1135: Sandra Matz | How Algorithms Read and Reveal the Real You

Coming up next on the Jordan Harbinger Show. Facebook in 2015 was actually accused of predicting whether teenagers on their platform were struggling from anxiety, depression, low self-esteem, and then they were selling them out to advertisers. So this is like someone at their most vulnerable state, not only are they suffering from anxiety, they're also teenagers, they're still figuring out the identity. So the moment that you tap into this vulnerability, the damage that you can do, I mean, it's very obvious. Welcome to the show, I'm Jordan Harbinger. On the Jordan Harbinger Show, we decode the stories, secrets and skills of the world's most fascinating people and turn their wisdom into practical advice that you can use to impact your own life and those around you. Our mission is to help you become a better informed, more critical thinker through long-form conversations with a variety of amazing folks, from spies to CEOs, athletes, authors, thinkers, and performers, even the occasional cold case homicide investigator, hostage negotiator, gold smuggler, or Russian spy. And if you're new to the show or you want to tell your friends about the show, I suggest our episode starter packs. These are collections of our favorite episodes on topics like persuasion and negotiation, psychology and geopolitics, disinformation, China, North Korea, crime and cults and more. That'll help new listeners get a taste of everything we do here on the show. Just visit jordanharbinger.com/start or search for us in your Spotify app to get started. Today, my friend Zandra Matsch shows us how companies steal our data, use our data to target us, not only to sell us things, but how they can essentially read our moods, almost read our minds, how AI and computers get to know us on such an intimate level, what they can do to predict things like depression, which unfortunately they monetize instead of helping us solve, and how we leave millions of digital footprints each day. This is a bit of a deep episode on the data we leave, and how it is used a lot of fascinating details in here. Algorithms can tell if someone is gay, 81% of the time, just using their face. That's pretty interesting. They might have better gay, dar than humans. How computers actually test their assumptions about us, and when having a low-phone battery means about you. And yes, we already know you're one of those. Now, here we go with Zandra Matsch. I think everybody knows that companies collect a lot of data on us, but I guess I didn't personally realize how granular they could get with psychological targeting. Can you explain what that is, first of all? Yeah, so psychological targeting is, in a way, taking all of the digital traces that you leave. So that ranges from what you post on social media to use wiping your credit card to your smartphone, capturing all of these very intimate things like where you go based on GBS, like you making taking calls, and then translating those footprints into meaningful psychological characteristics. So anywhere from your personality, your values, your political ideology, sexual orientation, so really painting a picture of the person behind a data. You know, I just occurred to me. I wonder if they collect data in the same way when I pay with my phone and the answer is, now I'm just telling Apple and the credit card company everything that I buy. And I know that they know this because it'll pop up like your Mx has been charged $47 for eating at this restaurant. So of course they are logging that, right? They're eventually used against me. Exactly, it must go through a bank. Yeah, it goes through a bank, but it's also, I'm telling Apple, which is what business do they have about what I'm buying? And the answer is they're in the business of data just like everybody else, probably. Yeah, and it's funny because Apple is one of these cases where they shut down the third party tracking, but they still collect all of the data. So at the end of the day, they benefit because now they holding the monopoly on the data that they capture. Yeah, I noticed whenever you install something new, it's like this app wants to track you. It's ask app not to track. It doesn't say we won't track you. It says, we're not going to let them track you. Don't worry though, we're still tracking you. Obviously, we're still tracking everything that you're doing. It's like location data. Somebody told me the other day, oh, I turned off location data so they don't know where I am at. And I'm like, they know they're just not sharing it with you and your friends and your mom. It's not like the FBI can't find you. Come on. And also, it depends on what you turn off, right? You might be able to turn off the GPS. You still need to be connected to a cell tower. Otherwise, your phone doesn't work. The fact that you turn off GPS doesn't mean that you're not trackable. So I was doing training for some journalists a couple of months ago in another country. And one of the things that they did was they put all of our phones in like a safe box. And I said, oh, I'll just turn it on airplane mode. And they were like, oh, sweet summer child. That's not going to do anything to stop intelligent services or whatever from finding or turning on the microphone to hear what we're doing. And I thought, oh, that makes sense. Especially if they want to get the location data, they can do that. If the phone is in a led box, then it just sort of vanishes. And I guess they had us put it in the box in one place and then we moved to another place so they just couldn't track us that way. But that was the only method they had. And you had to put your watch in there, everything. Oftentimes people take it even a step further, right? Well, I'm not using social media. Nobody can really track me across the internet. It's so short-sighted. Because obviously you use your credit card, your smartphone, and there's CCTV on pretty much every corner. So people will find you. It's hard to escape. I do want to address those kind of arguments in a little bit. But first, I want to scare the crap out of everybody a little bit more. The way that computers and AI get to know us at an intimate level is really hard to describe to people who have lived with some level of privacy their whole life. Growing up in the 80s in a medium-sized town, some people were all up in my business but they didn't really know that much about me. I could hide stuff from parents pretty easily. But you had the ability to do that. You, on the other hand, grew up in a really small town and everybody knew everything about you. This analogy was actually really good to illustrate the idea here. It is a tiny, tiny town. So 500 people, my parents keep reminding me that it's grown to 1,000. Oh, big time. It's 100% bigger. It doesn't make any difference because it still meant that everybody knew everything about me, right? Who I was dating, what I was doing on a weekend, which music I was into. And what Village Neighbors do best is then make inferences about who you are. They saw me running to the bus every morning. They probably figured out that I wasn't the most organized. And then it doesn't stop there. Village Neighbors are not just there to poke around in your life and to your psychology. They then try to meddle with your life. They are not really trying to figure out who you're dating. They want to influence who you're dating. And sometimes that's really helpful because they know you and you get this feeling of there's someone who truly understands me and when they have my best interest at heart, they're going to give the best advice that I can possibly get. But also oftentimes it felt a lot more manipulative behind my back without me necessarily having control or appreciating the support that I was getting in anyway. Oh, man, that would be really irritating because it's like your parents theoretically have your best interests in mind. But the woman who lives five doors down who babysat you twice when you were little, they don't really know you. They think they know you because they've made all these assumptions