Ep. 267: “No Price Tag for You!”

“What the Hack?” is DeleteMe’s true cybercrime podcast hosted by Beau Friedlander.

Derek: In some cases, there were five different prices for the same exact Wheat Thins or bananas or milk.

Beau: The only thing that changed was who was looking.

Derek: Instacart has been experimenting on people.

Beau: You’ve seen this on the news.

News Clip: The US Senate set its sights today on the ever-growing use of so-called surveillance pricing… that practice in which businesses use consumers’ personal data mined online to charge different prices to different people.

Josh Hawley: This is one of the biggest scams in American history… AI surveillance pricing is the unholy trinity of everything Americans hate, spying on people, ripping them off, and taking away jobs.

News Clip: People deserve to know that the price that they pay is not different from the customer who walked in just before them.

Beau: Welcome to the throwing of the second harpoon into the slippery and fearsome leviathan that is surveillance pricing. This week, how the price tag quietly died. Quietly? Anyway, it’s dead. Why a retailer might charge you more for peanut butter than it charges me, and the U.S. states that decided enough is enough. And yeah, high time because it’s costing people real money.

Beau: I’m Beau Friedlander, and this is “What the Hack?” The podcast that asks, in a world where your data is everywhere, how do you stay safe online? Last week I talked to Chris The producer, a YouTuber who investigated firsthand how surveillance pricing is affecting everyday life. Well, specifically his everyday life.

Chris: We’re kinda getting screwed right now and we gotta do something. I don’t know what, but like we gotta do something.

Beau: Now, if you haven’t listened to that episode, you should. If you haven’t checked out his video, I think it has like 1.2 million downloads. I’m impressed. But that’s not where the story of surveillance pricing ends.

Chris: Oh, we’re just getting started.

Beau: If you haven’t listened to that episode, it’s called “How Much is Surveillance Pricing Costing You?” It’s over $8,000, I mean, if you’re Chris. I think it’s more like 1,200 bucks for a lot of us. Anyway, you really should listen to it. Today, we’re gonna talk to Derek Kravitz, investigative journalist. He’s been nominated for three Pulitzer Prizes. And he’s a person who spends his days figuring out exactly which companies, and how those companies reach into your wallet, often without you ever noticing it. And, I mean, actually, his beat is the ones you don’t notice. He’s currently at Consumer Reports.

Derek: My official title is Deputy Editor of Special Projects at Consumer Reports.

Beau: Thank you so much for joining What the Hack.

Derek: Yeah. Thanks for having me.

Beau: I think not everybody gets nominated for a Pulitzer, and the reason you have is because you’re doing super important work protecting consumers. And the work that came to my attention at first was the investigative report that you did in conjunction with Groundwork and More Perfect Union.

Derek: Yeah.

News Clip: This is pretty stunning new research that could help explain why your grocery bill is so high. It shows that someone buying the exact same food as you, same time, same store, might be paying a different price as new technology changes how retailers are setting prices. Part of this small case study from Groundwork Collaborative and Consumer Reports, nearly 200 shoppers, four US cities, they all went on the app Instacart. They shopped for the exact same basket of foods at the same time. What they found, at one store in Seattle…

Beau: Now on the Instacart story that subsequently resulted in Instacart changing their practices, let’s start there. I think most people would be shocked to know that two people buying the same things on Instacart might leave with a totally different total.

Derek: We had hundreds of people across the country look at prices on Instacart at all of their favorite grocery stores, and look at the same items at the same time and price them out. And what they saw was, in some cases, there were five different prices for the same exact Wheat Thins or bananas or milk. And that’s because Instacart has been experimenting on people in terms of trying to gauge their price sensitivity for years and very out in the open. So we decided to test it.

Beau: You had 437 volunteers divided into four groups, right? And they were looking at 18 to 20 goods from a variety of retailers.

Grace: Yeah, so Consumer Reports investigated grocery sales on Instacart’s platform.

