18 January 2017 6:55 PM (politics)
Since the United States election, a few people have pointed me to Sean McElwee's thoughtful argument for the censorship of hate speech wanting to know my thoughts. From the title of this post you can infer that I disagree. I believe wholeheartedly and fully grant that the world would be far better if it contained no expression attacking the dignity and social status of any person. I grant completely that hate speech does harm. However, laws of the sort Mr. McElwee advocates would have little to no benefit and introduce harms of their own.
Mr. McElwee likens those who oppose hate speech laws to ‘free-market fundamentalists’. While not correct, this analogy suggests a way to evaluate proposed laws. I am by no means a free-market fundamentalist given my long-term preference for democratic forms of non-Marxist communist technocracy. This does not mean that I support every proposed market intervention.
The world would be a better place without heroin abuse, but I oppose drug prohibition. I do not hold the market sacred nor do I think heroin harmless. I oppose drug prohibition because it has conspicuously failed to reduce the harms of drug abuse while adding new, exciting harms of its own.
As Mr. McElwee wrote, legal limits exist on speech that most people accept. His examples were child pornography and defamation. Child pornography is a bad example and defamation is complicated, but given that our society isn't completely transparent some form of defamation law is warranted. Laws regulating advertising and political campaigns are valuable limits on speech, as are laws against inciting a riot or panic.
That a given kind of speech is intensely harmful and that we recognize the need for some legal limits on speech is insufficient to justify laws against that kind of speech. Consider anti-vaccination and homeopathy. There is no evidence for either. The dissemination of these ideas kills people. We should make no attempt to censor these ideas in general discourse, published works, television, the Internet, or presentations.
An attempt at such censorship wouldn't work. It might work in Canada where trust of civic institutions is relatively high, but in the United States trust in civic institutions is fairly low. Anti-vaccination advocates tend to have lower than average civic trust and will often justify their beliefs by positing conspiracies between pharmacology corporations and the government. When they speak, they also evince a dislike of authorities dictating how they raise and care for their children. Censorship would validate their conspiracy theories and give them a heavier hand of authority to resist.
Every advocate of hate speech laws focuses on the harms of hate speech. This is insufficient. They must make a reasoned case that the laws they propose will have some benefit that is not outweighed by the harms they introduce.
Are There Benefits?
This is the weakest part of Mr. McElwee's article. He claims that hate speech laws give minorities more positive freedom to express themselves but fails no support this claim. If there were no long-standing hate speech laws this would be fine. We would have an obvious harm, the claim that penalizing it would make it less frequent, and a suggested benefit. Opponents would have to argue convincingly that the claimed benefit might not materialize.
However, we have Europe. Mr. McElwee asks us to examine Europe for actual harms arising from hate speech law. He does not speak of actual benefits arising in Europe from hate speech laws. I have looked independently and failed to find any claims that there have been any. I would expect advocates of hate speech laws to reference any studies on the subject that existed, and while the European Union released papers on hate speech, they focus on legal theories underlying the law rather than social outcomes. Absent formal investigation, we can try to see if Europe is significantly less hostile or more inclusive in a way that could be plausibly traced to these laws.
European hate speech laws target the incitement of hatred against or denigration of various protected groups. Punishments are fines of one to several thousand Euros and there may be jail time, ranging from a few months to a small number of years. Successfully prosecuted complaints are high profile: interviews in widely read magazines, billboards, televised interviews, and political rallies. These policies have been around for long enough that we can examine them for both benefits and harms.
Currently, the EU is addressing hate speech on large social networks. A notice, review, and takedown regime has been in effect on the largest sites for a few months. Mr. McElwee's article focuses on Internet hate speech, and that is the domain where government intervention is so new that one cannot give reassurance based on existing practice. In 2015, the French government announced a heightened campaign against online hate speech. It introduced several measures into parliament several new powers, including the ability to remove entire websites without a court order, but few were passed.
Antisemitism is the easiest form of bigotry to look at, since the Anti-Defamation league has a worldwide survey of antisemitism broken down by region and nation. There isn't any obvious correlation between how recently a country adopted hate speech laws and its levels of antisemitism. Greece is the obvious outlier, having both high levels of antisemitism and hate speech laws dating back only to 2014. France, however, has some of the oldest hate speech laws on the continent and has a higher level of antisemitism than every nation but Greece. Historical factors dominate. The Great Recession caused an upswing in antisemitism throughout Europe and it (combined with the Euro crisis and a decade of Austerity) is the major factor behind the level in Greece. Long-standing hate speech laws in France seem to have done nothing to blunt the economic antisemitism that results in anti-Jewish feeling during financial shocks. Further, if the ADL's numbers are remotely meaningful, hate speech laws have done nothing to create a safe and inclusive environment for Jewish people.
