Alex Stamos: Asking tech companies to police hate speech is “a dangerous path”

Chats and phone calls aren't neutral in the same way as the content of a newsfeed or timeline. Who calls you or texts you is hardly a random population sample.

I’m not entirely up to date, but I believe there are calls for regulating WhatsApp communications

> No one is seriously asking Apple to police hate speech in iMessage or Facetime, for instance

Many people are demanding that Facebook do exactly this.

> They're not. And the specific reason they are not, is the algorithmic timeline and content suggestions.

I think the more pertinent reason here is that these platforms have broadcast capability (immediate communication with many people) as opposed to p2p capability (traditional SMS or phone calls). Even if Twitter was strictly chronological, without any algorithmic mutation, we'd still presumably be insisting they police content, right? I agree with your conclusion that they're publishers, but to me, what makes a publisher a publisher is not content curation or mutation, but is simply broadcast capability. And so our drive to regulate follows quite naturally from similar drives to regulate the press and media.

> Making decisions about what content your audience sees is an act of publishing, even if it's executed via complex algorithm. The companies doing this need to accept their responsibility for what they decide to serve and promote.

Repealing Digital Safe Harbour would be a good first step. If you are responsible for what people see, you are responsible for the content.

Beyond that, they need to be paying people, micropayments of course, for the content. That way they truly are considered a publisher and they employ independent contractors to create content.

> What people are concerned about are the newsfeeds and timelines, specifically. Companies like Facebook and Twitter and YouTube love to pretend that their newsfeed/timeline products are just like chat apps or phone calls--neutral messaging platforms.

> They're not. And the specific reason they are not, is the algorithmic timeline and content suggestions.

They're not because they're public, akin to broadcasting. One to many. In the past such has always been more or less under careful control. Public broadcast TV and radio were under control of dogma and moral, and it wasn't feasible to make your own. Publishers could opt not to release a manuscript if it didn't fit their ideology.

Consider the following thought experiment: "Twitter and Facebook were exactly as popular as they are now but they'd show everything only chronologically (last on top). Do you recon the control problem would be solved at that point?"

Now consider the following thought experiment: "Twitter and Facebook are only private 1:1 conversations. Do you recon the control problem would be solved at that point?"

In example #2 (regardless of it being chronologically shown or via an algorithm) the communication -whatever it might be- only goes to one person, not the general public. This contains the strength of propaganda (such as fake news or hate speech) greatly.

Also, remember that there are all kind of biases [1] even while we're not aware of them or when we are weak to fall for them.

[1] It is worth summing them all up but I am by no means an expert on this subject. I'm currently reading the book "The Confidence Game" by Maria Konnikova and it explains various of them in detail.

> Instead of pretending they don't make choices, they should be introspective and thoughtful about the criteria they are using to make those choices.

They're stuck, however, because many of the people who bought into the myth of objective, non-biased algorithms have gone down the rabbit hole of the garbage recommended by those algorithms. To those users, attempts to cull the garbage will be interpreted as censorship.

Add to that there is no way to improve the recommendation systems (from an ethical standpoint of "improve") without hurting engagement.

Add to that HN's allergy to government regulation.

It's going to be quite a rollercoaster ride over the next few years. :)

>Making decisions about what content your audience sees is an act of publishing, even if it's executed via complex algorithm.

This is true. And it is more difficult than most people want to even come close to dealing with. Let's say that you wave a wand and you have a bot that will 100% eliminate every racist claim ever made on the service, burying it and showing it to no one. You have just made it impossible to talk about racism. By killing the discussion, you will enlarge the groups of people who would never think to use a racial epithet, but who harbor deep convictions that different races have genetically-driven differences in capability and some need coddling. That is racism. But you can't even point that out, let alone discuss why it is factually wrong, on a sanitized platform.

On a sanitized Twitter, Megan Phelps-Roper and her sister would still be members of Westover Baptist Church, protesting gay funerals and spewing vitriol. They might not be able to do it on Twitter, but they'd be doing it elsewhere. Because Twitter was NOT sanitized, and because it WAS possible to confront people with total refutation and challenge to their most closely-held beliefs, Megan Phelps-Roper was convinced that her own position was wrong and destructive. And because of that lack of censorship, that permission to offend and call out, Westover Baptist has 2 fewer people working daily to hurt others. Anyone calling for a sanitized online platform is calling for a death of discussion, a death of social progress, and a death of any opportunity for the ignorant to learn.

In the 1960s, it was profane, disgusting, and obscene to suggest that interracial marriage should be allowed. It wasn't just 'a different opinion.' It was a view that made people sick, that riled up violence, that led to name-calling and hate. And it was only because the public forum was able to bear that hate, those insults, etc, that progress eventually happened.

