HARM AND OFFENCE

The need to minimize legal constraints is suggested by Joel Feinberg's durable “generic contrast between harmful conditions and all the various unhappy and unwanted physical and mental states that are not states of harm in themselves,” a distinction that “can be made out in quite parallel ways for hurts, offences [and other, miscellaneous forms of unpleasantness].”⁷

The distinction lends itself to discriminating between using words hurtfully and inflicting harms that warrant legal punishment. When it is neglected, the boundaries of politically acceptable behavior become unnecessarily confused.

In trying to define the moral limits of the criminal law, Feinberg identifies several “coercion-legitimizing principles” of which the two most important are the “harm principle,” sanctioning measures effective in preventing harm to others, and the “offence principle,” justifying prohibitions against offending others. Feinberg endorses both. He defines “the liberal position” as holding that these two principles exhaust the acceptable reasons for criminal prohibitions and distinguishes “the extreme liberal position” as the view that only the harm principle states a good reason for such prohibitions.⁸

While Feinberg views harm and offence as distinct categories that provide separate grounds for legal intervention, the extreme position has much to recommend it. As we note below, appeals to offensiveness are inherently more disputable than ascriptions of harm, the most important offences can be subsumed under the harm principle, and that principle is sufficient to identify the freedoms of expression that deserve protection.⁹ We do not disallow the appeal of offence in all social contexts, but we do maintain that clarity is served by defining responsibilities for hateful speech on the relatively firm and narrow foundation of the harm principle. Hurtful and offensive speech acts provide the most compelling reason for legal coercion when they result in wrongful harms.

A great strength of the harm principle is that harm has fundamental importance in moral judgment, since it diminishes one's capacity for responsible action. Generically, we understand harm as damage, including impairment and disability.¹⁰

An instrument is damaged when its function is impaired. In the case of human beings, harm impairs freedom of agency, rendering the person harmed unable to carry on as well as would have been possible had no disability been created.¹¹ If harm is wrongful and the responsibility of another person, then it should be redressed, for moral agents need opportunities to act well or badly. Even slaves and members of rigid communities normally have these options. Moral indignation and other sentiments of justice are thus appropriate responses whenever actions occur that undermine the conditions of responsible behavior. In short, wrongful harm is morally offensive.

While moral requirements refer generally to the conditions of responsible behavior, in a decidedly liberal society justice requires protecting independent agents' ability to pursue self-determined projects that comply with the constraints of the harm principle rather than simply enabling fulfillment of responsibilities within communal roles. Hence offence, hurt, and the like are harmful to members of a liberal society when they adversely affect their autonomous objectives.¹²

Harm has occurred when bodily injuries or psychological damage restrict one's ability to act as one otherwise could have acted, interfering with pursuit of one's interests. Affronts to dignity and hurtful speech can have some of these effects, but this is not to say that pains and attempts to pain are themselves harms. In contrast to forms of unpleasantness and adversity that are spurs to useful action, pain must restrict one's capacities in order to justify that equation. When the black mayor of Chicago was falsely and cruelly depicted wearing nothing but frilly underwear, a group of aldermen were moved to take the painting from the exhibition, but in general, “Injured sensibilities do not constitute a right to remove the agency of injury.”¹³

The doctrine of liberty protects expression but can accept that, if others are responsible for speech-related damage to a person's ability to pursue her personal projects, then the harm done can deserve legal remedy. The doctrine thus protects autonomy as a component of moral responsibility in a liberal society, but the defense of free expression is based on the harm principle, not on a principle of autonomy. In this respect, our argument is consistent with Susan Brison's exploration of a number of versions of the autonomy defense, all of which she finds wanting. We also cautiously agree with her that speech is not “so special as to be worthy of protection, even when it is conceded to cause … harms which, if brought about by any other means, would be considered unjust and legitimately preventable by governmental intervention.”¹⁴

Caution is required, however, lest this line of argument extend actionable harms to a point where freedoms of speech are unduly compromised by conflating the injurious consequences of physical violence with the more cognitively complex effects of distressing, upsetting, humiliating, degrading, and offensive speech.

