ACLU reaches settlement with Facebook over discriminatory ads | Hacker News Reader

ACLU reaches settlement with Facebook over discriminatory ads

axios.com

I've never really understood how targeted ads, aside from things like job or housing opportunities, would be considered harmful discriminatory. Yes it's discrimination in the basic sense of the word - the same way that my company discriminates between men and women in our work experience surveys so that we can see any differences in trends.

If I'm selling high heeled shoes, why should I not be able to target my ads to women? If I'm trying to solicit support for, say, a charity focused on helping South American immigrants why should I not be able to target Latin immigrants if I have a belief that they will be more invested and likely to follow up? If I built a new church why shouldn't I be able to advertise to Christians?

Certainly, using such categories for job and housing advertisements should not be allowed. But if I'm reading Facebook's announcement correctly targeting on race and gender is being completely removed [1]. I'd make more sense for me to just forbid it for jobs and housing. If that's too difficult, explicitly limit it to selling people consumer goods like shoes, concert tickets, and video games.

1. https://www.facebook.com/business/news/keeping-advertising-s...

> aside from things like job or housing opportunities

That's what this was about. The ProPublica article explains it better than Axios: https://www.propublica.org/article/facebook-ads-discriminati...

Exactly. Here’s another article that makes it clearer: https://www.cbsnews.com/news/facebook-ads-will-no-longer-dis...

This case was about targeted ads for housing, credit cards, etc. You can't target people of a gender or color for such things by law.

Facebook already forbade using those categories for targeting ads for those things:

> For over a year, we have required advertisers we identify offering housing, employment or credit ads to certify compliance with our non-discrimination policy.

It looks to me like they're removing this as an option for all types of ads:

> We’re committed to protecting people from discriminatory advertising on our platforms. That’s why we’re removing over 5,000 targeting options to help prevent misuse. While these options have been used in legitimate ways to reach people interested in a certain product or service, we think minimizing the risk of abuse is more important. This includes limiting the ability for advertisers to exclude audiences that relate to attributes such as ethnicity or religion.

Facebook forbade those only after reports. There were further issues followed by ACLU which ended in agreement that requries further changes.

Chain of events: 1.) A series of investigative reports over the past two years have uncovered ways that users can abuse Facebook's ad platform in order to discriminate against some populations, mainly people of color, when it comes to certain marketing for certain opportunities, like housing, employment or credit.

2.) Following the reports, the Department of Housing and Urban Development (HUD) filed a complaint against Facebook [...], charging that the social network allows advertisers to illegally discriminate in housing ads by excluding some groups from seeing the ads.

3.) Shortly after, Facebook removed over 5,000 ad-targeting options to prevent that capability.

4.) In September, the ACLU and its law firm, representing the Communications Workers of America and several individual job seekers, filed charges with the Equal Employment Opportunity Commission against Facebook and a number of employers, alleging that they had unlawfully discriminated against certain populations via their ad targeting platform.

5.) As part of the settlement, Facebook will pay $5 million to several groups, [...] It will also take three new steps to prevent advertisers from engaging in unlawful discrimination around employment, housing, and credit ads on Facebook and its subsidiaries (Instagram, Whatsapp, Messenger, etc.):archive for housing ads,

"It will also take three new steps to prevent advertisers from engaging in unlawful discrimination around employment, housing, and credit ads on Facebook. [..] As part of this process, Facebook is further cutting the number of options advertisers [in housing, employment, credit] can use to target ads"

For housing, would it make sense that an apartment complex inside a Chinese community targets ads on Chinese speakers, who are much more likely to live in there? Is this considered as discrimination based on color?

What if the landlord only speaks Chinese?

One of the goals of the law is to hinder ethnic ghettoization like that.

An ethnically homogenous area doesn't always imply it's a ghetto though?

The original meaning of "ghetto" was to be ethnically homogenous, it gained the additional meaning of being ramshackle or of poor condition.

Why should it be the goal of the law to hinder prosperous but ethnically homogenous areas? Personally I like my diverse neighborhood, but I also like that Chinatown is just down the street.

Because the reality of racial restriction covenants and redlining practices is that they weren't used to develop great locales; they were used to systemically exclude people of undesired races who otherwise had the means to live in said area.

This doesn't mean that ethic areas just melt away, though. Chinatown can remain Chinatown on the basis that they value being close to Chinese community centres and Chinese grocery stores more than the average resident, meaning they're willing to pay more and price out other people to live there. They can't remain Chinatown by literally refusing to rent to someone because they're black, though.

This allows ethic areas to remain if they provide a benefit to their community, but melt away to redistribute real-estate resources more efficiently if they don't.

isn't the point of things like rent control and affordable housing to specifically prevent people from paying more and pricing out people who want to live there?

