Comment moderators should focus more on hate speech than profanity, study says
niemanlab.org
niemanlab.org
> Participants were exposed to a social media post that contained either hate speech or profanity. Then they were exposed to a post from a moderator — either a human or an algorithm — that deleted the initial post because it was offensive. This message either explained specifically why the post was deleted, gave a general sense of why it was deleted with a clickable link to community guidelines for the site, or offered no explanation. Afterward, participants answered questions about how fair or legitimate the deletion was and how transparent or trustworthy the moderator was.
> In all countries, survey participants thought that removing hate speech was more fair and legitimate than removing profanity, and found the moderators who did so and offered a detailed explanation of why to be more transparent.
:)
Seemed funny that calling someone a “f**ing idiot” is not banworthy but calling someone a “[insert your least favorite racial slur here]” is. So irrational race-based hatred is worse than just run of the mill irrational hatred.
That overlooks the degree of offense and threat. Some things are far more offensive and threatening than others. Clearly, hate speech is exceptionally offensive and threatening to many people.
Fuck. Shit. Asshole. My writing these words is rude, but has little impact. Even if I stuck them into my comments ('the fucking interface ...'), it would just be a little rude.
Hate speech can be a personal attack. Also, hate speech is often tied to power: If a powerful group uses hate speech against a vulnerable one, it can represent a serious threat: Humanity's history of the powerful abusing the vulnerable is long and terrible, and hate speech indicates, often intentionally, the willingness or intention to carry out such acts.
I’m not really offended by profanity or blasphemy, but many are. So I don’t think it’s based on number of people offended.
I think I understand why hate speech is particularly dangerous, but it seems funny that other offensive speech is deemed acceptable. For example, HN allows profanity and blasphemy but not racial slurs (whether targeted or not).
> it seems funny that other offensive speech is deemed acceptable. For example, HN allows profanity and blasphemy ...
I'm interested to know what your experiences are with this? I don't recall seeing a complaint about blasphemy in an Internet forum. However, I haven't quite yet read the entire Internet ...
What blasphemy is on HN? I could imagine statements that unknowingly violate a theological tenant - for example, to some using the word beginning G... for the monotheistic deity is blasphemous [edit: to some religions] (which, given the circumstances, I'll avoid). But I am not aware of any knowing blasphemy - there is so little religion here that it's not even an issue.
Hate speech directed at a religion would (or should) be outside the guidelines, but that's regardless of the blasphemy of it.
- “Jesus Christ, Use a Password Manager Already” - “ Revision 2020 Demoscene Event Live – Praise Jesus Christ” - “ So, just by looking at the box of pizza, you're going to judge that the pizza tastes bad? Jesus Christ man...”
Is calling someone a "fucking idiot" even any more offensive than just calling them an idiot? I don't see why.
So from a world view perspective, someone saying “Jesus Christ, you’re an idiot” would be part of an inferior perspective on people who think that blasphemy shouldn’t be done.
I think this is different from profanity, but I’ve worked with people who thought use of the “F word” was so offensive they would feel uncomfortable working with people who say such things.
I think it comes down to those views as not being reasonable, but everyone agreeing that being offended by racial slurs as being reasonable. So it’s unreasonable to ever use hate speech, but it’s reasonable to use blasphemy and profanity.
As an important nuance, if the speaker knows that someone will be offended, I think it assumes the views of the offended person are unimportant. That's different, though not necessarily any less bad, than seeing that person as inferior.
I can't anticipate everything that will offend anyone. But if I know someone is offended by something, I certainly want to respect them as people and avoid saying it. If I know someone is of a certain religion, I'd avoid saying anything about it that I wasn't sure of.
> I think it comes down to those views as not being reasonable, but everyone agreeing that being offended by racial slurs as being reasonable.
There's hate speech against religions too, and that's as verboten as hate speech against racial groups. In fact, for a moment I considered using an example of religious hate speech, but that would be completely inappropriate (and rightfully so, IMHO).
Blasphemy is a different matter than hate speech.