Beau: This is Grace Ghedye, also from Consumer Reports. I spoke to her last week too. She’s a policy expert focused on AI and, well, this.

Grace: Instacart partners with a whole bunch of major grocers across the country, and we tested grocery prices on several of them. So We got people together on Zooms, and we would have them look at the exact same store. Choosing a specific Safeway in a specific location in Seattle, Washington, or at a specific location in Ohio, for example, all at the same time, and then give them a list, a basket of goods to go through and take screenshots of every single one of those goods, hand them over to researchers who double-check what the secret shoppers say they found, our volunteers, our members and then do some data analysis and figure out, okay, how often are we seeing this pricing variation?

Beau: What were some of the findings in the Instacart study?

Grace: Nearly three-quarters of all the grocery items that we tested on Instacart showed different prices to different shoppers, again, shopping simultaneously at the same store. And some items actually carried five different price points at the same time. On average, we found the difference between the lowest and highest prices offered for the exact same item was about 13%. And tallying up how a family of four would shop for groceries on Instacart over the course of a year I think based on Instacart’s numbers, that kind of difference between the prices people are seeing could translate into a cost swing of about $1,200 a year for a family of four.

Beau: That’s significant. And did you collect any demographic information about the people participating that would give you some insight into, for instance, what their credit was like or what their employment was like, or why they might be getting targeted with higher or lower prices?

Grace: I think we collected some information and it just wasn’t a robust enough data set to establish really firm correlations. What Instacart said is that what we’d observed was the result of randomized price testing. Their claim was this is not based on people’s personal information. It is based on this randomized testing experiment we’re running where we’re assigning different prices to different people randomly.

Beau: We just wanna see what happens. You want butter, you want butter, you want butter. Let’s see who buys the cheapest butter. It’s ridiculous.

Grace: That is what they say what was happening behind the scenes. Interestingly enough, after that investigation came out, there was quite the outcry from public officials, lots of sitting members of Congress, US senators writing letters to the company, and they in fact decided to halt that pricing experimentation. But they did say that they weren’t halting it for– so people might still see different discounts for the same item.

Beau: Instacart is used a lot in rural areas and food deserts where people don’t have otherwise access to shopping. You know, getting especially a wide variety of products. And so does this include people on SNAP benefits or do they get a carve out here or was Instacart just targeting everybody?

Derek: Instacart really makes a big point of targeting SNAP benefit users. They market directly to SNAP holders and users. And they also, a large number of people that use Instacart are disabled or the elderly, having trouble navigating brick-and-mortar store locations. We spoke to several of them for our investigation. For them, it’s less a “want to” than a “have to,” right? They don’t really have many other options. So for them, this really does matter. And it matters for everyone, but for them, it’s especially acute.

Beau: So we’re talking about a difference of like maybe $10 or so on a checkout of several items. $10 For somebody who’s on SNAP benefits is no joke.

Derek: Right. It adds up over time. Like if it’s a basket of, say, $110 to $120, let’s say somewhere between 10% and in some cases we saw up to 23% difference. That’s real money, especially groceries, essential items over months or years. So that’s why it rang true for a lot of people and why there was such a reaction.

Beau: In case you’re not doing the math out there, it’s $1,200 price swing in some cases for people, and you can read about that on Consumer Reports. It’s pretty stunning.

David: Yeah. That’s a crazy part of this.

Beau: This is David Dayen, executive editor of The American Prospect, author, he’s written all over the place about really important topics. I spoke to him last week. Actually, with the exception of Zephyr Teachout when she was working for the state of New York, I believe he’s the first person to actually float the term surveillance pricing. He was talking about it before it became a common practice. So we’re gonna start with just a look at what we’re talking about here.

