Facebook knows you’re gay before you do | Hacker News Reader

Facebook knows you’re gay before you do

americablog.com

145 pointsfollowmylee88 comments

I wonder if anyone here has actually tried running a Facebook ad campaign and has seen the results? I have, and Facebook is not nearly as sophisticated as this article and many commenters are making it out to be.

The article makes a claim with a very unsubstantiated sense of certitude:

> Matt’s two comments, along with other data points ... led Facebook to estimate that if you lined a hundred people with characteristics identical to Matt’s, a significant number of them will be closeted gays

At best, the article can state Facebook might have estimated that Matt was a closeted gay. I seriously doubt that. Facebook (on the ad manager) only estimates that 14 million people are interested in the "Gay" topic. And it only estimates that 4 million are interested in the LGBT topic. Looking at those numbers, those figures are probably drawn off like pages. Even if Facebook were trying to sneakily indirectly profile everyone, they're doing a bad job of it.

The more likely explanation is that this is a coincidence due to someone targeting their ads at people interested in liberal politics (which Facebook would know from liking a politician).

The paranoid, Occams-razor-butchering explanation is the one the article gives.

I'd wager that for every closeted gay that was targeted this ad, 50 other not closeted-gays saw this ad and thought it was weird and then moved on.

What I wonder is, what kind of attitude will people who grow up with those kinds of analytics have towards them?

A little while ago in a conversation about privacy an acquaintance told me that he feels good about the fact there is someone or something out there that cares about his interests, even if it's Google [1]. Thinking back to this brings up in my mind the conversation with the AI Morpheus in Deus Ex [2]; looks like it is still a tossup whether it will turn out to have been prescient. I don't consider data mining to be inherently unethical but I'm not sure what to think of this.

[1] To put this in context, at the time the person in question also reported feeling depressed and lonely.

[2] https://en.wikiquote.org/wiki/Deus_Ex#Morpheus_2

I'll quote the juiciest part:

"Morpheus: Human beings feel pleasure when they are watched. I have recorded their smiles as I tell them who they are.

JC Denton: Some people just don't understand the dangers of indiscriminate surveillance.

Morpheus: The need to be observed and understood was once satisfied by God. Now we can implement the same functionality with data-mining algorithms.

JC Denton: Electronic surveillance hardly inspires reverence. Perhaps fear and obedience, but not reverence.

Morpheus: God and the gods were apparitions of observation, judgment and punishment. Other sentiments towards them were secondary.

JC Denton: No one will ever worship a software entity peering at them through a camera.

Morpheus: The human organism always worships. First, it was the gods, then it was fame (the observation and judgment of others), next it will be self-aware systems you have built to realize truly omnipresent observation and judgment.

JC Denton: You underestimate humankind's love of freedom.

Morpheus: The individual desires judgment. Without that desire, the cohesion of groups is impossible, and so is civilization."

Someone really needs to tell fiction-writers that "Contemplate the inevitable/eternal" tropes have gotten very, very old. I can't speak for all readers, but I for one have stopped seeing these conversations as any kind of serious thematic debate and started just waiting for the hero to punch Mr. Zen Cleverdick in the face.

Perhaps a Herbert reference is appropriate. "The most fundamental principles of the universe are accident and error."

I remember a while back that Target had to back of a similar approach when it began sending out coupons to women that they suspected (w/ apparently a high degree of accuracy) were pregnant. The main anecdote was about a father who found out his daughter was pregnant because they started to receive the coupons. Target eventually backed off this approach (at least for newly pregnant women) because many of their customers complained that the marketing campaign was creepy.

I don't think they stopped it. Just adjusted it so the offers we're alongside random stuff (eg discounts on lawn mowers). That way people wouldn't freak out (presumably because they didn't know how much the company really knew about them)

http://www.nytimes.com/2012/02/19/magazine/shopping-habits.h...

Edit: Worth posting the whole quote (just found it).