The strength of the harm principle is accentuated by an ambiguity in the offence principle. “Offence” may refer either to offensive actions or to offending actions as a justification for regulating behavior. That is, it may apply either to actions deemed deserving of offence through appeal to moral principle or to actions that merely offend individuals without sufficient reason for that response. We take it as established that merely offending actions cannot normally overcome liberties of expression, but identifying offensiveness by reference to principles separate from harm is unlikely to succeed. The contrary view is that acts of racist and sexist speech are offensive and subject to regulation because they constitute forms of moral subordination and are thus unacceptable on grounds of social equality.¹⁵

It is then their violation of human dignity that makes them worthy of offence, justifying restriction of expressive actions that diminish a person's moral status even if no one is offended. This line of thinking lends itself to an extreme Kantian alternative to Mill's principle that “the only purpose for which power can be rightfully exercised over any member of a civilised community … is to prevent harm to others.”¹⁶ Meir Dan-Cohen argues that in a liberal society it is not the harm principle that should operate, but a dignity principle, “the view that the main goal of the criminal law is to defend the unique moral worth of every human being.”¹⁷ We nevertheless favor regarding the traditional interpretation of Mill as canonical for liberal societies. The moral offensiveness of harmful actions has already been recognized, and the addition of a further principle of political offensiveness poses serious problems of interpretation and scope. Our emphasis resembles that placed by David Brink on Millian principles, but the harm principle will remain basic through the course of our defense of expressive freedoms. Just as that defense takes seriously but does not rest upon a principle of autonomy, it takes seriously but does not rest upon principles of equality or deliberative values.¹⁸ We will return to this point at the end of the discussion, distinguishing our approach to hateful speech from Brink's.

It is difficult to avoid the harm principle. Dan-Cohen's objective is aptly restated as the wish to avoid harms to dignity. How the state might carry out this task without simply protecting persons from harm is unclear, though. Many international covenants and conventions mention the inherent dignity of human beings in their preambles but become largely silent about dignity as soon as matters of application arise. That is easy to understand when, in issue after issue—birth control, euthanasia, abusive speech—the importance of human dignity is appealed to on either side, with claims and counterclaims effectively challenging one another and undermining the capacity of dignity to advance fundamental political debate.¹⁹

Moreover, as a matter of simple logic, the argument from equal dignity can be deployed by abusive speakers as well as by their targets. These problems do not arise when moral responsibility for harm to others is coupled with epistemic responsibility for responding to indignities that should fall short of harm. We will try to explain and establish this point by revisiting a famous counterlibertarian argument.

In rejecting the proposition that “harmful speech should be answered not by regulation but by more speech,” Stanley Fish asserted that “that would make sense only if the effects of speech could be cancelled out by additional speech, only if the pain and humiliation caused by racial or religious epithets could be ameliorated by saying something like ‘So's your old man.’”²⁰

While insults can be more unpleasant than injuries, the capacity to trade gibes does not imply debilitating pain. In general, therefore, hateful speech does not preempt further speech, whereas conflating pain and harm simply begs the question in favor of political regulation. Conflating harm and humiliation can do this as well, while leaving unaddressed the fact that humiliation is subject to tests of reasonableness. It is this fact that explains the rhetorical force of the familiar question: “Are the effects of speech and of physical action really indistinguishable? One might wish to grant that we can be hurt by certain words and that such words cause bruises, but are such psychic bruises indistinguishable from bodily ones?”²¹ As the discrete chapters on the liberty of expression and action in Mills's On Liberty suggest, they are not. The distinctiveness of psychic bruises can be displayed analytically through forms of distress that inherently contain judgments, such as the feelings of humiliation, fear, guilt, and grief. All of these painful feelings include characteristic beliefs—for example, the belief that one is unworthy, in the case of humiliation, and the belief that one is endangered, in the case of fear.

Judith Thomson makes a well-known distinction between belief-mediated distress and non-belief-mediated distress, but it is too broad.²²

The nature and scope of belief-mediation as she describes it do not distinguish mental distress from physical distress whose severity is determined by one's beliefs about its importance. A broken finger will likely have more significance for a professional pianist than for a retired poker player. Her distinction also fails to show why anyone is more responsible for avoidable mental injury than for avoidable physical injury. The fact of mental mediation is nevertheless important when it is narrowed to forms of mental distress that entail mutual responsibility. Our examples provide this focus by identifying distinctively emotional forms of distress, where by “emotional” we refer to affective states that include beliefs inherent to them. These distressing states are inherently fit matters for discussion. Argument may show that fear or humiliation are on some occasions reasonable but on others unwarranted. By contrast, the pain of physical injury does not include characteristic judgments in this way. Although physical distress may be accentuated by a variety of beliefs contingently related to it, it is not inherently susceptible to change of belief, as when a person in fear finds that there was no danger after all. Disabusing someone of a threat should result in the disappearance of potentially damaging distress, but simply correcting one's beliefs does not similarly protect against physical injury.