David: A good example was a situation with Staples. So, Staples has an online store and you can purchase products there. And somebody did a study that looked at a zip code-level analysis of what the prices were. And zip codes in poorer areas were being charged higher prices by Staples, and the reason was those people didn’t have a hardware store outside their door. So that was the correlation. Because you don’t have a place to get a stapler or to get a hammer, you had to go online to purchase that product. And Staples knew that, and they were gonna charge you more if you didn’t have that business in your community.

Beau: It’s juiced up supply and demand. It’s actually just putting your thumb on the scale of supply and demand.

David: And it’s a form of redlining, right? I mean, it’s like, okay, that product isn’t there in that community, so the only way that you can get it, the only recourse is through us, and we’re gonna charge you more for it. It’s price gouging on an individual and daily scale.

Beau: 100%. My daughter was living in a just newly gentrifying neighborhood in Brooklyn, which is to say it was a food desert. And I took her grocery shopping, and she wanted organic eggs.

David: Mm-hmm.

Beau: The same organic eggs where I live maybe five bucks were $20 in the food desert, because they knew if somebody wanted them there, they were gonna pay for them. And so Gary Shteyngart wrote this book called Super Sad True Love Story, and I know you know about it ’cause you referred to it in an article that you wrote. He envisioned this world with credit poles around the city like lampposts everywhere. And if you walked under them, it’ll say what your credit score is, so you could see whether or not you wanted to date that person or if they were worthy of you. I mean, we’re kinda there.

David: Yeah, only it’s the companies that have the credit pole. You’re constantly, when you walk through the door, whether a real brick and mortar door or a virtual door, you are triggering that credit pull, and they know everything about you, and they’re able to deal with that. And I think we also have to say here that we really are just scratching the surface of this. A lot of this has been created without the benefit or at least, the significant benefit of AI. And there are now consultants that are fully AI-based that are doing millions of tests, billions of tests on prices in a fashion that might not even necessarily need your personal information to know how to price things in ways that maximize profits for that company. And that’s the really scary stuff. There are these companies like Fetcher and others who are using these AI tools. Delta Airlines in an investor call a few months ago said that I think it was something like 20% of prices by the end of the year were going to be generated through AI. And what they meant by that was that these millions of data points were going to be used, and in a real-time context on every booking inquiry, to facilitate what the proper airfare should be a particular time. Now, this gets a little bit mind-bending because it’s sort of outside the level. It’s still surveillance pricing to an extent, but it’s using just really fine-tuning and sophisticating aggregate data to know that the person querying right now is going to want this particular travel at this particular price. And it’s a very strange way of thinking about it. But we might be in a situation where there’s enough outrage at the state or federal level to ban surveillance pricing. But we don’t get out of the woods because there’s this use of AI pricing that almost supersedes the way in which data is being used. So there’s a whole next frontier here. And another part of that is the use of AI agents to shop for you and that we’re starting to see this at a much wider level where you interact, or you give over your shopping list to an AI agent who then does shopping for you at a particular online store with the benefit of another agent, and maybe those agents are working for the same pricing consultant, or they’re working for the same company. I mean, there are a whole lot of other factors here that we have to think about, that have changed the shopping relationship in fundamental ways.

Beau: When we come back, how these other factors and surveillance pricing are changing the way we think about the price tag.

Fictitious Pricing

Beau: Can you talk to me about fictitious pricing?

Derek: Sure.

Beau: Here’s Derek Kravitz again.

Derek: So basically, fictitious pricing is the idea that you maybe artificially inflate a price. Say you have a set price at sixty-two dollars and you increase it to eighty-two dollars, and then you mark it down, you strike it through, and you bring it back down to that sixty-two dollar original price. That’s a fictitious price or a fake discount. The FTC, the Federal Trade Commission, has been to some degree regulating this for fifty years. But they stopped really enforcing it decades ago. And so it’s commonplace in retail. But there are state consumer protection laws, they call them UDAP, Unfair and Deceptive Pricing Practices, that typically police that and regulate that. But it differs. It’s a patchwork of state laws there. It’s fairly straightforward when you think about it, but in the era of algorithmic pricing, rapidly changing dynamic pricing, e-commerce apps, it is harder to police, right? Because to basically track that baseline price or that original price, you have to be constantly monitoring and looking. And regulators haven’t really kept pace or caught up to what’s going on in the marketplace. We did an investigation into Uber and Lyft, and we saw a lot of fictitious pricing, fake discounts there. About 12% of all advertised discounts on those platforms had a fictitious price attached to it.

