What Facebook Knows About Us

 

So I’m reading this opinion piece in The New York Times (Yes, I confess! I read it!) and thinking — hmmm.

Sandy Parakilis, who used to work for Facebook, writes that she knows this behemoth from the inside because she led the effort to fix privacy problems on its developer platform before its 2012 IPO. And she knows for sure, she says, that they don’t give a toss about your privacy. Now, you can judge her arguments on their merits (I’m sure she’s right), but they’re not my point. My point is about this quote:

Facebook knows what you look like, your location, who your friends are, your interests, if you’re in a relationship or not, and what other pages you look at on the web. This data allows advertisers to target the more than one billion Facebook visitors a day. It’s no wonder the company has ballooned in size to a $500 billion behemoth in the five years since its I.P.O.

The more data it has on offer, the more value it creates for advertisers. That means it has no incentive to police the collection or use of that data — except when negative press or regulators are involved. Facebook is free to do almost whatever it wants with your personal information, and has no reason to put safeguards in place.

Okay, no surprise there. Of course I’m especially interested in this, for many reasons, but one of them is that I’d like to see serious, well-constructed research — not idle speculation — that would allow us to form reasonable assessments about whether this kind of targeted advertising on Facebook is effective. Might we be able to form non-totally-random conclusions about the probability that a Russian Facebook-ad buying blitz (or a Trump buying blitz, or a Hillary blitz) persuaded so much as a single voter to change his or her mind about anything, no less the way he or she would vote?

Don’t get distracted by “the Russians,” though — that’s a tangent: I’m just explaining why the question especially interests me. I assume this advertising works, or people wouldn’t spend money on it. But do I actually have any evidence of this? No. Facebook probably does. But does anyone else?

And here’s where I get especially doubtful. I don’t have access to their data, or to the data of other people they’re showing ads, but I do know what ads they show me. Judging by these (which, admittedly, is not the right way to judge, it’s a tiny anecdote in an ocean of data), their targeting is ridiculously bad. I reckon any human, even a child, who reads my Facebook page regularly would do way better. What might we hypothesize from this about the real state of their ability to microtarget ads effectively based on the extensive data they collect?

I put this question on my own Facebook page yesterday and then kicked myself, because obviously if I want a useful answer, Ricochet is the place to ask. The question below is roughly what I wrote. Maybe you guys can at least steer me toward the state-of-the-art academic research, or suggest who might know where I’d find that research.

Given that Facebook does know what I look like, my location, who my friends are, my interests, whether I’m in a relationship, and what other pages I look at on the web — not to mention what I buy — why on earth do they think Claire Berlinski’s going to go big time for “Zanzea Women Plus Harem Loose Pants?” (And I’m actually offended by that. They do have pictures of me, after all. “Woman Plus?”)

Why do they think Claire Berlinski would want to “Discover the London cultural season with the fashion historian Amber Butchart?” Why would she be entranced by “l’interprétation du Tourbillon de la vie par Keira Knightley pour la plus irrésistible des collections de CHANEL Joaillerie,” or keen to know more about “The Incredible Story Of The 61 Year Old Makeup Artist Turned Super Model Who Is Now Changing The Way We View Beauty with Her New Makeup Line?”

What on earth have I ever said or done on Facebook — or anywhere on the Internet, or anywhere in the world, for that matter — that would give anyone, even a machine, the impression that I’m remotely interested in that? Those of you who know me in real life will be doubly doubled-over with laughter at this point. You know that I’m to fashion and makeup as kryptonite is to Superman. I am not only the least chic woman in Paris but in all of the Milky Way. The only reason I don’t regularly head up Mr. Blackwell’s list of “worst-dressed women” is that you have to at least try to be well-dressed to make it. As anyone who knows me in person will confirm, nothing comes between me and my Calvins — just so long as by “Calvins” you mean “some cruddy old pair of sweatpants that isn’t so badly covered in cat yak and cat fur that I’m genuinely worried I’ll be sectioned if I go out to buy a loaf of bread wearing them.” (That’s my standard: “Will I be locked up in a mental hospital if I wear this?”)

As for expensive jewelry, you’ve got to be kidding me. The only valuable jewelry I’ve ever owned, my great-grandmother’s wedding ring, was stolen by Maoist-Kurdish-separatist-PKK-slave-urchins. (Probably.) I’ve never replaced it. Make-up? I’m allergic to it. That’s why you never see me in it, unless I’m on television — and it’s why whenever I’m on television my eyes are so red I look like I’m on PCP. Even if I weren’t allergic to it, I hate it. I own the same stick of concealer I’ve owned since I was fifteen. I keep it in case of a major zit and a television appearance. If I have to put it on, I count the seconds until I can wash it off.

Anyone who knows me at all, in person, would know these are ridiculous ads to show Claire Berlinski. What’s more, as far as I can see, my friends are no more interested in these things than I am. I mean, I don’t know them all in real life, but I sure know that if any friend of mine starts nattering on about celebrity makeup artists, I’ll lapse on the spot into a coma of boredom.