The crucially different relationship of beliefs to physical and emotional distress is instructively illustrated by emotions that have become unresponsive to evidence. As the result of traumatic abuse, one may exhibit fears that persist even when one realizes intellectually that no danger exists. In this case, one has clearly suffered injury, and it may be solely the responsibility of other persons, but such contradictory emotions defy epistemic rationality in a way that physical injury and distress cannot. The contrast accentuates the point that the form of belief-mediation typical of rational fear, humiliation, and hatred must be encouraged if individuals are to be psychologically sound and their societies coherent.

The feature shared by physical and psychological harm is the impairment of capacity. It leaves intact the important difference expressed imprecisely in the distinction between words and deeds. In suffering physical harm at the hand of others or being damaged by theft or libel, one has been deprived of something for which loss one bears no essential responsibility. The injury occurs independently of one's interpretation of another's actions. Even if one has exposed oneself or one's property to injury for want of observing adequate precautions, it is another person alone whose actions have resulted in the loss. In the matter of the emotional harms of speech, however, one's judgment of events is normally part of the full story of responsibility for the injury. It is not generally necessary to accept the viewpoint of the person who utters humiliating words. While the object of hateful speech is an unwilling victim in such cases, rational persons do have resources that are not available when suffering direct harm to one's body, property, or reputation. The capacity for epistemic responsibility is central to the issues arising at the intersection of the harms and liberties of speech.

The view that people should be held no more accountable for the harm they suffer from verbal assaults than from physical injuries discounts epistemic responsibility. In a liberal society this is to undermine capacities that make its liberties admirable. In the case of liberties of speech, these capacities include the responsibility of citizens for their beliefs, the alternative being government by those who know better or the rule of the strongest. One of the ironies of this responsibility is that some persons suffer from those who fail to cultivate it well, but there is no escaping misuses of liberty in a free society. Rather, certain harms inflicted upon some persons by others become a shared responsibility. The principal difference between the harms does not lie in what the victims could have done to protect themselves from physical and verbal assault, since we would then hold equally responsible those who suffer physical injury for failure to learn self-defense and those who suffer emotional disability for failure to learn that racial invective is false. We do not do that because of the discursive capacity to avoid the damaging effects of verbal actions.

Freedom of speech is thus connected with a presumption of responsible belief. Moral agents often have a capacity to avoid the harms of speech, with implications for their accountability and that of the state. Developing this case convincingly will depend upon defending it from the charge of expressing a viewpoint that enables the powerful and privileged “to dismiss conduct they could not imagine being harmed by as being simply a matter of the construction the alleged victims chose to put on it.”²³

Although it is not simple, the generic contrast between distress resulting from battery, theft, and libel and that resulting from emotional causes coincides with the need to distinguish the verbal assaults against which responsible agents have cognitive defenses from the physical attacks against which one should not have to carry weapons or practice karate.

Of course, presumptions of reasonableness can be defeated in particular cases where differences between forms of distress have no practical importance. To this extent, complaints about disjoining verbal and physical action are justified. Protecting freedoms convincingly depends upon renouncing absolutism in this respect. We must recognize the harm that speech can do and be wary of the general strategy that attempts to minimize the adverse consequences of free expression. This strategy discounts the effects of unwelcome speech upon those who experience them and thereby attempts to remove issues of speech from ongoing political debate.²⁴

Rather than silence this discussion, we will weave two further presumptions into our exploration of the emotional pains that are potentially debilitating effects of tolerating all discourses. The presumption of multiple responsibilities notes that the censure of a community may be effective where the force of law is not appropriate, making it fallacious to suppose that what communities may rightfully condemn, the state may condemn as well.²⁵ The presumption of enablement over control notes that it is generally preferable for the authority of the state to be otherwise expressed than in coercion and prohibition, so that attempting to remedy problems has priority over suppressing their agents.

