Comment by RickS | Hacker News Reader

Comment by RickS | Hacker News Reader

What's the expected fix here?

I was unfamiliar with the bands and most of the people in those targeting lists.

The way I see it, these are the ways you can handle this:

1) Facebook builds this data the hard way. They staff a team of experts on "undesirables", who research and implement custom blocklists at facebook's scale. Insanely cash and time intensive, to say nothing of the "who decides what's undesirable" problem.

2) Spread cost and effort by amassing a central repository of known baddies, and all the orgs contribute and share access. The government does something like this with hashes of sex trafficking imagery, so that eng teams can filter against a blacklist. I think this topic FAR more nuanced and less binary than "does this picture contain illegal pornography or nah". Who maintains this list of undesirables? You're at "social credit score" in a hurry.

3) Algos. You let software extrapolate commonalities from known-bad actors – school shooters, confirmed russian propaganda branches, etc. And let the machine learn their language and flag accordingly. This is going to be coarse and stupid in the way ML always is, and local business owners with names like Heinrich are gonna get their livelihoods smashed accidentally here and there. Not great.

4) What Simulacra said – you just turn the whole targeting infra off. Facebook stops making money. This is great, I'd love to see it as regulation, but it's a big stretch, and very lofty when phrased like this.

5) Some kind of adtech equivalent of finance's KYC (Know Your Customer) regulation. Tie ad buys to confirmable, prosecutable identities, and rather than filtering before launch, aggressively follow up after launch. You run an ad campaign for nazis? Cool, your LLC and its primary stakeholders are permabanned. Facebook has already tried light versions of this, but it was lip service.

IMO 4 and 5 are the places to spend effort. I think we nee to start having conversations that do away with the idea that humans are autonomous and impervious to influence, and start having the discussion in a new context: When and how are you allowed to manipulate the minds of citizens at scale, and what kind of paper trail does it leave?

I don't understand why we don't let such platforms adopt a more laissez faire approach to such situations. There's an inordinate amount of pressure to curb free speech these days which seems very un-American.

I was in this camp for a while but the current political reality has shifted me towards viewing the normalization and acceptance of such hate speech as a negative to society. When I was growing up nazis and white supremacy were always framed in a negative light to make it clear they were wrong (this is mostly relevant in the developing years, less than 14 or so where children don't have a developed moral compass) now a days the extreme right is being treated as just another opinion you can have, normal adults recognize the danger associated with unrestrained nationalism and hate speech but young people are unexposed or unfamiliar with what it can lead to.

This stuff is dangerous, it can warp the way you view the world, any information glorifying it should always be accompanied with explanations and warnings as to the hate it is imbued with... It is un-American to hate another person because of their origin or their religion, it's also un-American to squash an open discussion on that topic but... that discussion needs to happen between mature adults in a setting that makes it clear how unacceptable it is to lean on racist tropes.

Can we be clear on this, please? It's inhuman to hate another for their origin or religion - not just un-American. This problem isn't about whether or not you're a good American, it's about whether or not you're a good human.

I absolutely agree with this while acknowledging that in some circles being inhuman carries less weight than being un-American. I'm currently in Canada anyways[1] so... good luck down there folks!

[1] Expat, still a US citizen.

Hating people from outside your tribe is as human as any behaviour gets. It's a behaviour we have to work so hard to suppress precisely because it's so natural.

It's a throwback to our feral ancestry. I'd contend that the urge to overcome our animal nature is the bedrock of what makes us human.

Nazis and white supremacists are still shown in a negative light. In fact, I'd say it's even more negative now with people openly suggesting violence against them ("punch a nazi"...). I think the problem with all of this, and one that we're experiencing now is the widening of the definition of nazi and white supremacists. That's where the danger lies and that's where free speech is most valuable. Once you say it's ok to censor nazis then the game changes to redefining your opponents as nazis.

A lot of people will say people who wear MAGA hats are white supremacists for instance. It's a slippery slope, and a dangerous one.

An easy way to encourage unthinking hate towards a group is to falsely tag them with the term "nazi". Be careful here.

Uhh, that's what we have been doing the whole time. That's the status quo. We're talking about changing it because it's caused massive societal problems.

I generally agree with you, and I'm trying to parse through this myself. I think one difference in the modern era is that the Internet and social media have blurred the lines between broadcast speech and individual speech. Historically, we recoil at any restrictions on individual speech. They exist (can't yell FIRE in a crowded theater), but they are generally considered an unfortunate necessity. However, most people agree that broadcast speech should be censored.

Now we're in an era where speech can start out as individual, and then be broadcast. Or perhaps it's better considered a false dichotomy in the first place. Either way, we definitely haven't figured out how to handle this as a society.

Social media is in a way a hack similar to money and property law, where you can carry or own more than you actually could in natural way without losing it or having to leave behind, only that this time it's with the local space that hooks into the global space and vice versa.

Smashing natural boundaries like that creates anomalies that we have to learn to deal with. I think it's also something that will enable us to spread to other planets sooner or later, hopefully.

Hacking natural boundaries and overcoming the self regulating mechanism that otherwise would take place could be painful in a way and bite us back. But just think about the global exchange of knowledge, people getting interested in things they never thought they'd have access to under pre internet circumstances. A hell of a ride indeed..

