speech that can be effectively countered by opposing speech defines the realm of freedom of expression, that realm may be quite small, depending upon what resources we believe government must commit to making counterspeech effective. And because withdrawing resources from other goals in order to devote them to countering advocacy of harms is itself harmful (to those goals), the government is put to committing some harms in order to prevent others. Or at least this is so if the right of freedom of expression requires it to employ only counterspeech, and not prohibitions, to combat advocacy of harms. But that means that advocacy of harm, if not prohibited, will be harmful one way or another. It will either be harmful by forcing government to devote resources to counterspeech - which even then might fail to be totally effective. Or it will be harmful by persuading the audience to engage in harmful acts.

This line of thought suggests that laws criminalizing corrupting vices such as gambling and drug use might fall within the scope of freedom of expression if their rationale were that the proscribed vices lead people to hold incorrect beliefs about what is valuable - i.e., that they corrupt people through the mechanism of inculcating false beliefs.

⁵⁶ See Brandenburg v. Ohio, supra note 30 (distinguishing constitutionally protected “advocacy” from constitutionally unprotected “incitement”).

⁵⁷ Strauss recognizes that ordinary criminal solicitations fall within his principle. See Strauss, supra note 32, at 338-9.

⁵⁸ As Strauss recognizes. See id. at 346-7, 367.

Brandenburg v. Ohio^[134] sets forth a test for when advocacy of harmful acts can be punished without violation of the First Amendment, namely, “where such advocacy is directed to inciting or producing imminent lawless action and is likely to incite or produce such action.”^[135] The test is unsatisfying for several reasons. First, it suggests the speaker’s intent is material to the protectability of the expression (“is directed to inciting”). Although the speaker’s intent may be relevant to the culpability of the speaker, it is difficult to see why speaker culpability should offset the status of the expression with respect to freedom of expression. Suppose a speaker on a television show is aware that certain words in his prepared text will be interpreted by a group within the audience as a code calling for them to blow up a building. The speaker does not intend for such an act to occur and regrets that it will, but he refuses to refrain from giving his prepared text. If the government could legitimately suppress the speech if the speaker intended the harm - say, by blacking out the program - it is difficult to understand why government could not do so in the situation described. The speaker’s state of mind might be relevant to concerns about “chilling effects” of sanctions, but it appears irrelevant to the protectability of the expression itself.^[136] (Likewise, it is surely immaterial whether the speech uses words that literally “advocate” as opposed to words that are interpreted as advocacy.)

Second, the imminence requirement is not only vague but is also of doubtful relevance to freedom of expression. I have already argued that speech advocating harm frequently cannot be effectively met by counterspeech. To make this point more concretely, consider a shipment of pamphlets that the government has interdicted. They bear a message that is literally innocuous but that the government is sure is a code designed to activate a sleeper cell of terrorists and have them blow up buildings at some indefinite future time within a year. Is the lawless action “imminent” within the meaning of the Brandenburg test? If so, the imminence requirement is toothless. But if not, and the government must allow the pamphlets to be distributed, then the imminence requirement is pernicious. Although the government may have sufficient time to ferret out the sleeper cell, and may have the resources to do so, it may fail in that endeavor or, if it succeeds, it may sacrifice other values in the process. (The resources consumed in finding the terrorists might have been used to avert other crimes or to fund health care, disaster relief, and education.)

We might combine these first two points and consider the publication of information about how to build an atom bomb in your bath tub, or how to poison a city’s water supply.^[137] The article uses no words specifically advocating harmful acts. Indeed, the author is merely trying to be informative and would detest anyone who used his article for nefarious ends. Finally, although the chances of someone so using the information are quite high, when that use will occur is completely uncertain. In any event, the earliest one could wreak harm based on the information would be several months after publication. The article would therefore not be advocacy; it would not be directed to inciting; and the lawless action it would almost certainly trigger would not be imminent. Yet surely there is as much justification for suppressing this article as there is for suppressing ordinary criminal solicitations.

Finally, consider the requirement that the advocacy be “likely to incite or produc[e] such action.” Likelihood of harm does seem relevant to whether the speech is protected by freedom of expression. As I said, however, ordinary crim­inal solicitations are frequently deemed punishable even when they are quite unlikely to cause harm - as, for example, when the solicitee is an undercover police officer.^[138]

One final way to carve up the domain of speech that causes harm in two steps is to distinguish, not false factual assertions from assertions of incorrect opinions or values, but false factual assertions that we hold the audience re­sponsible for believing, and hence acting upon, and those that we hold only the speaker responsible for making, not the audience for believing. Thus, we might deem the audience faultless for believing the National Enquirer’s proclama­tion that “Hilary is having an affair” but fully responsible for believing an Al Quaeda operative’s assertion that suicide bombing is righteous. In the former case, because government cannot legitimately penalize those who believe the story about Hilary and act toward her accordingly, it may attempt to prevent the harm to Hilary by penalizing the National Enquirer. In the latter case, govern­ment can legitimately hold the audience responsible for acquiring false beliefs about the morality of terrorism, and thus it may attempt to prevent terrorism by punishing those who act on such false beliefs, but not those who instill the false beliefs.