PAINFUL KNOWLEDGE

The importance of ascribing moral responsibility for harmful speech cautiously and recognizing that epistemic responsibility can mitigate the effects of offensive messages is nicely illustrated by a set of disagreements between Wayne Sumner and Joseph Raz. Without speaking directly to one another, they provide an instructive study of antithetical positions on freedoms of expression. Sumner is able to support significant regulation of speech by conflating the power to harm and the ability to offend, identifying toleration of expression with a form of approval and equating the political and the social order. Raz carefully avoids all such reductive moves. Their contrasting views are useful in another way as well. Sumner has since repudiated his conclusions as the expression of “pretheoretical convictions” that have not survived his philosophical scrutiny.²⁶

As urged below, we should hesitate to allow our thinking to be led astray by our hatred of hateful speech.

Sumner makes three suggestions important to an argument for criminalizing certain kinds of speech.²⁷

(1) “[T]he public expression of hatred or contempt for minorities can cause those groups demonstrable harm.” (2) “The immediate result for the target groups will be diminished self-esteem and a questioning of the extent to which they are valued members of the social order.” (3) To tolerate hateful speech “in the interest of freedom of expression would inevitably be to confer upon it a certain degree of legitimacy.” As the foundation of an argument for restricting freedom of expression, the philosophical case does not support its rhetorical expression, since, by reasonable standards of adequacy, all of these points are simply false.²⁸ The harm asserted in (1) is rarely demonstrable; the result described in (2), when it occurs, is not immediate; and the effect of toleration described in (3) is not inevitable.

Raz's contrasting positions bear out this judgment.²⁹

(1′) Of “content-based censorship or criminalization” he says that “the evil to be avoided is … a harm to … those whose interests are served by the condemned way of life.” A subtle argument supports this claim, but our point is only to reinforce the propositions that harm is a demanding criterion and that attributions of harmfulness can cut in unexpected ways. Insofar as wrongful harm is demonstrated, those subjected to censorship can rightly claim to be its victims on the grounds that they are made unable to pursue their self-determined projects. For (2′) “The very act of censorship insults the censored, denies their rationality, treats them as means rather than ends in themselves, etc.” Moreover, (3′) “Whereas prohibition … condemns … prohibited conduct, permission does not endorse what is permitted, it merely tolerates it.” Raz's claims require development. They do not show that those subjected to censorship are wrongfully insulted any more than those punished for physical assault are treated merely as means. Nonetheless, the tensions within each of these pairs of contrasting statements raise prima facie doubts about Sumner's assertions. It is a simple matter to accentuate these doubts without diminishing respect and concern for the targets of hateful speech or forgetting that those who speak abusively also have epistemic responsibilities.

(1) Sumner's thesis that public expressions of hatred for minorities can cause them demonstrable harm deserves as sympathetic a response as possible. He observes:

Presumably it is the neo-Nazis who are the thugs in this list of racists, but their particular disposition to violence makes it more difficult to connect biased attitudes generally with clear harm. Even the risks attendant upon racist propaganda do not seem to have been very thoroughly tested by social scientists. There is good evidence that psychological abuse leads to physical violence in close domestic relationships, but this causal pattern is not well confirmed for the sphere of relatively anonymous public relationships.³¹ The appeal to ostensible connections between verbal abuse and harm is also challenged by moral principles that assign responsibility to immediate agents of action rather than to their remoter antecedents in hateful attitudes and propaganda.³² Responses might be made to all of these challenges, but the additional burden of argument required bears out the fact that the connection between hateful speech and harm is less than compellingly demonstrated where it needs to be for Sumner's argument to be accepted as stated.³³

Demanding strict philosophical demonstrability here might seem inappropriate. Sumner's statement about harms to groups is made in the context of section 1 of the Canadian Charter of Rights and Freedoms, which subjects these rights and freedoms to “such reasonable limits prescribed by law as can be demonstrably justified in a free and democratic society.” Legal justification here is a matter of weighing the limitation of an important right against legislation whose purpose in this case is to “protect racial, ethnic, and religious minorities against the harms which are likely to result from the spread of contempt or enmity directed toward them.”³⁴

The task does not lend itself to logical demonstration but to judgments balanced between the values embodied in law and the facts of particular cases. Since case-based rather than universal reasoning can be at work, and notice can be taken by judges of legislative arguments that establish rational connections between hateful and harmful speech, it would be an error to insist upon an unduly strong reading of “demonstrable harm.” Causation in the law can be reasonably demonstrated within the context of particular legislation, if not beyond it.