about you. But they could be wildly off base. And also, oh, I think she should date that boy. Why? Because they both have dark hair. What the hell does that have to do with anything? You're not exactly using your genius matchmaking skills. It's like, the boy she's dating now, I don't like him because one day he dropped something on my lawn by accident. Okay, 20 years ago he dropped something on your lawn by axis. Dog peed in your yard, so he's a bad kid now. It's so idiosyncratic. It's actually what I find fascinating about the shift to the online world is we're doing it a lot more systematically. So your neighbors, they had their own biases. They had their own perspective on the world. And they were filtering out of the data that came in through their own lens and their own incentives. Algorithms don't have the same incentives. They essentially do whatever you tell them to do. They optimize for the goal that you set for them. So the way that I've been thinking about essentially that we live in this digital village, where algorithms now replace our neighbor with essentially a digital neighbor who takes all of the data traces and makes the same predictions. And for me, the important part, and this is coming back to something you said, is it used to be the case that what happened in the village stayed in the village. So maybe it travels to the next town, but if I wanted to escape, I just moved to Browland, or I moved to a bigger place in New York. And that's it. But it's no longer true for the digital space. Once your data is out there, everybody has access to it. - It's funny, because people will ask me something like, wow, your life is really not that private because you have a podcast and you have this online brand. But the difference is I have been for the last 18 years, is thinking about what goes online, because I realize, oh, if I pose this, it's part of my brand. If I pose that, it's part of my brand. Not that everything I do is branding, but it's just, I'm consciously aware that like I can't start talking about, I don't know, buying a gun without people being like, okay, so now you're this political, this, or maybe you're this guy, or maybe you're that guy, or maybe you're having a midlife crisis, or whatever. Right, you're doing something like, whereas normal people, they post on Facebook and they go, I'm just telling my friends that I got a new rifle for hunting. But what you're doing is telling the entire world for all of eternity that you are now a certain type of person, or with 50% probability, or maybe this person. You forget you posted that last Monday, but the algorithm never forgets that you own a browning with a scope that you can use to shoot 100 yards. They will never forget that for the rest of your life. - The thing is that even if you don't explicitly put this out, there right, so a psychologist, we think of data in two categories. One is these explicit identity claims, which is like posting on social media. Right, that's you telling the world, here's a person who I am, here's how I want to see myself, and here's how I want other people to see me. No, that's a very intentional signal that you're sending. But then there's also all of the other traces. You don't have to post about you buying a gun. I could just by tracking your GPS records, figure out that you probably went to a shop where most people buy guns, and if you do this repeatedly, or you go to a shooting range, then my assumption is with like a high percentage accuracy that you're probably on a gun. And for me, that's in a way the very intrusive part that we oftentimes forget is that it's not just this explicit signaling. It's like all of these behavioral residue that we create without really intentionally thinking about it. - That's interesting. I hadn't even thought about the fact that you don't have to post it. It just comes out of the data exhaust or whatever they call it. - Yeah, I think it's exhaustive. - Where it's like location data, oh, we don't need that. And that's actually now we can tell where this guy goes to lunch every day. Oh, that's useful. We can advertise similar restaurants in the area. That all becomes useful as soon as they figure out what to do with the mess of data that they have, which is what AI is getting better at doing in these kinds of things. One thing I felt was super interesting in the book was that these clues or data can predict depression. And lots of people are depressed. I don't think that's a big surprise. There's like a million suicides a year. Is that global? I assume that's global because that's enormous. - I think it's global, yeah. It's an insane amount. - That's a huge number of people, no matter which way you slice it. 280 million people give or take are suffering from depression. I don't know how they figured that out or if it's an underestimate. But so if we can predict depression, I have to assume that these companies are doing everything of their power to help users who are at risk as soon as they get word. - Of course. What else would they be doing? - Not telling me a leather jacket and telling me it's gonna make me feel better. - It's not hypothetical. So Facebook in 2015 was actually accused of predicting whether teenagers on their platform were struggling from anxiety, depression, low self-esteem and then they were selling them out to appetizers. So this was a slight that was actually circulated. And you can imagine, right? So this is like someone at their most vulnerable state, not only are they suffering from anxiety, they're also teenagers. They're still figuring out the identity. Then there's also potentially beneficial use cases of that kind of tracking. - It seems like if you're not a complete psychopath, profiteer, I get it, advertising to people as profitable. It seems like if you find out the teenagers are depressed, one of the best ways to get every parent on your side is a tech company would be like, hey, imagine this headline. Facebook saves 10,000 teenage lives per year with depression tracking, notifying teachers or caring adults, parents, doctors, healthcare people, authorities, whatever it is, every year by tracking them online. Parents would be like, here's your new phone so that you can use Instagram because this is the only insight we've really have into your life and they're keeping you safe. As opposed to the current narrative, which is the complete opposite. 100% and I was recently talking to a mother, whose son attempted to commit suicide and it's traumatizing and for me, we actually have this opportunity there to catch it early because as you said, typically how it works is you enter a full-on depression, which, first of all, even for an adult, when you commit to it is really difficult to get out because that's the point when you're inward turning, you're not necessarily seeking out help and it's really hard to work your way out. What you would ideally do, and this is where the tracking actually comes in handy, is you catch it early and what you can do with your phone, for example, is just looking at your smartphone sensing data, maybe you're not leaving the house as much anymore as you used to, maybe there's much less physical activity, you're not taking as many calls anymore, so there's this deviation from your typical baseline, and again, it might be nothing, maybe you're just on vacation and you're having a great time. But it's like, I can send you this early warning signal that says maybe to people you nominate, maybe if I know that I have a history, which is oftentimes the case, suffering from depression, and I see this coming in the future, I could say I'm gonna nominate my spouse, and when this happens, I want you to notify me and I want you to notify my spouse. It's not a diagnosis, right? It doesn't replace a clinician coming in and going through all the questions, but it's at least one way of saying, just look into it. Maybe it's nothing, but to be on the safe side, why don't you try and get some support? And I think that's a total game changer. - It seems like this would be so easy to implement because obviously they can trigger advertising if you're