"And knowing the Facebook sales playbook, I cannot imagine the company would have concocted such a pitch about teenage emotions without the final hook: “and this is how you execute this on the Facebook ads platform”. Why else would they be making the pitch?
The question is not whether this can be done. It is whether Facebook should apply a moral filter to these decisions. Let’s assume Facebook does target ads at depressed teens. My reaction? So what. Sometimes data behaves unethically."
This is the key quote for me. It's clear that if this data can be abused, it will be.
> This is the key quote for me. It's clear that if this data can be abused, it will be.
And it's so fucking sad that this is true. This kind of technology could enable Facebook and entities like it to predict when people are feeling depressed, hopeless and suicidal and instead of doing something like presenting a simple message like "Hey are you alright? Do you need some help?" nope, their thought is "this person is RIPE for some well placed marketing."
I get that this guy is likely a product of his industry and he didn't set out from childhood to prey on people's emotions to sell them shit they don't need, at least, in all likelihood. That said I certainly hope there is a hell if for no other reason than this man and all like him really, really belong there.
Advertisers and Facebook don't have to try to target emotionally troubled teens in order to target emotionally troubled teens. All they have to do is have an advertisement that works well on emotionally troubled teens, have sufficiently correlated targeting tools, and run an optimization algorithm like gradient descent.
To take a slightly different example: it doesn't matter if Facebook doesn't offer racial targeting if a payday loan company runs ads segmented by zip code and picks the best performing areas that happen to be predominantly black.
EDIT: This should teach me to read the entire article first, since I made up an example that was literally used in the article. My bad.
The main problems they're talking about are PR problems, not ethical problems (if you assume that advertising itself is ethical, which I don't necessarily think is true). Specific types of targeting are then bad because of the optics, not because of their efficacy.
Put another way, there's no ethical problem with recommending that people who follow Jay-Z should follow Obama, but there's a PR problem there.
> Put another way, there's no ethical problem with recommending that people who follow Jay-Z should follow Obama, but there's a PR problem there.
I agree with that specific statement, but I would say there is an ethical problem with the similar situation of recommending payday loans to poor people.
The difference is recommending Obama to Jay-Z followers isn't harmful to those followers, but recommending payday loans to poor people is.
Targeting seems, to me, to be ethics-neutral in terms of first-order effects. It amplifies the effectiveness of both positive-value and negative-value advertising. Like, roads can be used both for transporting troops to invade a country as well as to transporting food to relieve a famine, and better roads improve the effectiveness of both.
Like, advertising payday loans to poor people may be an acceptable cost of advertising if it gets out valuable mesothelioma information as well.
I mean, is it any less ethical than recommending that poor people purchase things in general? The point of advertising is to convince people to make purchasing decisions that they wouldn't have otherwise made, which is almost always damaging financially.
To be clear, I'm not arguing that you're wrong- I think that the argument you made applies to advertising in general, and that it's correct.
> The point of advertising is to convince people to make purchasing decisions that they wouldn't have otherwise made, which is almost always damaging financially.
That's not really true. People purchase things because they decide that the benefit is worth the cost. Advertising makes people pick one brand over another when they already plan to buy something, and it helps sway someone's decision when they're not sure if it's worth buying. It also brings their attention to products that they wouldn't otherwise have known about. But this sort of advertising is almost never "damaging financially", unless you think the act of buying something counts as financial damage.
"The fact that Facebook could easily throw the election by selectively showing a Get Out the Vote reminder in certain counties of a swing state, for example, was a running joke."
The idea was telling only those voters in certain counties (who predictably would vote a certain way) to vote; i.e., a less-blatant version of telling exactly the subset of people who are registered to vote for a given party, to go out and vote. That is a partisan action.
Perhaps, but it'd then just be on the same level as any other broadcast medium that puts out such a message. I presume the more partisan of the cable news channels are still allowed to say "hey everyone, vote!"?
That's correct. Get-out-the-vote efforts are seen as partisan. I volunteered to help register people to vote (as a non-partisan without any messaging at all) and was verbally abused by Republicans on the street.
Republicans also regularly impose laws to disenfranchise voters. An example is Georgia's law where voters are purged from the rolls every 3 years. Minority voters often only vote every 4 years.
The list goes on and on. These voter suppression tactics are very, very targeted at minorities and Democrats, often to the point that the Supreme Court has to strike them down as unlawful.
Yeah, basically if everyone always voted, the Democrats would generally win. That's why Republicans don't want to encourage voter turnout and also want to suppress ease of voting which usually harms poorer people who tend to be Democratic.
I think they should be, once they get to the point where they are ubiquitous enough to be considered an official identity and information provider (in the way that newspapers and TV have been), something which they've brought upon themselves in advocating a "real name" policy.
Though of course it's a different thing when a county inadvertently affects vote results by unintentional bad design, versus when a centralized entity intends to have an effect.
Indeed. Zuck was forced to backtrack his denials following the election, and is now committed to tackling fake news. They sing a very different tune now.
The problem with any platform like Facebook (though Facebook happens to be have the richest seam
of user data, at scale and across devices) is that the segment with the highest clickthrough rate will inevitably be one which draws on shady factors.
It will be the segment deliberately concocted to find that most vulnerable / subsceptible to buy. If you're looking to ship garbage self help books, you're going to target people who like garbage self help book writers. Rinse and repeat for preppers etc.