Of course, identifying a legislative rationale falls well short of establishing that it is sufficiently well founded. Raz offers a criterion for sound legislation in this area, suggesting that it should recognize a distinction between public expression and expression by members of the public. In contrast to “authoritative condemnation of the views censored…, criticism, hostility or neglect by individuals or sections of the public express only their hostility or condemnation. They do not carry the authoritative voice of society.”³⁵

To be sure, it is not possible to draw this distinction with complete precision. Is there much difference between a uniformed police officer watching someone utter racial epithets and that police officer uttering them personally? In the former case, it may not be clear whether the official is condoning the behavior or simply tolerating it. Such boundary questions can be addressed, however, without implying that the objectionable aspects of the situation owe themselves to their harmfulness. A jurisdiction may refuse to legislate against racist remarks on grounds of freedom of expression but still insist that an officer not ignore such remarks because failure to object is deemed to constitute inappropriate public expression.³⁶ Since demonstrable harm is no part of this picture, Sumner's thesis remains incorrect.

(2) The difficulty of demonstrating the harmfulness of hateful speech is evident from the way in which Raz's point (2′) challenges the statement that “the immediate result for the target groups will be diminished self-esteem.” Just as official censorship insults the censored, possibly with good reason, it is independently open to the targets of hateful speech to view their abusers as dupes or fools. To suffer disabling psychic pain instead may require having already been subjected to a loss of self-esteem that has made one dependent upon recognition from unworthy others. Since one could in principle rely upon one's better judgment, there are prior events that render the harmful effects of expression upon self-esteem less than immediate. The original problem residing elsewhere than in expressions of hatred alone, it remains unclear that the problem can be effectively or appropriately addressed through restricting such expression. In short, where distress embeds beliefs, there is no immediate or necessary connection between suffering racial insults and losing the ability to trust others or becoming fearful in the presence of members of dominant social groups. The law need not therefore be invoked in one's defense, as could be reasonable if there were a straightforward, unmediated connection between other persons' utterances and one's self-esteem.

The division of responsibility for psychological distress recognizes the pains of demeaning speech but also includes a point of analysis implying that legally coercive measures are not the preferred mode of recognizing the claims of minorities and allowing their ways of life to flourish. Of course, some utterances and symbolic actions—such as threats of beatings or murder and cross-burnings—may merit proscription as crimes if they create fears of injury that reasonable people would find warranted. Credible threats are harmful because the fears they arouse impair one's self-determined objectives. Such fears may occur when bigots confront newcomers to pluralistic societies who have not learned to distinguish ethnic taunts and other ritual abuse from intentions to do harm. They clearly need social support, but (as we discuss further below) in this case the need—on both sides—is more educational than legal. In any event, the quiet dignity with which members of an abused group can ignore a jeering crowd reinforces the deeper point that while expressions of hatred can sometimes be taken as implied threats, reasonable fears of harm rarely follow immediately from an expression of hatred or hostility. Similar reasoning shows that there need be no immediate loss of self-esteem and no immediate questioning of the extent to which one is a valued member of the social order.

The capacity to reject unreasonable views and feigned threats is an important personal and cognitive resource and a barrier to the immediacy Sumner and others assert.³⁷

It is better to cultivate this capacity than for people to be vulnerable to harm through being unable to recognize unreasonable views for what they are. Harms of this kind belong to a larger class of cases in which the law should clearly decline interest in any event. Among these cases are those in which a reasonable interpretation of another's unbigoted statements may result in psychological injury—when, for example, one is rightly called a cheat or coward. Although Sumner does not discuss such cases, they suggest a qualified form of his second point: diminished self-esteem may occur immediately when it is the reasonable result of contemptuous speech for the object of that speech. Because harms resulting from such reasonable beliefs should not be addressed through criminal sanctions, though, the qualification does not advance Sumner's case. Rather, it only confirms that there is no general right to be spared emotional distress.

The implications of epistemic responsibility also extend beyond the humiliating knowledge of one's own failings to the painful knowledge that others unjustly despise you. As long as this knowledge is not menacingly conveyed, a wish to avoid it would not invite legal entitlement to remain in ignorance of others' hostile attitudes.³⁸

To the contrary, if there were a plausible case for legal prohibitions of such ignorance, then they would reasonably extend to polite discourse that conveys the same knowledge.³⁹ While a regime that tried to suppress civilly expressed contempt as well as overtly hateful speech would be faithful to a principle of equality, distinguishing between subtle disparagement and permissible disagreement would also call for unusual powers of discrimination whose exercise would often be contentious. A claim to be spared unwelcome knowledge is not one that merits general endorsement.