It comes down to this issue of whether or not more housing is available. In New York, there's effectively a finite cap on housing, so you can't just say "well buy somewhere else". Pricing caps and assistance programs make sense there, but those also come with less ability to control your neighborhood.

When there's tons of cheap housing available, then having small areas with expensive pricing isn't per se a problem (but could be an issue in context, say a small midwestern town excluding black people from buying houses by over-inflating prices.)

>you can't just say "well buy somewhere else"

You can, for the most part, if you don't make it harder than it needs to be to build apartment complexes.

You can't put 256GB of ram in your desktop and expect it to run faster.

You can't replace one house with 16 apartments and expect it to work as well.

Parking, noise, litter, etc. It's not fair for a developer to push their direct costs onto the neighborhood instead of eating it themselves.

The answer is obviously contingent on your values and maybe your prediction of sociological outcomes.

It's important to notice that this is as pure as racism gets. I mean, why shouldn't you be able to ethnically cleanse a place if the landlords prefer a white clientele?

It's racist to advertise to the group that is most likely to buy your product? And advertising is equivalent to ethnic cleansing? This hyperbole is ridiculous.

> Certainly, using such categories for job and housing advertisements should not be allowed. But if I'm reading Facebook's announcement correctly targeting on race and gender is being completely removed

The problem is "such categories" is squishy. What about elective medical treatments? What about vacation deals (often partially sponsored by real estate interests)?

There is a broad, broad spectrum between "elective frivolity" and "hidden opportunity" in advertising, and who are you to make that decision? Step out of the obvious counterexamples and scary edge cases abound.

I mean, yeah, advertising "soda for women" or whatever isn't going to harm anyone who doesn't see the ad. But would you say it's OK if I pitch my startup camp to asians only?

> What about elective medical treatments?

Should companies have to advertise mammograms to men? Or vasectomies to women?I would object to this - not because of targeting certain populations but on the grounds that ads for medical treatment should not be run at all.

> What about vacation deals (often partially sponsored by real estate interests)?

Why not? I don't see how people would be negai impacted by not being exposed to vacation deals. Jobs, credit, and housing are necessities and if a group has lessened exposure to those opportunities they are at a disadvantage. Trips to Vegas are not necessities and not being exposed to deals to go to Vegas are not harmed (one could even argue that the ones being advertised to are the ones harmed here).

> Should companies have to advertise mammograms to men?

That's the same strawman though! There are an endless list of products that are clearly targetted in some kind of inherent way. My point is that there is an equally large list of ones that are gray areas, and that if your policy is "advertise whatever to whoever you want" you're going to be engaging in an awful lot of accidental discrimination.

So sure, make your exceptions for tampons and birth control. Just let them be exceptions.

> Why not? I don't see how people would be negai impacted by not being exposed to vacation deals.

You say that now, until, oops, it turns out you've been running ads for condos in a de facto whites-only resort community. Who could have known, amirite?

Whats wrong with running ads for condos in a whites-only resort community? If their internal numbers show that this community has a really high chance of conversion, why should they target something else on a finite ad budget? Can't a company decide who it wants to sell to?

Are there de jure whites-only resort communities in the US?

To answer your questions:

> Whats wrong with running ads for condos in a whites-only resort community?

- profiting from racists

- ghetto creation

- perpetuating harmful discrimination

- assisting in the longevity of a community which exacerbates the prior 3 issues

> If their internal numbers show that this community has a really high chance of conversion, why should they target something else on a finite ad budget?

They shouldn't. But also it shouldn't be available to them, by law.

> Can't a company decide who it wants to sell to?

In the general case, yes, but since the civil rights movement, you cannot refuse goods or services to someone based on race. That includes housing. I was under the impression this was widely known.

All ads for medical treatments should be illegal

So ignorant consumers become totally reliant on getting medical information only from doctors?

Or independent reporting. Or public information offers. Ads are not a good way of distributing good information. And people who argue otherwise are naive, or doing so in bad faith.

There's no way to distinguish an ad from those things. It's analogous to distinguishing between porn and art.

Paying for placement is a clear boundary between one and the other, isn't it?

What if somebody slanders your product?

Figuring out astroturfing isn't a perfect system, but kicking out TV/radio/magazine/banner/streaming ads is straightforward and beneficial.

You think people are getting better medical information from ads than from doctors? I don't know about you, but if I started taking a new medication every time I saw an ad saying I needed to I would be on a lot of boner pills

I consider my doctor an adviser, not an oracle.