I don’t buy these things. I don’t post articles about them. And I never click on these ads. Because I am just. Not. Interested. So what does Facebook know about me that makes them think I’m not the woman I think I am?

If I wanted Claire Berlinski’s money, and if I had all the data Facebook’s got about her, here’s what I’d do: First, I’d look at all the book reviews she’s read lately and all the books she almost bought — to the point of putting them in the basket — but which she decided, at the last minute, not to buy. (They should easily be able to figure out, too, from my purchasing history and what I’ve done since my last impulse-splurge, that the reason I couldn’t pull the trigger was because I thought, “No, Claire, you can’t buy any new books until you finish reading the last ones you impulse-bought and never so much as opened.”)

If I were Facebook I’d be showing me ads, over and over again, not only for those books, but for the ones they could easily figure out — based on what I keep reading and writing and posting about — that I’d find near-irresistible. Why are they showing me Oversized Harem Pants when they could be tempting me with this?

I mean, you guys know me well enough that you know full well that if I see that ad often enough, I’ll crack, right? It’s so obvious. I’ve given them so much information that adds up to, “Show her that ad. Over and over. She can’t hold out more than a week. She’ll hit the button.”

One of my worries, based on the amount of data they clearly do have about me (because I no longer make even a desultory effort to keep it from them, or anyone) is that they actually do know something about me that I don’t. I mean, could they be right? You know how Facebook can supposedly figure out that you’re pregnant even before you do? What do they know about me and “Plus-Sized Loose Harem Pants” that I don’t? Are they checking out the number of times I order pizza and counting the minutes I spend reading The New York Times’ latest pumpkin pie recipes and figuring, “She’s gonna need them soon enough?”

Is there any way they could be checking my photos out against my friends’ profiles, and adding it up with my Twitter feed, weighing it up with time I spend almost-buying books on Amazon, looking at all the sites on the Internet where I waste my days flipping through article after article desultorily, and figuring, “Typical. Age 49. By February she won’t be able to keep her (increasingly fat) fingers off ‘l’interprétation du Tourbillon de la vie par Keira Knightley pour la plus irrésistible des collections de CHANEL Joaillerie.’ Start priming her now, because she’s so headed there. Didn’t you see the way she gobbled up that article about Martin Schulz’s fight for the future of the SPD? Not one woman in that demographic who read that whole boring article from start to finish failed to buy the Chanel jewelry within six months. And she devoured that “Stop sugarcoating the housing market” article in the Economist (which she read not just once, but twice), and she nearly — but didn’t — post it to her Facebook page. That’s solid. We’ve run this scenario often enough to know for sure that 87.3 percent of the white women in Paris with seven cats who read that article (twice) and who nearly (but didn’t) post it to Facebook will buy anything Keira Knightley flogs. Get Keira to sell her something expensive, too: She’ll be in for something big, don’t bother with the ads for a revolutionary new cat litter box or a microwave, no matter how much she thinks she needs a new microwave.” (And surely they know I need that: How many more times to I have to search “why is my microwave making that scary noise” and “microwave+consumer reports” before they get the hint?) But they never show me the ads. “Oh, and guys: check out her photos: The makeup? Seriously? We’d be doing her (and the world) a mitzvah if we got her to do something about that oily T-zone, that is is so totally unnecessary. Send in the 61-Year-Old-Makeup-Artist. She’ll crack. After all, she did post that article about the NSA’s data trove on Americans — twice! — and all the data tells us that women who read Conor Friedersdorff when they’ve got really bad insomnia and worry about the NSA hoovering up metadata on Americans just cannot resist make-up tips from a 61-year-old supermodel. Odds on a purchase: 77.86 percent.”

So am I doomed to buy this stuff? How exactly do they know?

And what on earth do they know — it’s clearly something I don’t — that made them show these ads to me today?

I mean … they do have access to information I don’t. So how do you imagine I should bring this up? “Honey, you seem great to me, but Facebook thinks you’re emotionally unavailable. And it kind of sounds like they think this thing we’ve got going here is a dumpster-fire-in-prospect. That’s a little worrying, since they know what you do when I’m not around. And I don’t. Maybe we should talk?”

But seriously — and some of you guys would probably have pretty good insight into the state of Facebook’s AI and the way these algorithms are apt to work — what on earth is keeping them from taking the information about me that any human who looks at my Facebook page would, putting it together with all the other information they have about what I read, what I buy, and where my attention lingers, and showing me ads that I would find utterly irresistible?

I bet every one of you could guess what ads might tempt me to push the button. Give it a try in the comments. And I bet none of you would think, “Oversized Harem Pants.” And you’d all be right. 

What ads do you see on your Facebook pages — or on the other parts of the Internet where these ads are supposedly so perfectly integrated with your personal data, in such a diabolically cunning way, as to to make you unable to resist clicking on them?

Are they things you might ever buy? Do they even make sense?

If not, why do you think that is?

 

Published in General
Like this post? Want to comment? Join Ricochet’s community of conservatives and be part of the conversation. Join Ricochet for Free.

There are 92 comments.

