Short Wave - The Indicator: Destroying Personal Digital Data

Episode Date: April 6, 2022

Today, we present an episode of NPR's daily economics podcast, The Indicator from Planet Money. It's filled with one of our favorite topics: Data.Algorithms are the secret sauce for many tech platform...s. With user data, they can help a company tailor a subscriber's experience and make the product better. But what happens when the data that feeds an algorithm is obtained through less than legal means?We learn about the curious case of Kurbo, the weight loss app for kids that the feds say illegally collected data to generate that secret sauce.Email the show at shortwave@npr.org. See pcm.adswizz.com for information about our collection and use of personal data for sponsorship and to manage your podcast sponsorship preferences.NPR Privacy Policy

Transcript
Discussion (0)
Starting point is 00:00:00 You're listening to Shortwave from NPR. Hey there, shortwavers. Aaron Scott here. Do you ever have that experience where you're thinking about how much you want a donut? And then later, you're scrolling through your phone and you see an ad for donuts, as if your phone has read your mind, invaded your private space up there in your brain. Well, your phone is not exactly reading your mind, but the apps and online services on have algorithms, and these algorithms use your data to personalize ads and recommendations.
Starting point is 00:00:36 And sometimes they can be so good, it feels like telepathy. This episode from The Indicator is about how government regulators are starting to scrutinize this and why it's actually really difficult for companies to get rid of that data. All right, I'll leave you in the good hands of Waylon Wong and Adrian Ma from The Indicator. This is The Indicator from Planet Money. I'm Waylon Wong. And I'm Adrian Ma. The other night, John Verdi watched a classic film, a movie I am also obsessed with, the 1991 action film, Point Break. Now, is that like the 100th time you watched Point Break, 50th time?
Starting point is 00:01:20 It's well into the dozens. Oh, my God, Point Break. I just saw this movie like a couple of weeks ago for the first time. Oh, congrats. For those of you who haven't seen it, it stars Keanu Reeves as an FBI agent who goes undercover with a gang of surfers who also rob Banks, led by Patrick Swayze. Totally plausible. Plot. So many great scenes. John watched Point Break on HBO Max, and afterward, the service recommended some other early 90s crime thrillers. One movie called Rikashe, starring Denzel Washington, and another one called Rush about cops going undercover with drug dealers.
Starting point is 00:01:54 And this is an experience most of us are familiar with, right? I mean, we shop online or we stream music. And these apps or websites, they say, hey, you liked that thing. Maybe you'll also like this thing. These are examples of algorithms at work. And John is finally attuned to this plumbing of the internet because he's a digital privacy expert. Algorithms are computer programs that make decisions based on inputs from users. So all of us use algorithms every day, whether we know it or not. Algorithms are like the secret sauce of a lot of apps and online services. This potent blend of computer code and data generated by you and me, you know, regular folks doing stuff. online. And these secret sauces, they have shaped our online experiences into something more personal,
Starting point is 00:02:41 more targeted. But increasingly, they're attracting the scrutiny of government regulators. Scrutiny is one thing. But what would a company have to be accused of doing to get the feds to say, you have to pour your secret sauce down the drain? The Federal Trade Commission announced it had taken action against a company called Curbo. With Curbo, making smart food, choices is as easy as understanding traffic signals. So its parent company is WW International, formerly known as Weight Watchers. And Kerbo is a weight loss app that's aimed at children and families. The app collects personal information like weight, physical activity, what you eat. After a meal or snack, open up Kerbo's tracking feature. Enter the name of the food, color, a number of servings.
Starting point is 00:03:31 And then it provides personalized feedback based on that data. John Verdi is senior vice president of policy at the Future of Privacy Forum. It's a think tank in Washington, D.C. And he says algorithms were an important part of Curbo's service. Curbo, in this instance, used algorithms in order to better make recommendations about health, fitness, and weight loss. Those algorithms were not just based on the knowledge and the expertise of individuals of the company.
Starting point is 00:04:01 They were also based on an analysis of user data. In other words, the Curbo app wasn't just dispensing generic. advice like, eat more fruits and vegetables. It was offering personalized advice powered by algorithms built with data on customers' diets and activity. And what got Curbo in trouble with the FTC is that there is a law that spells out how data about children should be handled. It's called the Children's Online Privacy Protection Act, and it says that services that collect and use personal information from anyone under the age of 13 have to get permission from their parents. Now, the FTC alleges Kerbo marketed their app to kids as young as 8 years old and did not get their parents' permission.