To prevent this you essentially either have to ask marketers to hobble their own search terms even when they know there is likely a higher CTR out there (like payday loan companies willingly deciding not to advertise on daytime TV), or asking Facebook to hobble its own platform.
Until there is some kind of regulation or FTC pushback I can't see the situation improving.
It's not a title, it's a headline, and it's written by the editors. This is how newspapers work. Media title inflation is standard. Every 'senior' this and that you've ever read about is also bullshit, in case you hadn't noticed.
To be fair "Executive" means "A person or group having administrative or managerial authority in an organization."
Seniority of the product manager role depends on a number of factors. Sure, most FB product managers are not considered execs these days, but he was their first Product Manager responsible for ad targeting and then their advertising exchange all the way back in 2011. That probably brought with it a team, a good salary and a lot of sway.
Unlike the cookie-cutter 9-5 corporations of yesteryear, startups have been pretty loose with standard job naming convention. You can be fresh out of a BA and get called "CMO" or you can be a 20-year veteran leading a team making executive decisions and just be a "Product Specialist."
The responsibilities and compensation define executives far more than titles.
> There are alcohol and gambling ads on Facebook? Those two in particular often aren't allowed on ad networks.
Gambling? Loads of them, and for what it's worth Farmville and friends IIRC even invented their own virtual currency.
Alcohol I haven't seen yet but every major beer and spirits brand has its own Facebook page.
Ad networks usually forbid alcohol due to underage-drinking-ad-laws (which FB can avoid due to their real-name policy and them requiring you lots of times to submit a government ID) and gambling due to a host of even more country-specific laws (which FB can certainly spend a lot of lawyer money on to safely navigate them, given that gambling ads usually pay huge sums for well working spots). And I believe that CC processors for ad networks don't like both due to the same issue; with a giant like Facebook, no CC issuer can afford to not deal with them, so they trust FB to not get them in legal trouble.
I think that is part of the mistake he makes. "We do <morally wrong advertising thing>, but we also do <other advertising> how is that different?". It isn't, but the logical conclusion of this isn't that they are both "good", they might also be both morally wrong.
On the other hand, I've found that if you're suffering from health or mental problems, the ads that you see are those offering medical and psychiatric help. Which I thought was pretty welcome, contrary to worries of exploitative targeting.
Wow, I can't decide if this guy is more morally or intellectually bankrupt. Seriously, "sometimes data behaves unethically"?
Data doesn't do anything. Not even analysis or deep learning does anything. People (or programs written by people) execute decisions on the basis of data. It's still the people being unethical. Big data isn't some magic wand that absolves you of moral responsibility, but it's extremely interesting that at least one marketer at Facebook seems to believe it does.
And that's just one of the many things that manage to be simultaneously offensive and just plain incorrect in what is frankly, a pretty short article.
You ought to continue to read past that. Because you missed his point. He goes on to basically question "ethics" entirely:
>What did this tool start spitting out? Every ethnic stereotype you can imagine....People who liked Jay Z were more likely to like Obama – it was one of the statistical truths Facebook couldn’t be seen espousing.
I disagreed. Jay Z is a millionaire music tycoon, so what if we associate him with the president? In our current world, there’s a long list of Truths That Cannot Be Stated Publicly, even though there’s plenty of data suggesting their correctness, and this was one of them.
African Americans living in postal codes with depressed incomes likely do respond disproportionately to ads for usurious “payday” loans. Hispanics between the ages of 18 and 25 probably do engage with ads singing the charms and advantages of military service.
Why should those examples of targeting be viewed as any less ethical than, say, ads selling $100 Lululemon yoga pants targeting thirtysomething women in affluent postal codes like San Francisco’s Marina district?
That's not questioning, that's rejecting wholesale without a proper argument as to why. And frankly I don't at all see the ethical analogy between luxury athleisure and payday loans, and it seems to me you'd have to be pretty damn fucking self-deluded to think they're remotely comparable.
it's a good that a demographic buys. that's the way that ethically bereft corporations see it. i would even guess that this is how zuck sees it. people who buy jay-z buy obama. people with lower income buy payday loans. there is nothing ethical about any of this unless you want to add the ethical dimension. corporations are expected not to. and yet the data is allowing them more and more to target these "opportunities".
i wanted to catch some flies over the weekend. i used some soap and vinegar as the vinegar attracts them and the soap breaks the surface tension to drown them. corporations act much the same way. perhaps we need to see them as predators rather than someone looking to solve our problems, much like flies should see me. i don't provide what they want, but they don't see it that way until it's too late.
we sell bad food to overweight people. facebook can help with that. we also sell diets to overweight people. facebook can help with that too. either we add an ethical dimension to business or quit criticizing marketers for doing what they are expected to be doing.
i would even go as far as to argue that a system based on desire is not a good system. as the case with the flies, we can be led to death through our desires and various drives that have another purpose, but have been subverted to achive different objectives. whether as individuals or as a civilization.
Corporations will sell you just enough rope to hang yourself unless the government steps in and tells them not to through regulation. But that's the antithesis of business, neo-liberal economics will argue. Canada still sells asbestos to third world countries for instance.
Asbestos opponents and those weary of seeing Canada's mesothelioma rate rise celebrated in 2011 when the country's asbestos industry ground to a standstill. Canada's last two remaining active mines, the Jeffrey Mine in Asbestos, Quebec, and the Lac d'amiante du Canada in the nearby town of Thetford Mines, Quebec, shut down because of financial, labor and development issues — the first time in 130 years that the Canadian asbestos production stalled.[1]
Are you suggesting that pay day loans should be illegal or advertisements for them should? If not why not advertise them to the audience that is apparently interested in it?