depressed. They could easily trigger an email, a phone call, a notification in your spouse's use of that app, or whatever. It could literally just call the police so there could be some sort of central way to handle this. It surprises me that they haven't done it. Now, maybe that's naive because they're saying, oh, you'd say, that's not profitable. Why would they do that? It seems like it would long-term be such a massive PR win and create almost an incentive for parents to get these products in the hands of their children that it would be ROI positive. I don't know, what do I know? - I'm not sure if Facebook were to offer this tomorrow. I still am not sure if I would want that. So I would rather have a dedicated entity that's not Facebook. Facebook has like all of these market incentives, it's committed to, and you don't know what the leadership looks like tomorrow. There's this saying in the book, data is permanent and leadership isn't. So even if you had a CEO who kind of today, things were going to use it to help people, who knows the data is going to be out there and they could use it in very different ways tomorrow. So I'd much rather have a dedicated entity that doesn't even have to collect my data. There's now ways in which you can track locally on the phone and you just send your intelligence that says, if these patterns show up, you alert locally on the phone and I never have to even collect the data initially. No, that's not Facebook's business model. Facebook's business model is wrap as much data as you can and then you see how you can commercialize. So even if Facebook offered that, I personally would not trust them. - I don't think I trust them, but this is because of what they do with the data that they're already getting. If they years ago had said, "Hey, we can use this to target advertising." Hey, and by the way, this looks like you might be starting down the path of having an eating disorder. We're gonna notify somebody who you told us to notify or your school authorities. Then maybe we would have a totally different opinion of Facebook instead of, remember early it was like, wow, I can keep in touch with all my friends from school. This is incredible. I know what my aunt is up to. I only talked to her like once a year. Now I see her photos every week. This is such a glorious product. I don't see what could possibly go wrong with this. And then it was like two years, three years later, it was how the hell do they know so much about me and why are they trying to influence what I purchase or the elections that we have? It was such a rapid downfall. - Also didn't turn out to be connecting the world, so there's that. - Yeah, exactly. How many digital footprints do we leave each day? Can you take us through a typical morning? 'Cause a lot of people, again, they're gonna say, I don't post updates on Facebook and I don't let the phone track my location or I don't even take my mobile phone with me when I go to the gym, whatever it is. They're gonna have some sort of reason that this doesn't apply to them and the amount of data we create is insane. - It's just mind blowing. So to start with like average person generates about six gigabytes of data every hour. That's just already the sheer volume. And then when you break it down, it just really taps into all of these different parts of your daily routines in life. So if you wake up in the morning, probably what those people do is they grab their phone, which means that now just you unlocking the screen means that someone knows you probably woke up. The phone was stationary, maybe it was dark, so no ambient light. You didn't open it. The moment that you unlock it, someone knows that you're up. Then you're checking websites, you're sending messages so you kind of know exactly who's connected to whom, what you're interested in. My morning routine is essentially just going to the deli, getting a coffee, which means that I sweat my credit card. Again, someone knows that I've been out buying something in a specific location. If you have a fit bed or some kind of tracking device that counts your physical activity, also sees when you deviate from your routine. So if you have a typical routine and sometimes you don't do the same thing, people might have a sense that something is up. Even if you don't have a fit bed, take your phone with you on the walk or on the way to work. There's cameras and with facial recognition. Someone again knows what you do, where you go, and so on. So all of these traces, your car now has like sensors in it that track anything from your speed. Maybe you're going over the speed limit. Maybe you're not a great driver. You're going from A to B. So this idea that it's just social media that is really tracking us and coming back to Facebook is just so many data traces. It's really impressive when you think about it that way. To give people who aren't tech nerds in idea of how much data this is, this is a MacBook Pro storage every week or maybe two. So if you bought a laptop, it'd be full at the end of the week or halfway through the week, depending on how much storage you get in that thing. And companies are storing this because there's so many people. So are they really storing a terabyte of information on every storing 52 terabytes of data on me personally every week because where is all of that data? It's a great question. Some of it is you don't even need to store everything, but if you think about GPS records, oftentimes what you want is you want to extract the insights and you don't necessarily need to store the longitude latitude. What you want is, yeah, I kind of get the places that you visit. Maybe I can map it against Google and see what happened in these places when you were there. So a lot of the companies that extract insights that can then be used to tap into your psychology don't require the storage of the raw data. But then there's also other companies who have these massive servers. So still, I think even with that amount of data, like storage is so cheap that it pays off at the end of the day. And you might be deleting it at some point, but just the longer you can keep it, the more of these behavioral trajectories you can actually generate and create about people. Yeah, I suppose you don't really need every shred of data, right? If you have, say, someone's path that they walk every day to go get breakfast and then they go to the gym and then they come home, you can just say, it's basically this times a thousand. They don't have to get every time you cross the street. And you can even store, well, there was a deviation, right? So even knowing that, so if you know, here's the typical and now there's something that seems off, now you can trigger more data collection. So there's also ways in which you can say, we've seen that it's a repeat pattern. And we're going to just break the data collection at the point that we see that it deviates. So yeah, I would say this time seven, but on the eighth day, she went across the street real quick and came back. But then it was this again, 10 times a row. I mean, that's not that much data, right? It's not every step, every longitude and latitude repeated over and over again. And you're right, storage is super cheap. If you've got some sort of data farm or whatever in Iceland that's buried underneath the snow for cooling purposes, that just gets cheaper by the day. So it doesn't really matter that much. It's just a crazy amount of data given that it's one terabyte per week times hundreds of millions of people using these things. The amount is just bananas. I think by now there's this estimate that there's more points of data in the universe than stars. To me, it's just insane. It's quite impressive. How did they count that's already intriguing? Right, yeah. I suppose it's all just math. It's a way above my pay grade. Facebook knows us better than our closest friends and families. You wrote this in the book and that is not terribly surprising, a little scary because I'd like to think my family knows me pretty well, but Facebook does know me better. They'll show me some clothes where I'm like, oh, I have to