Starting point is 00:04:42 On top of that, the FTC says the company actually encouraged kids under 13 to, you know, fudge their ages when they signed up. The FTC says the failure to get this permission means this data on users under 13 years old was actually illegally collected. I reached out to Kerbo for comment, and they said they dispute that their historical practices violated the law. Eventually, Kerbo and the FTC reached a settlement, which included a $1.5 million penalty, which the company says is not an admission of wrongdoing. But that's not the part of the settlement that caught people's attention. It's what else the government required Kerbo to do. The company is required to delete any data that was illegally collected from children under 13 and to destroy any algorithms built with that data.
Starting point is 00:05:30 Pour that secret sauce down the drain. This settlement is the third time in under three years the feds have required companies to destroy their algorithms because they say they were using ill-gotten data. The first time in 2019 involved a company called Cambridge Analytica. You remember that controversy? Yep. That centered on targeted political advertising on Facebook. And then in 2021, the FTC settled with the makers of a photo sharing app that the government said improperly used customer photos to develop facial recognition algorithms. John says the Curbo settlement confirms the FTC is serious about using algorithmic destruction as a deterrent.
Starting point is 00:06:09 This sort of remedy by the FTC can look really, really scary to companies. There's a good line in a James Bond movie that, you know, once has happenedstance twice as coincidence and three times as enemy action. And these three examples, they're just investigations that resulted in some high-profile settlements. John says the FTC is watching tech companies really closely. Let me be clear, vast majority of FTC investigations are non-public. So we see settlements like the Curbo settlement. And those really are the tip of the iceberg. There's a huge iceberg under the water of investigations.
Starting point is 00:06:43 Companies are quite serious in terms of not wanting to be on the bad side of the FTC. So the FTC has signaled that it means business. But making sure companies actually follow through with deleting their data and algorithms, that gets tricky. Deletion is just far more complicated than people realize. Kinesa Ahmed is a partner with Alita Consulting, a firm in San Francisco that advises clients on privacy, security, and data. It's not just a matter of like, oh, we automatically know which data fields need to be deleted. No. It's not a Control F, control F delete. Sadly no. Sadly no.
Starting point is 00:07:19 Kinesa says data can also get copied and sliced up and moved around different parts of the company. Like if the marketing team is using one data set over here and the customer service team is using a slightly different data set over there. In the case of a company like Curbo, their trove of data is going to have information the FTC says was illegally collected, mixed in with information that was collected in the right way. But it's all potentially mixed together and used in the company's algorithm. How do you separate legitimately sourced data from the illegitimately sourced data? It's really challenging to track that data lineage, and a lot of companies don't have the full resources to automatically track this. So it could get quite complicated behind the scenes at Curbo, which is why Kinesis firm advises their clients to be proactive about good data practices, map out exactly how data is being used
Starting point is 00:08:10 across a company, and be thoughtful about what data they do collect and hold on to. Because if the FTC keeps using deletion as a penalty, that could really upend how tech companies operate. Destroying algorithms could mean erasing the very thing that makes money for the company, or makes the experience of using one app better than its competitors. As for users of Curbo, their experience might also change in the wake of the FTC settlement, because these algorithms may have found their way into many parts of the app. John Verdi at the Future Privacy Forum says, depending on how widely those algorithms were used,
Starting point is 00:08:47 getting rid of them could mess up how the app works for a whole bunch of customers, not just the ones whose data was considered unlawfully collected. When you're looking at a circumstance like Curbo, there are lots of lawful users of this service, folks 13 and over, who may end up in a situation where a service that they enjoy and rely upon, and in some circumstances, paid for, ceases working because the algorithms themselves become unlawful. If more cases of algorithmic destruction pop up, that could lead to an internet that's a little less personal, a little less anticipatory of our every need and desire. at least until tech companies land on a new and improved secret sauce recipe. Or they catch the next big wave, like Patrick Suisse's character in point break? Viacondios, Adrienne, Viacondios. Viacondios.
Starting point is 00:09:39 Viacondios. I wouldn't have understood this reference two weeks ago. This episode was produced by Nikki Willett with help from Isaac Rodriguez. It was fact-checked by Corey Bridges. Our senior producer is VLA, our editor is Kate Cancanon, and the Indicators of Production of NPR.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.