Benjamen Walker's Theory of Everything - Honeypot

Episode Date: October 25, 2016

Your host decides to follow back a Joy Division T-shirt that is following him around the internet. Plus the surveillance that powers behaviorally targeted personalized advertising.

You are listening to Benjamin Walker's Theory of Everything. At Radiotopia, we now have a select group of amazing supporters that help us make all our shows possible. If you would like to have your company or product sponsor this podcast, then get in touch. Drop a line to sponsor at radiotopia.fm. Thanks. episode. Why is there something called influencer voice? What's the deal with the TikTok shop? What is posting disease and do you have it? Why can it be so scary and yet feel so great to block someone on social media? The Neverpost team wonders why the internet and the world because of the internet is the way it is. They talk to artists, lawyers, linguists, content creators, sociologists, historians, and more about our current tech and media moment. From PRX's Radiotopia, Never Post, a podcast for and about the Internet.

Episodes every other week at neverpo.st and wherever you find pods. You are listening to Benjamin Walker's Theory of Everything. This installment is called Honeypot. Late last summer, someone on one of the music lists I subscribed to posted a link for rock and roll t-shirts that were being sold at Barney's, a high-end department store. I clicked, and sure enough, Barney's was selling distressed vintage t-shirts with logos from Black Flag, Pink Floyd, David Bowie, The Cure. They were all super expensive.

Only one, the Van Halen shirt, was under $100. And then, I noticed the Joy Division shirt. It's the classic white-on-black image of radio waves. They used it on the cover for their 1979 album, Unknown Pleasures. I think I paid $15 for this shirt when I bought it at Wax Trax Records. I wore this shirt a lot in high school. Well, here on the Barneys website, they were selling it for over $200. I gasped, I laughed, I think I even tweeted out a link,

forwarded it to a few friends, and then I moved on. But then, this shirt started following me around the internet. Whenever I would click on an article, say from the New York Times or other news sites, it would be there, taunting me. Even on my phone, I would click on something, a link in Twitter, and this damn shirt would just pop up and fill the screen. Now I use ad blockers, so this shouldn't happen, but a number of ad exchanges, especially Google, have recently started paying ad blocker toitelist their ads or let them through.

I tried to ignore this shirt, but after a few weeks, I started to buckle. Perhaps this was a sign. Perhaps the universe was trying to tell me something. Perhaps I just needed to own this thing. Hey, how's it going? Is that it for the band t-shirts? Yeah, you guys had this Joy Division one that I saw.

Oh! Yeah. I took the train uptown to Barney's to see the shirt in person. So who makes this one? R13. And what's the label? I'm not familiar with it. It is based in New York.

They use Japanese fabrics. They're cool. They're cool? Yeah. the label and I'm not familiar with it. It is based in New York. They use Japanese fabrics. They're cool. They're cool? How much is it? $2.25. $2.25. I'll have to think about that.

Do you want to try it on? Yeah, in a minute. Yeah. If you try it on and you end up hating it, then at least you can just free your mind. I went to the dressing room and tried the shirt on. I'm not a Japanese fabric expert, but I can say that it was super soft.

It was also super distressed, like I couldn't imagine it lasting more than two or three washings. But the real problem was when I looked at myself in the mirror. This shirt transformed me into one of those creepy old guys you see at shows. The ones that are always pretending that they're like 22 years old. There was no way I was buying this thing.

Wait, hold on. You should at least take a picture for God's sake. Sure. I did let my friend Andrew take a picture though. You can see it on the TOE page at toe.prx.org. Okay, so here is where it gets super weird. You see, after I took this trip to Barney's, that shirt stopped following me around. I haven't seen an online ad for that Joy Division t-shirt since that day. I know, it's probably just a coincidence, but I can't help but wonder.