I'm not saying anyone should blindly trust their doctor, and I would advise anyone to do their own research before accepting any kind of medication/treatment, but I agree that advertising medication should be illegal and provides essentially no beneficial or unbiased information. The goal of ads for medication it to make money, not to help people

Good sources of health information exist outside of doctors. Lots of governments, health insurance providers, colleges, and so on have good information.

Not to mention pharmacists, nurses, and the like.

And, of course, doctors. Why the heck would you trust an ad for a medical advice over a doctor - or set of doctors - giving you care after having been educated in the medical field well longer than yourself?

There is also a slew of information the general public doesn't really understand. Like why New Shiny Drug X isn't really an improvement for you than Slightly Older Shiny Drug S. Or why the doctor is leaning towards disease X instead of Y even though the symptoms are similar.

And then there is the bottom of the barrel. The outright dangerous stuff. David Avocado Wolf, for instance. The "Clean Eating" bandwagon. Being vegan without proper nutrition information, which is often lacking in the material. Or anti-vaccine movements that claim measles are good for folks, lying to say it prevents cancer. If you are lucky, you'll just wind up with one of the "cures" that aren't cures, but not actually harmful. Airborne for colds, for example, or most homeopathy that actually is. Or you might get mildly unlucky and wind up pregnant for not realizing St John's Wart can interfere with birth control. Or maybe Zicam will cause loss of the sense of smell.

Considering this last bits of information... yeah, I wish folks would only get their information from accredited medical professionals (though this includes more than doctors).

For a historical example,

https://en.wikipedia.org/wiki/Tyndale_Bible#Reaction_of_the_...

The argument against Tyndale was essentially the same.

And what if a new hip eatery wants to target only white customers? What if thier adds are attached to a coupon? Such tools can quickly become the modern equivalent of a "whites only" sign above a door.

Ignoring the glaring hyperbole of your comment, if their main demographic is white people, then they would be doing themselves a disservice by not spending the majority of their ad revenue targeting white customers, and there's nothing wrong with that.

there is something wrong with that, though.

But there isn't. There is, however, something wrong with hanging a "whites only" sign above your door, but that was the hyperbolic part that I was ignoring, since it was not comparable to the first two sentences of the comment.

Where would "this coupon only good for whites" lie on your comparability scale?

It's the same effect, right?

Sure, you can keep moving the goalpost as much as you want, but offering a coupon in an ad targeted to a white demographic is not the same as refusing to accept that coupon from a non-white patron. Apple simply do not compare to oranges.

Why not? It's the same results, right?

>I've never really understood how targeted ads, aside from things like job or housing opportunities, would be considered harmful discriminatory.

The question to me is, if it is so clearly wrong to do in these cases, then there must be a justification for it. To what extent could that justification be applied to other areas of life. Even worse, we do sometimes allow such discrimination, such as only allowing people 55+ to live in a retirement community is legal, so there must be some justification for why.

Take a boy who wants to play with barbies but sees it only being advertised to girls. This feeds into the social expectation and can result in both negative internal feelings as he sees the social expectations and compared it to his own desire as well as creating a social more than he is now punished for violating. It is thus harmful. The question then remains what level of harm reaches the point to make it immoral or illegal.

> Take a boy who wants to play with barbies but sees it only being advertised to girls. This feeds into the social expectation and can result in both negative internal feelings as he sees the social expectations and compared it to his own desire as well as creating a social more than he is now punished for violating.

Sure (save for the negative part, which I'll expand upon later).

> It is thus harmful.

Citation needed. You're approaching this story with the a priori conclusion that any gendered social behavior is bad. If you're trying to convince me that boys are harmed by the fact that we don't advertise high heels and Barbies to them you're going to need to back this up with strong evidence.

You're also assuming that a significant number of boys want to use such things but the advertising prevents that. I recall a couple studies in which children were raised from birth or early childhood in gender neutral environments, but the kids displayed gendered preferences and behavior much like the rest of society.

Advertisers don't have machinations for society. They just want to make as much money as possible. A very probable explanation is that these behaviors occurred on their own, and advertisers perform targeting in response in order to maximize sales.

>You're approaching this story with the a priori conclusion that any gendered social behavior is bad.

No, I'm saying that the enforcement of the social norm on the child is wrong. Social norms are enforced through punishment, though the punishment is often very light (such as being the butt of a joke or even just being asked to justify one's behavior, but these are separate treatment than those who conform receive). At worse levels it can lead to ostracizing and bullying.

>If you're trying to convince me that boys are harmed by the fact that we don't advertise high heels and Barbies to them you're going to need to back this up with strong evidence.

You seem to be reading this backwards. It isn't the lack of advertisement, it is that the existing advertisement feeds into norms which are then enforced against those who don't meet those norms.

>You're also assuming that a significant number of boys want to use such things but the advertising prevents that.

