Hacker News Reader

Comment by brundolf | Hacker News Reader

The unintuitive thing here (for the general population at least, possibly not the HN crowd) is that almost certainly nobody at Facebook saw, and approved (much less designed) this marketing strategy. It naturally emerged from the goals the system has been given. That doesn't excuse it: in fact, it indicts automated targeted-advertisement as a category.

This software is working exactly as it was designed to. I'm sure somebody could (and probably will, now that it's made the news) add a rule that makes an exception in this case. But this is going to keep happening, and keep happening, and keep happening, as long as we continue allowing systems to automate our content-feeds, optimized for getting us to spend money.

View on HN

> This software is working exactly as it was designed to.

Exactly this is why I think micro targeting should simply be banned by governments.

I mean:

- It's fundamental meant to manipulate people.

- It's easy to abuse and not really possible to control.

- Discriminatory micro advertisement, weather on a gender, race, religion or similar basis are pretty common for affected people, but at the same time hard to detect for Facebook or any control organization.

- Category combinations which indirectly target addicts are repeatedly used to target them with ad's promoting what this people are addicted to, e.g. gambling.

- *The value it provides for the consumer is questionable.*

- *The value it provides for a ethnically acceptable free marked are questionable.*

- As far as I can tell the only ones noticeable profiting (and as such majorly losing out from a ban) from it are such which act without any regard for the health and dignity/well being of the consumer and society in total (well and big ad platforms).

> Exactly this is why I think micro targeting should simply be banned by governments.

Any site that doesn't allow opting out of micro targeting should be banned by government.

The gulf between opt out and opt in is enormous, and the entire problem. Opt out is not informed consent.

Micro Targeting can absolutely be abused and you have done a good job outlining the dark side of targeting.

But to say nothing would be lost from banning it because its always used without regard for health and dignity is way to far.

Brands like Zappos Adaptive which sells adult shoes that don't need to be tied to people with Autism. Its very hard to do this kind of sales without targeting people affected by disability. The same goes for companies innovating in Wheelchairs, diabetes care, etc..

Would "people who bought this book also bought books X,Y,Z" count as microtargeting?

Because I actually find these recommendations valuable.

I think it would, and it's in a similar category to YouTube constructing an alternate reality based on videos I watched. The difference is that it takes longer (and I personally have to overcome more inertia) to read a book than to binge watch outrage media or something on YouTube.

I think we might need to differ between advertisers choosing the micro targets and automatic context dependent microtargeting . And for the later we probably would need to differentiate between basing the automatic general purpose targeting based on the current context (the site you are on + what subsection of the site you are on + what you have in your shopping car and similar) with the actual context + history + ML based micro categorization.

In general what I am against is advertisements where

- the advertiser can choose targets based on too fine grained categories (micro categories)

- people get classified with too fine grained categories

(Both points are mostly the same.)

So "people who bought this book also bought books X,Y,Z" is IMHO fine, even if it can sometime lead to bad situations as it can be abused much less. (EDIT: But that partially because it shows real products, "people who have seen XY news also see Z news" would be a different thing.)

> It's fundamental meant to manipulate people.

Manipulation implies that you're lying to people to get them to buy things, but is it easy to make that argument for FB? What would be the leading argument here?

And why should we think that automated processes tend towards manipulation (lying) as a strategy in the marketplace?

> Manipulation implies that you're lying to people

No it means you are trying to change their behavior, usually in an unscrupulous way.

These ads aren't lying. But they are certainly pushing a narrative. Which at best, is normalizing the violent actions taken. And at worst, is recruiting people for that cause.

How does manipulation imply lying? When you're manipulating someone you're trying to change their behavior from something the might not want to do to something you want them to do.

You don't need to lie to do that.

That's why the solution is liability: Companies will keep leaving unchecked algorithms responsible for this until they can go to jail for doing so.

Adtech is a cancer and it's time to cut it out using every legislative and regulatory option we have.

Put Mark Zuckerberg and Sundar Pichai behind bars and the problem will correct after that.

Liability is tricky because a) it seems like it would be hard to prove, and b) it could get too broad, hampering internet expression in general. I tend to think a specific law against automated content-curation would be better - that's the true source of problems - though it's obviously still really hard to define (arguably anything a computer does is "automated").

The simplest solution might be privacy protections, which (among other things) would indirectly kneecap targeted content curation, due to a lack of data.

People have been taught by big tech companies' PR teams that any regulation will hurt the Internet, but I don't believe it's remotely true. Internet companies currently enjoy near complete immunity to content they distribute, but when such things have gone, the Internet will readjust.

Either we legislate big tech or big tech will drive/own the process that creates legislation (arguably the latter is already happening).