“With the pregnancy products, though, we learned that some women react badly,” the executive said. “Then we started mixing in all these ads for things we knew pregnant women would never buy, so the baby ads looked random. We’d put an ad for a lawn mower next to diapers. We’d put a coupon for wineglasses next to infant clothes. That way, it looked like all the products were chosen by chance.

“And we found out that as long as a pregnant woman thinks she hasn’t been spied on, she’ll use the coupons. She just assumes that everyone else on her block got the same mailer for diapers and cribs. As long as we don’t spook her, it works.”

Right (I forgot,) thanks for the link.

Off-topic, but since when did Target start selling lawn mowers?

The only lawnmowers I've seen are the rotor-bladed push-powered mowers. They usually only have them in the spring when all the garden stuff is out.

Offering alcohol to someone you know is pregnant, ugh.

Last I knew wine glasses were non-alcoholic. (From the quote) it seems they have carefully chosen not to market them alcohol directly.

You do know that the medical profession has figured out a small amount of alcohol (a drink or so) after the first trimester isn't a problem, right?

Bullshit.

Studies have not been able to show negative effects from moderate drinking in pregnancy. The medical community tells women not to drink because they have not been able to show that any level of drinking is "safe".

http://palatepress.com/2012/02/wine/light-drinking-during-pr...

The medical profession is positively paleolithic and incredibly patriarchal when it comes to pregnancy. There is no rational sense of weighing risks by probability, instead it's a "you can never be too safe!" attitude. My wife researched this issue heavily during pregnancy, and what really ground her gears were all the people who say "the safest option is to just not drink at all." But the fact of the matter is that these same people continued to engage in incredibly risky behavior like the wife continuing to drive during pregnancy.

Inconclusive either way is different from 'shown drinking isn't a problem'. So technically, he's right, even if a little too aggressive about it.

speaking of probability, doesn't having a drink dramatically raise the probability of having a drink again soon? isn't human behavior the least reliable piece of the metaphor, not the spotty scientific research?

You can say that as much as you would like... then go do a Google search. No one is saying to get drunk with a baby, but a glass (max) is fine.

People have doing it (drinking a glass once in a while, while being pregnant) for ages without any problems. It's just BS American parents who are fussy with everything.

Oh, and "BS"? Really? It's from medical science.

You missed the point. From the quote: "Then we started mixing in all these ads for things we knew pregnant women would never buy..."

In other words, wine glasses are on par with lawn mowers as far as pregnant women are concerned. They know women aren't interested in them.

Because it's inconceivable that a pregnant woman might be buying things for someone not pregnant. Anyhow, as the others pointed out, it's part of the randomness disguise.

But this type of marketing is creepy and invasive. No one likes this because we're basically powerless to choose an alternative. My fiance is marketed tampons on Facebook during the correct time of the month. Because of things she bought on a shopper's card. Disturbing. The real impact will be when your rates on all types of insurance are affected by this data. Or when employment decisions are made.

Tesco was the first company to really do this right - there's a book called Scoring Points[1] that outlines this in details. By using the detailed profiles of customers, they learned that giving discounts on products customers cared about was much more valuable than just giving discounts based on vendor preference. A fan of Coca-Cola will come in to buy a Coke, but won't buy Pepsi, no matter if it's on sale or not.

[1] http://www.amazon.com/Scoring-Points-Winning-Customer-Loyalt...

For those with sub to Safari, Scoring Points is available there:

my.safaribooksonline.com/book/-/9780749453381

Personalization feels like the SEO of 2013. In general, personalization is very hard in terms of "what customer wants" -- look at Amazon's "products for you." In the case of Facebook, this feels like an association algorithm based on affinity rather than personalization the way Target was doing it.

Here's a link to the Target story: http://www.forbes.com/sites/kashmirhill/2012/02/16/how-targe...

This is a bit creepy but it is the kind of thing that could be guessed from facebook activity so it is kind of to be expected.

Lately I've been seeing ads that correlate with things that happen to me completely outside Facebook. About things on sites I only access through google for example. I find this more disturbing.