Eric Schmidt in his book 'A New Digital Age' makes the argument that people like himself should take the reigns and kill discussion so that he might make the decisions for society. Were he around in the 60s, he would be fighting to lock down discussions about interracial marriage. He, and many like him, see the public having heated discussions and roiling in conflict and conclude they are mindless and incapable of policing themselves. This is a view as old as time. It's Conservatism. The old kind. The kind that backed kings, pharoahs, chieftains, etc. The kind that said some people are simply Better and destined to lead, while others are Lesser and destined to follow. Don't be surprised, but many are comfortable to accept that role as a follower if it means less responsibility or need to think. Conservatism died out near the end of the 18th century and through the 19th but there is no reason it couldn't re-establish itself with a fresh coat of paint and maybe with the help of some automation.

>No one is seriously asking Apple to police hate speech in iMessage or Facetime

facebook DOES censor Messenger.

> "Tech companies" is just way too broad a designation to use here. No one is seriously asking Apple to police hate speech in iMessage or Facetime, for instance, or Verizon to police hate speech in SMS.

Sure they are. Microsoft is even monitoring their service for "bad words".

https://boston.cbslocal.com/2018/03/27/microsoft-ban-offensi...

> What people are concerned about are the newsfeeds and timelines, specifically.

No. That's a small part of what primarily the left want to censor.

> It's silly to worry about giving these products "the power to determine what people can—and can’t—say online." They've already seized it for themselves--by deciding for me which content will show up in my newsfeed/timeline/suggested list. They decide which content gets promoted to me.

Which you can choose to ignore or bypass.

> So if you build a publishing system designed solely to maximize engagement, it's going to become a system that preferentially serves content that feeds negative emotions.

Then why aren't you demanding CNN or the NYTimes be censored?

> The companies doing this need to accept their responsibility for what they decide to serve and promote.

They are. They are serving what their customers want.

The only people who are complaining about it are authoritarian and selfish individuals who want to control what people see and say. It's no different than a prude whining about the porn people watch.

In countries like India, people are asking for Whatsapp to censor content.

10-15 years ago, SMS forwarding and bulk SMS were marked as a problem. Indian government had put in laws in place where they can and do ask telecom operators to jam mobile or even internet signals in sensitive areas.

In which case, asking technology companies to toe the line is the next logical step.

Even Mastodon instances can get in on the banning.

    > we are taking the power to limit speech (which the
    > founders felt was too important to be wielded by the 
    > government) 

Someone spray-paints a swastika on your car. Do you think the founders would mind if you painted it over?

>This is problematic in ways that used to be obvious to people in free societies, but for some reason seems lost now.

I don't think this is true. I think some people always got it, and some people still do. The difference is that the internet allows anyone to post their opinions, but it used to be a lot harder to reach other people.

We still have people that we hold up on pedestals for saying the right things, and the people that we remember from the past are the ones that said things that were incredibly right, or incredibly wrong. We don't remember what every common Joe used to say daily.

The issue is that the wealthy and also foreign adversaries are exploiting the algorithms to amplify speech that serves their interests. That typically does not serve the interests of average people.

The issue is how to avoid exploitation and manipulation.

When the KKK marched several decades ago, it got coverage in newspapers and media proportional to its influence in society. Today, the wealthy and foreign opponents can weaponize hate speech like this to fan flames of division for their own purposes. That is the problem.

Personally, I think it's probably less likely that the world has somehow lost comprehension of an entire "obvious" concept, and more likely that people would have always reacted more or less as they do now, and what has increased is the quantity, extremity, and sincerity of hateful expression (in general, not just legally defined hate speech), partially because some people are becoming more loudly hateful, and partially (mostly, I hope?) because mass communication is so dramatically different now.

> Putting aside that it's a slippery and malleable label to apply to undesirable speech, there is an elitism hiding in these ban-hate-speech arguments.

It's also about control. The biggest whiners about "hate speech" are the media and news companies. Because they want to control what people see and hear. They don't care about hate speech, they just want to be able to control which hate speech the masses get to see.

It's patronizing, it's paternalistic and it's also puritanically authoritarian. It's evil.

>Cockroaches thrive in the dark corners

Viruses thrive through exposure to new hosts.

See, you can prove anything using an analogy.

There is evidence that deplatforming works (https://motherboard.vice.com/en_us/article/bjbp9d/do-social-...)

Bleach is a better antiseptic than sunlight.

There's some evidence deplatforming works:

https://motherboard.vice.com/en_us/article/bjbp9d/do-social-...

> “We’ve been running a research project over last year, and when someone relatively famous gets no platformed by Facebook or Twitter or YouTube, there's an initial flashpoint, where some of their audience will move with them” Joan Donovan, Data and Society’s platform accountability research lead, told me on the phone, “but generally the falloff is pretty significant and they don’t gain the same amplification power they had prior to the moment they were taken off these bigger platforms.”