I am assuming no such thing. There needs to only be a single individual harmed by the social expectations fed by the advertisement for it to be harmful.

>I recall a couple studies in which children were raised from birth or early childhood in gender neutral environments, but the kids displayed gendered preferences and behavior much like the rest of society.

Yes, kids do tend to have gendered preferences in what toys they play with. That you bring up something unrelated to my point makes me think there has been some misunderstanding of what I'm trying to say. So let us give a different example and skip gender since that seems to lead to a lot of incorrect assumptions being made about what I'm saying.

Go with an advertisement campaign that shows kids who play with a specific brand of soccer ball as being better than those who don't, either because of lack of access or lack of interest. If the marketing is aggressive enough, which isn't hard when dealing with young children, it can lead to children discriminating and engaging in social punishments against those who do not have the status symbol.

While the exact advertising leading to it is far more complex than a single commercial, this is a real life problem with brand name clothing and displays of wealth among school age children (especially preteens and teens). Children who cannot afford the brand named goods to display to peers are treated negatively compared to those who do, which is a harm created by the advertisement. It can lead to some extreme cases of discrimination depending upon how strong the message sent to children is. The classical example being the teacher who told students of different eye colors that some were naturally better than others, a message that is often largely mimicked in advertisement with purchase of the brand swapped for eye color.

>A very probable explanation is that these behaviors occurred on their own

I've never known any animal other than human have a preference for Nike over New Balance.

I'm sure sneaky landlords can get around this. There are plenty of demographic categories that correlate with race that I'm sure are still available on FB.

Definitely, and that will make it harder to detect, but if you suspect it's the case and manage to sue, it becomes pretty obvious. They'll have a tough time explaining to the court why they chose to target their apartment ads to, say, people who buy sunscreen. And trying to pick a more subtle demographic will come off as even more ridiculous.

I can imagine seemingly justifiable proxies. Targeting people who buy lawn care products and people who buy wild bird food is both arguably enriching to the neighborhood (birds and grass look nice) and probably correlates to WASPs fairly well. I remember the bird food example being brought up as something that credit ratings are influenced by, but I don't remember the source.

Edit based on sketchy memory: s/organic/wild

Those demographic correlates of race are already illegal as substitutes in similar domains (like qualifying for a mortgage). If they aren't illegal already for (e.g.,) landlords, it seems like an easy expansion of current regulations.

One of the measures is a public archive of all ads in these categories and their targeting.

That allows third parties to analyse ads for any remaining bias.

> The big picture: Unlike television or radio, internet platforms that sell ads aren't regulated.

I'm surprised that this continues to the case, especially in examples like this where EXISTING regulations seem to fit pretty squarely with "new" problems created on tech platforms. There's some really interesting work on how differently tech companies are regulated from other kinds of companies (e.g., here: https://innovationforallcast.com/2019/01/23/amanda-lotz-what...)

Along a similar line... I'm always surprised we don't see more generalized less-targeted advertising on the web for regular products like Coca-Cola or Charmin, etc like you do on TV. Perhaps with click-through to online games for engagement or coupons, etc.

Proposed reason that just popped into my head, and could be wrong: advertising that cannot be targeted to individuals is less effective, so there is less competition for it. Thus prices are low relative to ads that can be targeted. Thus companies with a wide target demographic can buy these spots cheaper, while companies that are targeting a specific niche get more bang for their buck from relatively expensive targeted ads.

The size/scale of Twitter or Facebook, etc may be an example where it might be better for generic inline advertising over the "promoted" messages. I can understand where a smaller site may do better with random targeted ads... I'm just a little surprised it hasn't been done or tried much that I am aware of.

I don't see why scale should affect whether targeting makes sense (with the exception of entities like facebook that have collected so much data on their own that they can target better than smaller competitors, which leads targeting increasing with scale, not decreasing). Television isn't targeted because it wasn't designed to be, and hasn't been updated (yet) -- they would need to deliver different ads to different homes. This is already done for different localities, but not individual homes (that I'm aware of). It's not a scale issue. Targeted billboards are coming[0], albeit not based on individual browsing habits (yet). I don't think this is about scale (mostly), I think it's about the technical feasibility of tracking and targeting customers on a given platform in $CURRENT_YEAR.

>random targeted ads

I'm genuinely confused by what this is supposed to mean. To me, random is antithetical to targeted. In my reading I decided to drop the random, but I should note that I may be parsing your intent incorrectly as a result of that.

[0] https://www.google.com/amp/s/www.bloomberg.com/amp/opinion/a...

FIVE MILLION DOLLARS?

Seriously? That's it? What a fucking slap in the face. Mark Zuckerberg's home remodel's PARKING FEES in the Mission were about that.

Reply on news.ycombinator.com