I checked the "Block third-party cookies and site data" in Chrome so they shouldn't be able to track my web browsing through ads. I'm not sure where they get their info.

For example, last week I bought a new kind of barbecue sauce at the grocery store and the next day Facebook was asking me to like the specific brand's page. It made me wonder. Are banks funnelling purchase information to Facebook?

I never researched this sauce online although I had looked up recipes on how to make homemade sauce through google. But facebook seemed to know the specific brand I had bought.

A similar thing happened to me when I bought replacement windshield wipers for my car a few months ago. I don't know, maybe facebook was just following the weather and knowing that there had been a lot of wiper destroying freezing rain at the time, was suggesting liking major wiper sellers to everybody around here but I still found it suspicious. I had not looked up any wipers online or mentioned that I needed some anywhere and the ad showed up just after my purchase. I know this is all very unscientific. Anybody else have similar experiences?

I don't know whether or not facebook is doing this, but it is worth knowing that supermarket purchase data is often merged with online profiles. For example, my ISP (Verizon FIOS) buys this data, merges it with subscription information and household data and then "shares" it with their internet advertising service bureau, who happens to be google. They don't need 3rd party cookies in these scenarios because they can match ip to subscriber name & address.

I don't believe there are feeds that are real time enough to support next day purchase correlation, but there certainly will be in the future.

Note that even if you've given them bad or no data when you obtained the affinity card they still match you with your credit cards that you've used at the grocery.

Is there a way to opt-out of this without paying cash for everything (and avoiding "store discount" cards)?

I think switching where you shop is probably the most practical - it's rare for independent groceries to have such a program, and the smaller the chain the less likely they are to be selling shopping data on. Under a certain volume it's apparently not appealing to bother doing the deals to buy it.

don't tie a store discount card to your real name/email/phone numbers?

Target ties purchases to your credit card, without needing a store discount card.

http://www.nytimes.com/2012/02/19/magazine/shopping-habits.h...;

yeah, if you want to do that. and I do - we get 5% off everything.

> last week I bought a new kind of barbecue sauce at the grocery store and the next day Facebook was asking me to like the specific brand's page

Facebook has teamed up with datalogix to track users' offline purchases, but as far as I know this is just for a/b testing and not for targeting/ad serving.

http://www.slate.com/articles/technology/technology/2013/03/...

It's also quite possible it's just a coincidence; that specific brand happens to be running a campaign on facebook, and facebook either got lucky or predicted you would like their sauce (i.e. did their job well).

>and facebook either got lucky or predicted you would like their sauce (i.e. did their job well).

Or he saw the sauce on Facebook BEFORE he bought it, bought it unconsciously because of having seen it, and then only really consciously noticed the ad after he had already bought the sauce.

Which is much more probable.

Blocking third-party cookies does practically nothing to prevent such tracking. Web-bugs and supercookies, just for starters, are immune to that.

Install Ghostery to get all the low-hanging fruit - you'll probably want to tweak-up the settings from the default. There are still sophisticated trackers out there beyond what Ghostery can stop. For firefox the Request Policy add-on will give you very good control over that stuff, but it requires a "mechanic's" level of understanding to use without pulling your hair out.

Frankly I am a bit surprised by your naivete - this sort of profiling is big business. Huge. I kind of thought that everyone on here on HN was aware of just how intrusive these tracker/stalkers were.

Online tracking I can understand, but I consider myself to be pretty on top of these types of issues and I was surprised to find out about that Facebook might be tracking offline behavior too. Even though it seems obvious in hindsight - credit card and loyalty card companies know everything you buy, and Facebook knows your personal info, so just connect the two - I don't think it's the type of thing that occurs to most people.

Thank for for making me feel even better about deleting my Facebook account over 2 years ago.

Edit: also, you may want to look into using Ghostery: http://www.ghostery.com Many (most?) of these trackers are not blocked by the "third-party cookies" setting in your browser.

Another hypothesis is that Facebook had been showing you the barbeque sauce ad all along, which is why you bought it. You only noticed afterwards.