buy that even though I know objectively that I'm going to be disappointed as soon as it arrives. And I find out that it was made by small children in Bangladesh or whatever. Yeah. That sounds like me. My husband keeps making fun of me for that. Total impulse buyer. Yeah. But yeah, if you think about it, like it's Facebook, but also Google, you type questions into Google that you don't feel comfortable asking your closest friends or sometimes even spouse. It's not so surprising to me that all of the digital traces that we create can paint this picture of who we are in a more accurate way than the people around us. I was doing a show yesterday with my producer. We are doing some work. And I was like, isn't there a different word for pedophiles who are attracted to people at different stages of puberty? And he's like, one, how do you know that? And I was like, oh, it was a bit from a comedian. Let me just look it up. And he goes, please tell me you didn't just Google different types of pedophiles. And I was like, oh, yeah, shoot. I did. I did. Yeah. I still remember what I was doing my PhD with is one guy who was doing research on porn websites. Oh, man. And I remember his seminar talk where he wants to open a website and just pulls up and it's all porn websites. So, yeah, you got to be careful on what you type into that search bar. And now, generally, if I write people ask these large language models, the most obscure, absurd questions that are super intimate. There's stuff that I've asked chat GPT where I feel the need to tell it. A friend of mine has asked the following question asking for a friend of a friend. Exactly. Because in 20 years, when it's like, here's what you searched this day in 2025, I'm like, I do not want it to pop up. I just don't want that to show up anywhere. And if it does, I want it to be like, you asked on behalf of a friend and I'm like, see, it wasn't me. Sure. Sure, pal. The chat GPT, it's amazing. Because people don't even ask questions. It's just statements. I think there's like some research. People just say random stuff to a chat GPT because they want to get it out of the system. They just need to tell it to someone. And they don't want to tell it to the people who they think might be judging them after. Well, we have an AI chatbot on our website where people can use it to search for things that are inside episodes. And we get an occasional report of what people have searched for just so we can make it more useful. And it's not my team that's behind it. It's an AI company. And they'll say, hey, FYI, this was a pretty funny month for searches. Here's our top five favorites. And some of the stuff that people are searching is they're trying to get the AI version of me to tell them how to commit a crime because they think maybe Jordan knows how to get away with this. It's interesting because obviously I'm not liable for that because it's not really me telling them how to do something. But it's a little scary that somebody can get inside my brain or something I would never tell them. And my AI version is sure, I'll tell you exactly how I would hide a dead body. And it's like, why are you letting my AI brain tell it this? To me, it's fascinating because if you don't feel comfortable asking another human being, that's one person who has to keep a secret. But you're asking a server. You're asking essentially open AI. Now your question sits there for all eternity on a server. It might be passed around. And I think that's something that people don't realize. Somehow this intimacy of the screen feels like it's just not a person on the other side. If anything, it's probably more intimate and more dangerous to ask a question there. - Yeah, it's not a person on the other side. It's theoretically infinite number of people that can look at this at any time with no context and zero ability to defend yourself in the moment because you've been dead for 20 years. Yeah, if you're lucky. How do companies like Mehta and other social media companies? How do they get to know our preferences? Is it just me telling them that I like something when somebody posts something because I don't always like the things that I like. Does that make sense? - Makes a lot of sense. You're just probably gonna be nice to people or cynical. - Great vacation. I couldn't care less, but you're gonna feel bad if I don't click like on this and we're friends. I'm supporting you. But no, you look like an ass. - Now you told them on the podcast. - This is 87 selfies of your vacation. These are completely uninteresting. You're clearly an artist here. Here's my like. - Well, yeah. - No, but you're absolutely right. And it's coming back to this distinction between identity claims, right? So you liking something, you posting something about your vacation, you're following a certain page that you want other people to see that you follow. And those are all these explicit identity claims, but then there's all of this other stuff that they capture all the way from how much time that you scroll through the specific ad that they're showing you, a specific piece of content. To here's like some of the more subtle nuances in the way that you use language. So coming back to this topic of depression, for example, it's not just you talking about symptoms and feeling down and maybe having these physical symptoms, even like the use of first-person pronouns is a sign of depression. And that's not something that you put out there intentionally, right? It's like it's hidden in some of the cues that you generate either by you posting or by you just browsing the website. Now Facebook goes a step further 'cause they also, first of all, buy third-party data. So they also buy extra data to know you even better. And they even have data on people who not using Facebook to contrast and see how they could potentially bring them in. So Facebook really goes far beyond you liking or not liking the vacations of your friends. - What was that about first-person pronouns indicating depression? What does that mean? Because I'm not sure I understood what you just said. - Yeah, it's actually one of my favorite examples in that space. It's essentially the use of first-person pronouns, which is I, me, myself. What we know is that is empirically related to depression, so emotional distress. And I remember when I first heard about this, it was like, I don't understand why this makes sense. I would have assumed it's narcissism. As you mentioned, right, if you talk about yourself and your vacation and what you've been up to, it's probably a self-focused and maybe your narcissist. But what we know is that it's a signal that you're currently very focused on why am I feeling so bad? How am I gonna get better? Am I ever going to get better? And because we have this inner monologue with ourselves and we can't constantly control it, that just creeps into the language. So people who are suffering from any type of emotional distress, they're just much more focused on the self and that leaks into the language. And again, in your post about vacation and everything that's going on, you don't explicitly, intentionally use first-person pronouns more when you're not feeling great. It's just something that leaks to the other side and leaks into your language. So in theory, that happens when we're talking to people in real life also, or is this mostly online communication? - Yeah, no, it's also talking in real life. It's a pretty substantial effect. I think 40 times more than when you're not feeling depressed or emotionally distressed. 40 times more. - It's 40%, it's not 40 times. That's unmistakable. - Yeah, it's like pretty substantial. - Wow, so in theory, even a smart TV or my phone, which is listening, even if I don't want it to be or my Amazon Alexa thing, that could tell me if I'm depressed just by hearing what I'm talking about in the house or overhearing a phone conversation. - Yeah, it's just like this passive listening into not just what you're saying but how you're saying it. - You mentioned in the book that within 300 likes of me liking things on photos or whatever. The platform knows me better than my spouse. 