Not for nothing is Jeremy Waldron known as a rigorous thinker who poses, grapples with, and seeks to answer difficult questions based on critical reflection and nuance. And he is quite justified in not just accepting at face value, without further supporting documentation, my claim that free speech tends to lower violence.
Unfortunately, the study mentioned in my lead essay is currently only available in Danish. So Waldron can be forgiven for basing his critical questions on some rather sweeping and specific assumptions about the findings of our study that might have been devastating if we had made such bold claims. However, we are considerably more humble about our own conclusions than what Waldron assumes. We explicitly acknowledge that “our analysis is not perfect” and highlight the problems with establishing causal relationships between free expression and various outcomes. Nor have we claimed to settle the debate but merely to “add an extra layer of depth”; likewise we have not argued that the justifications for restriction on free speech advanced by Waldron and others are without merit, only that they “at best [are] lacking in nuance.”
Initially I should say that we have used data from datasets including Freedom House on freedom of the press, Varieties of Democracy on freedom of expression and PEW on religious hostilities, and we have looked not only at democracies but at most states with a population size of 1 million or more. We also do not focus exclusively on hate speech. Hate speech laws may be included as one factor among many when determining a country’s overall score in Freedom House and Varieties of Democracy measures of press freedom and freedom of expression, but we have not isolated our coding specifically to hate speech laws, nor have we claimed to do so. As such the study deals much more broadly with the likely outcomes of free speech protections overall. We have also reviewed the broad empirical literature on free speech, social conflict, genocide, terrorism, and radicalization to draw from the widest possible base of scientific knowledge.
Back to Waldron’s criticism:
As I understand it, the study shows that there is a negative relationship between free speech protections and social conflict: the more free speech there is in a society, the less conflict there is likely to be.
That is only partially correct. What we find is that this tends to be true in liberal democracies, while in the most authoritarian states, liberalization of the press is sometimes associated with a higher prevalence of conflict. Moreover, our finding is one of probability, rather than iron laws that hold true across the board in every country at all times. As I wrote in my essay: “This is not to say that speech will never lead to instances of violence or conflict, only that overall we should expect political and religious violence to increase rather than decrease when liberal democracies fight extremism with censorship.”
Another problem with Waldron’s response however, is that it seems to assume that our study is narrowly based on the question of hate speech: “But Mchangama and Andersen say they have found that in fact untrammeled hate speech does not exacerbate conflict of this kind.”
This is a misunderstanding. What we find is that strong free speech protections are associated with lower levels of violence, especially in liberal democracies. That is very different from the claim that hate speech cannot exacerbate conflict. Many types of hate speech are in fact tolerated as free expression in liberal democracies, but hate speech is often much more prevalent in societies with lower levels of free speech protection, and often it is peddled by politicians in power (think of Myanmar, Rwanda, and Nazi Germany).
Even if Waldron should reach the conclusion that our study is “poorly designed” or that our findings are insufficiently robust, a number of other studies should interest him. A 2016 study shows that only in former authoritarian states does partial liberalization of free speech increase the likelihood of armed conflict, while armed conflict is least likely to occur in liberal democracies with strong free speech protections. Another study (using a different dataset than ours on media freedom) finds that higher degrees of media freedom are associated with conflict in countries with a high degree of social intolerance, whereas media freedom tends to lower the risk of conflict in countries with a low degree of social intolerance. Since most liberal democracies fall into the latter category, this study too seems to support our findings. And as already mentioned, a recent study of right-wing terrorist violence at least partly supports our conclusion, specifically when it comes to consolidated north-western liberal democracies, where immigration is an especially hot button issue in recent years. Let me quote more extensively this time:
[M]ilitant mobilisation may in turn have been fuelled by extensive public repression and stigmatisation of radical right actors and opinions in countries such as Sweden, Germany and the United Kingdom….in other words, these findings suggest that while repression and stigmatisation may discourage some people from joining radical and extreme right groups, they may also push some of the most ardent activists onto more clandestine and revolutionary paths, ultimately leading to violence and terrorism.