Or more likely to me, the barbecue sauce company recently launched a marketing campaign that influenced you to buy one (and the campaign is continuing on Facebook where you see it now).

Facebook also might know that the store only sells one brand of barbeque sauce.

> Lately I've been seeing ads that correlate with things that happen to me completely outside Facebook. About things on sites I only access through google for example. I find this more disturbing.

Though you say completely outside Facebook, at least it is still just taking signals from other parts of the Internet. Imagine now what Google Now or Glass will be able add!

Seriously though, I think Perfect Audience (YC S11) https://www.perfectaudience.com does stuff like this. Here is a thread describing it: https://news.ycombinator.com/item?id=4615408

Edit: Didn't see that you bought the BBQ sauce at the store exclusive without any online behavior. Unless you searched for it or use Mint or a similar service, I am at a loss.

Facebook has friended Datalogix which tracks grocery purchases via loyalty cards. I would not be surprised if that's how facebook found out he bought the sauce.

http://www.dailywireless.org/2012/09/24/facebook-data-linked...

They have a bunch of disclaimers about how they don't share personally identifiable purchasing information but the entire industry has basically redefined what is "personally identifiable" to something much less inconvenient for them.

https://www.privacyassociation.org/resource_center/the_chang...

This is meant as a cynic humorous comment:

What do you wonder about? Facebook used advertisements that were relevant to your interests and wonderfully subliminal at that so you did not even get annoyed by them. You bought the barbecue sauce because it was perfectly advertised to you and you just couldn't resist.

This actually has happened to me before and I dismissed the same reasoning because it was not verifiable. I've convinced myself that it's just selective memory and coincidence. I am really not sure what to make of all this since I stumbled upon your post.

Don't assume everything is pure science and not just mere coincidence. Facebook sometimes gives me military related ads that specifically call out military people -- let's assume the marketer knows enough to put some sort of targeting on their ads, then why would I be getting those ads? There are too many variables at play to know. Also, I've seen some bugs where people get ads specific to one location even though their IP address and their stated location on their profile is something completely different. We could chalk this up to coincidence or maybe there was enough data on his profile to assume Matt is gay. At the same time, I don't think it's creepy because the advertisers don't see the names/identities of the people they target on Facebook, so the ads are still private until the user actually signs up for a product or service.

100% accuracy is never going to be possible. In fact I wouldn't be surprised to see a price structure for targeted advertisements based on confidence intervals.

Of course. So when someone posts an article with anecdotal evidence towards the way Facebook uses data, we can assume it might be true in thousands of other cases or tens of millions of cases. It's hard to know which though.

A different pricing structure would be interesting to see sometime in the future.

I made a point earlier that the advertiser chooses the target audience and I don't believe FB does what this article claims. I've seen retargeting creep into FB lately but again someone is making assertions rather than an algorithm IMO. Some 3rd party ad networks let you really drill down.

I think FB can do what this article claims but is not actually doing it.

I can't say what FB did here with much confidence - its an anecodte. But I'd like to point out that it isn't hard to reverse engineer the information out of FB. Buy a targeted ad and then watch the click-throughs. Cross reference the click-throughs from that ad with data from another tracker that knows identities and you've now been able to link whatever your FB ad criteria was to the identities of anyone who clicks through - even if they don't do anything else besides click through.

The article title is inaccurate - "Matt" did know he was gay, he just wasn't out yet. Facebook's ad platform didn't figure it out before he did. Just as in the Target story, the young woman knew she was pregnant, she just hadn't told her father.

The fear mongering is a little high in this article. However, we should stop and think: why is it so scary for others to potentially know that we're gay? Or pregnant? Neither of these is a shameful state.

Nothing wrong with being gay/lgbt. But in many states in the US you can legally be fired or denied housing on the basis of being lgbt.

really ? i thought USA was the land of the free ...