Edit: Maybe I'm a bit off-topic though. As for the question of whether two hateful comments should be weighted differently because one is based on race/gender/etc and one is just generalized anger, you could reframe that conversation as one about whether you agree with the concept of protected classes or not. Does hate against blacks produce more societal harm than hate against people who said somebody's preferred car manufacturer was shitty? I think so, even if the hate is just as loud and irrational in both cases. But it's definitely not a clear cut answer in every case.
If someone is not in a vulnerable position, than hate speech isn't much of a threat. If a Swede is locked in a room of Norwegians and says 'Norwegian scum should be taught a lesson', people might back away a little, but it's not much threat. But vice-versa, if a Norwegian says 'Swedish scum should be taught a lesson', then it's a very threatening situation to the Swede. (I'm picking two groups which, as far as I know, have little real-world animosity.)
Not easily. cf "Scunthorpe Problem" https://en.wikipedia.org/wiki/Scunthorpe_problem
> What counts as "hate speech" is necessarily impossible
Most problems like this can be broken into "the easy bit" and "the rest". Focusing on the thorny problems in "the rest" is a mistake if you want to do something about the easy bit (and a tactic, if you want to avoid it).
There are a few reasonable ways to approach this, free-speach maximimalist only one of them. I'm sure there are unreasonable ways too.
(The point being that determined posters can work around regexes trivially and that regexes have a high occurrence of false positives.)
A week later a new sign was being installed, “Thai Food”
So I guess a translation error, and no one thought to say something until after it was installed.
Expensive mistake.
For example, is the word 'Karen' hate speech? It is used derogatively against white women (see https://en.wikipedia.org/wiki/Karen_(pejorative): "the term was increasingly being used in media and on social media as a general-purpose term for middle class white women"). It seems to fit the definition - pejorative, discriminatory, based on colour, gender (and economic class?)... But it is the fashion for this term to be in widespread use and for it to be applied generally to dismiss and demean this demographic (The Guardian even calling 2020 the Year of the Karen).
Introducing rules based on subjective standards will make it impossible for them to be impartially applied. It is difficult enough apply laws even in cases where it should be far easier to judge rights violations (workplace bullying, rape, domestic abuse). Subjective rules will be open to abuse.
(Some caveats:
To talk about the elephant in the room, we both know we are engaging in an old debate about double-standards and 'reverse discrimination'. Let's not pretend it's some novel thing we're proposing.
Also, I don't know about the credibility of the first citation's definition of hate speech, and I think Wikipedia has little credibility (no idea why I'm suddenly seeing so many people citing it on HN) - it might as well have been written by you or me. So I'm not commenting on those.)
Some points:
* There are degrees to threats and danger, of course; communication depends on words, context, emotion, the parties involved, etc. Every insult is not a threat and does not rise to the level of hate speech in a significant way. Perhaps we should use the term 'threatening hate speech', to distinguish it for these purposes.
* There is a clear standard necessary to 'threatening hate speech' that is omitted from the reverse discrimination argument: The vulnerability of the target. To reuse an example I gave in another comment, if a Swede is locked in a room with 30 Norwegians and says, 'Norwegian scum should die', people might back away a bit but it's not much threat. But if a Norwegian says 'Swedish scum should die', that's a threatening and dangerous situation for the Swede. In U.S. law, this standard is expressed as 'protected classes' (afaik). Hate speech is only a double-standard if we omit this factor. The consequences to vulnerable people are very well established and very dangerous, arguably the worst events of human history; we can't just ignore that because we lack the perfect solution. Doing nothing is not the null hypothesis, it's an alternative with clearly worse results.
* Subjective judgments are less desirable than objective measures, but that's life; the world is just too complicated and in almost every case we don't have time for scientific accuracy. It's a theoretical argument to require objectivity; reality, as everyone can't help but observe, requires and uses endless judgment. That doesn't make all judgments good or equal (another theoretical argument). Note, for example, that courts have a complicated system designed to output good subjective judgments by judge and jury; they expect and accept subjectivity. But to the extent a rule requires subjective judgment (which is every rule, as every one has edge cases), I agree that it would be better to do it objectively, we should do that where possible, and that we have to be careful.