The law may still be available to address the effects of speech that causes injury by prejudicing third parties. These cases include the revelation of embarrassing information about another person, breaking confidences, and uttering demeaning racial remarks, all of which may undermine one's security. Whether instances of such types of action are legally actionable will depend upon a number of background facts about proprietary information, contractual rights, and actual dangers. Two conclusions are warranted. First, such cases entail no general claim upon remediation. Second, since the effects of the actions lie beyond both immediate harm and immediate loss of self-esteem, they too fall outside the scope of Sumner's argument.

(3) Sumner's third contention is also false as it stands. It is not the case that to tolerate hateful speech “in the interest of freedom of expression would inevitably be to confer upon it a certain degree of legitimacy.” Raz concedes that toleration and legitimation cannot always be distinguished, since “in illiberal societies … only those who express approved views are allowed to express them,” so that “the fact that an opinion is tolerated shows that it has authoritative endorsement.”⁴⁰

However, it is a social fact of free societies that views can be expressed even when they are not approved, so that toleration in such circumstances is not tantamount to endorsement. One would have to contend that all societies are inherently illiberal or that liberal toleration is veiled endorsement in order to maintain that toleration and legitimacy are inevitably linked. These are counterfactual presuppositions, however. The point is clear in the context of the several responsible voices of society to which we turn now in order to reinforce the centrality of epistemic responsibility in issues of hateful speech.

THE DIFFICULTY OF TOLERANCE

The harm principle connects naturally with issues of knowledge and belief to suggest that the social flaw represented by abusive speech should be contained through means more effective than “additional speech” alone but less draconian than legal compulsion. Circumstances may arise in which responses to hateful speech must be legally coercive, but exercising the more complex and various powers of the state and society is usually preferable. Moreover, in many cases the appropriate social role is enabling rather than coercive. When the state asserts its enabling roles aggressively, it addresses psychological abuse through civic entitlements and educational practices rather than through regulations or criminal proceedings; and when the modes and limits of appropriate political responsibility operate effectively, they should discourage hateful speech from causing damaging loss of self-esteem. In a well-ordered civil society, public and private disapprobation, rather than political coercion, should also play a preventive role. Shame is an effective tool for controlling the behavior of people who belong to a liberal community that enforces its views of right action in ways that are legitimate when state sanctions would not be. The public embarrassment of a popular New Zealand radio broadcaster upon calling the secretary general of the United Nations a “cheeky darkie” is a nice case in point.⁴¹

As long as distinctions are observed between politics and the practices of approval and condemnation characteristic of voluntary associations and defensible cultural traditions, the support of social relationships that can be identified outside of political life again moderates worries about loss of self-esteem.

A contrary view holds that free speech means “free from state pressure” but that expression is never free in this respect, so that “there is no [free] speech and therefore nothing … to protect” against the government's intervention.⁴²

However, this assertion is plausible only as long as it suppresses a difference between social and political regulation that removes the inevitability of “state pressure.” Unwritten rules abound in social life, defining tacit understandings about what one may and may not say. Any institution has purposes with which certain behavior will at some point conflict and not be tolerated, but the control of expression that runs up against the tacit rules of the dominant social institutions does not necessarily occur through state pressure or political prescription. Forms of expression unacceptable to parents may be censored by families, certain language may be shunned within churches, and certain symbolic dress may be banished from schools. It is another matter entirely to suppose that the suppression of expression by the state is inevitable. Unlike many other institutions, a pluralistic society lacks agreement on a distinctive defining purpose and thus has no internal point whose protection may justify exclusions apart from those necessary for sustaining the security of persons. Without some further argument there is no obvious raison d'état for censorship. The idea that expression is never free from political pressure is then sustained only by ignoring the difference between social and political regulation, that is, by construing “politics” as any attempt to implement a partisan vision of any kind. This makes scientists who contest one another's theories, neighbors who promote their literary preferences over the backyard fence, and citizens who assent to constitutional protections for speech all equally political agents, making “political action” a synonym for persuasive action generally. In assimilating to “politics” any effect of “being influenced by whatever notions are current in the culture,” the significance of saying “there is no such thing as free speech” becomes as thin as the sense of “politics” is broad.⁴³ Polities become indistinguishable from communities, speciously justifying the coercive weight of the state even where no communal consensus exists and there are preferable alternatives to suppression.