Become a member to join the conversation. Or sign in if you're already a member.
  1. James Gawron Inactive
    James Gawron
    @JamesGawron

    Bereket Kelile (View Comment):
    I can also be certain the Russian ads had no significant impact on the election because of the size of the spend. A gubernatorial campaign in California would spend $100k in a matter of days and the shelf-life of whatever your communicating only lasts so long. The idea that $100k could’ve had even a measurable impact is stretching reason beyond its limits.

    Bereket,

    Thanks for this input. I would suggest that those who are still obsessed with the Russian ADs read your short paragraph over and over if necessary. They can verify it all they want. The Russian ADs weren’t all pro-Trump either. Also, the Russian ADs were poorly crafted. The Russians couldn’t stay in business as American campaign consultants because their copy was so clumsy.

    Hillary needs to drink some more white wine and this nonsense needs to stop.

    Regards,

    Jim

    • #61
  2. Claire Berlinski, Ed. Member
    Claire Berlinski, Ed.
    @Claire

    Aaron Miller (View Comment):

    Claire Berlinski, Ed. (View Comment):
    I’m still kind of weirded out by that ad warning me of “emotionally unavailable partners.”

    Otherwise known as cats.

    Point taken. But I reckon too subtle for an algorithm, no matter how advanced ..

    • #62
  3. Claire Berlinski, Ed. Member
    Claire Berlinski, Ed.
    @Claire

    Okay, three questions:

    1. Any pointers to the literature on this? Who’s actually studying (besides Facebook) how effective this form of advertising is, or might be? We could design plenty of experiments that would at least give us a clue, instead of leaving us either with “Big Tech is totally evil and using your data to turn you into their mind-controlled bot” or “Big Tech is Totally evil and convincing its witless customers that it can use your data to turn you into their mind-controlled bot–even though it can do no such thing.”
    2. What’s going wrong with their ads? I’m not kidding: With that much data about me, they should be able to show me ad after ad that might be tempting to me, and they really aren’t. Sounds like they’re not doing a better job showing you tempting ads, either. What could account for that? Are they just not willing to put the work into figuring out how to build a bot that could gather, from the data they have about me, what I might like to buy? Is it much harder to build that bot than I understand? What are the challenges? How come I can’t find any good articles that explain this to me? Are they being paywalls, or is no one else even asking?
    3. Are they collecting all that data? Yes, no question. Are they’re good at targeting ads based on that data? No, doesn’t seem so — though the jury’s still out. So maybe we’ve misunderstood: Maybe they’re not collecting all that data so that they can get better at selling advertising? Maybe they’re collecting it for some other reason. What would that reason be? Is it possible even they don’t know yet, but they’re thinking, “Somehow, someway, for sure, we can monetize this data?”
    • #63
  4. SkipSul Inactive
    SkipSul
    @skipsul

    Claire Berlinski, Ed. (View Comment):
    Okay, three questions:

    1. Any pointers to the literature on this? Who’s actually studying (besides Facebook) how effective this form of advertising is, or might be? We could design plenty of experiments that would at least give us a clue, instead of leaving us either with “Big Tech is totally evil and using your data to turn you into their mind-controlled bot” or “Big Tech is Totally evil and convincing its witless customers that it can use your data to turn you into their mind-controlled bot–even though it can do no such thing.
    2. What’s going wrong with their ads? I’m not kidding: With that much data about me, they should be able to show me ad after ad that might be tempting to me, and they really aren’t. Sounds like they’re not doing a better job with any of you. What could account for that? Are they just not willing to put the work into figuring out how to build a bot that could gather, from the data they have about me, what I might like to buy? Is it much harder to build that bot than I understand? What are the challenges? How come I can’t find any good articles that explain this to me? Are they being paywalls, or is no one else even asking?
    3. Are they collecting all that data? Yes, no question. Are they’re good at targeting ads based on that data? No, doesn’t seem so — though the jury’s still out. So maybe we’ve misunderstood: Maybe they’re not collecting all that data so that they can get better at selling advertising? Maybe they’re collecting it for some other reason. What would that reason be? Is it possible even they don’t know yet, but they’re thinking, “Somehow, someway, for sure, we can monetize this data?”

    I think it’s a case of them not easily being able to distinguish the signal from the noise.  I’m with @guruforhire on the notion that this is going to blow up big time on them in the not too distant future.

    • #64
  5. Phil Turmel Inactive
    Phil Turmel
    @PhilTurmel

    Claire Berlinski, Ed. (View Comment):
    What’s going wrong with their ads? I’m not kidding: With that much data about me, they should be able to show me ad after ad that might be tempting to me, and they really aren’t.

    Don’t discount the possibility that advertisers with the wherewithal to run the analysis have enough information to decide you are too low a probability of sale to spend on ad space in front of you.  Leaving the clueless advertisers and the shotgunners to put theirs in your face.

    • #65
  6. RightAngles Member
    RightAngles
    @RightAngles

    Phil Turmel (View Comment):

    RightAngles (View Comment):

    Nick H (View Comment):

    Phil Turmel (View Comment):
    It’s not just cookies. The collection of techniques is generally called browser fingerprinting, and is remarkably effective without cookies or IP addresses. With cookies, you are 100% trackable. With IP address and browser fingerprint, you are very nearly 100% trackable.