Did I stop seeing ads for this shirt because they know I went to Barney's like they think I bought it? Or do they know I tried it on and didn't like it? I know that sounds crazy paranoid, but well, when it comes to the surveillance apparatus that powers online advertising, I have a seriously dark mental model. We all have a mental model for this stuff. Well, at least those of us who use the internet a lot and are aware of the data tracking that goes on when we click on news sites or use platforms like Google and Facebook and Twitter. People like us, dear listener. So I'm kind of curious, what's your mental model for this stuff? If you don't mind, take out a pen and a piece of paper. Okay, ready? Please draw a picture

of how personalized online advertising works. That's Darren Stevenson, a researcher from the University of Michigan and the Stanford Institute for Internet and Society. Over the past few years, he's been studying our mental models for how online advertising works, mainly through focus groups that he's run in Chicago and in London. And as you just heard, Darren kicks things off by asking his focus group participants to draw a picture or diagram for how they think online personalized advertising works. So when they think that an advertisement on a website or app has been personalized for them, what are they thinking about the process that takes place in order for that ad to get to them?

And what I'm really trying to do is sort of draw out the mental model that people have for how personalized advertising works. The drawing he saw most often was one in which the user sets the whole process in motion. So the person clicks to do a web search, then a marketer personalizes an ad, and then they ignore it or click on it. And this sort of makes sense, right? It's logical that people would reason about advertising personalization using the most visible way that they encounter it. This idea where they visit a product's web page and then later see ads for that product. Darren also got a lot of pictures diagramming for instance, Google, Facebook, Twitter. They also put companies, brand advertisers, and then computers and different items. But then all of them are connected with lines. Sort of each item in the network is connected to each other item. was an interconnected model where people are making these connections that it's not simply a linear process, that there's a lot of sharing of information between what call them nodes in the network. And a number of participants drew the picture I would have drawn had I found myself in one of his focus groups. They personified personalized advertising as a stalker. They

said it's following me everywhere. I could be at work, I could be on a trip, I could be on my cell phone, and the ads keep following me. And what this is, is people are picking up over time, they're learning about advertising personalization, and they're discerning what's typically referred to in the industry as cross-device tracking. Yeah, when it comes to advertising and surveillance, it's not crazy to be paranoid. In fact, it's crazy not to be paranoid. So I had a participant who talked about they felt as though their phone was listening to their conversations in daily life, that their smartphone was recording their conversations and then personalizing the ads they saw on their smartphone based off of their conversations. This might border on paranoia, but then at the same time, just this year in the U.S.,

the Federal Trade Commission cited a company who was engaging in a very similar practice, recording audio from people's smartphones, selling this data onto marketers, which was then used to personalize advertisements. So whether or not this person who was thinking that their smartphone was listening to them actually knew that it was, they perceived that it was. And that is what is important in some of this, is not how these technologies work, but in looking at people, how people perceive them to work. That actually matters. We know that people's perceptions of systems impacts how they use the system. This is why I use AdBlocker, because I perceive that all of my consumer technology is tracking and listening and monitoring me. But what's really dark about my mental model is how opaque it is. Like I said, I don't even know how well my AdBlockers even work.

My mental model doesn't really explain how or why or when I should see an online personalized ad. But pretty much every single one of the mental models Darren Stevenson mapped out in his research share this problem. They just don't work. I think the models that people are relying on, at least in this work that I found, tend to disappoint them or fail. So we would say the success or the usefulness of any model is how well it predicts the system. A lot of times the models that people appear to be relying on are not very good at predicting how ad personalization works or how ad tech works. In order to get to a mental model that works, a mental model that isn't opaque, we are going to need to spend some time talking about ad tech.

It's going to come up a number of times in this series on surveillance. But first, we're going to start with a close look at the sites where we mainly encounter personalized ads. Sites like the one that served me that Joy Division t-shirt. Sites like the New York Times, Vice, BuzzFeed, the Washington Post, and the Guardian. Media sites. News publishers. They're the ones putting the ads on the page in front of users.