Yeah I get that it is unfortunate to say the least that people need/want these loans. However, what is the right thing to do about the loans? If they are predatory, maybe they should be illegal. But who am I to tell people what they can and cannot do with their money? That would be rather patronizing. If we assume that people getting these loans are grown adults why shouldn't we allow people offering them to advertise to the people most interested in them? Is it ethically "better" if even wealthy women from the Marina see the ads despite everyone knowing they don't care about pay day loans for purposes other than discussing then on HN? I think it's honestly a really hard ethical question that I don't see a obvious answer to.
If you're an advertising platform, you can choose what sort of products are advertised on it. All of them do. You can, quite reasonably, decline to advertise things that are unethical. For instance, predatory loans. You are choosing not to profit off the misery of others. This is not nearly as complicated as you make it out to be.
Will the target demographics universally and unequivocally benefit when online access to such loans is cut off?
Other choices might include an even more predatory local check-cashing place and a guy named Vinnie who seems to own a lot of baseball bats for someone that never plays baseball.
This doesn't 'cut off access'. Last year Google stopped advertising predatory lenders entirely. Targeting ads for predatory lenders at potentially vulnerable people is even worse. Again, this is a pretty straightforward thing, not a complex logic puzzle that requires careful derivation from fundamental axioms.
Not to detract or digress, but oxygen is damaging to cellular life as well, just long term. If nothing else gets you beforehand, with age as all mammals "we caramelize (glycation) & we go rancid" (oxidation). When photosynthesizing microorganisms emerged and oxygenated the planet, it likely wiped out 90% of life as it existed pre-oxygen (not multicellular back then --- which could well predominantly be an adaptation to protect weaker "inner" cells against oxygen, anyway --- the other great adaptation was proto-mitochondria getting an actual massive energy boost from burning oxygen rather than being burned by it, and here we are today!)
yes, most things are goods that people can buy at their own discretion. However, payday loans are predatory and should not be considered in the same category as unhealthy food, expensive clothing, etc.
i could argue that addictive foods are similarly predatory. they have long lasting impact on the health of a person the same way that payday loans have on their economic health. predatory is arbitrary, the question is what is the impact of the harm. and can people be expected to control themselves or at least regulate themselves when presented with the option, so as to act in their best interest. for many things in the economy i don't believe this to be the case.
yet business gets done and the system at least appears to continue to function. people learn from their mistakes and move on if the damage isn't too great. or they get caught up and stuck and that's a problem that needs regulation.
>I don't at all see the ethical analogy between luxury athleisure and payday loans
The "athleisure" market has bilked people out of $35 billion dollars a year using offshore sweatshops to turn $1 into $20 a pop, mainly by amplifying people's insecurities day after day after day through mass advertising.
man, this hits home. i just recently got a pullover made by "Nike Golf" at work. It's a nice pullover. It's listed at somewhere between $37.50 and $150 on Amazon.
The box it came in? The same box that I used to see stuff delivered to the 99 cent store across the street at my old apartment.
It's amazing what athlete endorsements can do for brand value:
It's more than large profit margins, it's the constant deluge of propaganda every day that gnaws at people's insecurities. I'm too fat, I'm too skinny, I'm not strong enough, I'm not active enough, I'm not healthy enough, I don't exercise enough, I don't hydrate enough, I don't eat right enough, I'm not pretty enough, I'm not handsome enough, I'm not popular enough, ad nausium. Every day, to an entire population, for generations.
There's a saying in business, you don't sell product, you sell emotion. They are selling, "I want to feel better about myself." They can't sell that until they knock people down several notches. That's what it takes to get people to pay $40 for a shirt then can get for $7.
Just about everyone alive has grown up with that. Let that roll around in your head a while. I think it's worse in scope and what it does to the population.
I've never been screwed by a payday loan service, but I've certainly been subject to that type of advertising for as long as I've been alive, and it's gotten progressively more sophisticated.
Well to answer his question: because in the examples of African Americas and Hispanics, there's a power structure with its roots in the racist history of American capitalism, and profiting off of that power structure is definitely less ethical than getting an affluent woman to spend some of her disposable income.
Oh believe me, I caught that bit. And my summary of the entire article stands.
To paraphrase what he's saying here: why is selling things to people who can easily afford them "ok" when exploiting the poverty of historically disadvantaged groups and getting them possibly maimed or killed is not ok?
The fact that he even thinks this approaches an interesting argument reinforces my point: intellectually and morally bankrupt. The guy isn't smart enough to realise that his unanswerable rhetorical questions have obvious answers.
> Big data isn't some magic wand that absolves you of moral responsibility, but it's extremely interesting that at least one marketer at Facebook seems to believe it does.
I suspect a large portion of this attitude comes from the fact that the marketers (tasked with enacting campaigns and strategies based on results of data science/machine learning), likely don't understand all of the statistical intricacies and pitfalls around underlying assumptions, and a lot of people already view data science as producing "magic" results (and assume that if it tools them so, it must be true).
Of course, this guy is likely morally bankrupt anyways, which doesn't help either.
