Hacker News Reader

Comment by koenigdavidmj | Hacker News Reader

Seems to me that the courts and the public are becoming much less tolerant of companies operating on Google scale, where most things are decided by algorithms and a human never sees it at all. This might serve as an upper bound on the size of future companies: if you might have liability for every single ad or comment you let someone post, you’re going to want human review of all that content.

View on HN

I think that the public and the courts are correctly recognizing that the excuse that a human never sees something doesn't exempt an organization of the regulatory burden borne by companies where the work is done proportionally more by humans.

If Facebook, Google, et. al want to run advertising companies they have to properly conduct themselves in the current regulatory environment, if their competitive advantage really just boils down to the fact that they are ignoring the rules then there should be a market correction based on them being subject to actual enforcement.

(This is separate from how anyone feels about the actual rules, I think it is reasonable to say that everyone should agree that if there are rules they should be universally applied.)

>I think it is reasonable to say that everyone should agree that if there are rules they should be universally applied

What happens if the rules are bad rules? E.g. if it's illegal for Facebook to allow advertisements for marijuana dispensaries on a federal level? Clearly that's not something we'd want to see universally applied.

What exactly is your argument? That if there's a bad law on the books, we shouldn't have laws?

My argument is just that not all laws are good, and following the law does not equate to doing the right thing.

And they’re free to take that position and face the consequences.

Whilst that is totally correct morally, law enforcement must be very careful about applying that. There is a trade-off between having a stable and predictable set of laws (i.e. strict enforcement) and having room for judgement calls when the letter of the law is unfair (i.e. discretion in enforcement).

I understand when regulators decide to side with strict enforcement when a companies only advantage comes from not following the rules. Especially when it is disputed the rules shouldn't be there.

Or perhaps that if the law is bad, we should engage in civil disobedience.

Giant multinational corporations are squarely last in the "queue of parties that ought to commit civil disobedience."

Only if you start with the assumption that everything to do with them is evil. At the end of the day, they still employ and support thousands of people. There certainly are many cases in which they do the wrong thing, but their power for good should also be embraced.

I agree in this case, but I’d disagree if it were Apple refusing to unlock iPhones for authorities, so I’m not sure I can agree this is universal.

Yes, Facebook should commit civil disobedience by explicitly allowing housing ads to not be shown to black people. Great plan.

I think that if it were the consensus view of Americans that dispensaries should not be permitted to advertise then such a regulation is ok.

I think that if corporation wants to engage in a form of civil disobedience then that act should be recognized and treated by the authorities appropriately, not treated like it is not happening in the first place. The point of an act of civil disobedience is that it has consequences that you are willing to withstand to serve the greater good, just getting away with something is not the same thing.

> , if their competitive advantage really just boils down to the fact that they are ignoring the rules then there should be a market correction based on them being subject to actual enforcement.

I highly doubt that the majority of advertising business for Facebook and Google comes solely from discriminatory housing ads.

I also doubt that it is a majority of their business, but I think it would be interesting to find out just how much of their revenue is represented by discriminatory advertising across many domains (housing, hiring, for example), political advertising that may be dubious or illegal, and outright solicitation for scams. And then to think about how their revenue would fall and their costs increase if they are obligated to make a sincere effort to eliminate these items from their portfolio.

The problem here is that this is the exact opposite of what's going on - the competitive advantage of many smaller companies these days is that they are not even close to being compliant with regulations but platforms often tolerate them because they are not responsible for counterparty compliance.

If you do however put that onus on the platforms, then it's entirely predictable that platforms will cut ties with small businesses. This is more or less what happened with GDPR - lots of small ad-tech players became large liabilities for others in the value chain, leading to more consolidation in the industry. Outside of technology, it's common for certain types of services or tools not to be made aware to small-time customers. The technology industry at large and the large tech companies in particular have so far bucked this trend in a way that created a huge amount of value for the world at large, but this doesn't have to be this way.

In this case who are the smaller companies you are referring to specifically?