65 likes, it knows me better than my friends. - That doesn't seem like that much. - It's so little, right? So I remember when my colleagues published a study, I think the average note, and this is 10 years ago now, the average number of likes was 230. So back in the day, the computer was already better than everybody except for the spouse and you can very easily project into it here and now, where you have a lot more data, you have a lot more sophisticated models. So by now the computer is probably better than the spouse. And again, it sounds so intimate, but then if you think about the fact that a computer has access to the entirety of your digital life and some of the aspects that you're potentially trying to hide from other people you don't necessarily intend to signal, it's not a surprising. - Yeah, that's not super surprising. I guess if you think about it this way, and I'm sure this isn't a one-to-one kind of comparison, but if I had to remember 300 things that my spouse likes at the same time, I don't know if I could do that. That's a lot of different things. How do they measure that? - Is it like the newlywed game where you get asked questions and the computer just gets it right more than the spouse? - So in this case, it's actually you complete a questionnaire. So you tell us, here's how I think of myself when it comes to personality, and it's all kind of asking you about behavior. So how often do you enjoy socializing? To what extent are you making a mess of your environment? And then the spouse completes the same questionnaire. So on your behalf, so I think Jordan would answer strongly a creature to question, I make a mess of things. Not sure, hypothetically. - Wow, yeah, that's interesting. And the computer gets it right more than the spouse. Again, though, I think trying to remember 300 different things about your spouse at one time, it's a lot. It's just a superhuman feat of memory alone, let alone knowing someone that well. - And you also have a certain bias, like certain ways in which you want to see your spouse. So once you have a certain way of seeing them, the way that you integrate new information is just almost aligned with the perception that you have anyway. So it's much harder for humans to update just because it's in a way functional to stick with the impressions that we have. - Now it's time for us to hopefully monetize you. We'll be right back. - If you're wondering how I managed to book all these amazing thinkers, authors, creators every week, it's because of my network, the circle of people I know like and trust, and I'm teaching you how to build a network for yourself for free. Whether for personal or professional reasons, whether you're retired or just entering the job market, I have a course over at sixminutenetworking.com. The course requires no payment. I don't want your money. I don't even need to know who you are. It can be totally anonymous. It's all super easy down to earth, non-cringy. It's about your relationship, building skills, and just a few minutes a day, you can binge this course, practice a few things from it, it will change the way that you relate to others. And that's a whole idea. And many of the guests on the show subscribe and contribute to the course. So come on and join us. You'll be in Smart Company where you belong. You can find the course again for free at sixminutenetworking.com. All right, now back to Sandra Mott's. (upbeat music) How do computers test their assumptions about what they know about us? I know I'm anthropomorphizing computers a little bit, but whatever. How do they test and see if they're right because they have to do that somehow, right? - Uh-huh, that's actually how they learn. Right, so machine learning is called that way because they learn by trial and error. So the way that we train a model for example, to predict your personality from, say, Facebook likes, is we give it a lot of data where people completed a questionnaire, giving us answers of here's how I think about myself in terms of personality. And then they have access to all of the likes and they just played a trial and error game. So maybe if you liked the fan page of Lady Gaga, maybe that makes you more extroverted. Did I get it right or wrong? Got it, okay, I'm gonna update my belief of what Lady Gaga actually means. Same for the fan page of CNN. Maybe that makes you more conscientious and organized and reliable. So essentially you just throw a lot of data at them. And the beginning, they're just randomly guessing and over time they become a lot better 'cause you give them feedback. You tell them, yep, that was a good guess. No, this was a terrible guess. - I see, so it's just tons of trial and error. You have a really good analogy in the book about chick-sexing. Don't worry, still say for work folks. This is not the chick-sexing that I tried and vained to accomplish in my 20s. This is the kind of chick-sexing that happens on a farm. Yeah, tell us about this 'cause this is a very good metaphor. - I love it as an example just to explain machine learning. So there's like a profession that is essentially it's called chick-sexers, that's their name, which is amazing, right? I imagine you going to the conference and they ask for your title and you just say like I'm a chick-sexer. I mean, that's a life goal on your bucket list. But anyway, the point is that in hatcheries, you very quickly want to determine whether a chick is male or female. 'Cause for all the vegetarians out there, you're onto something, the males, they get shredded pretty much right away 'cause they don't produce eggs. So they mostly keep the females. And it's really difficult to tell whether a chick is like male or female 'cause they're generally sort of tiny, right? It's like a tiny baby chick. So what chick-sexers do, they essentially learn over time by having someone supervise and their actions. They just pick up a chick, they look at their little vent and say, "Oh, I think this one is a female." And then the supervisor says, "Correct." They put it in the one basket and they move onto the next. So they go through this trial and error game many, many times and it's not that it comes with an instruction manual. So it's not that they sit down for two week training course to see how to distinguish males and females. They just start with a 50% baseline where they might get it right or they might get it wrong. And then over time, they develop this intuition and they see these patterns that sometimes they might not even be able to explain. And I think of like machine learning the same way. Instead of looking at whether a chick is male or female, you might try to predict personality, you might try to predict gender, sexual orientation, political ideology. And the input is essentially people's digital footprints. And then over time, they just learn how to distinguish. Now the interesting part is actually that this is how it used to be. So we used to train these models specifically by supervising them. Large language models were never trained explicitly to make some of these predictions. They were just trying to predict the next word in a sentence and use the entire internet to do that. And they can still make similar predictions. So if I now give Chechi B.T. access to, let's say your social media posts or your credit card spending and I ask, what do you think is the personality profile of the person who generated the data? It's almost as accurate as these supervised models that we trained specifically for the task. So that's a totally different game because now anyone can use it. You don't even need that data. You don't need the training crosses. That's really interesting. And for someone like me with a thousand or 2,000 hours of audio content out there, that's just a bonus for these companies. They'll eventually be able, I mean, I'm sure they already are, ingesting all of that. There's a company right now that's making an AI clone of me, whatever that means. And they're using all of the data from the show. But they're only going to use about a thousand hours, but apparently that's more than enough. I'm curious to see it