Of course data can behave unethically. For example, data might show that people of a particular race in a particular neighborhood have high concentration of bad credit scores. Though using this information to approve/deny loan applications might be illegal. And when algorithms are run on large datasets, are closed source, and their downstream effects are difficult to measure, then yes data/algorithms can behave unethically.
>Why should those examples of targeting be viewed as any less ethical than, say, ads selling $100 Lululemon yoga pants targeting thirtysomething women in affluent postal codes like San Francisco’s Marina district?
Because one can cause poverty and the other doesn't.
the problem lies with payday loan itself not with the advertisement, though. If we know that payday loans are harmful for society, why not to outlaw them right away, rather than dance around it being advertised.
Ethics is a human construct. So data could very well show that unethical (by some definition) behaviour is profitable.
I think this is the point being made: the data shows, beyond statistical doubt, that payday loan ads aimed at a certain demographic have better conversion, ergo profit. The data also shows, beyond statistical doubt, that [current fashionable brand] yoga pant ads aimed at a certain demographic have better conversion, ergo profit. The data doesn't care. If you are an ad-funded company (as both facebook and google are), you can't afford to ignore the data, it's your differentiator.
I'm not saying I agree with it, I am saying that ethics is an opinion, data is fact. I long for a time when we as a species evolve beyond being so predictably influenced by propaganda.
Well, they are. Addicted to immediate and frequent social validation. Telling people heroin is bad for them isn't going to make them quit if they are addicted to it. They have to witness or experience the hardships that come from such an addiction.
Telling people Facebook is bad for them, telling them why it is bad for them, isn't enough. We have to show them indisputable evidence of social media having an extreme negative impact on their lives.
We have to pull them away long enough for their brains to stop craving the dopamine release they associate with browsing their feeds and narcissistically posting several times a day.
Most importantly, we have to care and understand what they are going through. Being angry at them or thinking they are stupid for not knowing better is alienating and divisive behavior. They will rush to Facebook to alleviate their negative emotional response to what you are telling them. Even if they think that you're right.
This is just semantics. You get to use Facebook in exchange for having the info you share there used for advertising, and having those ads show up on the side of the page. Who cares what you call it.
It gets more complicated when you consider the concept of "social currency" and the utility of one's profile in matters such as... gettin' laid for example.
I view Facebook on one level from your perspective. They sell to an advertiser eyeballs of potential customers. It's user performs the immaterial labour required to create content which drives eyeballs to the site and sort the customers into demographic groups.
But there really is a much more complicated Web of transactions taking place on Facebook. Exchanges which can convert the labour performed on a user's part into a social profit.
In your statement - "you are making the decision", "you" can be replaced by algorithms. Algorithms are taking decisions (e.g. giving loans, showing ads) based on data. Algorithms can have bias. Data can have bias.
I think you and he have sort of the same point: data has no ethical orientation. If you unpacked that sentence, he would probably clarify it to "data behaves in ways that, if there was a human making that call, would be unethical".
What makes the situation interesting is that there's not necessarily a person that makes the decision to do potentially damaging ad targeting: if you want to prevent it, you probably have to specifically prevent it in each case. Is Facebook responsible for preventing every single unethical ad targeting? Given the unfeasibility of that, does this problem mean we should call it quits on targeted ads entirely?
Exactly. There is a long [0], long [1] history of people being held accountable for problems enabled by their carelessness.
[0] Hammurabi's Code, 229. If a builder build a house for some one, and does not construct it properly, and the house which he built fall in and kill its owner, then that builder shall be put to death.
[1] Exodus 21:29 If, however, the bull has had the habit of goring and the owner has been warned but has not kept it penned up and it kills a man or woman, the bull is to be stoned and its owner also is to be put to death
I don't agree with his claims that the following are among "a long list of Truths That Cannot Be Stated Publicly". I don't think anyone would care; I care very much about racism and can't imagine anyone I know objecting to these.
* "people who liked Jay Z were more likely to like Obama"
* "Hispanics between the ages of 18 and 25 probably do engage with ads singing the charms and advantages of military service."
(I'm skipping the other example because it's an issue of economic exploitation, not ethnicity.)
Yeah, basically if you have any kind of oppressed category of people, you can take advantage of them by exploiting their vulnerabilities (lack of education, lack of access to credit, etc). If you aren't careful with how you use your algorithms, you can come up with all kinds of systems that reinforce systemic racism.
Well, if you're still there can I ask you about Silicon Valley's approach to the law.
You say that lawsuits are merely part of "a well-scripted conflict narrative between corporate entities".
Here in Australia our Federal Court recently hit Chevron for $300 million in a tax case that some say will have implications worldwide for corporate tax structures.
Couldn't a new model of law based on that kind of questioning of corporations and corporate governance regulate the Valley (and Wall Street) effectively?
Dumb article is dumb. Ex-product manager = / = Facebook exec, this person also hasn't worked there in four years. "Targeting emotionally troubled teens" without detailed explanation as to how someone could implies simplicity in the matter, but it's not like an advertiser is typing in "show my ads to emotionally troubled teens please."
Dumb reader is dumb. It's an op-ed not an article.
For the details, perhaps open FB's ad creation UI. You are indeed correct you don't target ads like you enter a search query.
I am surprised people are shocked by this. Personally, I live in Berkeley - within a 20 mile footprint you have a nice sampling of exorbitant wealth and abject poverty.
As you drive through you can see drastic changes of roadside advertisements transitioning between English, Chinese, and Spanish. In the poor areas of town bus stations are plastered with high-interest payday loans. Drive 5 miles North and you will see advertisements for expensive art/music events. Drive slightly more North and you will see endless advertisements for divorce and accident lawyers.
