Tentative Thoughts on Speech Policing

post by kbog · 2020-01-06T19:20:36.485Z · EA · GW · 7 comments


  There are systematic problems in the way harmful speech is generally identified
    can have incorrect views on the nature of bad speech
    too readily assume that their opinions are right
    have not rigorously investigated which kinds of speech are harmful
    result: a good portion of targeted speech is probably beneficial
  Speech policing tends to succeed
  Much of the harm of speech policing is a matter of incidental damage
    or distorted careers
    time and resources
    politicization of everything
    trust in institutions
  Many arguments about speech policing are flawed
    Paradox of Tolerance argument is often misapplied
    of ‘cancel culture’ miss the point
    slippery slope is real – but only in certain contexts
  Speech policing could be procedurally improved in a few ways
    policing should not be so personally disruptive
    policing can actually be better when it’s done by the government
    need fair backchannels for deciding which speech to suppress
  We have too much speech-policing
  The side to worry most about is disability activism

In my previous post [EA · GW], I briefly, informally addressed the specific question of which kinds of speech are harmful.

In this post, I now address the further question of whether speech policing and deplatforming are actually good or bad.

First I make some assorted key observations and arguments. Then I give my general conclusions.

There are systematic problems in the way harmful speech is generally identified

Speech-cops can have incorrect views on the nature of bad speech

There are some cases where people’s philosophical views on defining harm are incorrect. They’ll focus too much on their own nationality, race, gender, or species while neglecting broader social impacts. Their views can also be distorted by nonconsequentialist or contractualist intuitions about how to weigh harms and benefits for different people. Finally, they can abandon the idea of harm altogether and instead make more philosophical arguments about speech being innately ‘prejudicial,’ ‘un-Christian,’ ‘degenerate’ or other things like that.

Speech-cops too readily assume that their opinions are right

Few speech-cops seriously take into consideration the question what if I’m wrong about the actual issue? In some cases, like terrorism-and-Muslim-immigrants stuff or the metaphysics of gender, a left-leaning person could reasonably decide “even if my opponents are correct about the underlying factual issue, we still ought to keep it hushed,” because sharing the opposing ideas can lead to bad outcomes even if they are correct. But in other cases, like the James Damore controversy, I have a hard time seeing how you can justify speech policing without assuming that you are correct on the science, because they are ideas that – if actually true – would probably be used to help achieve better arrangements for society in general.

Of course, sometimes it’s okay to assume that you are correct on the science. However, that really requires robust arguments from multiple lines of inquiry. Theory and empiricism, personal experience and scientific study, mainstream academia and the Intellectual Dark Web, etc; if multiple avenues line up to a single point of view then speech policing makes more sense. But on controversial topics, I generally don’t see people considering that. People have reasons to believe that they are right, but there aren’t higher-level debates of “have we proven this to a strong enough level that we can really assume that it is right?”

Speech-cops have not rigorously investigated which kinds of speech are harmful

There is generally not enough serious discourse and research on what ideas and speech acts are actually harmful.

And when someone argues that apparently harmful speech actually isn’t harmful, that in turn is sometimes considered harmful (or at least offensive) speech, as happened with Carl (2018). This is a circular fallacy and can put society in an epistemic rut.

The result: a good portion of targeted speech is probably beneficial

When I look at instances of recent speech policing, I’d say that a fair chunk of the controversial speech is beneficial. Occasionally this is just because the general point of view of the disparaged speech is correct. More often, it seems to be an issue where the facts are in dispute and society ought to hear things out better before we can make a strong decision. Rarely, it’s a case where the speech is wrong but it would still be healthy for people to hear the other side of the story and come to a better understanding.

Speech policing tends to succeed

AFAICT, speech policing tends to be somewhat successful. Banning people and ideas off social media and deplatforming them from colleges is effective. Academic mobbing lends fame and notoriety to a particular person and their views, but also sends a message to the rest of academia. When someone is the victim of an academic mob, tons of allies line up to say that they defend their right to academic free speech, but not many will actually come out and say that they genuinely agree with the controversial statements. The point is not to convince a particular person, but to keep their views marginalized.

A different pattern occurred in the James Damore controversy. Lots of people defended Damore by saying that his memo was correct. However, the controversy also placed a huge spotlight on critics of Damore, who had a big opportunity and a strong political motivation to say that his memo was incorrect. There was more open discussion and debate over Damore’s ideas, but it might have reduced the popularity of his views.

Most of the cases of speech policing do not get much media attention or are not even known publicly. Therefore, it’s more effective than you would think given the histories of a few high-profile cases in the news.