Beau: 12 percent’s no joke. How is this different from supply and demand? How is this different from surge pricing?

Derek: So, supply-demand, dynamic pricing, been around for years, decades. When you think about hotel reservations, airline tickets, concert tickets, those are very traditional examples of dynamic pricing in the digital marketplace. What we’re talking about is a little bit more than that, right? It’s not simply supply and demand. It’s not the number of people looking at something and then the number of drivers out there for Uber and Lyft and time and estimated arrival time and all of that. It’s more than that. It’s hundreds or thousands of different variables all sort of working together at the same time in algorithmic black box and then spitting out an upfront price or a number at you. Now, the other variables that go into that, when we go to the companies and we share our data, which we do, every single investigation, we crunch all the numbers, we put it in a readable file, and we send it back to the companies and we say, “What say you?” And typically, the responses we get are, “It’s way too complicated for you to understand.” Here’s some of the things that we think are at play for base prices, which are supply, demand, weather, estimated arrival time, GPS signal, network latency…

Beau: So they just throw “you couldn’t possibly understand this complexity” at you.

Derek: Yes. And by the way, “all promotions and discounts on our platforms are personalized.” So just stepping back a little bit on that, they’re saying base prices we don’t personalize, but promotions and discounts we do. And we can use any number of variables: who you are, your billing zip code, where you’re going, your account history, your purchase history. We can use all of that to personalize a promotion or a discount to you. And if these companies are increasingly using promotions and discounts, and they’re giving this person fifteen percent and this person ten percent and not giving a promotion discount to this person, the net effective price is all different, right? And so those promotions and discounts are essentially the new number, the new price. And if you employ that in to such a frequency, to such a degree, it changes the entire pricing landscape. There’s no true price anymore. What Uber and Lyft will say is there’s no baseline price, ’cause each price is unique and different and special to that individual moment, and the next moment, seconds later, it’s different.

Beau: Is the price tag on an item now an obsolete thing?

Derek: Yeah. In the world of algorithmic prices, and that includes electronic price labels in store where you see the little digital price tags, it’s a different world. It can change in seconds. And so there is no true set price anymore. And that’s a problem for people because we’ve come to expect a level of consistency or at least transparency when it comes to a marked price.

Beau: Now, but I’ve heard tell, and I don’t know if it’s apocryphal, anecdotal, that it’s possible to walk through a store that has that kind of pricing where there’s an electronic tag on the shelf. And through some dark magic of the internet, probably AI-assisted, the store will know that a certain customer is nearby and change that price on the fly.

Derek: So, ESL is the acronym they use, electronic shelf label, and the companies that make them, Fusion Group and Handshower are the two that are the biggest. They say, “No, we don’t do that.” The devices that are near the price tag, you have to be millimeters apart, and it’s not reading that, right? But if there’s an electronic shelf label, and there’s a QR code on that label, and you scan the QR code, and it pulls up a promotion or discount, yes, in that particular case, that label, and by extension the company, knows that you yourself are interested in that item, right? And so, yes, when that happens, you are directly connected with that product, and for time immemorial, right? I mean, from then on, you are linked. And so that’s where people see that happening.

Beau: And that’s what gets me because, for instance, I don’t care if anyone knows this, but I don’t let my supermarket know this. They wanna track what I buy; I’ve always said no. But I buy…gosh, I must buy three bottles of peanut butter a week. I buy Teddy’s. It doesn’t have any sugar in it. It’s great. It’s organic. And that’s just a staple for me that I use all the time, which means that if I were to scan a QR code and they had my number, I’m gonna get a different price than you are, in theory.