For clarity: liability solutions are civil and aren't going to put anyone behind bars.

But yes, that's absolutely part of the puzzle. Companies who made money by peddling armor and ammo and gear into an extremist community they were simultaneously trying to moderate need to be held partially responsible.

And that goes for the apocalyptic end-times mess that has infested the gun and gun-peripheral industries too. If you spend your advertising budget telling your customers they need your products to "protect their freedoms" then you don't get a pass if they end up sacking congress to do it.

Criminal liability isn't an impossible thing to do.

Are the ads illegal? Or the products they are advertising? The website for Soldier of Fortune advertises similar products, should they be shut down? How about the print magazine, should that be shut down as well?

There are a lot of sites out there that publish much more objectionable content. Should we shut all of them down as well?

> Are the ads illegal? [...] should they be shut down?

Probably not, but insurrection is. The ad isn’t the problem here, the context & juxtaposition is the problem. It’s not a call to shut down ads, it’s a call to avoid accidentally (or worse, intentionally) supporting illegal speech and violence.

> Probably not, but insurrection is.

If Facebook is distributing illegal messages inciting, or that constitute part of a conspiracy aimed at, insurrection, that is obviously a problem.

What legal ads it shows next to such messages is, if a legitimate concern at all, orders of magnitude less important.

If the content is not illegal, and the ads are not illegal, I’m not sure what the issue is at all.

Advertisements are also messaging.

Take a news article about a governer starting a new coronavirus lockdown. This article having ads for hand sanitizer, and this article having ads for zip-ties, are two very different messages.

>If the content is not illegal, and the ads are not illegal, I’m not sure what the issue is at all.

Have you ever considered that what is illegal can be changed, and that the commenters you are replying to may feel that some part of this industry should be made illegal? It used to be legal to advertise cigarettes to children, and now it's not.

If you want to argue that it will be difficult to write and pass legislation that actually accomplishes anything meaningful, I'd agree. But to say "no laws are being broken so who cares?" is just so unproductive. Something has to be done here, and I think I would rather have it be done at a government level so that I can at least feel like I had some say in it.

> Have you ever considered that what is illegal can be changed, and that the commenters you are replying to may feel that some part of this industry should be made illegal?

Sure, and I’ve seen plenty of decent arguments for moving the line of legality on the main content (less so on the content of the ads at issue, though there might be some.) Feel free to replace “should be illegal” for “is illegal” in the grandparent post.

Doesn’t really change the basic point, which is that I fail to see much argument for why the juxtaposition should be of concern that doesn’t depend on one of the items (most likely the main content rather than the ads) being of much greater concern to start with.

> the context & juxtaposition is the problem

This seems like an issue that cannot be solved (yet). No human editor would place insensitive adverts next to related content, and I think we're a way off ML knowing the myriad subtleties that could be involved.

A sibling content mentioned adding rules to stop this particular case in point, but that just sounds like an endless game of whack-a-mole.

The site https://www.sofmag.com/ currently shows advertising for gun accessories next to mostly political news and opinion pieces. Does the legality of this advertising depend on whether a person or an algorithm selected the advertising?

Questions like these are probably why the bill Josh Hawley wrote is clearly targeted at Facebook and Google, what with the $1.5bil revenue number: https://outreach.senate.gov/iqextranet/iqClickTrk.aspx?&cid=...

(huh, is that a click tracker? URL from https://www.hawley.senate.gov/senator-hawley-introduces-bill...)

> [site] currently shows advertising for gun accessories next to mostly political news and opinion pieces.

I guess this is still somewhat straw man. Ads near politics and opinion on a politically oriented site also isn't what was being called out exactly. I guess that it could be summarized as advertising the ammunition to carry out a violent insurrection next to the call to create a violent insurrection (illegally), on a site that is primarily social, mass market, and supposedly politically neutral, that was the problem being called out. Trying to draw broader lines than that makes any argument specious at best.

> Does the legality of this advertising depend on whether a person or an algorithm selected the advertising?

Currently, no. Currently the advertising is legal regardless. (And note, again, the legality of the ads wasn't the issue being discussed.)

> Put Mark Zuckerberg and Sundar Pichai behind bars and the problem will correct after that.

Why not the CEOs of the companies using the targeted keywords/demographic information?

The "it's not the gun that kills, it's the people who kill" argument?

How is that argument? The people advertising and selling are not the people buying and using them. It's different people.

Yes, but Facebook and Google are neither of those. They're just the platform where the advertisers and their victims connect. I think the parent was saying this is less to do with the algorithm, and more due to the keywords that the advertisers chose to link their product with.

