2. The main elements of chapter 2 of On Liberty

Mill's conclusion in chapter 2 of On Liberty is that there should be absolute freedom of discussion in stable, cooperative societies.Footnote ⁹ The main argument is that whether received opinion is possibly false or certainly true, the harm of restricting free discussion always outweighs its benefits. If received opinion is possibly false, Mill says, restrictions cost us the opportunity to exchange falsehood for truth, or worse partial truth for better partial truth. If received opinion is certainly true, we still lose something important: the opportunity to recall the rational grounds of, and to keep vital, true belief.

Two further claims deserve special mention. First, Mill argues that any indefinite restriction of discussion involves social or political authority in an “assumption of infallibility.” In other words, on the topic being restricted, censors take up a position of epistemic superiority with respect to future decision-makers that they are not entitled to take up (i.e. one which only infallibility could justify).Footnote ¹⁰ By contrast, in protecting free discussion, decision-makers “take precautions against their own fallibility.”Footnote ¹¹ Second, Mill argues that freedom of discussion is required for social or political authority's decisions to remain based on reasons and not on their mere “likings or dislikings.”Footnote ¹²

These arguments show that discussion is vitally important as the mechanism that ties decisions to the lessons of experience and reasoned justification, facilitating individual development and social progress. “[T]he source of everything respectable in man either as an intellectual or moral being,” Mill writes, is that he “is capable of rectifying his mistakes, by discussion and experience.”Footnote ¹³ Discussion is also needed to hold decision-makers accountable to public opinion. Like Jeremy Bentham and James Mill, John Stuart Mill was concerned to maintain an “identity of interest” between the rulers and the ruled, and for all of them free discussion is vital to any effective democratic check on social and political authorities.Footnote ¹⁴ To silence discussion, then, is to undermine these key liberal and democratic goods.

3. Does Mill's argument rely on the liberty principle?

Notably absent from my summary of chapter 2 is any mention of Mill's famous “harm” or “liberty” principle, according to which society may consider interference with individual liberty only when there is risk of nonconsensual harm to others. Doesn't Mill claim that the expression of opinion or discussion is, strictly speaking, harmless? Some recent commentators, including Macleod, John Skorupski, Jonathan Riley, and myself, have argued that Mill simply doesn't make his case that way.Footnote ¹⁵ That is, he does not say that expression of opinion or discussion never risks nonconsensual harm to others (and is therefore outside of society's jurisdiction), but only that society should not interfere with discussion because its benefits always outweigh its costs.Footnote ¹⁶ If that is correct, then chapter 2 is independent of liberty principle reasoning – though it is still part of Mill's broader articulation of the value of liberty.

Miller's contribution to this symposium, “The Place of ‘The Liberty of Thought and Discussion’ in On Liberty,” provides valuable pushback against this general approach. Appealing to textual cues, especially from chapter 1, he defends a “continuity interpretation” on which Mill's defense of freedom of discussion in chapter 2 follows naturally from the introduction of the liberty principle in chapter 1. The continuity interpretation turns (to a large extent) on the claim that “as a general rule – in the great majority of cases, if not literally all – discussion is self-regarding,” that is, it does not risk nonconsensual harm to others.Footnote ¹⁷ If expression of opinion is self-regarding or harmless, then it falls outside of society's jurisdiction. This is not to deny that there are some “cases in which expression is directly harmful,” but Miller argues that, on Mill's view, these are “infrequent departures from the norm.”Footnote ¹⁸

The nature of Mill's reasoning is important because it affects our sense of the limits of his argument and of his openness to revising his conclusions. One source of skepticism about Mill's relevance today just is the thought that expression of opinion often risks harm that his theory doesn't take into account – something Miller addresses in the final section of his article. As we shall see, Bell makes this issue the centerpiece of her Millian reclamation project in “John Stuart Mill's Harm Principle and Free Speech: Expanding the Notion of Harm.”

4. What kinds of expression amount to harms that might be restricted?

One curiosity of the literature on Mill's defense of free discussion is the general failure to focus on the many cases in which he believes expression of opinion on balance should be restricted, and the need to draw some lines between those cases and whatever is covered by chapter 2. Let me summarize some of the evidence, which extends beyond On Liberty.

In “Mr. O'Connell's Bill for the Liberty of the Press” (1834), a robust early defense of free discussion, Mill is clearer about limits. He writes first that “Discussion on the subject of religion ought to be perfectly free.” But with respect to politics, he refines the point: “[I]t shall be lawful to controvert any political doctrine, or attack any law or institution, without exception; in any manner and in any terms not constituting a direct instigation to an act of treason, or to some other specific act to which penalties are attached by the law.”Footnote ¹⁹ He does not specify which illegal acts he has in mind, but one could imagine instigation to violence or threats to public safety as the next obvious candidates.

These are not the only issues, however. In that same article, Mill argues that private libel, concerning conduct that does not bear upon “the public character” of a person, ought to be restricted even when the allegation is true. It is an invasion of privacy; only when conduct becomes public in one way or another should it be open to public disapprobation or censure.

In On Liberty itself, as we saw before, Mill introduces the corn dealer case as an example of incitement or “instigation to some mischievous act.”Footnote ²⁰ He there also asserts that instigation to tyrannicide (despite its often being considered a virtuous act in the end) “may be a proper subject of punishment.”Footnote ²¹ Just as Mill did not specify the acts “to which penalties are attached by the law” he had in mind, he does not specify all “mischievous” acts either. But it bears mentioning that both turns of phrase suggest that more acts than just threats or incitement to violence may be included.