It would be a miracle if all of this store location/advertisement placement was coincidence.
It's such a shtark reality you can stand at the Bus Station, read the bus outside advertisements, and estimate if it's likely to be going where you are. But once you get inside the bus where fewer people of wealth can see, you no longer can.
Everybody I know and his cat has read the book. Give us another book. How about reprinting the old adgrok blog posts about why SF is better than nyc because person in NYC is significant only if they showed up in the Sunday marriage section, whereas the SF guy can build something out of thin air and climb the rings of Fame. I dug that.
> I’ll illustrate with an anecdote from my Facebook days. Someone on the data science team had cooked up a new tool that recommended Facebook Pages users should like. And what did this tool start spitting out? Every ethnic stereotype you can imagine. We killed the tool when it recommended then president Obama if a user had “liked” rapper Jay Z...
The author left Facebook in 2013 [0], so I had assumed by now Facebook could do a proper job of recommendations. It's been awhile since I've "Liked" a page, so I just tried it (In-N-Out Burger) and was extremely surprised to see that Facebook apparently still doesn't provide a popup for related pages I may also "Like", in the same way that Twitter does whenever you follow any user.
That's really surprising. For starters, Twitter has had user suggestions for several years now, and it has much less structured data to work with than Facebook does -- for example, a FB page can specify its category, address, and many other attributes; Twitter has nothing to go by other than user's name, account screen name, biography, a location field, and connection metadata (i.e. geolocated IP). Yet it's been awhile if ever since I've seen Twitter recommend someone truly off, e.g. President Obama and Jay-Z.
In contrast, Facebook provides a related stories feature for news stories a user shares in the feed - or at least it did before the fake news thing blew up. Is this a signal that Facebook doesn't really care about user engagement with brand pages? Or did the fuckup that the author describes make it so that this was a feature that no product manager wants to be associated with?
(Or it could be that I'm just missing the Related Pages feature. Maybe it shows up in mobile and not the browser version)
edit: It's strange that the feature doesn't exist in 2017, but this anecdote, as portrayed by the author, reflects badly on both Facebook's data science and the author. I mean, how is it possible for so many smart people to not figure out how to fix such undesired associations? There are so many other signals for clustering related users beyond similar followers.
edit 2: This feature does exist -- you have to like a post in your news feed (on the website), and then Facebook will append a "Related" box in your newsfeed. Unless your browser window is open particularly high, the fact that the "Related" box was triggered by the Like action is obfuscated (because it is out of view). Still doesn't explain why they couldn't make the feature work in 2012-2013.
edit 3: Appears author is wrong about the tool being "killed". Seems to still exist, though maybe it was only temporarily "killed"
Actually existed for years and made a pile of money, and was only closed down when their entire programmatic stack was shut down (LiveRail, the test DSP, and effectively Atlas). You clearly don't know dick about ad tech.
This is a tricky problem in pretty much the entire consumer-facing ML space. For example, even if you don't feed in 'protected classes' - you can say 'the algo finds a way'. It might figure out a certain zip code won't pay back a loan, so it will down weight them accordingly - even though that zip may predominantly be one ethnicity. Is that discriminating on a protected class? Well, kinda. It's discriminating on a proxy for a protected class which, if we're getting technical, is really a (poor) proxy for cultural, genetic, and socio-economic backgrounds.
So it's in some ways 'more fair' but it does take away our ability to 'socially engineer' certain decisions by prohibiting use of one feature or another.
In those years FB ads was a toddler compared to the platform of today, I worked with FB ads at the time and until 2013-2014 too (author went to Nanigans after FB)
For starters, today's platform is completely mobile focused. FBX (product the author worked on) was entirely desktop inventory.
Did you even read what I said? You're literally repeating it. Funny, and you're citing Custom Audiences, for which I was the first PM, as some sort of counter-example. LOL.
>> you can also target people who search for ways to suicide on Google
Not that either companies' employees seem to care, of course.
Once again, a little reminder about my suggestion to create a public dataset of all employees of Facebook and Google for improving the state of the art of ML and DL algorithms (consider it a public service).
If you obtain private information and exploit it, it is called doxing. If a megacorp does the same, it is called a business plan.
We should probably start with Antonio Garcia-Martinez as the starting node, and do a BFS. Or is DFS better? I just get super stoked about all these technical riddles, I can't really help it!
So yeah obviously Facebook like any other large advertising platform is gonna use its data for seedy means. What's crazy to me is that the author seems to be using the word 'ethics' synonymously with the concept of political correctness. As if he actually does not know the difference.
Just started reading Chaos Monkeys. Thoroughly enjoying it so far (am up to Zuck's inner circle of desks). Guardianistas are a funny lot aren't they. They write as if people who can endure the push and shove of business long enough to try to improve things from the inside (unlike themselves) are going to be anti-business (like themselves).
139 comments
[ 1.9 ms ] story [ 209 ms ] threadThe question is not whether this can be done. It is whether Facebook should apply a moral filter to these decisions. Let’s assume Facebook does target ads at depressed teens. My reaction? So what. Sometimes data behaves unethically."
This is the key quote for me. It's clear that if this data can be abused, it will be.