Ultimately, the winners of speech policing are IDW types, "anti-antifa", and others who focus on defending free speech. Actual white supremacists and researchers of controversial topics lose.

Much of the harm of speech policing is a matter of incidental damage

Damaged or distorted careers

Cancellation doesn’t just suppress certain ideas, it damages whole careers, causing people to be fired or more subtly disregarded. Thus, we lose some of their potential contributions to society. A worrying example of this is the Maya Forstater case because she worked for a global poverty think tank, as I noted in the previous post.

In some cases, individuals appear to gain from cancellation, as they can go on political and media tours talking about the experience. However, this is not the same as a long-term career, and it’s not socially productive in the same way as their ordinary jobs.

Wasted time and resources

There is now a burgeoning industry across media and social media where people (me included, with this post) devote effort towards investigating or arguing these issues. The outrage on all sides of the spectrum is a big loss of time. Imagine if people spent that time on their jobs, or their education, or their personal interests, or their families. Social media, published media, academic research, blogs, and college clubs could put more focus on more productive things.

The politicization of everything

Speech-policing contributes to broadly turning more of our environments into political ones. This worsens our quality of life. Most people don’t want to think about politics all the time. They want to do their job or enjoy their own interests in peace. Popular suppression of certain kinds of speech inevitably creates a threatening, disheartening environment for ordinary people who were just listening. Chris Arnade summarized it best – “leave people alone.”

Lost trust in institutions

Speech policing diminishes trust in institutions. When institutions police speech, they are perceived as less trustworthy by free-speech advocates. When institutions are expected to police harmful speech but they choose not to, they lose standing among speech-cops. Finally, when institutions themselves are considered to have said something harmful, they broadly lose popularity.

General loss of institutional legitimacy is not a good thing. It inhibits progress across humanity’s collective projects. It empowers populist outsiders in politics, who tend to be worse (see appendix). People who want to prevent fascism or similar things should especially focus on maintaining institutional strength and legitimacy.

Many arguments about speech policing are flawed

The Paradox of Tolerance argument is often misapplied

Popper’s argument is that tolerant people have a right to be intolerant of intolerant people, lest liberal society be destroyed. It’s generally used in defense of speech policing. However, it only really applies against people who are organizing hateful or delusional ideas, like the alt-right and conspiracy theorists.

In cases where a reasonably rational and tolerant point of view is being suppressed, the argument should be applied in the reverse direction. The mob or institution which suppresses speech is the intolerant side who is shutting themselves off from rational debate. So they arguably should not be tolerated.

Dismissals of ‘cancel culture’ miss the point

Many people think that none of this is really serious because people like James Damore get so much attention and support afterwards. But given the incidental harms pointed out above, this misses the point.

The slippery slope is real – but only in certain contexts

At this point, I think it’s pretty clear that concept creep, the politicization of everything, and victimhood culture do in fact create a slippery slope in the popular context of speech policing. Speech suppression by the left and the right have increased and gotten more ambitious. Fears of the political-correctness treadmill have been generally vindicated.

However, I don’t see any good evidence that government speech codes are a slippery slope. Democracies like Germany with hate speech laws do not show a slippery slope. Autocratic regimes like the USSR have often become more liberal over time. Nor do institutions like companies and universities go down a slippery slope deliberately; rather, they are pulled by changing popular pressures.

So the ‘slippery slope’ seems to be an outcome of dynamic systems with competing actors in certain conditions. It’s not the case that individual people and institutions have a repressed tendency for authoritarianism that will get unleashed whenever they deviate from a hard, fixed rule.

Speech policing could be procedurally improved in a few ways

Speech policing should not be so personally disruptive

It seems clear that people who get cancelled for bad opinions should eventually be integrated back into the same circles, perhaps if they make a simple retraction and apology. For someone to lose their career or academic status for decades is too excessive. Yes, harsh norms can theoretically provide a stronger deterrent, but the effectiveness of that is dubious. We generally don’t think that locking up minor criminals for decades is a good way to deter crime; we are beginning to understand that restoration is a better approach to criminal justice. The same should apply here.

Additionally, speech policing should focus more on simple elimination of particular ideas and documents, rather than trying to destroy people’s reputations and characters. It should be more about censorship and smoke-and-mirrors deception, rather than mobbing and outrage.

Speech policing can actually be better when it’s done by the government

Government hate speech laws have more legitimacy, predictability, and due process protections. They don’t spark the same kinds of problems and collateral damage as popular mobs. In a democracy, they are much less excessive due to the nature of the political process. Of course, they are also much more effective at consistently enforcing the restrictions, and can exact harsher penalties, which can be worrying.