> There’s not a ton of research on this, but the work that has been done so far is promising. A study published by researchers at Georgia Tech last year found that banning the platform's most toxic subreddits resulted in less hate speech elsewhere on the site, and especially from the people who were active on those subreddits.

https://mashable.com/article/milo-yiannopoulos-deplatforming...

That's essentially John Stuart Mill's argument in On Liberty. Suppression of bad speech and thought only leads to a tyranny of the majority and can eventually lead to us forgetting what is bad versus good, just as we can't tell what darkness is if we only live in the sun.

I wish this were true. Sadly, evidence may not bear this out. If it did, then e.g. YouTube wouldn't be in a constant battle to take down ISIS recruiting content...

Cockroaches are also hardy and reproduce at an incredible rate. We've seen some cockroaches like white nationalism, that would've atrophied as demographics change, spread more rapidly. We now have a solution for a problem that we wouldn't need without the solution. Great.

Ideas aren't cockroaches, they're plants. And sunlight helps poison ivy grow just as much as it helps tomatoes and flowers.

There's no plausible mechanism by which talking about an idea more and more, makes it disappear. That's all an idea is: something to talk about.

Now, if you want to argue that no ideas should ever disappear, go ahead, but don't cloak it in "antiseptic" language. Antiseptic kills things. If you're talking about antiseptic for ideas, you're talking about killing ideas.

> Ideas are the only counter to other ideas and how we communicate those idea is via speech.

A lot of holders of bad ideas have no interest in rational debate. For those people, sunlight is an energy source, not an antiseptic.

When ISIS posts a video of a beheading, there isn't a lot of reasonable discussion going on in the comment section.

Not only is this analogy bad (sunlight helps plenty of bad things grow, things even detrimental to our survival as a species), but it assumes that you're dealing with rational actors. Stop naively assuming this. It's a bad assumption. There are people out there who will argue in bad faith and have no interest in reasoning or truth. And there are people who will fall victim to those bad ideas. Religion, terrorist factions, general cults, hell we can barely keep our own actions in check while we actively harm the environment.

The idea that if you just say the right, True (TM) thing then people will flock to it so not only naive it's so obviously wrong if you look at a laymen's perspective on basically any subject. It's also just a waste of time for the subjects typically in question. The communication of ideas has changed. It's a chaotic free for all. We've over done it. It's time to have a serious and reasonable conversation about the current state we find ourselves in, least we shoot ourselves in the foot with blind, headstrong optimism about ideals that don't much the reality of human nature.

In before someone references 1984 blindly and displays the ever-popular dystopia-prediction fetish that's so prevalent in these conversations.

The primary issue with this argument is the imbalance of the situation. There is an enormous difference in the amount of effort it takes to make a false claim (that sounds nice enough for people to believe without evidence) and the amount of effort it takes to refute said claim, especially if the refutation hinges upon mechanics that most people do not understand. In the unlikely scenario that most of the people who listened to both arguments happened to accept and understand the right one, there is still another huge hurdle. The fact is, the short, wrong claim is optimized for remembering and regurgitation. It is easier to remember that than to remember the complicated refutation, or even remembering that there was a refutation. Even in the ideal scenario where all other parties accept and remember the good argument, the person espousing the false claim can simply leave and continue to make the claim. When also taking into consideration the fact that most people do not have the knowledge and eloquence to make the good argument in the first place, it seems intuitive that the bad argument would still easily make its rounds.

I like to think of facebook type algorithms as amplification, not suppression. Its not what content gets deranked, its what content gets promoted to the top.

The question is, should "bad speech" be amplified, promoted, propogated, broadcast, surfaced, and repeated, ON PURPOSE; just so it can get rebuked, debunked, dismissed, and exposed?

(I agree with you, the answer is closer to "The Remedy to Bad Speech is More Speech ... Marketplace of Ideas" however, the question being discussed is more akin to "Theres a limited amout of space at the front page, and people have limited amounts of attention to give, WHO gets the megaphone, and for how long". Its the inverse of "the robust debate principle recognizes that sometimes in a crowd of speakers it is necessary to turn down the volume of certain loud and clamorous speakers in order to give others a chance to speak." Facebook and the algorithm DO DECIDE who to turn the volumn up on, who to promote to the top. They already arent neutral, they already exhibit preference and bias for certain ideas.)

Others are arguing that this is flipping the argument, BUT we are talking about algorithmic placement moreso than true censorship. If someone is allowed to post something but it NEVER makes it into someone elses Newsfeed, is it as good as censored?

Speech is either free or not free, there is no middle path.