And it's so fucking sad that this is true. This kind of technology could enable Facebook and entities like it to predict when people are feeling depressed, hopeless and suicidal and instead of doing something like presenting a simple message like "Hey are you alright? Do you need some help?" nope, their thought is "this person is RIPE for some well placed marketing."
I get that this guy is likely a product of his industry and he didn't set out from childhood to prey on people's emotions to sell them shit they don't need, at least, in all likelihood. That said I certainly hope there is a hell if for no other reason than this man and all like him really, really belong there.
To take a slightly different example: it doesn't matter if Facebook doesn't offer racial targeting if a payday loan company runs ads segmented by zip code and picks the best performing areas that happen to be predominantly black.
EDIT: This should teach me to read the entire article first, since I made up an example that was literally used in the article. My bad.
Put another way, there's no ethical problem with recommending that people who follow Jay-Z should follow Obama, but there's a PR problem there.
I agree with that specific statement, but I would say there is an ethical problem with the similar situation of recommending payday loans to poor people.
The difference is recommending Obama to Jay-Z followers isn't harmful to those followers, but recommending payday loans to poor people is.
Like, advertising payday loans to poor people may be an acceptable cost of advertising if it gets out valuable mesothelioma information as well.
To be clear, I'm not arguing that you're wrong- I think that the argument you made applies to advertising in general, and that it's correct.
That's not really true. People purchase things because they decide that the benefit is worth the cost. Advertising makes people pick one brand over another when they already plan to buy something, and it helps sway someone's decision when they're not sure if it's worth buying. It also brings their attention to products that they wouldn't otherwise have known about. But this sort of advertising is almost never "damaging financially", unless you think the act of buying something counts as financial damage.
Woah. They're asking to be regulated.
http://www.washingtonpost.com/wp-dyn/content/article/2007/11...
https://www.propublica.org/getinvolved/item/if-tv-stations-w...
https://sunlightfoundation.com/2015/11/10/everything-you-nee...
Real question, I'm in Australia and don't know.
Republicans also regularly impose laws to disenfranchise voters. An example is Georgia's law where voters are purged from the rolls every 3 years. Minority voters often only vote every 4 years.
The list goes on and on. These voter suppression tactics are very, very targeted at minorities and Democrats, often to the point that the Supreme Court has to strike them down as unlawful.
The idea that UI/UX can change the outcome of an election, though, isn't terribly new. Consider 2000's butterfly ballots: http://www.asktog.com/columns/042ButterflyBallot.html
Though of course it's a different thing when a county inadvertently affects vote results by unintentional bad design, versus when a centralized entity intends to have an effect.
It will be the segment deliberately concocted to find that most vulnerable / subsceptible to buy. If you're looking to ship garbage self help books, you're going to target people who like garbage self help book writers. Rinse and repeat for preppers etc.
To prevent this you essentially either have to ask marketers to hobble their own search terms even when they know there is likely a higher CTR out there (like payday loan companies willingly deciding not to advertise on daytime TV), or asking Facebook to hobble its own platform.
Until there is some kind of regulation or FTC pushback I can't see the situation improving.
Though completely believe FB will use their data in any way possible to make money, ethical or not.
Seniority of the product manager role depends on a number of factors. Sure, most FB product managers are not considered execs these days, but he was their first Product Manager responsible for ad targeting and then their advertising exchange all the way back in 2011. That probably brought with it a team, a good salary and a lot of sway.
Unlike the cookie-cutter 9-5 corporations of yesteryear, startups have been pretty loose with standard job naming convention. You can be fresh out of a BA and get called "CMO" or you can be a 20-year veteran leading a team making executive decisions and just be a "Product Specialist."
The responsibilities and compensation define executives far more than titles.
The guy trumpeted the fact that he lied to his investors as a point of pride:
https://www.reddit.com/r/IAmA/comments/5b4tqe/i_am_antonio_g...
"Trust me, they're lying, I know because I'm a liar" doesn't seem terribly convincing.
So what if I can target recovering alchoholics with alchohol ads?
So what if I target gambling addicts with my gambling platform?
So what?
Gambling? Loads of them, and for what it's worth Farmville and friends IIRC even invented their own virtual currency.
Alcohol I haven't seen yet but every major beer and spirits brand has its own Facebook page.
Ad networks usually forbid alcohol due to underage-drinking-ad-laws (which FB can avoid due to their real-name policy and them requiring you lots of times to submit a government ID) and gambling due to a host of even more country-specific laws (which FB can certainly spend a lot of lawyer money on to safely navigate them, given that gambling ads usually pay huge sums for well working spots). And I believe that CC processors for ad networks don't like both due to the same issue; with a giant like Facebook, no CC issuer can afford to not deal with them, so they trust FB to not get them in legal trouble.
Data doesn't do anything. Not even analysis or deep learning does anything. People (or programs written by people) execute decisions on the basis of data. It's still the people being unethical. Big data isn't some magic wand that absolves you of moral responsibility, but it's extremely interesting that at least one marketer at Facebook seems to believe it does.
And that's just one of the many things that manage to be simultaneously offensive and just plain incorrect in what is frankly, a pretty short article.
>What did this tool start spitting out? Every ethnic stereotype you can imagine....People who liked Jay Z were more likely to like Obama – it was one of the statistical truths Facebook couldn’t be seen espousing.
I disagreed. Jay Z is a millionaire music tycoon, so what if we associate him with the president? In our current world, there’s a long list of Truths That Cannot Be Stated Publicly, even though there’s plenty of data suggesting their correctness, and this was one of them.