They're the ones responsible for things like the number of advertisements on a web page and who their partners are. They're the ones who have the audiences. So they very much control the supply of the inventory for marketers. That is Joshua Schnell. He's a former tech journalist who's now a content marketing manager for an ad tech firm called Buy Sell Ads. He told me that if he found himself in one of Darren Stevenson's workshops, he would draw a giant sticky honeypot.

A honeypot designed and built by publishers. A lot of publishers online these days have built really complicated and confusing ad installations on their websites. There's a number of reasons for that. I mean, if you're giving them the benefit of the doubt, some small to medium publishers might not realize what they're doing. But the larger publishers, the tier one A-list publishers, are installing several types of ads from several different ad networks on their sites with two real goals in sight. The first is obviously monetization. The more ads on a site, the more likely someone's to click that, either intentionally or unintentionally. So that experience that they're building becomes sticky in that way. And then the second way is

to collect data, to collect browsing history data, track cookies, drop a cookie on your site, figure out where you're going and where you've come from, what your interests are. So then it becomes sticky in that way too. Using his own Facebook news feed as a starting point, Joshua ran a series of tests. He clicked on articles that friends had shared with him, articles from the Wall Street Journal, the New York Times, and the Washington Post.

I clicked on an ad every single time I visited an article from those publishers. I had zero intentions of clicking an ad every single time I visited an article from those publishers. I had zero intentions of clicking an ad. Not only did I have zero intentions of clicking an ad, the ads that were being delivered to me in those situations weren't even remotely advertisements that I was interested in. I mean, taking the creative aspect out of it, I had no use for any of them. And yet, in all three instances, I clicked on advertisements.

They do not make it easy to get by an advertisement without accidentally clicking on it. Now, all three of these sites have a lot of money and engineering power behind them. It shouldn't be this way, especially the Washington Post, which is owned by the same guy who owns Amazon. As Joshua Schnell points out, they kind of know how to make a good website.

You know, they have the infrastructure and the technical knowledge to deliver an optimized ad experience. But when you land on a Washington Post article, for instance, from any third party source, you know, like you're in Facebook and you see a Washington Post story go by and you click on that article and now you're shunted over to the Washington Post and then you start to scroll. And the next thing you know, you're on some other website that you had no intention of going to because the latency in the advertisements being loaded has then shifted the page at the perfect moment.

You know, when you're starting to scroll and you've clicked on an ad. Everyone's experienced that, right? It's so persistent that it's hard to argue that it's unintentional. Really, their main objective is to get you to click on the ads. Now, there is value for a publisher when they get us to click on an ad. They get paid on a per-click basis. They get paid even when we click accidentally. But our misclicks also generate data.

And it's this data that is truly valuable. It's become more about the collecting of data and selling the data and for an advertiser buying placements based on that data than it's been marketing. As I mentioned earlier, if Joshua had to draw a picture of how he believes online personalized advertising works, he would draw a giant honeypot. A honeypot is an idea that comes from Internet security. It's a server that you set up with open ports, a server that looks like it's a vital piece of your network.

But it's really a trap to spy on and catch intruders. So the whole premise is to actually capture them and get them stuck in that system. They think they found a payday, you know, everything's wide open, and they're going to poke around and do whatever they do, you know, try to install malware. But in reality, there's not much they can do, and there's not much on that server. For Joshua Schnell, a close look at the advertising infrastructure that media publishers have built reveals that the open web itself has become a worldwide ad honeypot that spies on and entraps users. media companies are on some level trying to trap people and encourage them through really horrible placements to click on ads so they're pushing us to buy stuff.

But what if they're also changing who we think we are? More on that after the break. Beginning with this new series here at The Theory of Everything, we're going to start running mid-roll ads, which I want to be awesome. I want them to sound awesome, and I want them to be awesome. So in order to do that, I have asked my friend and fellow co-conspirator Andrew Calloway for help. You might remember Andrew. He was the intern slash insta-surf that I had working on the show when we did our sharing economy series a while back. He's now moved to New York.