Derek: Yeah, if you’re scanning a QR code and they’re gonna give you a 10% or an 8% or a 34% discount, yeah, you are getting a different price than the next person.

Beau: But if we know Beau has three bottles of peanut butter a week, we know he’s gonna pay whatever price that we’re charging.

Derek: Right. So then they don’t extend you that promotion discount, right? Your willingness to pay is higher and they’re gonna charge you full freight.

Beau: Now, I ask you this. In your work as a journalist, I don’t wanna know your opinion ’cause you’re not supposed to have one. Is that okay? Is that moral? Did the people that you spoke to who were finding out about this on the fly as part of your studies, did they think this was cool?

Derek: You bring up a good point. Journalists historically are not supposed to have opinions on these things, and the way I’m approaching this is we do a lot of testing, and we have to approach it like you would a scientific method. So our methodology has to– we have to run through it with a statistician, with subject matter experts. We take all of our findings and we share it with them, and they analyze it, and they come up with their own conclusions and interpretations. We send it back to the company. So that’s all to say, it’s not just me opining or thinking this. Every single expert that we go to in this space and every consumer we speak to, ’cause we do surveys too, and we recruit volunteers to do this. Every single volunteer we’ve talked to, and it’s self-selecting, sure. These are CR members or people that are interested in civic science journalism. But they are deeply offended by the concept and the practice of algorithmic pricing. And they don’t like when it’s not transparent and when it’s not upfront and when they don’t know what’s happening and they don’t know what factors are going into a particular price. And the subject matter experts say this is happening everywhere. You think about a retail industry that has an e-commerce or an app presence, and they are likely doing this. So it’s everywhere.

David: The sheer amount of surveillance pricing that is being used in the marketplace.

Beau: David Dayen again.

David: A company called Plexure is the company that runs the app for McDonald’s, and they do this not just for McDonald’s, but IKEA, and 7-Eleven, and a whole bunch of others. And I saw a slide presentation that they have given to investors and prospective clients, and it shows what kind of targeting they have, how they can predict what you’re going to buy next, what information that they use. And one of them says, “Relevance to key moments, i.e. payday.” So, the app learns when you are paid on a regular basis. And so on the day before you’re paid, maybe it offers you a Big Mac at $2.50, but the day after you’re paid, you have money in your pocket, so maybe it offers it to you at $3.50. And multiplying that makes a huge difference.

Beau: [inaudible] Thousands, thousands of dollars.

David: Oh, easy.

Beau: Let’s go to vulnerability though, because that’s the next step, right? If an airline knows that I have somebody in my family who’s sick or perhaps just died, maybe more to the point, they know I need to fly.

David: There was a celebrated case about just this factor, just what you’re talking about. Somebody went onto social media and said, “Hey, JetBlue, I’m just trying to get home for my aunt’s funeral, and the price keeps going up. Why is this happening? I’m just trying to get home.” And the response on social media from JetBlue was, “Hey, why don’t you go into incognito mode, when you purchase this product? Because then we won’t know that you’re getting home for a funeral.” So, inadvertently perhaps, the social media of JetBlue was informing this customer that, “We are looking at your email. We are looking at what you’re searching for, and we have figured out that you are, as a customer, looking to get home for a funeral. We’ve gotten rid of bereavement fares. Remember those? Those are way in the past. And now we’re going to gouge you.” They as much as admitted it.

Beau: It’s gouging. It is gouging. You can call it whatever you want, but it’s gouging. It’s $30 PPE in March of 2020. And that’s all it is. But it’s legal, or it’s not legal; it’s not getting caught because of this architecture that is built around… All the data privacy protections are out the door when you’re dealing with first-party data collection.

Derek: So the good thing about this topic area at the moment is that there’s a ton of different nonprofits and academics and journalists looking at it.

Beau: Derek Kravitz.