Great comment, thank you.

I agree.

I think the only lasting solution to this is to raise a populace smart enough to be impervious to the radical left and right, both fringes being the domain of sloppy thinking and emotionally driven agendas. And in a way that keeps the flywheel of education spinning. We do a bad job of that today, I think.

But that takes time, strength, money, and – most critically – unified vision that I'm not sure America has right now.

So how do you implement education reform that takes 50 years, when nobody even agrees that it's needed, and kids are getting killed and radicalized today?

Do you allow it to continue in defense of the underlying principle of truly free speech? Maybe. To abandon that principle is a terrifying slippery slope.

This is just one of the ways where tech and culture vastly outpace science and regulation.

I have no answers.

Really? I've witnessed zero efforts on behalf of the government to curb free speech. But if you're talking about private speech that happens on privately owned websites, then that's flat out what is allowed by the constitution.

Seems to me that the issue is actually with private ownership being able to censor. Which can only be addressed with, guess what, government intervention.

The US has always had effective limits on broadcast speech, a history of bans on pornography and "communist" literature, as well as a number of quasi-self-imposed rules (Hays Code, MPAA ratings, Comics Code, "color bar" on radio and live music, etc). Effectively you could have very free speech so long as it was small-scale, but anything sufficiently controversial or offensive on a mass scale attracted attention.

Un-American maybe, but they’re global. Here in Europe, their behaviour is highly controversial and I believe that regulation would be popular.

In the specific context of Facebook ad targeting, reports and investigations concluded that it was used as a propaganda channel in the US by foreign national agents attempting to tip the scales in the US Presidential election in 2016. "Why is that a bad thing" or "Why is that Facebook's problem" is a reasonable question, but it won't be seen as popular in the US if Facebook accepts those questions publicly as its business attitude.

"Inordinate" is in the eye of the beholder. Free speech is curbed all the time in the interests of maintaining smooth operation of society, and as even the most intelligent among us tend to forget, free speech is not guaranteed on platforms like Facebook. There are quite a few laws on the books about hate speech, and how that extends to, say, Facebook ads, is what's being discussed here.

"Free speech is curbed all the time"

No, it really is not.

"There are quite a few laws on the books about hate speech"

In the USA I believe there are zero. [1]

There's a misnomer that free speech only exists somewhat arbitrarily as a bill of right. In fact, we actually value free speech even when granted by a private entity.

1: https://en.wikipedia.org/wiki/Hate_speech_in_the_United_Stat...

Letting the government use its monopoly on force to stop speech is un-American. Holding Facebook accountable for the voices it chooses to amplify and profit from is the market/society at work. Free speech was never something that makes distributors not responsible morally or ethically for what they distribute.

Separately, hate speech crosses over into a form of speech many societies, America included, have decided can cause direct harm and needs special treatment (including legal restrictions). People can disagree about that, but should probably get that disagreement sorted before diving into the additional questions of speech versus distribution.

...6) Full transparency at scale. When someone buys an advert in a newspaper or TV broadcast, it is visible to anyone watching.

In contrast, thousands of adverts can run in FB environment without anyone but the target able to see -- completely under the radar.

Starting with #5 KYC, and adding a site where EVERY advert of every type is available for public inspection, along with its (verified) originator info and targeting parameters.

This would allow all kinds of scrutiny by journalistic and public interest groups (e.g., researchers tracking hate groups, etc.).

FB Is making a bit of a start at this publishing some ads, but until they get to the full transparency, it can't be trusted.

The full transparency would also be a huge benefit to researchers.

I'd also be happy to see it required as a regulation for all players, not only FB.

I'd argue the rule should simply be that, the moment money changes hands, you can either tell me who paid you for a service, or you're legally liable for performing that service yourself. KYC for ads would then just be a natural consequence of this.

I would expect some combination of (1) and (3).

Facebook already necessarily employ a small army of moderators to remove illegal and "undesirable" material; they must have supervisors who set the overall policy direction and deal with new, emergent problems.

Companies that use social media on a large scale, and those companies that run social networks disguised as videogames, employ "community managers", whose job it is to understand and communicate with the community, including keeping abreast of disruptions. It doesn't seem that Facebook itself has many of these.

Facebook should get itself some "machine anthropologists", to study the ant farm. They can then get a sense of these problems before they get in the press, and definitely before they get to the Parliamentary committees. And feed the existing algorithms.

> Facebook should get itself some "machine anthropologists", to study the ant farm.

This is such a cool job title, I love it. I've long dreamed of doing this job at Twitter. There are so many blatantly patterned spam attempts and whatnot. I would love to work on analyzing, for example, the patterns in follower graphs surrounding templated bitcoin scam tweets.

The only viable solution is #4. So long as Facebook can make money by targeting specific groups or behavioral features, their algorithms and advertisers will find facsimiles for protected groups, the vulnerable, children, or Nazis since this optimizes revenue and engagement in dramatic ways. Even the most well-intentioned actor -- and Facebook is very far from that -- could not win this self-imposed game a whack-a-mole.

One alternative to shutting off targeting altogether would be switching from a blacklist to a whitelist approach, where regulators provide the set of features or groups that are allowed to be targeted.

Reply on news.ycombinator.com