    For those who want to know more about browser fingerprinting (and how to limit tracking), I’d suggest https://panopticlick.eff.org/ as a starting point. It will tell you how unique your browser fingerprint is. There are also some good browser plugins like Ghostery that will block most obnoxious trackers. Combine it with an ad-blocker and your internet experience will be greatly improved.

    Thank you! I’m woefully uninformed. Recently, in my neverending attempts to thwart them, I was browsing with a VPN signed in from the Netherlands. I went to a company’s Facebook page, and up popped my own login screen with my username and password already filled in for my convenience. How??!?

    If you deliberately generate traffic to Facebook, the VPN is a very low hurdle. If you actually sign in to Facebook, you’ve completely defeated the reason to use a VPN. The fact that your username and password were filled in is a local browser behavior, not a Facebook thing. It is generally good to have your browser remember your passwords for you, as it lets you use a different, truly random password at every website. It is important to have a master password in your browser to protect the others, though, or you’d be vulnerable to local attacks.

    Oh, good thing I didn’t sign in then. I do like having my password remembered, because I use all different ones.

    • #66
  7. Bereket Kelile Member
    Bereket Kelile
    @BereketKelile

    Claire Berlinski, Ed. (View Comment):
    Okay, three questions:

    1. Any pointers to the literature on this? Who’s actually studying (besides Facebook) how effective this form of advertising is, or might be?…
    2. What’s going wrong with their ads? I’m not kidding: With that much data about me, they should be able to show me ad after ad that might be tempting to me, and they really aren’t. Sounds like they’re not doing a better job showing you tempting ads, either. What could account for that? Are they just not willing to put the work into figuring out how to build a bot that could gather, from the data they have about me, what I might like to buy? Is it much harder to build that bot than I understand? What are the challenges? How come I can’t find any good articles that explain this to me? Are they being paywalls, or is no one else even asking?
    3. Are they collecting all that data? Yes, no question. Are they’re good at targeting ads based on that data? No, doesn’t seem so — though the jury’s still out. So maybe we’ve misunderstood: Maybe they’re not collecting all that data so that they can get better at selling advertising? Maybe they’re collecting it for some other reason. What would that reason be? Is it possible even they don’t know yet, but they’re thinking, “Somehow, someway, for sure, we can monetize this data?”
    1. Not sure if there’s literature but I know that getting access to proprietary data is not usually easy, especially for use by any academic researcher. Plus, even with the data it’s difficult to measure the effectiveness when you’re targeting different audiences. Facebook collects data on how people respond to an ad but that doesn’t tell you if that’s a good ad to run. You have to combine the right message with the right audience.
    2. Advertising is still a creative endeavour and it’s not automatic that people click “like.” As an advertiser, the data I have on you is in the aggregate. I don’t really know you, Claire, very well. I have to build an audience off of the kinds of pages you’ve liked, where you live, your age, etc. but that’s not a comprehensive picture of who you are. At some point in time you exhibited an interest in something that I could use for targeting but I don’t know much more about you than that.
    3. As I mentioned earlier, you get data on how many people are seeing your ad, clicking on it, how long they’re watching a video, how many times they’re seeing it, etc. Again, that’s all in the aggregate so that gives you information more about the ads than it does about the individuals. If there’s continuity there then, yes, you’ll get better at it.

    My impression is the Russians just weren’t that familiar with the platform and how to use it effectively, based on what I’ve read in the news. When you run a repressive regime there’s really no need to develop skills in persuasion (the non-violent type, that is). These are the crafts we learn in a liberal, democratic order.

    • #67
  8. Jules PA Inactive
    Jules PA
    @JulesPA

    Nick H (View Comment):

    Phil Turmel (View Comment):
    It’s not just cookies. The collection of techniques is generally called browser fingerprinting, and is remarkably effective without cookies or IP addresses. With cookies, you are 100% trackable. With IP address and browser fingerprint, you are very nearly 100% trackable.

    For those who want to know more about browser fingerprinting (and how to limit tracking), I’d suggest https://panopticlick.eff.org/ as a starting point. It will tell you how unique your browser fingerprint is. There are also some good browser plugins like Ghostery that will block most obnoxious trackers. Combine it with an ad-blocker and your internet experience will be greatly improved.

    Dies this work on a phone too?

    She stupidly asked.

    ?

    • #68
  9. Chuck Enfield Inactive
    Chuck Enfield
    @ChuckEnfield

    Ekosj (View Comment):

    Aaron Miller (View Comment):
    We consumers are not “giving away the store” because we are offering mostly mundane information in return for a social toolset that is not available otherwise. We are well paid for offering faceless companies info we would not hesitate to tell neighbors if asked.

    “What did you buy yesterday?”

    “Socks and an HDMI cable.”

    “Ooh! Salacious!”

    In return, we keep in touch with relatives and acquaintances better than people did in the era of rotary phones. We keep family up to date with pictures and laugh with friends over amusing anecdotes.