"gun" = "platform"

Gun = Facebook platform

Hitman (because Gunslinger is too cool) = Facebook company

Targets = Capitol Rioters

Hirers = Gun companies

Bullets = Ads

The analogy works because although the companies were unintentionally blanket-targeting their victims, a real world equivalent would still be persecuted. Hence everyone involved should be punished - the rioters (who already are), Facebook the company, the gun companies (albeit in minor fashion like fines because it was stil unintentional), and Facebook the platform (which should be neutered).

That argument doesn't really hold up when Facebook literally serves the ads. In the metaphor you're suggesting, this would make Facebook the one wielding the gun, and companies buying ads would be whoever hired them to do the job. Hired hitmen are still criminally liable, despite not being the client.

But only real alternative to adtech is spam: cheap (because of very low effectiveness), flashy rows of banner ads, 1990s style. I am old enough to remember how obnoxious internet experience was ~20 years ago before targeted ads became a thing. It sucked.

I'll take "obnoxious" over "insidiously effective"

That’s a false dichotomy. The only thing that ad tech gets you is specific targeting on a per user basis. You can still advertise related products and services based on the context in which the ads are presented without needing to know the marital status age range and skin color of the person doing the viewing.

Thank you, i didn't quite grasp that difference.

Nothing's fundamentally changed, besides the flashing. Load up a site without Adblock, and you'll have a time warp moment.

People always act as though companies are somehow entitled to highly effective advertising. I'd prefer a return to those obnoxious ads, they were low-tech and everyone learned to ignore them.

> Adtech is a cancer and it's time to cut it out using every legislative and regulatory option we have.

That attitude is the problem: why do we go straight from "This is a problem!" to the government? Aren't there solutions other than making things illegal? Is it just because making things illegal is easier than talking to our neighbors and friends, and giving them peaceful solutions to their problems, doing business locally, and refusing to do business through online ads that use techniques that lead to results like this?

People who call for insurrection are usually plenty aware of where to get things like this anyway. It's a bad look for Facebook, sure; but are these ads likely to do real harm? I doubt it. There's plenty of time for sensible people to make a difference without creating more anger by trying to make even more restrictive laws.

One of the major specific roles of government in a civil society is to maintain and establish the rules of fair play which then allow for a functioning market economy. This represents an inherent compromise.We recognize and appreciate the value that integration and scale provides, and we use the scale of the government to keep it in check.

This isn’t just about going out and talking to people who disagree with you. Part of the issue is that the news bubbles that have been created out of news and stories and conspiracies that algorithms know we will react to chip away at our ability to have those conversations with each other.

You know, I watch a lot of Fox News, I have some deeply conservative friends and acquaintances and dude they live in a parallel universe. We need the government to collapse those universes together so we can have those conversations in the first place.

The ads to harm because they reinforce the structural walls between universes which prevent the necessary conversation from taking place. It’s pretty bad out there dude.

I disagree on your first paragraph, but that's less critical than my disagreement with your second and third. Don't you think that some personal engagement could help bring that parallel universe — and yours — back into sync with reality? You say the algorithms chip away at our ability to have conversations, but I see the opposite: real life has a stronger effect on people's perceptions of truth than on-line bubbles, and can help reassert reality.

The role of government is literally to establish rules for fair play in a free market economy. Like that’s the point of the United States code :-) that’s what being a rule of law country means.

Regarding the rest I’m not saying don’t talk to them, By all means invite crazy Susie over for dinner. Have those conversations. It’s tilting at windmills as long as they continue to get their reality defined by what they’re going to react to online and in media.

Humans aren’t really able to process an objective reality, we live in a world of abstractions. Our preconceived notions of what the universe and objective reality are define are personal realities. Reality is all around each and everyone of us, if objective reality were all it took, we wouldn’t be having this conversation in the first place.

Liability for what exactly? How about people who actually commit bad acts are the ones responsible for it?

It seems like you haven't really thought about this and are just personally against adtech and those companies. That's not how rights and legislation should be decided.

Fostering insurrection for profit is, I am pretty sure, illegal. And I'm pretty sure a prison or worse offense.

What about empowering people to protect themselves from violent insurrectionists? If I read in the news that a whole bunch of people planned to converge on my town and shoot at each other, I would certainly consider buying some body armor

So the options are a) Facebook sells tactical equipment to one side, and body armor to the other, or b) They ban these product categories altogether.

In context with the fact that Facebook also served up the precise content day after day, year after year, which put people (the whole population) into this inconsistent-with-reality, individually tailored emotional state which is nothing if not a constant reflection and validation of their worst tendencies and pre-dispositions.

Think of an honestly shitty person whose problems are of their own creation in every way (for the purpose of this thought experiment, you can even pretend it’s not you). Do you think Facebook ever gave them a series of posts that reflects that? Or would they log on to see all the ways other people are the problem? And that’s everyone.