If the small companies in question are gaining competitive advantage through non-compliance that makes them viable enterprises then it is good and correct that they should be forced to change their business model and compete with the same rules as everyone else, right?

I don't understand what you are saying with this: "Outside of technology, it's common for certain types of services or tools not to be made aware to small-time customers. The technology industry at large and the large tech companies in particular have so far bucked this trend in a way that created a huge amount of value for the world at large.."

Companies operating on Google scale should also have the scale to ensure that their ad platform doesn't enable people to target ads discriminately, especially when it's on something like housing where there are laws against that. They're probably more worried about their quarterly results and shareholders than they are about potential problems until it becomes a big problem.

Let this be a lesson to all shareholders that they should pressure companies to operate lawfully and that companies have a fiduciary duty to them to obey laws.

Right, except what does scale had to do with anything? The law is there regardless of the scale.

Scale is what gives these companies some cover from "not having human review". Not having human review is what made the scale possible.

last company i worked at we built a PaaS for banks and credit unions that handled mortgage applications.

you wouldn't believe how much time, effort, institutional knowledge, and expertise we applied to regulatory auditing, reporting requirements, and fraud detection per federal guidelines.

we were less than 80 employees total.

hardly a burden for the likes of google.

This is a very easy thing to avoid though:

if (ad target == anything to do with housing) {disable options to exclude ethnic affinity [and a bunch of other things]}

It's not that easy, because there are many proxies that correlate with protected categories. For instance, take a look at the greater DC area with census data on race. You can easily target regions without explicitly using race.

Compliance.

If they lock down microtargetting for housing related ads and have some code to flag housing ads posted in different categories they become compliant with the law. Nobody says they have to be 100% perfect to be compliment.

Sure, but if you think the law serves a purpose, then one may ask whether being technically compliant is success.

It's a start. They weren't even doing that.

Your use of the phrase "it's a start" suggests that further restrictions will follow.

No, it implies that they at least appeared to be trying, even if it wasn’t very effective.

As it is they don’t seem to have put in the smallest possible effort to comply and decided to ignore the law completely.

Is it more important that they "appear[] to be trying" than that the policy imposed on them actually achieve the ostensible policy goal?

For their legal case? I’d think so. It looks way better that you were trying to do something and weren’t very effective than that you decided not to bother.

For their need to comply? No. They have to follow ineffective laws as long as they’re on the books.

Where did people jump to the conclusion that following the law is incompatible with the goal of the law? Saying that it might be insufficient is a different matter.

You're responding to me as though I was providing Facebook's defense for not complying. I wasn't. I was saying that if Facebook technically complies, that isn't necessarily sufficient.

"if you think the law serves a purpose" Surely it is not controversial here to suggest that anti-discrimination laws serve a purpose and therefore if someone finds a loophole that does not mean it is just fine?

I was just reading in Jalopnik about how Toyota added reinforcement to their minivan to deal with small offset driver's side impacts, and then when a test for a similar passenger side impact was done, surprise surprise, there was no symmetric reinforcement on that side!

It's not that easy

If a small town newspaper with a staff of 30 can do it, Facebook has both the staff and resources to do the same.

If it cannot, or will not, it should not take advertisements for sectors that are regulated.

But small town newspapers are endangered by those costs! Facebook only makes about $6 ARPU. Automating the crap out of everything is what enables them to exist.

cry me a fucking river, Facebook.

Then it can't exist. Worse things could befall the world.

Elaborating a bit on the sibling comments, Facebook has no "right" to exist as a business. If the only reason it can exist is by ignoring laws with which other businesses are able to comply, it should be shut down (or alternately, forced to obey those laws and then "corrected" out of business by the market).

>Automating the crap out of everything is what enables them to exist.

If they're too big to be accountable then maybe they shouldn't exist.

What about this complaint would prevent Facebook from addressing it without humans involved? All they're going to do is revoke microtargeting abilities from housing ads.

How do they determine what's a housing ad?

There's two pieces to this a) Facebook makes this possible and b) Facebook damns themselves and promotes that they make this possible for housing. Part b is easy to fix: make it harder to do housing ads that are targeted. However, part a is much harder: what stops me from advertising my AirBnb listing as a URL on Facebook as a "generic ad"?