African Americans living in postal codes with depressed incomes likely do respond disproportionately to ads for usurious “payday” loans. Hispanics between the ages of 18 and 25 probably do engage with ads singing the charms and advantages of military service.
Why should those examples of targeting be viewed as any less ethical than, say, ads selling $100 Lululemon yoga pants targeting thirtysomething women in affluent postal codes like San Francisco’s Marina district?
i wanted to catch some flies over the weekend. i used some soap and vinegar as the vinegar attracts them and the soap breaks the surface tension to drown them. corporations act much the same way. perhaps we need to see them as predators rather than someone looking to solve our problems, much like flies should see me. i don't provide what they want, but they don't see it that way until it's too late.
we sell bad food to overweight people. facebook can help with that. we also sell diets to overweight people. facebook can help with that too. either we add an ethical dimension to business or quit criticizing marketers for doing what they are expected to be doing.
i would even go as far as to argue that a system based on desire is not a good system. as the case with the flies, we can be led to death through our desires and various drives that have another purpose, but have been subverted to achive different objectives. whether as individuals or as a civilization.
Oxygen and carbon monoxide are both gases. One of them is poisonous to humans.
Asbestos opponents and those weary of seeing Canada's mesothelioma rate rise celebrated in 2011 when the country's asbestos industry ground to a standstill. Canada's last two remaining active mines, the Jeffrey Mine in Asbestos, Quebec, and the Lac d'amiante du Canada in the nearby town of Thetford Mines, Quebec, shut down because of financial, labor and development issues — the first time in 130 years that the Canadian asbestos production stalled.[1]
1. https://www.asbestos.com/mesothelioma/canada/
Other choices might include an even more predatory local check-cashing place and a guy named Vinnie who seems to own a lot of baseball bats for someone that never plays baseball.
yet business gets done and the system at least appears to continue to function. people learn from their mistakes and move on if the damage isn't too great. or they get caught up and stuck and that's a problem that needs regulation.
The "athleisure" market has bilked people out of $35 billion dollars a year using offshore sweatshops to turn $1 into $20 a pop, mainly by amplifying people's insecurities day after day after day through mass advertising.
The box it came in? The same box that I used to see stuff delivered to the 99 cent store across the street at my old apartment.
It's amazing what athlete endorsements can do for brand value:
There's a saying in business, you don't sell product, you sell emotion. They are selling, "I want to feel better about myself." They can't sell that until they knock people down several notches. That's what it takes to get people to pay $40 for a shirt then can get for $7.
Just about everyone alive has grown up with that. Let that roll around in your head a while. I think it's worse in scope and what it does to the population.
I've never been screwed by a payday loan service, but I've certainly been subject to that type of advertising for as long as I've been alive, and it's gotten progressively more sophisticated.
To paraphrase what he's saying here: why is selling things to people who can easily afford them "ok" when exploiting the poverty of historically disadvantaged groups and getting them possibly maimed or killed is not ok?
The fact that he even thinks this approaches an interesting argument reinforces my point: intellectually and morally bankrupt. The guy isn't smart enough to realise that his unanswerable rhetorical questions have obvious answers.
I suspect a large portion of this attitude comes from the fact that the marketers (tasked with enacting campaigns and strategies based on results of data science/machine learning), likely don't understand all of the statistical intricacies and pitfalls around underlying assumptions, and a lot of people already view data science as producing "magic" results (and assume that if it tools them so, it must be true).
Of course, this guy is likely morally bankrupt anyways, which doesn't help either.
Because one can cause poverty and the other doesn't.
No, data doesn't behave at all. This is quite a cynical deflection of responsibility for ethically questionable actions taken by people.
I think this is the point being made: the data shows, beyond statistical doubt, that payday loan ads aimed at a certain demographic have better conversion, ergo profit. The data also shows, beyond statistical doubt, that [current fashionable brand] yoga pant ads aimed at a certain demographic have better conversion, ergo profit. The data doesn't care. If you are an ad-funded company (as both facebook and google are), you can't afford to ignore the data, it's your differentiator.
I'm not saying I agree with it, I am saying that ethics is an opinion, data is fact. I long for a time when we as a species evolve beyond being so predictably influenced by propaganda.
I have stated that fact so many times to so many people, and yet most people don't care. They continue to use Facebook.
It's almost as if people are addicted to Facebook.
Telling people Facebook is bad for them, telling them why it is bad for them, isn't enough. We have to show them indisputable evidence of social media having an extreme negative impact on their lives.
We have to pull them away long enough for their brains to stop craving the dopamine release they associate with browsing their feeds and narcissistically posting several times a day.
Most importantly, we have to care and understand what they are going through. Being angry at them or thinking they are stupid for not knowing better is alienating and divisive behavior. They will rush to Facebook to alleviate their negative emotional response to what you are telling them. Even if they think that you're right.
I'm the product to advertisers, but also to my friends.
Similarly I am the customer to the products that are my friends.
You're keeping your friends there, they're keeping you there, and now the group of you brings in the bigger fish.
I view Facebook on one level from your perspective. They sell to an advertiser eyeballs of potential customers. It's user performs the immaterial labour required to create content which drives eyeballs to the site and sort the customers into demographic groups.
But there really is a much more complicated Web of transactions taking place on Facebook. Exchanges which can convert the labour performed on a user's part into a social profit.