If you want free speech, you accept the consequences. If you want “regulated” speech, there are consequences.

That’s it. I would argue that the level of satire a society can cope with, is directly proportional to the quality of democracy the society has.

>There's still too many unsolved philosophical questions here.

It's only unsolved among people who don't understand what free speech is.

There are no "compelling answers" because the problem at hand is how to maintain the positive branding of free speech while removing what it means for speech to be free.

Personally I think hate speech should be given a far more narrow and less catchy term to capture actual issues <ethnic> intimidation with ethnic substituted for relevant group or axis. The whole point of hate crime charges: it isn't to make the group sancrosanct but that it isn't just the crime of vandalism/murder but violence against the entire group akin to sundown towns lynching ethnic minorities. That suppression is dangerous to freedom and can be suppressed to its benefit akin to the paradox of tolerance.

That's how you end up with echo chambers and parallel online universes.

Modest proposal: The government should appoint a Department of Truth to review and approve all social media posts in a country to eliminate election misinformation.

I just read an article today about a named tech company banning marketing accounts associated with said far-right candidate. But I don't think that will stop him from a landslide win. The knife was the opposition's best hope.

What should these tech companies have done to prevent this?

>Well, these unnamed "tech companies" are responsible for the proliferation of absurd lies that will elect a far right authoritarian candidate in my country, Brazil.

There was no reason why the opposition to that candidate couldn't have put up their own absurd lies, or, dare I say, disproved the far right candidate's lies, thereby achieving the same success via the same platform. The elections are all about effective campaigning. You can't blame the platform because the candidate you don't like is too effective at using it. Instead learn from your mistakes and use this platform to be just as effective.

That is a strange definition of responsibility - the voters aren't responsible? They lack agency completely? The propagandists who would and have used other means like newspapers, radio broadcasts and even just whispered malicious rumors? They completely lack agency as well?

Tech is just the easy scapegoat for society.

Why are those the only two options available?

Speech isn't "dangerous". You know what you do about people that say stupid things: you call them out on it. The answer to hate speech is more speech.

> Now everyone is in on it, so we're getting confronted by parts of society we could pretend didn't exist 10 years ago. I don't know why I didn't see it coming.

Did you have the same biases I did? At that time and age (my teens, mostly), I just assumed people with different values than mine were ignorant, and so naturally they wouldn't be capable of using advanced technology.

I'm not proud of that, but there's still a _lot_ of that sentiment kicking around, including in the form that giving people additional access to technology and knowledge will educate the masses into the "correct" set of values held by whoever's pushing for greater technological adoption.

Your definition makes no sense. "We should kill people from country X" isn't hate speech by your definition - it doesn't imply anyone is any less human than anyone else.

I feel that this comment dehumanized me. Therefore, you must remove it. See the issue?

Yes. It's very easy to create a personal definition of what's offensive. If you were the King of the world, I'm sure it would be very easy for you to tell everyone what is and isn't good.

>Speech that explicitly or direct line implicitly dehumanizes anyone.

Except, you've gone nowhere. You've substituted the intrinsic ambiguity and subjectivity inherent in 'hate speech' with the same intrinsic ambiguity and subjectivity inherent in 'dehumanization' .. in fact, I have a better idea of what 'hate speech' is than what 'dehumanization' actually entails. I've seen people argue that a model in a bikini is dehumanizing.

I really don’t think so. You just replaced hate speech by dehumanizing speech but have the exact same situation. Same question then, what is dehumanizing speech, who defines it? The definition that you offer doesn’t seem too different to me than saying hate speech is content that you find hateful.

Speech that outrages, offends or angers another person or group is actually the most humanizing concept available. Anger, being offended, being rude are human traits. No they are not very positive human qualities, but that doesn't make them "de-humanizing". De-humanizing is a completely different concept that can happen in a variety of situations, but one is where one group dominates society and prevents individuals from having (wrong or not) concepts that are contrarian or against their agenda. For example, falsely claiming offensive speech is de-humanizing and rejecting the ability for others to have "offensive" models of the world is in fact, a projection of de-humanizing wrongly re-wrapped as "de-humanizing" itself. There are other concepts in de-humanizing, that are beyond the scope of this. But cenosring speech or rejecting models (positive or negative) in the name of de-humanizing is actually a de-humanizing itself. Being angry is a right as a human. Imagine if we passed legislation making it against the law to be angry. That would be so outrageous and against every thing that is human we know.

"hate speech" and "dehumanizing" are implicitly political terms. Will you will agree with me that one's politics are simply deeply cherished opinions and nothing more?

It appears to me that we are excluding voices from public discourse because, in the opinion of a powerful group, they have been rude.

One person's rude criticism can easily become another person's hate speech.