    It’s fashionable to pretend social media is just cat pictures and people insulting each other. But if that’s what you see then you use the tools poorly or dwell in fruitless forums. Billions of people participate, including a few reasonable folks (such as nearly the entire Ricochet membership), participate in social media like Facebook, Twitter, and Pinterest because we do indeed get something valuable in return.

    Sometimes the information we won’t share in face-to-face interactions is quite reasonably shared with robotic advertisers. If your neighbor takes an interest in your underwear, that’s probably cause for concern. If Sears wants to know, why not?

    I beg to differ. Most folks participate because they are under the mistaken impression that it is free and don’t understand the value of the”mundane” information they surrender. Certainly the multiple billions earned by Google, Facebook and the rest should disabuse us of the belief that the data is “mundane” and without value. There is a huge market out there for this data and we are settling for chump change. Remember the “Seinfeld” episode where Kramer she’s the coffee shop? That’s us.

    I, too, beg to differ.  The raw data is plentiful and nearly worthless.   It only has value once aggregated and organized.  But even if you disagree about the value of the data, Aaron is nonetheless correct that it’s a value proposition and we’re each free to decide for ourselves.

    • #69
  10. Phil Turmel Inactive
    Phil Turmel
    @PhilTurmel

    Jules PA (View Comment):

    Nick H (View Comment):

    Phil Turmel (View Comment):
    It’s not just cookies. The collection of techniques is generally called browser fingerprinting, and is remarkably effective without cookies or IP addresses. With cookies, you are 100% trackable. With IP address and browser fingerprint, you are very nearly 100% trackable.

    For those who want to know more about browser fingerprinting (and how to limit tracking), I’d suggest https://panopticlick.eff.org/ as a starting point. It will tell you how unique your browser fingerprint is. There are also some good browser plugins like Ghostery that will block most obnoxious trackers. Combine it with an ad-blocker and your internet experience will be greatly improved.

    Dies this work on a phone too?

    She stupidly asked.

    ?

    The tracking? Yes.

    The browser plugin countermeasures?  Some.

    • #70
  11. David Foster Member
    David Foster
    @DavidFoster

    Claire Berlinski, Ed. (View Comment):
    What’s going wrong with their ads? I’m not kidding: With that much data about me, they should be able to show me ad after ad that might be tempting to me, and they really aren’t.

    Most of the success or failure in this isn’t due to Facebook, but to the advertiser (or their agency) who decides on the interests and demographics to target.

    FB tries to establish what your interests are by looking at what you have done in the past.  Try clicking on the three dots at the upper right of a FB ad, and you will see a ‘Why am I seeing this?’ menu item.  Click this, then click “Manage my ad preferences” and it will tell you what it (FB) thinks your interests are. In my case, it thinks I am interested in Entrepreneurship (true) and Penny Stocks (not true), among other things.

    These Interests are available to the advertiser as selection criteria when he runs his ad.

    • #71
  12. Randy Webster Inactive
    Randy Webster
    @RandyWebster

    David Foster (View Comment):
    These Interests are available to the advertiser as selection criteria when he runs his ad.

    I’ve not got any random ammo ads.

    • #72
  13. David Foster Member
    David Foster
    @DavidFoster

    Randy Webster (View Comment):

    I’ve not got any random ammo ads.

    Maybe ammo isn’t an interest category….I believe these categories are automatically created, but human-edited in some cases.  Here’s an interesting summary of some of the FB interest categories that do exist, or at least used to exist:

    https://www.theverge.com/2016/2/1/10872792/facebook-interests-ranked-preferred-audience-size

    …and here’s a plaintive plea for the addition of new categories:

    https://www.facebook.com/business/help/community/question/?id=10210106775051976

    • #73
  14. Chuck Enfield Inactive
    Chuck Enfield
    @ChuckEnfield

    Randy Webster (View Comment):

    David Foster (View Comment):
    These Interests are available to the advertiser as selection criteria when he runs his ad.

    I’ve not got any random ammo ads.

    Probably because the consequences of getting it wrong are so high.  My guess is that those inclined to prurient internet activities don’t get ads for porn on National Review’s site. Advertising with a significant likelihood of offending certain viewers is generally contained to sites frequented by people with that particular interest.

    • #74
  15. Sisyphus Member
    Sisyphus
    @Sisyphus

    Claire Berlinski, Ed. (View Comment):

    Justin Hertog (View Comment):

    Yes, I think Facebook is doing all those things with our data and selling it.

    I don’t doubt it for a second. I just wonder why they don’t seem to be as good at it — at least in my case — as I’d expect them to be.

    They think they are narrowly targeting us but they still err on the side of the mass market 90% for any given set of evidence.

    A woman who was living in Istanbul and now living in Paris must clearly be the plaything of some rich shah or sultan, work slavishly to maintain her air of desirability, and may even be competing for his attention against an 11 year old 1st cousin of his.

    Or the harem pants are for your cats.

    Drawing the right conclusions from all that data is harder to automate then it looks, but at least they aren’t pushing Viagra.

