[snowwrestler]

"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. No one expects AT&T to police hate speech in a phone call.

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.

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.

Yes they use an algorithm to do so, instead of human decisions. But guess who built the algorithm?

Companies that run algorithmic newsfeeds and timelines need to own their role as a publisher and a gatekeeper of content.

Instead of pretending they don't make choices, they should be introspective and thoughtful about the criteria they are using to make those choices. "Engagement" is not neutral criteria because emotions are not symmetrical. Engagement is higher on topics of fear, anger, rage, violence. That's down to our evolution; that's down to the amygdala.

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. There are articles and case studies where a person starts with a fresh account and sees what kind of content gets pushed to them; inevitably they get horrible conspiracy theories and fear-oriented content.

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.

[padolsey]

> 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.

[Fellshard]

Not really.

One involves a neutral role, in which subscribed feeds are delivered to users without modification or filtering.

The other involves an active role on the part of the platform for any number of reasons: increased engagement, removal of voices that may cause perceived damage or lack of trust in the platform itself, or other, more ideological reasons.

I've noticed an intentional avoidance of distinction, lately, between active and passive behavior on a number of fronts, from sexual activity, to medical advice and intervention, to social media publishing. It's a pretty crucial component in ethical analysis that I suspect is being intentionally blurred.

[padolsey]

The public wouldn't buy the 'neutral' aspect, as every mechanism that a platform provides, is in some way biasing the type of content that is broadcast. Twitter's RT feature, which requires no curation or modification of actual content by Twitter, still biases content to that, perhaps, which is most divisive or simplistic. Broadcast technology, even without algorithmic bells and whistles, is already a biased technology. I think we might in-fact agree, but to me, there is no crucial ethical difference between a chronological twitter and a algorithmically mutated one, for in either case, the very platform itself (its existence, its design) pre-biases the type of content that will flow through it, and thus we'd end up with content that would make us consider policing it.

So, in your words, I would say, it is somehow fundamentally impossible for a technology to be "passive".

[leereeves]

Retweeting doesn't bias content; it's rather the opposite: making it easy for people to share content they like removes a longstanding bias. The old models of broadcast media selected content to fit the biases of a few powerful media executives.

You may find the choices of the average person "divisive or simplistic" but the Retweet button doesn't dictate their choices.

"Policing" content, however, is all about dictating people's choices, motivated by the thought that the "police" know better than less powerful people, and reestablishing the biased filter controlled by the powerful (new-)media executives.

[Fellshard]

To an extent, but if they provide the information through a set of transparent protocols and mechanisms that users can comprehend, they are remaining as neutral and passive as possible. And that's key: Twitter, at its core, started out as something extremely comprehensible, and grew less so as it began to paint over its transparent abstraction.

[734786710934]

> 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.

[zaidf]

It’s even worse: many people are demanding Facebook monitor Messenger chats for abuses in Mayanmar while at the same time not wanting Facebook to monitor Messenger chats. This is just a perfect example of constant catch-22s I see in people’s expectations of tech companies. You can’t expect Facebook to both not monitor chat messages AND prevent chat messages that do harm. You can have one or the other.

[geodel]

This is the doublethink the most people like to indulge in. The free speech I like must be protected at all cost and that I do not like should be handled via glib statements: 'free speech is not free reach', 'private companies are not obliged to give platform to anyone' and so on.

[ceejayoz]

> You can’t expect Facebook to both not monitor chat messages AND prevent chat messages that do harm.

Not monitor, or not use for advertising targeting?

I'm fine with spam/malware/virus prevention in my Messenger or Gmail. I'm not enormously comfortable getting an ad for baby clothes after I send a private message to someone that I'm pregnant.

[dwighttk]

wasn't it WhatsApp and message forwarding that people had a problem with in Myanmar? Not p2p.

[pas]

India, but yes. And again, not exactly p2p, because large (meaning 100+ people group chats) that spread false alarms.

[dwighttk]

I was reading something about Myanmar[1], but I got it conflated with another article I can't find talking about WhatsApp message forwarding.

[1]https://www.nytimes.com/2018/10/15/technology/myanmar-facebo...

[pas]

Can you please show one such instance of impossible demand? I thought that's two very distinct group of advocates.

[jancsika]

> 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. :)

[gbear605]

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

[rashkov]

not one-on-one communications, as far as I've seen. However there are some calls to limit group chats:

"There are good ideas floating around for how Facebook could make life harder on WhatsApp propaganda artists. In an op-ed published in the Times this week, Brazilian researchers Cristina Tardáguila, Fabrício Benevenuto and Pablo Ortellado offered three ideas: restrict the number of times a message can be forwarded from 20 to five, which Facebook has already done in India; dramatically lower the number of people that a user can send a single message to, from its current limit of 256; and limit the size of new groups created in the weeks leading up to an election, in the hopes that it will stop new viral misinformation mobs from forming." https://www.theverge.com/2018/10/19/17997516/facebook-electi...

[thisisit]

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.

[basch]

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

facebook DOES censor Messenger.

[kazinator]

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.

[otakucode]

>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.

[rhacker]

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.

[qubax]

> "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.

[Fnoord]

> 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.

[51lver]

It absolutely is not broadcast as long as an algorithm is selecting posts for you. Air TV is broadcast. The radio is broadcast. Those platforms do not not selectively choose their audience to maximize engagement. There's no innocent one to many relationship here. Facebook actively gives extremist material to extremists because their robot thinks it will make them use the platform more. It's not the same message going to all subscribers. That is why they are responsible.

[Fnoord]

My point was that neither phenomenons are exactly new.

One-to-many relationships aren't innocent to begin with; they've always been under public scrutiny, a magnifying glass. Algorithms might make it easier to find what you seek (I can assure you they do not always as I've witnessed on Facebook, Google, Amazon, Netflix, Apple -- you name it).

Bubbles are also not new. If you were a Catholic in The Netherlands in 1950 or 1960 then you watched Catholic TV and listened to Catholic radio and went to a Catholic church on Sunday and listened to a Catholic preacher and a Catholic pope telling you what to think about atheism, abortion, HIV, anti-conception, marriage, homosexuality and what have you and you went to a Catholic dancing. You came home with a Catholic partner of the opposite sex. Oh and you went to a Catholic school. Protestants? They exist, somewhere, but not in your myopic world. [Full disclosure: I grew up as an atheist child of Protestant parents in a Catholic area.]

Its a matter of choosing your overlord(s)...

[golemotron]

> 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.