For info, the first citation is from the 2019 UN Strategy and Plan of Action on Hate Speech (https://www.rightsforpeace.org/policy-docs), so is a pretty influential body for this topic and may well affect policy. It is true that Wikipedia should not be relied upon as an authoritative resource, but the link to the Guardian article can be found here: https://www.theguardian.com/world/2020/dec/27/karen-race-whi...; how broadly the term is applied can be gauged from the article or other instances where it appears on social media.
Addressing each of your points:
1. I agree that every insult is not a threat and its severity is subjective. One person might be hurt or even feel threatened by seemingly innocuous language whilst another might shrug off deliberate, targeted insults; simply using a person’s name or words like ‘boy’ or ‘girl’ can be threatening if the other person is about to be violent. People can even deliberately claim offense to exploit the subjectivity in dialogue. In this instance we are in danger of granting a term a degree of recognition which it is actually too subjective to deserve and this is opening up opportunity for misuse.
2. The vulnerability of the target argument is again subjective; 1 violent criminal amongst 30 sheepish academics will feel zero threat whereas the 30 will all be intimidated by this person. My experience has been that bullies are generally a minority in a pool of people – but weak people will follow and support them. History is full of instances where subordinated groups have vastly outnumbered the minority which holds sway over them (the feudal system, caste systems, colonialism, crime bosses, corporate org structures, etc.). Power systems are not necessarily gauged by the number of people in the group – a lion will not be threatened by a herd of sheep.
3. I agree that centralised micro-management of this issue is not possible or even desirable. Will some people have to overcome prejudices against them irrespective of whether they have the same rights as everyone else? Yes. For example, women may be regarded as the weaker sex or more emotional, whilst men may be regarded as dictatorial, insensitive ‘mansplainers’. Should we try to even things out for them by introducing subjective rules that favour them above others? I do not think so.
> I agree that that using destructive blanket terms for groups and making caricatures of people is not novel ...
You are agreeing with something I didn't say. I said: "we both know we are engaging in an old debate about double-standards and 'reverse discrimination'. Let's not pretend it's some novel thing we're proposing"
Also, you keep using the world 'subjective' like it is disqualifying. You can see my GP post for why.
The idea that we don't know about many vulnerable groups in our society is absurd and theoretical; there is plenty of objective evidence and it's obvious who they are. I'm not even sure what you wrote really says anything to the contrary.
It would have been great to have a genuine conversation.
Why do I not accept that the attempt to classify things as 'hate speech' is about 'reverse' discrimination (i.e. I do not agree that this is something 'we both know')? Because this implies that it seeks to react to and target initial discrimination. But because it is subjective and open to abuse it can be used opportunistically against individuals who have done nothing to the other party (no initial discrimination to reverse). So, it is a label that can itself actively be applied in a discriminatory way.
Which are these 'obviously vulnerable groups' that you think 'it is obvious who they are'? Children - agreed. People with some disabilities - agreed. Some frail elderly people - agreed. Even with this set of groups, the degree of vulnerability will vary hugely. Are these the groups you refer to?
As far as your assessment of hate speech is concerned, I read it and was about to compose similar arguments to those of Andy_G11. I have little to add.
As far as what I personally think about your premise that certain speech from certain individuals should be classified differently whether in law or in general society, it's bunk. I don't know how you think any such law would functionally work in a world with free press, mass publishing, the Internet, etc. and, even if followed, I don't see how "hate speech" laws won't be abused to silence individuals. If you want to see how laws like this that "work" in practice, look up lèse-majesté.
Disagreeing with someone is not hate speech. Yet that is an easy way to shutdown your opponent.
That's obviously a simple example, but even in real examples, I don't find them hard to distinguish, as long as we ignore the endless trolling and theoretical arguments.
While concepts like "hate speech" very easily turn into arbitrary censorship. So it's easy to say people should focus on hate speech more, but there be dragons.