The style of argument that conflates political and social considerations particularly neglects modes of social influence that constitute expectations of epistemic responsibility. This form of influence is most obviously represented by education rather than law. If we view universities as parts of a desirable institution whose core rationale is the advancement of knowledge, then we place a high valuation upon correctly informed attitudes. Let us therefore assume that forms of expression that are not conducive to this purpose have no particular claim upon protection and that those in conflict with it may be controlled. Hence, within an educational setting, racial epithets are potential objects for social proscription on the grounds that, being essentially derogatory, they do nothing to advance relevant knowledge and may actively impede learning. If a university should suppress racial epithets, though, its proper epistemic rationale should be clear. It is not that hateful speech hurts or offends people, since on these grounds there may be words that should not even be mentioned and propositions that should not ever be uttered. When a university refers to its educational mission in suppressing some speech, it also needs to distinguish carefully between advancing knowledge—part of its purpose—and following public opinion.⁴⁴

Permitting the offence that is taken to words to weigh as heavily as reasoned belief obscures the objective that warrants institutional regulation.

The social responses represented by education could be taken to suggest that distinctions between high- and low-value speech make uncontroversial sense only in the context of particular social institutions. These responses certainly focus attention first upon incentives for learning rather than on penalties for unacceptable actions. Although all communities constantly remind those within them of what is considered right in the way of thinking, a respectful system of organization will work best when it is primarily enabling. Whenever actions have social meaning, they are regulated through generally understood conventions, but the rules of evidence and inference for educational communities describe activities that make discovery and illumination possible. Inappropriate actions are not censored but corrected, so that suppression is inherently secondary to enablement. For effective public policy as well, desirable behavior needs to be encouraged. The challenge is to allow encouragement to dominate and appropriate belief-mediated attitudes to prevail.

One might reject this challenge on the grounds that education, while important, is no panacea. It has been effective in reducing much harmful behavior—cigarette smoking is a good example—but cannot by itself eliminate injurious practices. A plausible inference is that the enabling and coercive powers of the state should be allowed to complement and reinforce one another.⁴⁵

So doing would also employ the symbolic value of law, albeit the state's authoritative condemnation of hateful speech should be balanced against the unfortunate symbolism of compulsory belief. However, even if one accepts the advice not to leave criminal law as a last resort, hateful speech is offensive to the liberally minded for the epistemic irresponsibility displayed by willful stupidity and unnecessary ignorance. These are the vices that primarily require attention, a priority lost when the law announces that expressing hateful beliefs will not be tolerated. The emotion of hatred expressed in abusive speech commonly includes beliefs about its objects that need to be unlearned. Refusing to tolerate them is contrary, not complementary, to education.

One might instead reject the challenge of enablement more radically, asserting that the distinction between toleration and endorsement actually displays liberalism's deep intolerance, since the difference is meaningless to those who suffer the painful consequences of toleration. This recurring line of thinking finds no major virtue in liberal education, which is already implicated in oppression. The complaint is reminiscent of important arguments by critical theorists of the Frankfurt School, who a generation ago complained that “what is proclaimed and practised as tolerance … is in many of its most effective manifestations serving the cause of oppression.”⁴⁶

Today, critical race theorists and others committed to working “toward the end of eliminating racial oppression as part of the broader goal of ending all forms of oppression” have forcefully renewed demands to regulate hateful speech politically.⁴⁷

A difficulty of the critical theorists' approach to “oppressive tolerance,” now as then, is that because punitive measures are themselves oppressive, they cannot succeed in ending all oppression. Coherence is lost by calling upon the coercive power of the state to protect against verbal oppression, thereby subordinating and marginalizing some members of society. In these circumstances, it may become unclear what behavior could fail to constitute evidence of intolerance and oppression. Within the libertarian framework, civility, notwithstanding its capacity for ugliness, is important for sustaining discussion; but when that framework is discarded, it can be asserted that advocating civility is also “evidence of deep-running colonizing strategies, by which persons (or groups) in a superior position maintain that position vis-à-vis some other persons (or groups).”⁴⁸

If everything is evidence for a proposition, civility and incivility alike, the position it states must be trifling or contradictory.