Derek: So with Delta, there’s a big investigation about Fetcher and their use of that in order to change prices. Bloomberg did a big project and series on that. We’ve also looked at Airlines Reporting Corporation, which is the ticket clearing house for all the airlines. And they collect a lot of information about the purchases you make. So, again, to your point, this is in a lot of different industries.

Beau: Now, my peanut butter thing is specific to me and it does have some meaning. Let’s go back to Uber and Lyft. A lot of people who are on SNAP benefits have jobs that are somewhat not close to their homes, and they also don’t have cars. So let’s say they work at Walmart or at Kroger or at one of these stores that is doing all this that was involved in the shenanigans which have now been clamped down on. So they’re using Uber to get to work. They’re using Lyft to get to work because they don’t have a choice. What did you find about the dynamic pricing or personalized pricing or surveillance pricing that was happening to people who were captive audiences on Uber and Lyft? Was there any noticeable effect?

Derek: When we looked at Uber and Lyft, we looked at 30 different routes across the country. Again, used volunteers looking at pricing things out at the same time. Rural, urban, different days of the week, different times of the day, across city, across county, across state, 17 different states. We also did in person. We priced rides in Portland, Oregon, and we all took the same rides together. Five different routes, all with drivers and riders matched together, if that makes sense.

Beau: It doesn’t. Can you actually explain that to me? So did you all sort of at one address and you got drivers though?

Derek: Yeah, picture an auditorium, a gymnasium, and you have 13 different riders and 13 drivers all together, and the riders are requesting their ride, and the driver is nearby, and they say, “Oh, we’ve matched.” And we go together, and we all take the same route the same time, and we price it out, and then we compare the receipts at the end of each trip.

Beau: What did you find?

Derek: We found the difference between the lowest price group, so more than two people getting a particular price, and the highest price groups was a little bit more than 42%. That can be significant, and again, we’re pricing the same trips with the same time. Now, of course, the companies will say there’s a lot of different factors at play, but what we found is not different than what other academic peer-reviewed teams have found when looking at this data. And these are two of the most scrutinized tech companies in the US subject of a lot of different oversight, house oversight, and FTC investigations at the moment. And what we found was no different, so not new, not novel, not any different than what others have found. And we work with a lot of academics on this because Consumer Reports has a brand, has a name to it, but we don’t have a monopoly on this information. So tons of great researchers working on this.

Beau: Are big retailers basically shadow data brokers now? Are they in the business of data, or are they in the business of selling stuff?

David: I think they’re in the business of both. Not only are they consumers of this data, but they’re producers and sellers of this data. Kroger, the big grocery chain, is known to sell the information that it has on customers to other parties. So there’s this data swapping going back and forth, the building of what is called a social graph, which is sort of a perfect set of information about you as an individual that’s being shopped around. And so yes, these companies are as much data companies as they are retailers.

Derek: We sometimes don’t think about how much we’re handing over to companies when we do business with them.

Beau: Derek Kravitz.

Derek: And at the end of the day, for the convenience and having this frictionless experience in the digital marketplace, we’re handing over a lot of information. And so when people ask, “Well, what do I do about it?” Well, aside from talking to legislators or regulators or others about changes you wanna see in the marketplace, also just if you can shop in person or go to retailers that don’t do this, maybe that are like a Trader Joe’s or others that do everyday low price and don’t have a loyalty rewards program that ingests a bunch of personal information about you.

Beau: So you heard it here first. If you don’t wanna be tracked, if you don’t want your data to be used in this absolute [BLEEP] show, go to a brick-and-mortar store. Use cash. You know, actually, remember, I hate to do this because I don’t like all these guys who are talking about cash all the time because I think their politics are usually very specific and they’re also buying gold, but you know what cash does? Cash makes it impossible for some of these data captures to happen.