The law hasn't caught up to today's social standards quite yet. Just like you can fire someone because you don't like how they dress, you can fire someone for being gay, as choice in clothes and sexual orientation aren't "protected classes":

http://en.wikipedia.org/wiki/Protected_class

It may not be pleasant, but it just isn't possible to have any sort of real freedom without also having the freedom to discriminate.

The moment one person is prohibited from discriminating against another, freedom is lost.

Setting aside the issue of denying to enter contract with somebody because he's gay (to remove any doubt, I think it is stupid), how the absence of laws that prohibit this behavior on the state level make US not "land of the free"? What the word "freedom" means for you?

Here that means freedom for the business owner to fire you.

The nature of privacy is the desire to control what others know about you and how they learn about it. Big Data is making that more difficult for individuals to control and it is making people uneasy.

My money is on the facebook widgets embedded in other sites this guy visits. Like buttons, login buttons, perhaps a comment widget or three.

Many users may not realize that you don't need to actually click these widgets in order to be recorded by them. Unless you're using a blocker (e.g. Ghostery or DoNotTrackMe), your browser is pinging facebook every time they're included on a page.

Visit a few gay porn sites that have those widgets embedded and facebook can make a pretty educated guess you're gay.

Do porn sites not embed these things? Ok, how about any number of LGBTQ blogs, news sites, advocacy sites, or even sites that just show things that appeal slightly more to a gay audience than average, such as men's fashion or disney showtunes.

Can you imagine what's going to happen the first time we have presidential candidates that have clickstreams stored from back when they were twelve?

That's going to be some powerful lobbying mojo - certainly more powerful than anyone's privacy policy.

Another thing that Facebook is doing is running a background service on your phone to track your location.

I went to a restaurant. Didn't open the Facebook app while I was there. When I accessed Facebook on my desktop, the first thing it showed me was an ad to like that restaurant.

This is very interesting, particularly because this capability exists without people knowing the little clues they leave can be put together in this way.

The EU data protection act contains the concept of sensitive personal data thanks to the abuse of people by the Nazis. It includes ethnic origin, religion, medical information and sexuality.

The idea is that once you can put together a list of people meeting a particular criteria it's very easy to round them up. While I don't think Facebook or the US are in danger of that kind of excess, neither was Holland before the second world war. Yet one of the reasons for the high death toll of Dutch Jews was the good state of the records:

http://en.wikipedia.org/wiki/History_of_the_Netherlands_(193...

I think it might be a good idea for Facebook and others to consider not just the data they have but also the inferences that can be drawn on it, and reflect whether that leaves them with certain responsibilities like easy opt out or not storing certain things.

This is amazing and creepy at the same time. It's no wonder Data Scientist is one of the highest paying jobs in tech right now.

Facebook (and other similarly sized companies) have petabytes of data - what can they do with it? Imagine!

Considering how many adult websites are adding social features (i.e. Like buttons), I assume 'Big Data' already has much more accurate information on a person's sexual interests than most people realize.

You should be running Ghostery and NoScript/ScriptSafe with JS turned off be default. It's your Internet condom from viruses and advertisers. Like a condom, not 100% effective but you shouldn't concern yourself with edge cases. People can still track you many other ways, but the point is right now they DON'T, so this will keep you pretty safe from this for now. I have any and all domains owned by Facebook and any other social networks I don't use blocked at the router.

Obviously, in reading the article, the primary problem here is that he's a Facebook user, so no amount of protection would help. You can't really use Facebook safely.

call me naive, but it has been my impression that Facebook ads are never that sophisticated. He liked a gay-friendly post and then he got served a gay-friendly ad that happened to be about coming out.

To put things into perspective, I told Facebook seven or eight years ago that I was gay and it hasn't stopped serving heterosexually-oriented dating ads in the mix ever since, presumably because a filter on "male NOT currently in a relationship" is close enough...

How facetious to post such an article on a web page that, upon scrolling to the end, will recommend to me an article about deadly meningitis hitting gay men in NYC ...

Even worse is the "Why not try X, your friend Y is."