    • #75
  16. Snirtler Inactive
    Snirtler
    @Snirtler

    Claire Berlinski, Ed. (View Comment):
    Okay, three questions:

    1. Any pointers to the literature on this? Who’s actually studying (besides Facebook) how effective this form of advertising is, or might be? We could design plenty of experiments that would at least give us a clue, instead of leaving us either with “Big Tech is totally evil and using your data to turn you into their mind-controlled bot” or “Big Tech is Totally evil and convincing its witless customers that it can use your data to turn you into their mind-controlled bot–even though it can do no such thing.”
    2. <snip>
    3. Are they collecting all that data? Yes, no question. Are they’re good at targeting ads based on that data? No, doesn’t seem so — though the jury’s still out. So maybe we’ve misunderstood: Maybe they’re not collecting all that data so that they can get better at selling advertising? Maybe they’re collecting it for some other reason. What would that reason be? Is it possible even they don’t know yet, but they’re thinking, “Somehow, someway, for sure, we can monetize this data?”

    Bereket pointed out that $100K in ad spending is small potatoes.

    Ethan Epstein in The Weekly Standard wrote a nice, brief article on the likelihood that Russian-sponsored ads made a difference in the election. Facebook bragged that about 126 million Americans saw the ads. He compared the reach of those Russian ads with a couple of social media ad campaigns:

    In 2013, Frito-Lay ran a contest in which people voted on new flavors of potato chips to offer. More than 1 billion Facebook impressions [industry jargon for how many times an ad is seen] were generated by the gimmick. Even niche products can draw a huge audience: More than 124 million people have viewed a Facebook ad for the Squatty Potty, a stool that sits underneath a traditional toilet.

    By a different metric, again the Russian ads appear to be small potatoes.

    Maybe it doesn’t matter that Facebook is not so good at “targeting ads based on that data”, because it’s nonetheless succeeding at making money off it. In 2015, digital ad revenue was $60 billion, 65% of which went to just two companies, Google and Facebook. And between the two of them, they account for 99% of the growth in digital revenue from 2015-2016. (source)

    • #76
  17. Claire Berlinski, Ed. Member
    Claire Berlinski, Ed.
    @Claire

    Sisyphus (View Comment):

    Claire Berlinski, Ed. (View Comment):

    Justin Hertog (View Comment):

    Yes, I think Facebook is doing all those things with our data and selling it.

    I don’t doubt it for a second. I just wonder why they don’t seem to be as good at it — at least in my case — as I’d expect them to be.

    They think they are narrowly targeting us but they still err on the side of the mass market 90% for any given set of evidence.

    A woman who was living in Istanbul and now living in Paris must clearly be the plaything of some rich shah or sultan, work slavishly to maintain her air of desirability, and may even be competing for his attention against an 11 year old 1st cousin of his.

    Or the harem pants are for your cats.

    Drawing the right conclusions from all that data is harder to automate then it looks, but at least they aren’t pushing Viagra.

    Yeah, you’re right: I’m not a good case study because so few people are in my demographic and it’s just too weird a demographic for it to make sense that they’d spend a lot of time studying it. That’s why I asked whether other people are receiving ads that seem as off-base — I’m still looking for anecdotal evidence, but I’m looking for more of it than “just me.” So far a lot of  people have said, “Yeah, it’s weird: I get ads that don’t in any way suggest that they’re able to use this data effectively.” But again — anyone who knows me is already in a slightly-weird demographic. So I’d really like to know what the academics who study this are finding (if indeed any are studying it).

    Still, you’d think the “Look at the books she nearly bought but didn’t” idea would be so obvious …

    • #77
  18. Randy Webster Inactive
    Randy Webster
    @RandyWebster

    Snirtler (View Comment):
    a stool that sits underneath a traditional toilet.

    An unfortunate choice of words.

    • #78
  19. Percival Thatcher
    Percival
    @Percival

    Claire Berlinski, Ed. (View Comment):

    Sisyphus (View Comment):

    Claire Berlinski, Ed. (View Comment):

    Justin Hertog (View Comment):

    Yes, I think Facebook is doing all those things with our data and selling it.

    I don’t doubt it for a second. I just wonder why they don’t seem to be as good at it — at least in my case — as I’d expect them to be.

    They think they are narrowly targeting us but they still err on the side of the mass market 90% for any given set of evidence.

    A woman who was living in Istanbul and now living in Paris must clearly be the plaything of some rich shah or sultan, work slavishly to maintain her air of desirability, and may even be competing for his attention against an 11 year old 1st cousin of his.

    Or the harem pants are for your cats.

    Drawing the right conclusions from all that data is harder to automate then it looks, but at least they aren’t pushing Viagra.

    Yeah, you’re right: I’m not a good case study because so few people are in my demographic and it’s just too weird a demographic for it to make sense that they’d spend a lot of time studying it. That’s why I asked whether other people are receiving ads that seem as off-base — I’m still looking for anecdotal evidence, but I’m looking for more of it than “just me.” So far a lot of people have said, “Yeah, it’s weird: I get ads that don’t in any way suggest that they’re able to use this data effectively.” But again — anyone who knows me is already in a slightly-weird demographic. So I’d really like to know what the academics who study this are finding (if indeed any are studying it).