Derek: I do think you’re starting to see… We did a thing about Kroger, which is the second largest grocery chain in the United States. We did an investigation with The Guardian and the Food Environment Reporting Network, and then we did a second investigation about their free loyalty program. And we got several volunteers’ loyalty dossiers, their profiles back from Kroger, and the inferences they made about them, how likely you are to take a cruise or be in the market for a new or used car or all number of different things. A lot of them were inaccurate; the inferences were wrong. And they use those inferences in order to serve you up promotions and discounts, and then they sell that data, in Kroger’s case, to 52 different third parties, including two tobacco companies, in order for them to go ahead and serve up promotions and discounts or offers. So that’s all to say, this information not only lives with companies, it then extends out in this sort of network analysis of different third parties. And so it’s one giant thing to navigate.

Beau: And it’s why you’re not getting ads that actually do you any good half the time, because it is wrong. Because if you’ve ever used an LLM and asked it a question, it can get it wrong, quite wrong. I’m guessing from your past at ProPublica and all the reporting you’ve done over the years, you’re familiar with the work of Yves-Alexandre de Montjoye on re-identification. There’s a lot of data scientists working on the problem of re-identification of anonymized data. And you would think it was for the right reasons, but often it’s being funded by people who are looking to know what’s what and have that God-eye view.

Derek: Yeah.

Beau: Doesn’t the privacy policy of these companies prohibit it, or are people not reading the policies and not understanding that it’s absolutely allowed?

Derek: We’ve done privacy policy checks on like automakers, to sort of see all the data that’s captured on your onboard navigation.

Beau: Like Mozilla did that report on how much cars were sucking in, and it was intense.

Derek: Privacy policies can be fairly limited even though they’re long and dense and technical at times. In a lot of states without state comprehensive privacy laws, the only requirement is that you provide a little bit of detail about the categories of information you collect, and generally speaking, the types of companies they might go to. But not specifics, not exact third-party clients or the exact examples of information collected about you. So that’s all to say, even with detailed privacy policies, there’s a lot of information you don’t know about how these companies operate and what they’re collecting.

Beau: We’re living in a shadowland of companies that are not asking for permission. They’re sort of ready to apologize after the fact.

Derek: You mentioned re-identification. The idea that you have for some of these companies a clean room, quote-unquote clean room, that is largely for liability reasons, but also for compliance, where you have someone maybe a team of people offshored sitting there getting these different disparate data sets and then joining them or merging them together and then serving up insights or reports back to third-party clients. There’s a business, there’s an industry that is interested in that, and that’s where you might see re-identification come into play, where you have, say, an anonymized data set, hashed names or identifiers, and then it’s re-identified and then re-anonymized again. It can get really fuzzy really quick. And when you speak to people that work in these clean rooms, they say it’s even messier than you might imagine.

Beau: That brings us back to something that you’ll hear a lot when you are interested in AI, which is the human in the loop. All good AI-assisted workflows will have a human in the loop to make sure that it’s not coughing up absurdities, that the work is correct, that it’s at least heading in the right direction. But you just said what you just said, and the clean room is messy. We’re living in a world where human in the loop can mean, yeah, great, you’re protected, but humans are also capable of bad behavior. And I guess that is kind of what Consumer Reports is in the business of finding out sometimes, where there’s compliance and where there’s stuff that needs to be called out and changed. Derek, we’ve hit a lot of different topics. People are probably super confused. Good. I’m glad. I want you to dig deeper. But what have we left out that people would wanna know in this area right now?

Derek: There’s a lot of academics who really dig deep into this and that we’re working with actually on a lot of different projects. Sorrell Friedler out of Penn and Haverford has the AI and Society Lab. Katie Wells out of AI Now Institute, formerly of Groundwork Collaborative, has her own initiative. Crystal Wilson and David Choffnes over at Northeastern have their own really impressive audit lab. They tested a bunch of our cars at our auto test track in Connecticut to see where all the data flows were going from vehicles in like a Faraday tent to isolate all the different traffic coming out of vehicles. That’s just all to say that there’s so much innovative, groundbreaking research being done in this space on the adversarial audit slash, nonprofit slash, consumer side. So again, if you’re interested in this, please check out those folks.