because it will supposedly talk like me, have the same reactions as me. It's basically as close as you can get to some sort of print out of my brain. And it's funny because it seems like such a waste to do it for me because the other people who have this are like Nobel Prize winners. I think I'm just low-hanging fruit because I have so much data out there. Interesting. And I was going to get there because I think to me, that's the next level, right? So far we're talking about we can make these predictions about your personality, but personality is helpful if you meet a stranger on the street and you know nothing about them. Knowing whether they're extroverted or introverted helps you understand how they might be thinking and how they might behave. But the people that you're really close with, like your spouse, for example, you don't think of them necessarily. She's the extroverted open minded. You kind of know them a lot more intimately. And I think that's what we're getting to with these digital doppelgangers. And then you can imagine once I have a second Jordan, I can ask, well, how do I best persuade you? Would you buy this? Would you like this, maybe not? So how do I get you to buy into this vision that I have or buy this product that I want to sell you? So I think it's just becoming more and more intimate. For us, I was talking to this New York Times column writer and she wrote this article of how she outsourced for a week her decisions to AI, digital twin version of herself. And she's like, it was pretty good. I got it right, like maybe 90% of the time. But it just turned me into this basic bitch where it was like always the same. And that I can totally see what it's trained to do is, yeah, what is the most likely thing that Jordan is going to go say next. But what makes you unique is, yeah, you don't always say the most likely thing. You still have like this unique element of like depending on who's on the other side, you'd come up with something new. So I think that's the world that I am worried about is I don't want to be boring as a digital doppelganger. I found it really interesting that the algorithms can tell gay men by their faces 81% of the time. I feel like a lot of humans are good at detecting this too, isn't it called gaydar, isn't that the whole idea? So this is like researched by one of my colleagues. And I remember that he stepped me down and he's like, well, computer can tell this like accuracy like of 80% I was like, I can do it too. And then I tried and I was barely battered in chance. So I think we oftentimes things that we know and maybe I'm just particularly bad, but at least in their research, they show that it's not just me. Most people are actually much worse than they think. - That actually makes perfect sense. I think the reason people think they're good at it is because when the really obvious examples come through, they're like, oh, I got that one. And it's like guy with the midriff shirt and walking a poodle in a sailor uniform. Okay, yeah, that guy's gay. Good job, Colombo. But like the quote unquote normal looking person and you find out their gay are like, oh, and you just like didn't know that about Tom, for example, it doesn't strike you as you would have gotten that wrong. You just weren't thinking about it at all. So yeah, it makes sense that the algorithm would be better, but 81% of the time is really incredible. - Yeah, and it's also the scary part is, I think we rely a lot on grooming signals, right? So the poodle or like the hairstyle, the shirt that you mentioned, this is all stuff that you can change. I think core of the research, and I should say that this is quite controversial. So I think there's a lot of people questioning it, but the fact that you can tell it based on facial features alone, both for like sexual orientation and also personality, that would really take the creepiness to the next level, 'cause you can leave your phone at home, you can't leave your face at home. So for me, this is really like one of these, if this is true, and there are potential reasons for it to be true, right? So we know, for example, that like hormones, they inflect testosterone, take testosterone, testosterone kind of influences what your face looks like, 'cause it makes you more masculine, but it also influences how aggressive you are. And so the fact that there is like these, for example, hormones that shape both is not completely delusional, I would say. And for me, that's really extremely creepy. - It is, although we're tiptoeing on the line of something that's gonna get me canceled, but whatever, there are plenty of aggressive, manly dudes who are also gay. So it's not just like, oh well, look at this guy's cheekbones. That's the whole like, humans can do this too. It's okay, the guy who looks like something out of a fashion show runway, like yeah, maybe that guy has a higher chance in your mind of being gay, you don't really know, but like the boxer whose photo is in there and has a big ol' beard. I don't know, I probably wouldn't be my first guest and I probably wouldn't ask, because I don't want to get punched in the face. - Actually, beard, I think beard was one of the, actually like, higher likelihood, which I didn't pick up on. - I wouldn't see that coming. Now, especially the trend is, oh you got to have a bushy, special forces beard and a trucker hat. And it's like, this is the pinnacle of manliness along with your tattoos leave. And now it's like, well, according to our algorithm, there's a higher likelihood that you actually like men. - For me, it's actually the interesting part of this entire prediction space is that there are certain signals that are somewhat universal and they are pretty stable over time. One of my favorite examples is that, if your phone is running out of battery, that makes you less organized. And that's probably gonna be true for the next 20 years. - For sure. So, okay, I love that because when I see somebody whose battery is at 30% or 40%, especially if it's before lunch, I'm just thinking, I don't know if I can work with you. But clearly your life is a mess. How did you wake up with a phone that's not charged? What is wrong with you? - That's what my husband tells me every single day. I think without battery sharing, more often than we're not. 'Cause he clearly is much more organized than me. But yeah, so that's a signal that's not gonna be different tomorrow. But then there's all these other signals that kind of cultural shifts, maybe something was a niche. Game of Thrones used to be like this. It's a fantasy and maybe it's just like these very nerdy open-minded people. Then it became like suddenly everybody was watching it. So, that's an interesting part for prediction. - That's true. When people told me it was a show about dragons set in the ancient, I was like, don't even finish your sentence. I've never watching this. And then enough people were like, you have to watch it, you have to watch it, you have to watch it. And I started watching it and I was like, oh, this is really good. And then I remember telling other people, I was like, do you watch Game of Thrones and they were like, I just can't with the dragons and the stuff. And I was like, no, no, no, no, I get it. I know exactly what you mean, but I'm telling you that it's really compelling. So fascinating how the window shifts. The battery thing makes me feel a lot better though because I was kind of like, am I psycho level of unreasonable because I judge people based on their battery status. At least it's not just me. - Maybe yes. - Geez. One great example of people prioritizing, I guess you'd say online clout over real life is the amount of people that die taking selfie photos. - I can't remember the exact numbers, but it's insane. I was like more than from shark attacks and other stuff. So again, it comes back to this question of what are we doing this for? And I think there's this fundamental need of humans to just talk about themselves. This is like why we see so many people posting on social media all the time. You mentioned that it's so annoying if your friend posts like a hundred pictures