It would be better to just have government suppression of a select few kinds of hate speech, rather than popular suppression of a broader set of opinions. If government laws against hate speech encourage activists and the public to calm down and worry less about offensive speech in general, then I think they would be a good thing. On the other hand, if such laws embolden partisans to go more harshly after more and more kinds of speech, then they would be a bad thing. Unfortunately I have no idea which is the case.

We need fair backchannels for deciding which speech to suppress

Deciding which speech is wrong or harmful should be institutionally separated from actually enforcing the norms. If there were more hidden ways for the most prominent offensive ideas to be freely discussed by careful, ethical, rational people, then speech policing would carry significantly less epistemic risk. Intolerant views could be debated in a context where they pose far less social risk. And speech-policing might obtain more perceived legitimacy, as people would know that the suppressed ideas had at least been considered somewhat fairly by informed people.

This is something that requires speech policing to be institutionally centralized. Speech suppression via popular mobs is not amenable to it, as far as I can imagine.

We have too much speech-policing

Such improvements as the above are pretty unrealistic in reality. We have to judge speech policing at is, not as we wish it could be.

The best and most popular targets for speech policing are white supremacism, Jihadism, and vicious racism in general. Assuming the deplatforming actually works, it should continue. Even antifa is okay, if it is effective and chooses the real targets without causing much backlash (I am personally skeptical).

Social media censorship on private platforms/groups is comparatively benign and effective, and can be worth extending over other harmful ideas besides outright hate, depending on the context.

Then there is always an opportunity for people and groups to internally exercise discretion for questions like “what research should I publish?” and “which speakers should I invite?” It’s okay for people to restrict more things besides proper hate in cases like this.

But generally speaking, American society is going much further than what I would recommend. We are erring on the side of too much speech policing, and will continue to do so for the near-term future. This includes excessive speech policing on both the right and the left. We should broadly retrench from these efforts. So for now, it seems fine to broadly promote the “classical liberalism,” “free speech” approach to contested speech, and to spread the arguments of people like Steven Pinker and Peter Singer. They are not 100% right, but they are pushing society in a better direction on the margin. And I find it highly unlikely that they will be so successful as to create a society where actually hateful views are met with open debate.

The side to worry most about is disability activism

The most concentrated risk on this general issue is that disability rights activists can shut down research and dialogue about treatments, cures and human enhancement, in particular by weaponizing the accusation of eugenics. These people also have a lot of antipathy for Peter Singer (for defending a parental right to terminate extremely disabled babies) and sometimes even antipathy for Effective Altruists by extension. This is more important and more neglected than the famous controversies about race-and-immigration, gender differences, and so on. We need to find some way to encourage institutions to fund the relevant research and to support people’s rights to freely modify themselves and their fetuses, while separating and jettisoning more controversial ideas.

Hopefully there can be a mutually satisfactory resolution along those lines. It would be much better to achieve a positive compromise than to have to go through the same kinds of pitched battles that we see between the IDW/right-wing and the SJWs. Disability activists themselves should greatly desire an early positive compromise, because many of the broader public are surprisingly open about supporting eugenics and infanticide, the hostilities and bigotries inevitably whipped up by an IDW-SJW type war could be dangerous, and military-economic pressures will push states to take the side of human enhancement. I am optimistic that with productive dialogue and leeway on both sides, we can take a better path.


Comments sorted by top scores.

comment by Khorton · 2020-01-06T22:21:59.490Z · EA(p) · GW(p)

I think we'd both agree that speech-policing is helpful in some contexts, but unhelpful if it extends across an entire society.

I think statements like this lack nuance: "Speech-policing contributes to broadly turning more of our environments into political ones. This worsens our quality of life."

I agree that speech-policing turns political situations into Political situations. It makes their political nature explicit. However, to some extent all human interactions are political, and there are many environments where explicit speech policing (or in these cases, censorship) can improve quality of life.

Example 1: I'm a member of a feminist career Facebook group with explicit rules that people are not allowed to state women are unequal with men. If they do, their post will be deleted, and they may eventually be banned. The Facebook group's goal is to encourage women to pursue high-powered careers and it occasionally attracts trolls who start bad-faith discussions. The rule supports us to pursue our goal of giving each other career advice, and stops us from being distracted by repetitive arguments. We enjoy the group and it feels more relaxing than other feminist social media because the censorship is enforced quietly and consistently.

Example 2: I used to teach primary school. I didn't teach children under age 10 about sex, or let them teach each other about sex. This allowed their parents to teach them about sex on their own schedule.