Yet this distinction seems to me just as unjustified as the prior ones. Why is an audience more responsible for believing the false assertions that would justify terrorism than the false assertions about Hilary’s love life? “Globaliza­tion impoverishes the Third World and should be resisted by force” combines assertions of false facts (about the actual wealth effects of globalization) with assertions of plausible facts (that movements that threaten the poor with even greater poverty are to be resisted, by force if necessary). And the false as­sertions may be more difficult for the average audience to unmask than false defamatory assertions like those about Hilary.⁶⁴1 can see no defense in terms of

⁶⁴ Consider in this regard a scenario in which A maliciously tells gullible B a convincing tale to the effect that C is about to detonate an atomic device in New York City and needs to be killed to save thousands of innocent lives. If B kills C, A will be deemed the murderer, acting through B as A’s “innocent instumentality.” See Model Penal Code, § 2.06 (2)(a). Compare that scenario to one in which Iman A preaches to gullible followers Bs that Cs, Americans, must be killed to prevent a Christian conquest of Islam. The Scanlon-Strauss principle arguably should apply to A the same way in both scenarios. And I am sure that neither Scanlon nor Strauss would insulate A from liability in the first one. And what would their position be with respect to doctors who recommend illegal drugs such as marijuana to their patients? See, e.g., Conant v. Walters, 309 F. 3d 629 (9th Cir. 2002).

Jim Wernstein, in private conversation, argued that in a democracy we must, as a matter of respect, treat citizens as trustworthy when it comes to assessing false and deceptive political speech, though not when it comes to false or deceptive commercial, medical, or legal claims. But surely people are no better at assessing the former than the latter. Suppose Michael Moore comes out with a “documentary” film making scurrilous and false allegations on the eve of an election. Is it disrespectful of the audience to punish the speaker for such falsehoods? For the audience to check out the accuracy of Moore’s claims would take time, resources, and attention that might be unavailable or that could be devoted to other worthwhile projects. And even with time, resources, and attention, the audience may lack the expertise to identify the central flaws in Moore’s claims. It surely seems no more disrespectful of the audience to protect it from being misled by Moore than to protect it from being misled by false medical claims, false claims about products, or false claims about bridge safety.

Of course, we want government to treat us as autonomous. But autonomous decisionmaking requires both information and the ability to process it rationally; and information and rationality vary from person to person and are always matters of more or less. We are never fully informed or fully rational. For these reasons, autonomy is a paradoxical and Janus-faced value that usually appears on both sides of moral dilemmas. If government protects us from misleading messages, it threatens our autonomy by arrogating to itself decisions that we might wish to make for ourselves. But government at the same time enhances our autonomy by removing the threat to that autonomy posed by misinformation. If we can, consistent with our status as autonomous agents, delegate to government the “soft” paternalistic power to protect us from false and misleading claims about quack medicines and defective bridges, it is a short and smooth path to delegating to government the “hard?” paternalistic power to protect us from false public policy and moral claims. There is no reason to think that unaided, we are better equipped to assess the latter than the former. In all of these cases, we may lack the necessary information and critical faculties required to avoid harm to ourselves or others. And see Susan Hurley, “Imitation, Media Violence, and Freedom of Speech, ”117 Philosophical Studies 165 (2004); Caroline West, “The Free Speech ArgumentAgainst Pornography, ” 33 Canad. J. Phil. 391 (2003) (both arguing that certain messages render the audience less capable of rationally assessing certain ideas).

audience responsibility for exempting on freedom of expression grounds those who advocate false values, but not those who defame or misrepresent facts.^[139]

So far, then, my conclusion is that the domain of speech that causes harm in two steps cannot be divided into regulable and nonregulable speech by the fact-opinion / fact-value, counterable-by-speech / not-counterable-by-speech, or audience-responsible / not-responsible-for-false-beliefs distinctions.

I am equally dubious about other suggestions for dividing up into regulable and nonregulable speech that causes harm in two steps. Some, for example, exempt incitement and solicitation of harmful acts from the scope of the right of freedom of expression by deeming incitement and solicitation to be so inter­twined with the harmful acts they incite that they should be deemed harmful acts themselves, not expression.^[140] Others would exempt incitement and solicitation by calling them, but not abstract advocacy of harmful acts, “situation-altering,” by analogy to speech-acts like promising or taking marriage vows.^[141] The idea is that like these other speech acts, incitements and solicitations change the nor­mative situation of their audience and thus are more akin to acts that tangibly alter the environment than to mere speech. Still others place incitement and solicitation of harmful acts on the harmful act rather than the expression side of the ledger by claiming that they trigger action through bypassing rational deliberation and evaluation.^[142]

With all respect, I believe that none of these attempts to sever incitement and solicitation from advocacy succeeds. Incitement and solicitation are no more or less “acts” than any other speech that may cause a harmful response.^[143] Nor do they alter the normative situation except to the extent that the audience’s values and receptivity to the message allow them to do so, which is true for all expression. And they bypass audience deliberation and evaluation, again, only if the audience permits them to do so. One may deliberate about and evaluate a request to commit a murder to no less extent than one may do so regarding the proposition that capitalism should be overthrown by force and violence.

Finally, Frederick Schauer has drawn attention to the close parallel between preventing legally and morally responsible persons from receiving harmful ideas - ideas that we believe are both wrong and, if acted upon, harmful, such as the idea that it is okay to torture animals for fun - and preventing legally and morally responsible persons from engaging in activities that are in themselves harmless but that raise the probabilities of the audience’s committing wrong and harmful acts.^[144] For example, we might prevent legally and morally responsible persons from possessing burglar tools, hand grenades, assault weapons, or other items solely on the ground that possession increases the chance that if such per­sons choose to act wrongly, they will cause great harm. If we can preemptively disable responsible adults from choosing to commit harmful acts by depriving them of means that are harmless if not misused, why, Schauer asks, can we not preemptively disable them by depriving them of harmful ideas? After all, the probability that receiving the message that torturing animals for fun is okay will lead some members of the audience to torture animals may be as high or higher than the probability that possessing, say, an assault weapon will induce the possessor to murder someone with it. And the autonomy of the audience to assess the message regarding torturing animals - the pivotal consideration for Scanlon and Strauss - is, says Schauer, no different from the autonomy of the possessor of the assault weapon to assess whether or not it should be used to kill.

III.