He's working at an advertising agency, and I thought, wow, he'd be perfect to help me. So thanks for coming out, Andrew. My pleasure, Benjamin. So I sent you an email. Our first campaign is a watch. It's a very big watch. I think it's a luxury watch. Well, can I take a look?'s a it's a very big watch um i think it's a luxury watch well uh can i can i

take a look you have it here um so i took it out of the box this morning since i knew i was gonna meet you and i went to the library and then i went to the coffee shop and somewhere in between i think i lost it hmm i mean it would have been nice if you had it for a little longer so you could, you know, have a more personal connection with it, you know? I was thinking we could just look it up online and talk about it. The thing is, it really helps if we can see how wearing the watch changes you, you know? But, okay, I mean, since we don't have it. It's Richard Mealy. Okay, Richard Mealy, like with the Y? No, with the E. wearing the watch changes you you know but uh but okay i mean since we don't it's richard mealy

okay richard me like with the y not with the e oh i see and it had a skull on it richard me okay i'll sort of it was i mean it was a very distinctive watch like it's beautiful had this big silver skull and this looks really impressive i mean i mean like punk rock. Maybe a... Wait. Is that... Oh. Is that the price? I lost a 500,000 fucking euro watch? Now that's a story.

We need to go find it. I may have lost my big skull watch, but you can find one of your own. Just go to richardmeleewatch.com and enter the code THEORY. Our visions of ourselves are perhaps a little more malleable than people like to think, because something as simple as a behaviorally targeted online ad can change how we see ourselves. The advertising industry does have proof that personalized ads or behaviorally targeted ads, that's their official name, work better than traditionally targeted ads aimed at broad groups of people, say women or folks living on the East Coast. But consumer

psychologists still aren't sure why they work. Rebecca Resick is an associate professor of marketing at the Fisher School of Business at Ohio State University. She's recently published a paper on some research that suggests behaviorally targeted ads not only push us to buy particular goods and services, they also influence how we perceive ourselves. A lot of the identity salience research in marketing has looked at this idea that an advertisement can serve as a cue that activates something that you already feel about yourself. So almost an activation of an internal sense of yourself. What we're doing is something a little bit different because when you get a behaviorally targeted ad, it's not necessarily cuing something internal. It's actually showing you that a marketer has made a characterization of you based on your past behavior.

So it's an external recognition that you are a particular type of person. And it's the power of being labeled by someone else who has observed your behavior. That's what really leads to this change in self-perceptions. That's what is resulting in behavioral change beyond just interested in ad, but it's actually really changing how you see yourself. Rebecca Resick and her colleagues ran a number of tests using Ohio State University students. They would bring them into the lab, have them browse the internet for a while, and then show them some ads. And then they'd explain to some of them that the ads were based on their browsing history.

The others were the control group. They were not told that the ads they saw were targeted. One of these tests involved an ad for an environmentally friendly speaker, something they could hook up to their smartphone or tablet. And after everyone saw the ads and believed that the test was over, the researchers sprang something else on the subjects. At the end of the study, we also told participants that as part of our behavioral lab, which is where our student participants come in to be part of studies, that we had a charity of the month that we were making donations to the Rainforest Alliance.

And we told participants there's a lottery that five of you are going to be selected to have an opportunity to win money for being in the study. But if you'd like, you can donate part of your money to the Rainforest Alliance. So for every participant we collected, how much of the potential money that they might win would they be willing to donate to the Rainforest Alliance. And our prediction was that because we expected behaviorally targeted ad in that condition to actually change how green consumers felt, so how green did these people feel as a result of getting a targeted ad, that should make them not only be more interested in buying the advertised product,

it should also make them engage in other behaviors consistent with that label as being a green consumer. So we expected they would be more willing to donate more of their money to the Rainforest Alliance. And that's exactly what we found. We found a difference in willingness to donate where the people in the behaviorally targeted ad condition were willing to donate more than those in the non-targeted ad condition. So that was really exciting to find because we found differences on self-perceptions, how you actually see yourself as a result of getting a behaviorally targeted ad, and differences on, you know, a measure unrelated to interest in purchasing the product, but related to this idea that people's self-perceptions had actually