Although I think broader society should allow conversations about whether men or women are better suited to particular careers, and conversations about sex, in these narrow situations censorship was really useful.

comment by MichaelStJules · 2020-01-06T20:40:25.739Z · EA(p) · GW(p)
These people also have a lot of antipathy for Peter Singer (for defending a parental right to terminate extremely disabled babies)

FWIW, this understates his position and the controversy. It's not just "extremely" disabled babies, but infants with basically any disability, due to a replaceability argument.

I don't expect disability rights activists to have much effect on selection and gene editing, though.

comment by willbradshaw · 2020-01-06T22:08:50.133Z · EA(p) · GW(p)

FWIW, this understates his position and the controversy. It's not just "extremely" disabled babies, but infants with basically any disability, due to a replaceability argument.

Can you cite this? I heard him talk about this in public (in Germany) and he focused strongly on the "extremely disabled" aspect. I'd be interested in how he makes the more general case, and how strongly he makes it.

comment by Khorton · 2020-01-07T01:55:44.150Z · EA(p) · GW(p)

This article is quite outdated, but it gives a lot of interesting information. I especially enjoyed the paragraphs about his friendship with a profoundly disabled woman, his regular donations to a disability charity, and his decisions in caring for his mother with Alzheimer's (all in the context of him essentially not believing that many people with disabilities have worthwhile lives and/or should be considered "persons".) Ignore the title. https://www.michaelspecter.com/1999/09/the-dangerous-philosopher/

comment by MichaelStJules · 2020-01-07T03:52:33.282Z · EA(p) · GW(p)
all in the context of him essentially not believing that many people with disabilities have worthwhile lives and/or should be considered "persons".

I think he doesn't (or didn't) believe infants are persons, regardless of disability status, and he argues that parents should be able to have their infant euthanized for any disability like they should be able to have their fetus aborted for any disability.

See this excerpt from Practical Ethics. Also, a specific quote from there:

In any case, the position taken here does not imply that it would be better that no people born with severe disabilities should survive; it implies only that the parents of such infants should be able to make this decision. Nor does this imply lack of respect or equal consideration for people with disabilities who are now living their own lives in accordance with their own wishes. As we saw at the end of Chapter 2, the principle of equal consideration of interests rejects any discounting of the interests of people on grounds of disability.
comment by MichaelStJules · 2020-01-07T03:58:13.608Z · EA(p) · GW(p)

From Practical Ethics. (It keeps going after this, but this is the basic argument.)

Given these facts, suppose that a newborn baby is diagnosed as a haemophiliac. The parents, daunted by the prospect of bringing up a child with this condition, are not anxious for him to live. Could euthanasia be defended here? Our first reaction may well be a firm 'no', for the infant can be expected to have a life that is worth living, even if not quite as good as that of a normal baby. The 'prior existence' version of utilitarianism sup- ports this judgment. The infant exists. His life can be expected to contain a positive balance of happiness over misery. To kill him would deprive him of this positive balance of happiness. Therefore it would be wrong.
On the 'total' version of utilitarianism, however, we cannot reach a decision on the basis of this information alone. The total view makes it necessary to ask whether the death of the haemophiliac infant would lead to the creation of another being who would not otherwise have existed. In other words, if the haemophiliac child is killed, will his parents have another child whom they would not have if the haemophiliac child lives? If they would, is the second child likely to have a better life than the one killed?
Often it will be possible to answer both these questions affinnatively. A woman may plan to have two children. If one dies while she is of child-bearing age, she may conceive another in its place. Suppose a woman planning to have two children has one normal child, and then gives birth to a haemophiliac child. The burden of caring for that child may make it impossible for her to cope with a third child; but if the disabled child were to die, she would have another. It is also plausible to suppose that the prospects of a happy life are better for a normal child than for a haemophiliac.
When the death of a disabled infant will lead to the birth of another infant with better prospects of a happy life, the total amount of happiness will be greater if the disabled infant is killed. The loss of happy life for the first infant is outweighed by the gain of a happier life for the second. Therefore, if killing the haemophiliac infant has no adverse effect on others, it would, according to the total view, be right to kill him.
comment by Bluefalcon · 2020-01-08T09:29:46.241Z · EA(p) · GW(p)

I think too many people ignore the "parents" part of singer's argument. If a parent is willing to kill their child, then either 1. That's a super f-ed up parent, or 2. it probably is the compassionate thing to do. Anyone who jumps to comparisons to racist euthanasia programs is missing the point entirely, because the kid's own parents are not going to act the same as some random Old South judge who's headed to the Klan rally after court gets out.