    Still, you’d think the “Look at the books she nearly bought but didn’t” idea would be so obvious …

    I get bombarded by “celebrity” gossip ads. I suspect most people do. I don’t know who 80% of these people are and won’t be moved to find out by a Facebook ad.

    The demographic of “people with bizarre senses of humor interested in history in general and ancient and medieval history in particular” doesn’t seem to be of much interest. I’m good with that.

    @snirtler‘s link to Ethan Epstein’s article also contained this gem:

    Indeed, an experiment run by PC World in 2013 found that Facebook ads have abysmal click-through rates—that is, the percentage of people who tangibly respond to an ad they see. That suggests they’re easily ignored.

     

    • #79
  20. Bereket Kelile Member
    Bereket Kelile
    @BereketKelile

    Snirtler (View Comment):

    Claire Berlinski, Ed. (View Comment):
    Okay, three questions…

     

    Ethan Epstein in The Weekly Standard wrote a nice, brief article on the likelihood that Russian-sponsored ads made a difference in the election. Facebook bragged that about 126 million Americans saw the ads

    I find it hard to believe $100k would reach that many people. One of the metrics advertisers follow on their FB ads is CPM, cost per thousand. Here the label accurately describes what it is, a measure of how much it costs to reach a thousand people. If my math is right, the CPM in that scenario would be around $0.80, which I’ve only ever seen in places like Africa, where apparently no one is advertising. If the Russians did accomplish that in the U.S. then they are the greatest advertising geniuses of our time. For context, if you can get a CPM of around $10 that’s pretty good and it’s more common for it to be up near $20.

    • #80
  21. Snirtler Inactive
    Snirtler
    @Snirtler

    Bereket Kelile (View Comment):

    Snirtler (View Comment):

    Ethan Epstein in The Weekly Standard wrote a nice, brief article on the likelihood that Russian-sponsored ads made a difference in the election. Facebook bragged that about 126 million Americans saw the ads

    I find it hard to believe $100k would reach that many people.

    All very interesting. One other thing Epstein points out–Facebook has a business incentive to overstate the effect of ads delivered through its platform.

    One of the metrics advertisers follow on their FB ads is CPM, cost per thousand. Here the label accurately describes what it is, a measure of how much it costs to reach a thousand people. If my math is right, the CPM in that scenario would be around $0.80, which I’ve only ever seen in places like Africa, where apparently no one is advertising. If the Russians did accomplish that in the U.S. then they are the greatest advertising geniuses of our time. For context, if you can get a CPM of around $10 that’s pretty good and it’s more common for it to be up near $20.

    Your math checks out. I’m not convinced either way about the claim of 126 million Americans. But if that’s anywhere close to being true, that lends credence to the characterization of Russia as a malicious international actor. They get to play dirty tricks abroad on the cheap.

    Can you say more about the $10-20 CPM for FB ads? Is that across industries, for a particular industry, for political ads in the US, or whatever?

    • #81
  22. Sisyphus Member
    Sisyphus
    @Sisyphus

    Claire Berlinski, Ed. (View Comment):

    Sisyphus (View Comment):

    Claire Berlinski, Ed. (View Comment):

    Justin Hertog (View Comment):

    Yes, I think Facebook is doing all those things with our data and selling it.

    I don’t doubt it for a second. I just wonder why they don’t seem to be as good at it — at least in my case — as I’d expect them to be.

    They think they are narrowly targeting us but they still err on the side of the mass market 90% for any given set of evidence.

    A woman who was living in Istanbul and now living in Paris must clearly be the plaything of some rich shah or sultan, work slavishly to maintain her air of desirability, and may even be competing for his attention against an 11 year old 1st cousin of his.

    Or the harem pants are for your cats.

    Drawing the right conclusions from all that data is harder to automate then it looks, but at least they aren’t pushing Viagra.

    Yeah, you’re right: I’m not a good case study because so few people are in my demographic and it’s just too weird a demographic for it to make sense that they’d spend a lot of time studying it. That’s why I asked whether other people are receiving ads that seem as off-base — I’m still looking for anecdotal evidence, but I’m looking for more of it than “just me.” So far a lot of people have said, “Yeah, it’s weird: I get ads that don’t in any way suggest that they’re able to use this data effectively.” But again — anyone who knows me is already in a slightly-weird demographic. So I’d really like to know what the academics who study this are finding (if indeed any are studying it).

    Still, you’d think the “Look at the books she nearly bought but didn’t” idea would be so obvious …

    In a society where 2% read more than one or two books a year, even those markers may be too arcane. Or Amazon and Barnes and Noble and Apple are keeping their own stats and not making them visible to Facebook and Twitter.

    I find Amazon ads funny because they try to steer me to more popular books in the same general domain. So I buy Pieper’s systemic Lutheran theology and they point me to Joel Osteen’s Live Your Best Life Now. An obviously demonic choice in light of Christian hopes of Heaven, even if Osteen outsells Pieper by several orders of magnitude. I bought a couple of historical novels by a distinguished professor and was rewarded for awhile with Harlequin style bodice rippers.