Global attitude surveys covering other forms of bigotry are lacking. It is also hard to disentangle people's economic and security concerns from their prejudice. One can be opposed to immigration for reasons other than racism, but an intense opposition to immigration seems to lead to an antipathy toward those who most embody one's stereotype of an immigrant. Similarly, a fear of terrorism can lead to an antipathy toward people resembling one's stereotype of a terrorist. Economic and political factors can thus manufacture racism or religious bias where they didn't exist before.
This is one area where hate speech laws fail. They target protected groups, so vulnerable people can be attacked even without obvious ‘code’. During the recent refugee crisis, newspapers referred to asylum seekers as a ‘swarm’, an ‘invasion’, and a ‘plague of feral humans’. Others advocated sending gunships to shoot them out of the water. The speakers might be Islamophobic. They might simply be upset about the level of their taxes, worried about them going up if they have to support refugees, and have some innate difficulty with the concept that people they see on TV have conscious experience. In truth, it doesn't matter. Arguing whether someone ‘is racist’ and assigning moral opprobrium is a waste of time, but the narrowness of hate speech laws requires one to do just that.
Germany saw anti-Muslim rallies by people worried about the imposition of Sharia. Levels of anti-Muslim discrimination in France are astonishing. Every time a new Eastern European country joins the EU and gains free movement of labor, shrill headlines pour out comparing them to leeches and vermin come to steal the benefits of hard-working natives.
Europe is not particularly intolerant. Indeed, parts of Europe are far ahead of the United States when it comes to treatment of women and homosexuals. That tolerance matches up very well with secularism and, in the case of gender equality, social services (state-funded childcare seems to be one of the biggest factors contributing to female advancement to high corporate office) rather than hate speech law.
I suspect that many people liked Mr. McElwee's essay because they thought that if the United States had laws similar to Europe's then Donald Trump's candidacy would either have never got started or that it would have been shut down. That is a fantasy. Shocking and tasteless as many of Trump's remarks have been, few if any of them rise to the level that could get a European politician convicted. His supporters in the alt-right may have been in trouble, but he didn't need them.
Hate speech laws have done little to prevent far-right nationalism. Several Alternative für Deutschland politicians have expressed the desire to remove all Muslims from Germany, and that country's hate speech laws are doing nothing to keep them from gaining more support on a tide of anti-EU sentiment. Finland's hate speech laws did nothing to prevent the right-wing nationalist Finn party from gaining a large share of the seats in parliament. France's hate speech laws have done nothing to prevent the rise of the National Front and the possibility that Marine Le Pen might do dangerously well in the second round Presidential election. Austria's hate speech laws did nothing to prevent a far-right, anti-Muslim ethnic nationalist from almost winning the Presidency. A conviction under the hate speech laws of The Netherlands hasn't harmed Geert Wilders in his bid for the Presidency.
The only benefit that can be claimed for traditional European-style hate speech laws is denunciation. In the words of David Cesarani:
Amid this anarchy, all that decent people can do is agree to reasonable limits on what can be said and set down legal markers in an attempt to preserve a democratic, civilised and tolerant societyThis is similar to the argument that we can't abolish drug prohibition because that would say it's ‘okay’ to start use heroin. This is no justification for a law. If you want to steer norms, pursue some strategy that works. One might claim that denunciation provides a positive reassurance to the targets of hate speech, however I doubt if Muslims in France have their experience of discrimination substantially improved by the knowledge that somewhere a movie star is being fined a few thousand Euros for having said that they're ruining the country.
European-style hate speech laws are ineffective tokenism. They are passed with the best of intentions, but their only effect is the pride that the liberal majority feels in having Done Something. They have no significant positive effect on disadvantaged minorities and any effort wasted on them would be better spent on combating structural discrimination.
Censoring Hate Speech is Harmful
Hate speech laws in the United States would marginally increase bigotry. In Europe, antisemitic groups have used laws against antisemitic speech and holocaust denial as recruiting tools, citing them as evidence for Jewish control of government and media. Anti-Muslim groups in France claim that Muslims use French hate speech laws to shut down criticism of their religion and advance their goal of ‘Islamification’.
Those most at risk of right-wing nationalism have low trust in civic institutions and view themselves as ruled by an elite liberal conspiracy. State denunciation will have a contrary effect and push them toward bigoted attitudes.
Prosecuting someone affiliated with the Trump campaign for hate speech would have made those who supported them do so more strongly and pushed more people their way. Geert Wilders's support increased after his conviction on hate speech charges and many voters surveyed cited the conviction as a reason for their support. They resented the idea that someone expressing views with which they were in sympathy would be prosecuted.
After the violent attacks against media that some Muslims have claimed is blasphemous, people surveyed Muslims about their attitude toward free expression. (Some Muslims just wrote about it on their own.) Those in the United States were, on average, more likely to favor unlimited free expression than those in Europe. There are many causes, such as better economic attainment and integration in the Unites States as well as that nation's cultural emphasis on free expression.