Beau: The consumer advocacy possible for someone interested in advocating for consumers right now is endless.

Derek: Yep.

Beau: Next week, we’re gonna take a look at something very specific, which is the curious case of rich people paying less than people who are not rich. Weird, right? Well, think about mortgages and you’re gonna be walking down the right trail to feeling really frustrated with this new thing called surveillance pricing. Now, there is some good news here.

Derek: The city of Seattle has introduced a new bill to ban surveillance pricing using your personal data to set different prices at grocery stores in that city. They just introduced it today. There was a rally this morning and then a hearing this afternoon. It’s actually one of several different places. Three laws have been passed in Connecticut, Maryland, and New Jersey banning algorithmic pricing at grocery stores or even other industries. Two bills are pending in California and New York. Another bill introduced in New York City, another bill now in Seattle.

Beau: People are paying attention for sure.

Grace: A retailer cannot change a price based on someone’s personal information. That’s surveillance pricing, and it’s prohibited.

Beau: But here’s the catch.

Grace: The data being collected and sold about each of us is extremely robust. Every interaction you have online, everything you hover over, everything you search is potentially being collected by some company. And either used by that company or packaged up and sold to another company or to a data broker who’s, again, packaging it up with other data about you and reselling it to other companies.

David: If we want to tamp down on this practice, do we go after the data itself and the data collection itself, or do we go after the ways in which that data is being used?

Beau: Next week, we’re gonna go further down the rabbit hole, so come back and join us.

News Clip: If we don’t intervene now and ban these sorts of price gouging and wage suppression right now, then I think it’s just gonna spread all over the economy. So it’s time for there to finally be a proposal in Congress.

News Clip: Kasar knows this is a problem in Washington, that technology may evolve faster than Congress can react to change the law to ban this practice. So this is something he’s trying to get on the radar to get people’s attention so they can move quickly before it’s too late.

Beau: Now it’s time for the Tinfoil Swan, our paranoid takeaway to keep you safe on and offline. And this week, we’re gonna ask you to do your own research. I want you to reach out to people at supermarkets or at the bodega or at the convenience store, wherever you happen to be, and say, “Hey, have you heard about surveillance pricing?” Don’t be creepy, but just ask, and then if someone shows any interest whatsoever, say, “Let’s do an experiment. Let’s both look up the price for…I don’t know, say a circular saw at Home Depot right now. See what the price is.” Or maybe a jar of peanut butter. That’s my thing, right? Whatever it is, pick a few things. Maybe it’s a rideshare from where you both are standing at that very moment to the same location, and see what you find. See if you guys are getting different prices and if you are, take a moment to sit there and soak it in with your new neighbor who didn’t know about this stuff, okay? Or maybe they do know about this stuff, but that makes them wanna talk about it too. I think this week’s job is to spread the word. So get out there and do some experimenting in your communities and have some good conversations. If nothing else, you’ll make a new friend, or at least there’ll be someone in your community who thinks you’re super weird. All right? Have fun with that, and we’ll see you next week to go further down this really, really odd data-centric rabbit hole. Take care. What the Hack is produced by a bunch of people. I’m one of them. Andrew Steven is another. And Sarah H helps out enormously from time to time. We’re available wherever you get your podcasts. If you are a regular listener to this show and you don’t use DeleteMe already, I want to tell you, you should. If you’re not and you want to, here’s what to do. Go to joindeleteme.com/wth. That’s joindeleteme.com/wth and get 20% off. I kid you not, 20%. 20% off. That’s joindeleteme.com/wth.

Learn More:

• Read our article that covers surveillance pricing and why personalized pricing is a lie.
• Hear the previous episode from What the Hack, which kicked off our surveillance pricing series.
• Listen to more from What the Hack.