changed. Now, what's really fascinating about this work is that these subjects have to be aware of the targeting. In other words, the group of students who were more likely to donate to the Rainforest Fund were the group of students who understood that they were being targeted and tracked, and thus labeled as being green. If a consumer in our studies got an ad for an environmentally friendly product and they were told that it was behaviorally targeted based on their past behavior, they actually felt more environmentally friendly conscious. They felt like greener consumers as a result of getting this ad because it's sort of like the marketer putting a label on them saying you are an environmentally

friendly person. It's only when you know that the ad's been targeted specifically to you that it carries information about how others see you. That's the part that changes your self-perceptions. It's important to note here, labeling isn't something unique to the internet. We get labeled every time we interact with someone. Bosses, co-workers, professors, bankers, police officers. But we resist and reject many of these labels because we believe that the person doing the labeling doesn't truly know us. But when we are aware that these labels are based on our online behavior, Rebecca Resick says we are more willing to accept them as totally valid. The reality is with these profiles that are created with the technology that we have to

track what happens online, we have an incredibly rich picture of what people do and what they are interested in on an ongoing basis, not just from recording them for an hour or even tracking them for a day. And so I think behavioral targeting is unique in that sense because it gives the labels that come from behaviorally targeted ads, yeah, legitimacy that I really don't think that we have in any other context because there are no other contexts that allow us to do that level of individual tracking. Some of these companies building profiles on us, Facebook, Google, at this point they have over a decade of our data, a decade of our searches, likes, and clicks. They may not know everything about us, but they definitely know who we are. And for the most part, these companies have worked to hide this from us.

Incomprehensible privacy policies, policies that are being constantly updated. But our relationship with our surveillors might be about to change. Already when we started this work, something called the Ad Choices logo or the Ad Choices icon was starting to appear on targeted ads online.

And what that is, it's a small blue triangle that appears in the upper right-hand corner of ads that have been behaviorally targeted. And it is an indication when you see it that that's what it means, that the ad has been delivered to you based on tracking your past behavior. So it is actually the case that as a consumer, when you get an ad online, you will know if it's been targeted based on your past behavior if it has that blue ad choices icon on there. But I think that disclosing that an ad has been targeted is actually it's a win-win. It's a win for companies because we see these positive effects of disclosure on purchase intentions.

But it's also a win for consumers because a lot of the concern around digital advertising, I think, is about consumers being concerned with how companies are using their information. So use of that icon, I think, is good for both companies and for consumers. There are over 200 participants in the Ad Choices program. All of the big names are there. Facebook, Yahoo, AOL, Microsoft, AT&T, Bloomberg, Comcast, Condé Nast, Dow Jones, Taboola, even Google lets its advertisers use the AdChoices logo as a third-party icon. But there are many, myself included, who view the AdChoices program as merely a cynical attempt to keep regulation at bay.

You see, there is no actual oversight or regulation when it comes to digital advertising and our data. To be honest, though, it's a lot easier for me to imagine the government paying for behaviorally targeted public service announcements than it is to imagine regulation. But like I said, I have a pretty dark mental model for this stuff. You have been listening to Benjamin Walker's Theory of Everything. This installment is called Honey Pot. Brendan Baker, Jesse Shapens, Mathilde Biot, and Julie Shapiro. Feel free to drop me a line if there's something surveillance-related you would like me to talk about.

You can get in touch through the website, toe.prx.org, or you can find me on Twitter, at Benjamin Walker. I've heard from a number of you out there already, and my mind has already been blown. The Theory of Everything is a proud member of Radiotopia, the world's best podcast network. Special thanks to our launch sponsors, The Knight Foundation, MailChimp, and listeners like you. Radiotopia from PRX