European Muslims who advocate criminalizing blasphemy cite hate speech laws, and many accuse the West of hypocrisy. Holocaust denial (or at least 'soft' denial) is common among European Muslims, with many believing that the Israeli government fabricated or exaggerated the holocaust to gain political leverage. Several who spoke or wrote on the subject were angry that it was illegal for them to express their historical and political views yet they were expected to sit back and silently accept what they viewed as an attack on their religion.
Mr. McElwee claims that hate speech laws provide historically disadvantaged groups more positive freedom to express themselves. Many in this historically disadvantaged group have said that such laws are a tool of oppression used by the dominant group to silence and subjugate them. Such laws have also harmed the integration of these minority voices into the Enlightenment ideal of a secular state with freedom of expression.
If passed in the United States, hate speech laws will very likely become a tool of oppression. Religion is a protected category, and fundamentalist Christians think of themselves as oppressed and have fought to be legally categorized thus. The limited success of the attempts to have discrimination against homosexual and transgender individuals protected as religious freedom makes it disturbingly likely that the religious right would succeed in having some forms of criticism of their beliefs and condemnation of their treatment of sexual minorities classed as hate speech once any such law was passed. Long court battles cost money, and the most vulnerable tend not to be the richest. Some foreshadowing of this possibility exists in that the 2015 proposals to radically expand hate speech in France targeted Muslims in the text of the law, and advocacy for them was often outright Islamophobic.
These harms arise from laws like those in the well-established European model, but Mr. McElwee's article is most concerned with online speech, for which there is no well-established example. The model that the EU is currently trying to establish would likely have many of the problems already listed. Since it focuses on taking down objectionable posts, it would have very little to no effect on harassment. It may make some attitudes less visible, but this seems unlikely. It will, like the current European hate speech laws, be a useless exercise that lets good, honorable people feel proud of themselves while accomplishing nothing. it will also lead to terrible outcomes.
These outcomes have nothing to do with hate speech, but with the realities of a notice, review, and takedown regime. The size of large social networks and a limit on response time guarantee that ‘review’ will cease to exist.
This is not mere conjecture; we have done this experiment. The Digital Millennium Copyright Act is a pure notice and takedown regime. It requires material to be taken down when a notice is received by anyone claiming to be the owner of any material they claim to own. Supposedly the user can file a counter-notice to have the material put back up, and if the claimant wishes to pursue the matter they can take it to court. Very small services, rarely, or paid services, sometimes, allow users to file counterclaims. This is the exception, and does not happen at all on the kinds of large platforms the EU is targeting.
Google is flooded with takedown notices. Some mistaken, many fraudulent (the DMCA has been used as a censorship tool repeatedly), most clearly nonsensical (Google keeps getting notices demanding lists of notices received be taken down). Many attack legally protected use like brief excerpts for critical purposes. It is not economical for Google or other large providers to let users contest notices so they don't. Recently, Google began selecting some users on Youtube and giving them the ability to make counterclaims against a limited number of DMCA takedown notices. If you aren't picked or someone slams you with too many notices, you're out of luck. Get enough notices, and they delete your account.
This will be the outcome of the EU's online hate speech regime. Review is more expensive than allowing counterclaims. If anything resembling this goes forward, it will be abused for all manner of undesired censorship. Trolls and bullies will effectively erase the accounts of their targets with an endless stream of hate speech claims.
In general, regulations of content on the Internet are harms in and of themselves because they contribute to the splintering of the network. That is not to say that they are never justified, but justifying such a law must yield a clear and unambiguous good.
Economic factors suggest that social networks above a certain share of the market ought to be classed as common carriers and forbidden from all but the narrowest restrictions on content. They are, first and foremost, loyal to their bottom lines. Even when they have good intentions, they do not enact them well. They seldom provide recourse because recourse is expensive and being seen to do something is profitable. One need only look at the history of Google locking people's accounts with no explanation and no chance to appeal or Facebook's embarrassing track record of taking down news footage and works of art. If we are to have censorship at all, it must be democratic (and I don't mean a vote of the shareholders), transparent, accountable, and contestable.
What is to be done?
I do not advocate inaction. I believe we should take a two-pronged approach. First, we must tackle harassment. I can say with great confidence that disagreeable people saying horrible things about women to each other on Reddit in an area especially set aside for that purpose can be blamed for precisely no amount of the under-reporting of rape. The idea is absurd. There is no plausible causal connection especially when compared to all of the known factors that contribute to the under-reporting of rape.
Most of the harm of online hate speech arises from it also being harassment. Harassment that does not target historically disadvantaged groups is still harmful and still to be dealt with. If our concern is to protect human dignity and the sense of safety in the community, then we must protect human dignity and the sense of safety in the community. Full stop.