Stop right there. Data has no behavior. You behave unethically. Data informs a decision, but you are making the decision. Take responsibility.
What makes the situation interesting is that there's not necessarily a person that makes the decision to do potentially damaging ad targeting: if you want to prevent it, you probably have to specifically prevent it in each case. Is Facebook responsible for preventing every single unethical ad targeting? Given the unfeasibility of that, does this problem mean we should call it quits on targeted ads entirely?
[0] Hammurabi's Code, 229. If a builder build a house for some one, and does not construct it properly, and the house which he built fall in and kill its owner, then that builder shall be put to death.
[1] Exodus 21:29 If, however, the bull has had the habit of goring and the owner has been warned but has not kept it penned up and it kills a man or woman, the bull is to be stoned and its owner also is to be put to death
* "people who liked Jay Z were more likely to like Obama"
* "Hispanics between the ages of 18 and 25 probably do engage with ads singing the charms and advantages of military service."
(I'm skipping the other example because it's an issue of economic exploitation, not ethnicity.)
You say that lawsuits are merely part of "a well-scripted conflict narrative between corporate entities".
Here in Australia our Federal Court recently hit Chevron for $300 million in a tax case that some say will have implications worldwide for corporate tax structures.
Couldn't a new model of law based on that kind of questioning of corporations and corporate governance regulate the Valley (and Wall Street) effectively?
If not what are the holes in such a claim?
If you ask Amazon "suicide how to", you get exactly what you asked for.[2] Is that wrong?
[1] https://www.amazon.com/Final-Exit-Digital-2011-Self-Delivera... [2]
As you drive through you can see drastic changes of roadside advertisements transitioning between English, Chinese, and Spanish. In the poor areas of town bus stations are plastered with high-interest payday loans. Drive 5 miles North and you will see advertisements for expensive art/music events. Drive slightly more North and you will see endless advertisements for divorce and accident lawyers.
It would be a miracle if all of this store location/advertisement placement was coincidence.
It's such a shtark reality you can stand at the Bus Station, read the bus outside advertisements, and estimate if it's likely to be going where you are. But once you get inside the bus where fewer people of wealth can see, you no longer can.
https://news.ycombinator.com/user?id=antongm
> I’ll illustrate with an anecdote from my Facebook days. Someone on the data science team had cooked up a new tool that recommended Facebook Pages users should like. And what did this tool start spitting out? Every ethnic stereotype you can imagine. We killed the tool when it recommended then president Obama if a user had “liked” rapper Jay Z...
The author left Facebook in 2013 [0], so I had assumed by now Facebook could do a proper job of recommendations. It's been awhile since I've "Liked" a page, so I just tried it (In-N-Out Burger) and was extremely surprised to see that Facebook apparently still doesn't provide a popup for related pages I may also "Like", in the same way that Twitter does whenever you follow any user.
That's really surprising. For starters, Twitter has had user suggestions for several years now, and it has much less structured data to work with than Facebook does -- for example, a FB page can specify its category, address, and many other attributes; Twitter has nothing to go by other than user's name, account screen name, biography, a location field, and connection metadata (i.e. geolocated IP). Yet it's been awhile if ever since I've seen Twitter recommend someone truly off, e.g. President Obama and Jay-Z.
In contrast, Facebook provides a related stories feature for news stories a user shares in the feed - or at least it did before the fake news thing blew up. Is this a signal that Facebook doesn't really care about user engagement with brand pages? Or did the fuckup that the author describes make it so that this was a feature that no product manager wants to be associated with?
(Or it could be that I'm just missing the Related Pages feature. Maybe it shows up in mobile and not the browser version)
edit: It's strange that the feature doesn't exist in 2017, but this anecdote, as portrayed by the author, reflects badly on both Facebook's data science and the author. I mean, how is it possible for so many smart people to not figure out how to fix such undesired associations? There are so many other signals for clustering related users beyond similar followers.
edit 2: This feature does exist -- you have to like a post in your news feed (on the website), and then Facebook will append a "Related" box in your newsfeed. Unless your browser window is open particularly high, the fact that the "Related" box was triggered by the Like action is obfuscated (because it is out of view). Still doesn't explain why they couldn't make the feature work in 2012-2013.
edit 3: Appears author is wrong about the tool being "killed". Seems to still exist, though maybe it was only temporarily "killed"
http://www.adweek.com/digital/facebook-suggests-similar-page...
https://www.facebook.com/addpage
[0] https://www.linkedin.com/in/antongm/
Would you please stop posting uncivil comments to HN? We ban users who do that repeatedly, and you've done it repeatedly just in this thread.
So it's in some ways 'more fair' but it does take away our ability to 'socially engineer' certain decisions by prohibiting use of one feature or another.
Not to nitpick, but the author worked at FB from 2011-2013. He was there when FB's ad platform really started to find legs.
For starters, today's platform is completely mobile focused. FBX (product the author worked on) was entirely desktop inventory.
Not that either companies' employees seem to care, of course.
Once again, a little reminder about my suggestion to create a public dataset of all employees of Facebook and Google for improving the state of the art of ML and DL algorithms (consider it a public service).
https://news.ycombinator.com/item?id=14245513
If you obtain private information and exploit it, it is called doxing. If a megacorp does the same, it is called a business plan.
We should probably start with Antonio Garcia-Martinez as the starting node, and do a BFS. Or is DFS better? I just get super stoked about all these technical riddles, I can't really help it!
Truer words were never spoken.