    Now if you were on Goodreads (owned by Amazon) and diligently documenting your buy decisions for the past few decades and you used your Facebook login on Goodreads and Goodreads was posting your activity to Facebook…

    • #82
  23. Don Tillman Member
    Don Tillman
    @DonTillman

    RightAngles (View Comment):
    On LinkedIn, a man appeared on my “People You May Know” list, and it was a man in Paris who I haven’t had contact with in 15 years. In addition, our last contact was through an email address I haven’t used since then. So how do they know I might know him?

    @rightangles , oh yeah, that can happen…

    When LinkedIn was starting out they were in the classic starting-a-marketplace situation; not enough people signed up to attract new users and give them a reason to sign up.  So they had to come up with something clever to jump-start the process and come up to speed.  And that was an offer to upload one’s personal email address book, and they’ll see who they can connect you to from that.  So perhaps somebody with your old email address did that.

    Also, second order connections.

     

     

     

    • #83
  24. Basil Fawlty Member
    Basil Fawlty
    @BasilFawlty

    Don Tillman (View Comment):
    When LinkedIn was starting out they were in the classic starting-a-marketplace situation; not enough people signed up to attract new users and give them a reason to sign up. So they had to come up with something clever to jump-start the process and come up to speed. And that was an offer to upload one’s personal email address book, and they’ll see who they can connect you to from that. So perhaps somebody with your old email address did that.

    Clever?

    • #84
  25. Don Tillman Member
    Don Tillman
    @DonTillman

    Claire Berlinski, Ed.:

    Are they things you might ever buy? Do they even make sense?

    If not, why do you think that is?

    If targeted advertising is worth anything, Facebook has the data and the capability to do it better than anybody else.  Facebook incentivizes people to provide them with an enormous amount of personal information, and they’ve amassed far more than ever before possible.

    Whether Facebook is currently making use of that information, or whether advertisers are taking advantage of it, is another question.

    But if you’ve got a product, and your prime demographic is middle-aged Presbyterian lesbian didgeridoo players from Iowa, Facebook has it covered.

    • #85
  26. Randy Webster Inactive
    Randy Webster
    @RandyWebster

    Don Tillman (View Comment):
    But if you’ve got a product, and your prime demographic is middle-aged Presbyterian lesbian didgeridoo players from Iowa, Facebook has it covered.

    Seems a pretty small market.  But I would have said the same thing about kazoos until everyone in my freshman class at Davidson bought one.  I might have mentioned that we had a hundred kazoo band play “Waltzing Matilda” for the Australian national basketball team.

    • #86
  27. Sisyphus Member
    Sisyphus
    @Sisyphus

    Randy Webster (View Comment):

    Don Tillman (View Comment):
    But if you’ve got a product, and your prime demographic is middle-aged Presbyterian lesbian didgeridoo players from Iowa, Facebook has it covered.

    Seems a pretty small market. But I would have said the same thing about kazoos until everyone in my freshman class at Davidson bought one. I might have mentioned that we had a hundred kazoo band play “Waltzing Matilda” for the Australian national basketball team.

    Life is stranger than Facebook.

    • #87
  28. SkipSul Inactive
    SkipSul
    @skipsul

    I have written something of a follow-up piece to this discussion:

    http://ricochet.com/472555/oh-brave-new-world-the-novel-world-of-big-data/

    • #88
  29. Front Seat Cat Member
    Front Seat Cat
    @FrontSeatCat

    My guess is that Facebook is hooked up to Twitter and anything else in your on line world (not you specifically but everyone). If you even type the words about ex-boyfriends or Turkey or anything on any social site or search, I believe the info is captured and shared. The makeup etc. is probably what’s popular at the moment and you are female, so it probably gets doled out regularly to female Facebook users.

    I read somewhere that if you are on Facebook, you’ve probably been hacked or will be. I was getting a lot of junk I’m my spam too, so for those 2 reasons, I deleted my account. I have a lot less junk, but an old friend told me she also deleted her account, and after about a year, put in the email and password and it came back up – all of it where she left off.  It was never deleted.

    Very good post – I didn’t answer your technical question, but I liked the story of the theft in Istanbul (not what happened) but it was interesting.  Who needs Kitty Genovese when they have the PKK! I wonder what it’s like now?  I wonder if you pondered it on Facebook if you will get bombarded with Turkish spice ads or obscenities from the PKK?? The on line world is a mystery – and Zuckerberg is a rich man because of it.

    • #89
  30. Bereket Kelile Member
    Bereket Kelile
    @BereketKelile

    Snirtler (View Comment):

    Can you say more about the $10-20 CPM for FB ads? Is that across industries, for a particular industry, for political ads in the US, or whatever?

    Sorry for the delayed response. I’m not sure how it breaks down across industries. It has more to do with the content of the ad, the audience targeted, and when the ads were running. Since it’s an observed figure it would be good to hear from marketers’ experiences in other industries.

    • #90
Become a member to join the conversation. Or sign in if you're already a member.