We should, in our implementation, pay close attention to the needs and well-being of those who have historically been the targets of attacks and mistreatment. Basic engineering sense should tell us to focus on areas that have had trouble, and members of vulnerable groups are likely to be particularly vulnerable to all sorts of harms. However, there is no justification for attempting to define other attacks on dignity out of existence. The greatest harm reduction will come from focusing on harassment generally.
The worst forms of harassment: death threats, rape threats, and threats of physical assault are already illegal. Law enforcement should be expected to deal with them and funded to do so. This may be impossible on the darknet or on purely anonymous sites, but most people aren't on the darknet and the people who go there do so with the intent of avoiding oversight for themselves and others.
It is impossible to outlaw anonymity and undesirable to try. It may be worthwhile to set a legal standard of identification. Users could identify themselves by providing some proof of legal identity to which their accounts would be associated. The law would require that these associations not be public. Since they would only be used for investigations, they should not be stored online. Users should be allowed to create multiple accounts attached to the same legal identity, and no information stored online should indicate that they have the same owner. This is very different from things like the real name policies some networks try to enforce in that you can assume whatever and as many fictional identities as you wish. The only information that would be public is whether an account is identified.
Users should be given the tools to curate their contacts and create safe communities. They should have the option to block unidentified accounts (with manually specified exceptions) from communicating with them in any way, to block specific users, or to grant and deny privileges in a useful way. I suspect that network effects would make most people identify their accounts where that is supported. It may be necessary to require large networks to provide an identification and curation framework, but smaller networks should be free to decide whether to allow, require, or simply not provide identification. This is not a hard and fast proposal that I expect people to adopt, merely a thought experiment for how one could deal with the enforcement problem in a way that balances competing goods.
A fundamental way to combat harassment is making objectionable people leave one alone or ejecting harmful people from a community. Therefore, we ought to provide legal penalties for ban evasion (returning to a group or speaking to people under a new identity when they have evicted one before) in the form of a substantial but not crippling fine. The goal should be to create enough self-policing and activity that the law is rarely invoked. In addition to threats of harm, we ought to criminalize suicide baiting or conspiracies to mobilize a many people against a target. Many states already have laws against online bullying which need to be enforced, though a federal statue would help for larger, distributed campaigns of abuse.
Networks large enough to merit common carrier status should be required to provide some algorithmic transparency and adjustment in their feeds; users should not to be harassed by an automated process. At the minimum it should be possible to get a rough model of what the system thinks it knows about one's preferences and change them, to see why it thought you should see a post, and to block phrases or topics.
This should, at least, help with online inclusion and expression, particularly for women, who have been the targets of the most vicious harassment. Any reasonable measure of effectiveness of an anti-harassment regime ought to focus heavily on the experience of various minorities and whether they feel more at ease.
Echo chambers where people tell each other bigoted things do little if any direct harm to individuals. The social ill is that they provide a self-reinforcing reservoir of bigotry which leads to actions and political opinions. It would be nice if we could eliminate this bigotry through hate speech laws, but the record in Europe suggests that we can't.
We should address this problem by addressing bigotry. We know that several things affect bigotry. The most powerful effect for decreasing it is normalized exposure. Bob Altemeyer argues that the biggest factor underlying a fall in upper middle class white prejudice against African Americans over time has been exposure on college campuses. As college becomes less affordable, this effect reaches the middle class less and less, so we should aggressively pursue school desegregation nation wide. Not only will it attack bigotry at its root, it also improves education outcomes for minorities.
Anti-Immigrant bias in Europe is, on average, highest where there are few if any immigrants. Anti-Muslim bias is highest in areas with few, if any, Muslims. I am not proposing that we try to import Muslims into Kansas, but positive education can combat a lot of negative stereotypes. We learn about Galileo and Newton in our science courses, why not mention al-Khwarizmi when students learn to solve quadratic equations? He wrote the book on the subject and gave his name to the field. We could mention the Muslims who invented astronomy (ever wonder why so many stars have names beginning with al-?) This sort of thing seems to work well if you use the flow of a subject matter to throw in stories about interesting people.
Bigotry is also increases by insecurity and loss of identity in the world. Anti-Immigrant sentiment comes from people with less secure jobs that compete with immigrants. Misogyny is hugely prominent in cultures with elements of machismo where the culturally accepted roles for men are better. Improving economic security in general, educating people, and recruiting male teachers to provide better role models for at-risks groups would probably help.
Obviously there are nine million other things. Anything that improves outcomes for a minority will cut down on bigotry, because members of a minority who do well are out there in the world, succeeding in whatever field they choose. It becomes less possible to think of them as a separate ‘other’ and people respect success. Surely this is an endeavor worthy of our effort.