This was especially fun in my case when, under the heading "You might like" I had: "<girlfriends name> just signed up on <dating site>". :)

I don't entirely mind them profiling me based on what I choose to put on Facebook. But leaking it to my friends is weird.

(I know you can restrict the audience for everything you do, and I actively do this, but it is soooo easy to slip up).

Can they predict how long before I finally am off the Facebook platform?

Never. I don't have a Facebook account, but it'd be pretty naive of me to assume they don't have a profile on me as well, especially since they've already admitted doing so[1].

[1]: http://usatoday30.usatoday.com/tech/news/story/2011-11-15/fa...

They can't track you if you use Ghostery.

Assuming the Ghostery team has identified and coded in all the trackers they use (and that assumes Ghostery isn't actually spying on its own users, and selling that to FB - I don't think that's likely, but who knows?).

Almost certainly. I wonder what their churn management interventions are.

I don't know all their transactional email strategies, but I've noticed one is those "Do you know [name] or [name]" emails that try to coax you back onto Facebook by asking you if you know certain people and want to friend them.

Annoying and manipulative, at minimum.

It's been about a year since I placed a FB ad but I was under the impression that whomever is placing the ad targets an audience not FB itself. If that is still correct then whoever was placing the "coming out" ad was acting more like an observant bartender mentioned in the article rather than an intelligent algorithm.

This is exactly right. I mean I think Facebook is as creepy as the next guy, but this article makes the unsubstantiated claim that Facebook guessed his orientation. Unless they have some secret new interface for their advertisers in this case it's unlikely. That said I'm sure they could predict this with a high level of accuracy if they wanted to.

You are not entirely right, the advertiser chooses what audience to target but only Facebook knows exactly who is in that audience, the advertiser probably was advertising to people who support gay rights, etc.

> And in the case of the phone company, if you had a little sexy talk with your significant other, no one was the wiser because you had a reasonable expectation your amorous dialogue wasn’t being stored for posterity.

Was that really an expectation? I'm in my early 30s and still remember having a shared party telephone line for several years of my youth and before machine switched networks, operators were also known to listen in on conversations and spread the latest gossip around town.

The focus on privacy seemed to not really happen until digital cell phones started entering the scene, which is very late in the life of the phone networks.

I suspect survivor bias. If you believe there is a billion people on Facebook (even if it's wrong by order or two of magnitude, doesn't matter), several millions of them must be gay. Some percentage of them would be closeted gay. Some of them would have friends that visited or liked "coming out" pages. So if Facebook just displayed random page which your friends liked - without knowing anything else about you - it would inevitably happen that somebody who is gay and thinks about coming out would see such ad sooner or later and would be creeped out about how much Facebook knows about him.

The "Like" button is tracking users all over the web. Isn't it possible that their algorithms are correlating patterns in his web history for ad placement on FB too?

That's what the article is saying, yes.

It seemed like there was more emphasis on his comment on a Rob Portman gay marriage story.

I'm totally pro-facebook: It's a nice, free platform where people can share pictures and chat (the chat is awesome since everybody has facebook).

But someday people are gonna realize how much information they have given facebook about themselves. It's like people go out of their ways to have such a complete profile as possible...

That sounded amazing but it's just a bad title. "Facebook knows you’re gay before your parents" would be true.

This is old news; researchers figured out how to do this years ago:

https://chronicle.com/blogs/wiredcampus/mit-students-faceboo...

[Edit: looks like the article mentions this]

I have the perfect solution ... Randomly click on articles you have no interest in. I'm sure (other than my nerdism) that my profile is so eclectic that they're totally confused. And if that doesn't work, I don't have a Facebook account.

On the contrary , they know you are the sort of person who randomly clicks on ads to confuse learning systems.

Not surprising at all. Ads are targeted based on user behavior on the entire web, not just Facebook. Cookies and such follow you every where and are collected by various sites.

FB big data will soon be able to predict your next bowel movement and recommend reading material based on your Likes.

Nope, facebook wasn't around yet.

Reply on news.ycombinator.com