> what stops me from advertising my AirBnb listing as a URL on Facebook as a "generic ad"?

By not permitting Airbnb URLs as generic ads.

Edit to ad: and placing a warning if it looks like it might be a housing ad: “this looks like a housing related ad and has been flagged for moderation”

"How do they determine what's a housing ad?"

You ask the user to say what type of ad it is, under penalty of violation of contract, and contractually obligate them to carry the legal risk if they lie.

There's no requirement that it be done via some fancy ML algorithm. I would expect that unless FB actively attempts to piss off the government, that would pass legal muster. Goodness knows I've signed enough agreements where I've agreed to very similar clauses just in normal day-to-day life.

I’m afraid that the law being used to accuse Facebook would apply in that case too.

Are you saying that as a lawyer who is aware of this area of law, or are you saying that as a random internet person? I'd be more than a little surprised if your guess is correct. Normally, in my understanding, if one makes a good faith attempt to follow the law, one is normally in the clear. One does not become culpable when one's efforts to comply are stymied by fraud on the part of a third party.

Then again, I'm just a random internet person. If you're a lawyer, or if you can provide chapter and verse, I'll shut up.

> contractually obligate them to carry the legal risk if they lie

The government is under no obligation to look at the terms of a civil contract when they decide who to prosecute.

Furthermore, Facebook has already tried this. After the first ProPublica article (last year, I think) they added a “I’m not breaking the law” checkbox.

How do they determine what's a housing ad?

The same way that newspapers, television stations, radio stations, magazines, and advertising agencies have for the last 200 or so years.

And they didn't have super-duper AI, the smartest people on the continent, and warehouses full of servers to help.

How do they determine what's a housing ad?

By doing it.

Facebook loves to try to play both sides, and claim to paying customers that it absolutely has the ability to do a thing, while turning round and saying to regulators that it is only a poor helpless social-media network utterly lacking in the ability to do such things.

If even a fraction of a fraction of a fraction of the capabilities Facebook claims in its materials for advertisers are genuine, it can figure out which ads are housing ads.

Have heavy fines in the TOS (communicated upfront) about illegal adds. This gives FB an incentive to police themselves, and lets the bad-actors pay for the enforcement.

I think you're conflating two unrelated dimensions - automation and size. Traditional ad-buying for TV operates at scale but without much automation. Meanwhile, there are plenty of players in the ad business that aren't particularly large but operate without much human oversight.

What this does is increasing the cost of serving small customers, while having negligible impact on the cost of serving large customers. This likely means companies like Google and Facebook will stop serving small businesses that cannot guarantee a certain budget. In a way, we live in a fairly rare time in that a small startup has access to AWS, Azure, GCP, FB ad manager and Google adwords, getting access to most of the same functionalities afforded to larger companies, while operating on tiny budgets without needing account management. This isn't true of many other industries or even services within the same industries. As is, I don't think small businesses are particularly profitable segments for these companies.

This is exactly what already happened with Youtube demonetization - the increased scrutiny on content safety and importance of having human content auditors made monetizing large groups of small-time publishers on Youtube entirely uneconomical for Google. So they got the boot. So in a sense, we should conclude the exact opposite here - automation is what allows a level-playing field for small players and if you take it away through putting a huge amount of pressure on platforms to police behavior of individual customers, it will accelerate consolidation in other industries.

On the other hand, this doesn't seem like an issue of algorithms at all.

I was reading the list you edited out and I was like "yeah I don't want to live next to any of those", I wish there was app for doing the reverse about your neighbors before you rent a place. Then I felt bad for thinking that thought.

For what it's worth, I edited it out because I couldn't figure out the formatting!

I think you’ve piqued everyone’s curiosity now.

It was just the list of offenses mentioned in the article.

Old and irrelevant laws have always been dampeners on company growth. Hopefully a Chinese Company might come up with a classifieds website that US citizens will be able to use.

Reply on news.ycombinator.com