The radical challenge has one further possible rejoinder. Rather than accepting that radical doubts about tolerance display logical and epistemological deficiencies, this alternative perspective might aggressively draw upon the deep conflict evident in case studies like that of Sumner and Raz. In one account, tolerating is entirely separate from endorsing. In the other, this is a distinction without a difference, while contempt for minorities is inherently harmful because harmfulness and hurtfulness are not well differentiated. More generally, from one orientation, a number of concepts that from the alternative standpoint are conceptually distinct are instead conceptually related. The libertarian thus regards all critical theories as conceptually lax, while the critical theorist regards the libertarian as spuriously discriminating in adopting a framework of expressive freedoms that disguises and preserves structures of subordination and weaknesses. This impasse arguably shows that the standoff between sharply divided intellectual paradigms leaves open the legitimacy of suppressing hurtful speech politically, since the theoretical alternatives can be represented as equally reasonable options. Given the stalemate, neither view can strictly refute the other, lending practical legitimacy to policies of political censorship. In the face of this argument, it is important to recognize that in liberal societies the competing views are not coequals.

The argument that deep connections exist between toleration and legitimation is best supported through analogy with illiberal and preliberal societies. In such societies, as noted above, toleration might reasonably be equated with legitimation—in the one case, because only approved expression is permitted, in the other, because there has not yet occurred the differentiation of social and political authorities that permits distinguishing public expression from expression by members of the public. Under those circumstances, important epistemic resources for spurning discriminatory utterances are not readily available, so that it makes sense to suppose that hurtful actions are ipso facto harmful. To this extent the radical position is correct. However, it provides no reason to superimpose the terms and conditions of life in such nonliberal societies upon political cultures that value expressive freedoms, leaving the position supported only by the counterfactual assertions identified at the end of section II above.

In sum, if hurtful and hateful speech is not considered separately from other forms of action that can do damage, then great personal and social resources are squandered unnecessarily and at great cost. We have noted the cost in logical and epistemological confusion. The cost to deliberative argument is also great. Of course, people's capacity to deal with unreasonable views discursively can be exaggerated, making it easier for hateful speech itself to have serious costs. The libertarian view, though, identifies building the capacity for deliberation as a necessary condition for the thriving of a political culture in which freedom is more substantial than freedom from unpleasantnesses. A liberal polity permits its citizens to express nonliberal views and thereby subjects itself to the risk of loss of political dominance if it cannot sustain the best of social argument. This risk is not avoided by punishing illiberal voices, endeavoring to remove certain forms of expression from political debate by banning hateful expression directed toward groups variously identified in terms of gender, race, ethnicity, and sexual orientation. To the contrary, extirpating such expression politically may be self-destructive, undermining the openness of liberal culture by regarding correct opinion on certain points as so undeniably established that promulgation of erroneous views and attitudes may be prohibited.

Perhaps equally costly, regulative regimes may do little to address corrosive relationships, such as the casual racism that sometimes marks the mutual enmity of ethnic minorities toward one another. Hateful speech is a more prevalent phenomenon than its most common illustrations suggest, extending, in further example, to childhood bullying of the small, weak, and different. Political regulation in such cases can be no more effective than attempts to control historical instances of fratricidal genocide that have defied all law. Limited to verbal abuse toward women and minorities by members of dominant cultural groups, political regulation may permit less-remarked but more pervasive forms of undue hostility toward others to work their evil effects.

There is an uncomfortable further cost in neglecting the distinctive issues of responsibility attaching to hateful speech. Hostility toward other groups may not be in every respect unreasonable. It is not only victims of abuse that can feel threatened. The hatred expressed in abusive language toward identifiable groups may itself spring from fear of the kind identified by Thomas Scanlon in supporting an “interest in ‘protecting society’ from certain kinds of change.”⁴⁹

Concern for the evolution of one's society is legitimate. It helps to explain the attitudes of bigots who fear the presence of equality-seeking groups that may upset established ways of life. It is not necessarily more reasonable to say that they should toughen up than that their targets should, since if the perceived threat is not creatively addressed, the causes of the fear and abuse are likely to remain. The real challenge for political responsibility lies here, in fashioning a domestic climate where legislators promote reform rather than feeling driven to questionable regulation.