that you don't wanna see, but there's something that's inherently rewarding is research that I think is fascinating that shows that talking about yourself activates the same areas of the brain than having sex or taking drugs. So it just essentially gives you this dopamine boost. If that's the case, it's not so surprising that a lot of people are like, "I'm just gonna use this time." No, it was five minutes that I have and post something on social media. So people are willing to give up money in experiments to just be able to talk about themselves. - Good Lord, that really sheds a lot of light on why I started a podcast. I had no outlet for anything anywhere else. (laughing) - It's just funny, it makes me think, "Oh, okay, maybe if I'd done more drugs, "this show wouldn't exist." Certainly if I'd been able to have more sex that the show wouldn't exist, I'd still be a lawyer, so thank my bad luck for that. It's very interesting to me that this phenomenon where say you're out with friends and you order some food, I don't know if it's a conscious rule, but I basically give 10 seconds for them to take photos before I start eating, 'cause if I start eating right away, I ruin it. But I'm also not going to wait five minutes while they get the right angle and the right lighting and they rearrange the food on the table. That's like the end of my tolerance for this. And I won't travel with people who take more than a couple of selfies at each place. I get you want one or two, you went to a castle, it is impressive, it's really cool. But if you're trying different poses and different angles, I'm just leaving you behind, you can take an Uber. - The food thing I've never understood, you're never gonna go back to these pictures. Hey, you post them on social media, you're never gonna go back and say, "Oh, I wish I could find this picture of the pizza "that I had on 72nd Street." And we even know the moment that you take pictures, you actually reduce the likelihood that you're gonna remember that moment, because you're now outsourcing your memory to your phone. He's like, "Okay, this is on my phone, I took a picture "so I don't need to remember." And so there is even something that's taken away by us taking all these pictures all the time. - That's like how you get worse at math if you only use a calculator and you never try to add, subtract or divide on your own. Yeah, oh, that's interesting. I didn't realize that you would remember something less because you took a photo. It's almost counterintuitive. I get the logic behind it, your brain says, "I don't need to remember this, I have a photo." But you would think that focusing on it for an extra few seconds, trying to frame it in your phone, camera, looking at it longer, that would make you remember it more, but actually it's the opposite. - I think that's the old school. I think you're still coming from the generation where you had like 24 pictures and you're like, "Okay, that's something "that's worth photographing." Whereas now it's a click, click, click, click, click, doesn't no longer has something that is worth a while. - It used to be like a dollar or two by the time you got the film, took it to Walgreens to have it developed and waited however long a week or whatever, three days to have it developed. It ended up being, I don't know, a buck or two, it was expensive. You're right, now I've got a Sony over here that I got to film my kids and film events and stuff like that and it holds, I think when I put the memory card in and it showed at the top, it was like, you can film 16 hours of 4K video or 9,999 photos and I was like, "Oh, it just stops counting "because there's probably 30,000 photos available on here." - Back in the day, you're like, "Man, this is like maybe one of these moments "where I should take a second one just in case." Now you have 10 by default. The new cameras that are out there for sports photography, I think when you hold the button, first of all, it sounds like some kind of machine gun from Terminator, but it'll take, I want to say a hundred photos in a second or something like that at maximum speed, which is great if you're trying to catch a jumper at the peak of their jump at the Olympics and you want to get the perfect moment of them going over a bar or something like that. That's when it makes sense, but when you're taking a picture of your kids to produce in the potty at home, it's like-- - Before they're smashing their head into the wall. That's what you capture. - Yes, I want the wave and the skin of the forehead when it smashes into the drywall. - Yeah. - Just as a memory, the good old days. - Exactly, oh my gosh. Tell me about Facebook status updates and word clouds. I miss Facebook updates. I mean, maybe they still exist, but it used to be, and I know I sound old AF when I say this, AOL Instant Messenger. Do you remember that? Did you use that? - Yeah, I do. I secute. - All right, you had your away message and you're like, okay, I think of something creative and fun. Like, off days, you just pick a quote from an author you like or something, but on days where you think of something really funny, you put that in there and everyone's checking everyone else's away message all the time to see if there's a-- And it was like, if you could do that day in and day out and make it fun, people were like, this guy's smart. In Facebook, the original Facebook status updates where you just typed in the box, what you were doing, that was like almost like a status game for good writers in college. And it was also informative. I think like right now, we're just posting, first of all, anything and like pictures of food and so it's lost its appeal. And it's not just Facebook updates, right? You can think of Facebook status updates as the same of like you posting on Twitter, even in the way Instagram pictures that we take in the way tell the same story, right? Can write about you going on a vacation or you can post a picture of you on vacation. There's a lot that we can learn. So we already talked about emotional distress, depression, all of the personality traits. And some of them are really obvious. Oftentimes when people talk about machine learning, AI, it's just like magic and a black box and we don't really know what it's doing. If you talk a lot about going out and parties and weekends, you're probably more extroverted than the person who talks about sitting at home, reading, gardening and interested in fantasy novels. So those are the obvious ones. Sometimes there's the ones that are a little bit less obvious and maybe more interesting for psychologists. For example, income. This is like one of the topics that I study is can we predict someone's income, someone's social, economic, standing based on what they post? And again, you see the obvious ones. Like the rich people post about luxury vacations and brands. Yeah, that makes sense. But you also see that people who have lower social economic status or lower levels of income, they are first of all much more focused on the present and they also much more focused on the self. And it's not that they're again, like these narcissists that just only can focus on the here now. It's just freaking them hard to think about the future and anything other than how do you make your ends meet if you don't have that much money. So there's these subtle cues that we can parse out when we look at what they talk about that are actually interesting beyond just prediction. How the rich and poor talk online is actually quite fascinating. The idea that people have lower socio-economic status or people who are really having trouble making ends meet, can we just say poor if you can't make ends meet, you're not doing so with it? I mean, I actually feel like that if we use labels, labels matter and I know why people don't like them, but it's most of the time I think you don't like them because they make them feel uncomfortable. No, you should feel uncomfortable because there's people who are poor and it's just a freaking hard life to live. - It's tough and I never thought about that because of course if I saw somebody who only talked about themselves and things they were doing that day, it would seem to me that they were not thinking long-term because of some character defect or they're not smart enough or something like that. But now of course it makes total sense that if you can't think far enough in advance because you're just trying to literally feed your kids or you don't have gas to get to work and you're that poor, it's not necessarily a character defect or you having screwed up your life. Rich people, is it really that obvious that they just talk about luxury brands and vacations or are there some more subtle cues that out people is high socioeconomic status? 