Perhaps the foremost expert on the relationship between speech and violence is Susan Benesch, who heads the Dangerous Speech Project. She argues that the occurrence of what she has defined as“Dangerous Speech” is a reliable indicator of outbreaks of deadly violence. But her definition of Dangerous Speech is significantly more narrow than most of those hate speech laws that Waldron supports. It requires among other things that the speech in question must be understood as an incitement to violence by its audience. Dangerous Speech also requires a “a powerful speaker with a high degree of influence over the audience.” This would rule out many of those ordinary people convicted under hate speech laws in Europe for nasty tweets or Facebook comments. More importantly, Behrens favors “counterspeech” over censorship, when seeking to prevent Dangerous Speech:
efforts must not infringe upon freedom of speech since that is a fundamental right—and when people are prevented from expressing their grievances, they are less likely to resolve them peacefully and more likely to resort to violence.
Moreover, a number of studies on counter speech seem to confirm that this liberal alternative to censorship has the capacity to work.
As I stated, our study is not perfect and many questions are yet unanswered. We may never be able to make sweeping claims about the specific effects of particular free speech restrictions on violence. Nonetheless, the statistical results from large datasets are valuable supplements to purely anecdotal or theoretical arguments. Such arguments unfortunately lend themselves too easily to cherry-picking of convenient examples and the omission of those not supportive of the writer’s agenda. Given Waldron’s justified skepticism about the claim that free speech tends to lower violence, one wonders why he didn’t ask himself the same tough questions before making sweeping statements about the harm in hate speech, with no attempts to document such harms empirically.
While I fully expected Waldron’s meticulous questioning of empirical claims, I was more surprised by his dismissal of the relevance of historical free speech precedents. After all, Waldron himself has written on Locke, Bayle, and Montesquieu in the context of religious hatred and modern hate speech laws. In the superb article just linked, Waldron explicitly bemoans that “There is a very considerable literature on hate speech… but most of it lacks an historical dimension,” and he explicitly tries to “bridge that gap.” And as Waldron knows much better than I, Montesquieu relied on ancient examples to provide context for his treatment of contemporary issues. In Chapter 13 of his Spirit of the Laws Montesquieu explicitly mentioned how the Roman emperors Augustus and Tiberius expanded the crime of high treason to cover writings, and that “nothing was more fatal to Roman liberty.” In other words, Montesquieu—like Machiavelli before him—saw the connection between ancient and contemporary that Waldron no longer (?) acknowledges.
But it would also have been interesting if Waldron had included more recent history in his writings on hate speech laws. In the Harm of Hate Speech, Waldron seeks to dispel worries about the censorious effects of such laws by pointing out that hate speech bans are included in a number of international human rights conventions. But as I have argued elsewhere, the internationalization of hate speech bans in human rights law owed much to the efforts of the communist states during the Cold War era, whereas the liberal democratic stares of the era generally opposed the idea. Not because they were free speech absolutists, but because they recognized the danger of including an obligation to restrict free speech in international conventions ratified by deeply authoritarian regimes. Eleanor Roosevelt put it well when she opposed the current article 20(2) of the International Covenant on Civil and Political Rights, since it was “likely to be exploited by totalitarian States for the purpose of rendering the other articles null and void.” She also feared that the provision “would encourage governments to punish all criticism under the guise of protecting against religious or national hostility.”
When the current wording of Article 20 was put to a vote in the Third Committee of the General Assembly, it was adopted with 52 votes in favor, nineteen against, and twelve abstentions. Those in favor were primarily the communist states of Eastern Europe, as well as non-western countries and those with very questionable human rights records such as Saudi Arabia, Haiti, Sudan, and Thailand. The nineteen countries that voted against included almost all western liberal democracies—such as the United States, United Kingdom, Canada, Australia, New Zealand—as well as the five Nordic countries and the Netherlands. Eighteen countries, including the United States, entered reservations to Article 20 upon ratification. A somewhat similar story played out when the Convention on the Elimination of Racial Discrimination was adopted. Yet today, the very states that opposed the internationalization of hate speech bans in human rights law have generally mainstreamed such instruments at the national and regional level.
Concluding, I would humbly suggest that the relevant data (such as it is), history, and practice suggest that in democracies, the robust protection of free speech is more likely to foster social peace than violent conflict. Waldron once wrote “where there are fine lines to be drawn the law should generally stay on the liberal side of them.” If he still maintains this position, I would respectfully contend that while he is right to be concerned about the harm in hate speech, he should also consider the harm in hate speech laws.