Finally, in “Thornton on Labour and its Claims” (1869), Mill considers the case of union members expressing their frustration at workers who benefit from the union's efforts but are unwilling to make sacrifices themselves. In line with the earlier discussions, he first says that while it is justifiable for those workers “to have brought before them, in an impressive manner, what their fellow-workmen think of their conduct” there should be no “threat of infringement of any of the rights which law guarantees to all – security of person and property against violation, and of reputation against calumny.”Footnote ²² He then addresses “picketing” as a difficult case, and writes:

Here, Mill explicitly cites the safety of the public peace and public decency as reasons capable of justifying restrictions on expression of opinion.

How should we make sense of these cases of incitement, threat, private libel, public safety, and public decency? One might argue that they are just a handful of specific exceptions to the protection of expression of opinion. But if we want to understand how to update Mill on free speech, our challenge is to see how he can both allow restrictions on a range of expression of opinion and argue for absolute freedom of discussion.

The first part of this challenge is, I think, answered by an underappreciated comment Mill makes about Montesquieu in his earliest writing on the topic, “Law of Libel and Liberty of the Press” (1825). He writes that Montesquieu saw correctly that “the only case in which the expression of opinions and sentiments could be a fit object of punishment” is when – and here he begins quoting Montesquieu – “the words that are joined to an act take on the nature of that action.” Montesquieu gives this example: “a man who goes into the public square to exhort the subjects to revolt becomes guilty of high treason, because the speech is joined to the act and participates in it [. . .] Speech becomes criminal only when it prepares, when it accompanies, or when it follows a criminal act.”Footnote ²⁴ Mill emphasizes that Montesquieu's argument could extend to cases beyond treason, and we have already seen ways in which he does extend it. But then, we are presented with the following question: are there other expressions of opinion that, in certain circumstances, become joined to criminal or harmful acts such that we may restrict the expression itself? If so, Mill should be open to those restrictions, at least in principle.

Bell's article picks up at exactly this point, arguing that “bigoted insults” are fit objects of restriction because of the harms they are tied to: “[B]ecause bigoted speech causes direct harm to individuals, and plays a central role in maintaining and perpetuating institutional oppression and injustice, it causes not merely subjective unpleasant mental states or offense, but also serious harms that usually remain unrecognized both in law and in public discussion.”Footnote ²⁵ This expanded sense of harmful expression, Bell argues, reflects new psychological and social scientific evidence that Mill did not have at his disposal. She concludes that, on his own terms, Mill should agree that bigoted speech is open to restriction because it is harmful, while offering none of the educative or democratic benefits he associates with discussion.

The second part of the challenge – how can Mill then defend absolute freedom of discussion? – might simply be answered by understanding “discussion” as any expression of opinion that is not joined to a harmful act in the way described. Is that right?

5. What is “discussion”?

So far we have seen that Mill's argument in chapter 2 concerns only expressions of opinion that are not joined to some harmful act justifying a restriction. But Mill must be saying something more, or at least more specific; it would be trivial to assert that where the benefits of restricting expressions of opinion do not outweigh the costs of restriction, the benefits of restriction do not outweigh the costs of restriction.

One way of approaching this question would be to return to the topic of Miller's article. Perhaps Mill's distinctive contribution is to argue that in a wide range of cases falling short of instigation, private libel, threats to public safety, public indecency, and so forth, expression of opinion is properly understood as “self-regarding,” and therefore interference is not even open to consideration. We would then want to understand what that range of expression is – something Miller addresses – when so much expression of opinion does seem to risk harm to others, on Mill's view.

Another way to approach this question would be to notice how consistently Mill uses the term “discussion” in defining the topic of chapter 2, and then to ask what distinguishes discussion from other expressions of opinion. Macleod's article, “Truth, Discussion, and Free Speech in On Liberty II,” pursues a novel version of this second approach. He leans on the distinctively epistemic nature of the arguments of chapter 2 to argue that, for Mill, “discussion” is an epistemic practice aiming at truth that consists only in “the statement and consideration of truth-apt propositions.”Footnote ²⁶ On Macleod's account, then, Mill's absolute support for free discussion in chapter 2 is only as an epistemic matter, and not an all things considered claim, despite appearances. That is, in cases of incitement, private libel, and so forth – when restrictions are justified – what happens is that non-epistemic reasons come to outweigh the epistemic reasons.

The key thought for the second approach, in general, is that Mill's argument in chapter 2 focuses on “discussion” understood in some limited way. What Mill would say about other forms of speech or expression of opinion goes beyond the aims of chapter 2 (consider that the corn dealer example is not introduced until chapter 3). This is not to say that he would welcome restrictions on other expressions of opinion in the absence of the sorts of reasons mentioned earlier. But on the basis of chapter 2 alone, we should not simply presume that manipulative or demeaning speech, such as Bell's “bigoted insults,” would be protected.Footnote ²⁷

One challenge raised by Macleod for his own particular account is that some discussion might not seem to be truth-apt, like flag-burning. Cejudo's article, “J. S. Mill on Artistic Freedom and Censorship,” very nicely addresses this issue in the context of works of art. Cejudo argues that, on Mill's own (sketchy) aesthetic theory, art has epistemic relevance despite being “non-assertoric,” because it can reveal to us truths about human existence. If this is right, then much artistic expression and other epistemically relevant forms of expression deserve the same protections as discussion. Cejudo allows that works of art may sometimes constitute incitement to harm or violate public decency. But he argues that the former cases are very rare, and the latter cases could usually be addressed by hosting the artworks where their audience would have to choose to engage them.