We all have flaws, but this is a profit-driven platform. And if you’re trying to sell someone something, making them feel like the true piece of shit they really are isn’t exactly in the ABCs of how to make a sale, much less at their machine-driven scale.

Facebook is basically the automated version of the guy rubbing a boxer’s shoulders in the corner of the ring, whispering in their ear a never-ending slew of made-up things the other guy said so that when the fight starts, he’s angry as can be.

Are you saying the posts, or the ads next to them, are the illegal fostering? I think the gear advertised is legal.

How is someone supporting and fostering a bad act not also to blame? I see from your profile you work in adtech, perhaps rights shouldn't be decided by people who seek to avoid blame because they profit from it.

You have no idea about my background so let's stick to the actual argument.

Selling items and running ads is not the same as supporting or fostering a bad act. People can similarly buy knives or many other potential and actual weapons, it's still the user that's ultimately at fault.

When you run an ad for weapons next to articles inciting retaliation against the government, it does become the same as supporting or fostering an insurrection.

Now, sure, it can happen because you unleashed an algorithm on the world and don't know what it does, but you're still responsible for it.

People are responsible for their own actions.

Perhaps you want the government to tell you what you can say but I would prefer to go in the opposite direction and completely divorce speech from action, and stop trampling on our rights over the actions of a few. This endless shifting of blame solves nothing.

Hm. OK. Let's compare this with how child pornography is handled, be it its distribution, or posession, not to speak of the production.

Do you still think your position has merit? :-)

Yes, because my position isn't discussing illegal items. They can't run ads for heroin either.

> How about people who actually commit bad acts are the ones responsible for it?

"Strangely" enough this does not apply when corporations' money is at stake, hence sites like The Pirate Bay are held responsible. Taking 2 opposite approaches for the same issue in order for the lobbying corporations to profit is a sign of acting in bad faith.

Facebook built this 2 billion people village, exploited them and collected taxes, then said "You know what? Policing it is hard so let the governments do it". And the governments should do it. EaaS - Enforcement as a service. And as any service it should cost. No personal liability needed, once money starts pouring out of the companies' coffers you'll see how quickly they find a solution.

Careful, with that argument incitement charges become impossible.

Why not ? If Parler/Gab can be de-platformed and sued for supporting violent insurrection, so should facebook/twitter for advertising violent insurrection.

This will never happen. Cigarettes are still being advertised and sold.

Exactly. If you joke about nods and stuffed dogs full of tannerite all day of course the algo is gonna try and show you adds for level "wish.com" armor plates and bulk ammunition because it's job is to show stuff to people who will buy that stuff and you've fit the profile of people who buy cheap body armor and bulk ammo to a tee.

People preparing for revolt buy revolt supplies, water is wet, more news at 11.

To make a counter point, in developing any technology that makes possible things that weren't before, you will eventually encounter problems that the creators never thought to consider. Fixing a problem once it's been identified, and thus refining the technology, is exactly what they're supposed to do. If you realize your ocean liners don't have enough life boats, you add more lifeboats, you don't ban ships.

I'm sure people will claim that this problem should have been foreseen, but it's really only obvious in hindsight. Just a few weeks ago, none of us would ever have thought to plan for the possibility of the US Capitol being stormed. Of course we could know that there would be problematic edge cases, but we couldn't have anticipated every single one, nonetheless handled all of them.

I think we sometimes forget that if Facebook were a person, they still wouldn't be old enough to drive. For all the billions of users and billions of dollars, this is still a technology in its infancy. I'll gladly hand out pitchforks if it's shown that Facebook has been dragging its feet fixing known problems, or if it's shown they've feigned ignorance about a problem they were warned would happen, but if we forbid any innovation that carries the risk of unforeseen issues, we will be banning all innovation.

What's very interesting about this is that it makes a great data counterpoint for the "AI will never just spontaneously decide to do [bad thing]." If releasing the HypnoDrones will help sell more paperclips, it's gonna happen.

It's the whole maximum paperclip problem with automation. There is a limit in how much you want to automate a task or else you end up with major unintended consequences.

You would think we would have gotten wise to this problem by now but nope.

IMO the entire corporate system is working as intended, maximizing profit and growth, by promoting engagement and generating clicks within bounds it’s given in case with social media.

It’s often said that companies aren’t people and not alive or conscious. I agree to the first one, but for the rest, is there a mathematical proof for those?

Or they should never allow military gear to be advertised at all given its main customers.

More of privatize profits, socialize costs. This is the end game to Capitalism, and it can't fix itself.

Reply on news.ycombinator.com