'Cause I can't name one single time where I've been like, just getting back from my business class flight to Turkey and staying at a five-star hotel, here's my dinner, I just don't do that. - Yeah, some of them are more subtle, right? It's oftentimes the opposite. So if poor people talk about the present, you might be like more future focused, say it's always a contrast the way that these models work, even the fact that you talk about going to the seashells or like an exotic place, just means that you don't have to be bragging about going to the five-star hotel on your next location, the fact that you can afford to fly outside of the country which most people haven't done in a lifetime, that alone is an indication that you're doing pretty well. - That is a good point. I hadn't even thought about that necessarily. There's still a stat that something like less than half of Americans have a passport or something like that. Yeah, you don't have to say, I'm going to a five-star hotel in the seashells. You just have to say, oh, immigration is so slow in, I don't know, India. Okay, well, you went to India, even though you're complaining about something. - And probably if you're complaining about that, just give us an extra boost in socioeconomic status. - Good point, yeah, I suppose if you're just excited that it's the first time you've ever left the country. You're not complaining about immigration status. You're like, I can't wait to eat after I get out of this six-hour line. You're just excited to be there. I thought it was quite fascinating about how these algorithms can tell if you're extroverted or introverted. You mentioned based on likes, if you like fantasy novels or if you like going to music festivals, that makes sense. There was a theory in the book or hypothesis in the book that attractive people become more extroverted and outgoing because of the attention they receive as kids. That makes a lot of sense. Or is that just pure speculation? - That's a real research finding. And it comes back to what we talked about earlier with face signaling potentially parts of your identity on a psychological level. So we talked about testosterone kind of being related to aggression. This idea that like your environment responds to you in a certain way, right? If you're kind of this beautiful kid, perfectly symmetric face, blue eyes and constantly smiling, people around you are probably gonna be a lot more kind of appreciative and they're gonna talk to you and they're gonna approach you a lot more often. And the fact that those kids then grow up to be somewhat more social and extroverted and craving the social affirmation and social stimulation is not super surprising. So it's like one of these ways in which actually who we are interacts with like our environment and that in turn again influences who we are. - This makes so much sense. And it seems like it might be something that you could encourage in kids regardless of how attractive they are just by interacting with them a lot, putting them in environments where they are interacting with other people, adults and kids. You're right, there's still that spontaneous element. My daughter, she's three. She loves to sing and dance and she'll be like, turn your chair around the show's gonna start. And I'm just like, where did you learn this crazy extroverted behavior? But then she's in music class and then the teachers paying attention to her and the other kids are paying attention to her. So it is a reinforcing cycle. My son who looks exactly like me, so he's very cute. He has it in sort of an almost like a negative way where he's like whenever I do bad things, people pay attention to me at school or otherwise. And I'm just like, oh no, this is not the reinforcing. - That's you were hoping for. - Yeah, this is not what we want. We want him to be reinforced. He's good, but he is not shy at all. It's crazy. He talks to the cops when they're here. He just has no fear at all. The interesting part is also, I think the way that we oftentimes think of personality is like, it's the static like you either extroverted or you're introverted, but it's actually a lot more dynamic than I think even personality psychologists assumed a couple of years ago. So it's not just that you can develop over the lifespan. So most of us become nicer, less neurotic. So there's like these trends that we see when people get older, but we also kind of very much fluctuate across situations. So like your son, depending on what the feedback is, it might be kind of more reserved or more extroverted. So I think there's also something that when we interact with our kids and I just add a kid, so he's like one year old, I just constantly think about how do I expose him to these different situations where sometimes I tell him, like, look, it's totally okay to be quiet and sit in the corner and kind of just think for yourself for a second. But then also I want him to have these other situations where he can be a lot more outgoing. So I almost think of it as like this repertoire where you have a certain tendency, right? There's a pretty substantial genetic component to personality, but then there's also you being able to adjust to different contexts. And I think that's something that we can teach kids and even tell them, look, if you behave differently across situations, that doesn't make you hypocritical, that can still be like the authentic version of yourself. It just means that you're adjusting to whoever is on the other side or what the context requires. - You mentioned though facial recognition can take faces from photos or videos taken by other people or just the CCTV that's present in whatever stadium you're in or on a straight corner if you live in China. So it doesn't really matter if you don't use social media. You're still a part of this surveillance, capitalism system or whatever we want to call it. - Yeah, absolutely. And for me, really intrusive part is that it's not just the ability to make inferences about who you are. Right, you mentioned China. The reason for why the Chinese social scoring system is creepy anyway is that it also influences what you can do and what you cannot do. So it doesn't stop it. I want to try and understand who you are. I'm also going to influence the path that your life can take. Maybe the choices that you're making. So in China, if the government predicts based on your data that you might have a higher likelihood of voicing the scent or protesting, you're not allowed into Beijing. Okay, I teach this class on the ethics of data, but that's what's happening in China. What do you think here kind of companies decide whether you might get a loan or not, whether you might get credit or not, what your insurance premium is or not? It's very similar. We try to understand how you might behave and then we shift the offerings that we have. We might try to sell you something that you don't need. So I think this notion that it's not just about privacy, it's really about the second step of people then interfering with your ability to make your own choices. For me, that part is almost creepier. It is creepy and there's not much we can do about it because if an AI is making a decision to give someone a loan or not, and by the way, you know damn well, it's gonna be like, this person's battery was 25% at lunchtime when they applied. We're not giving them a loan. They're totally irresponsible. It's not gonna say that's the reason. It's gonna say, oh, based on 20,000 factors that weighed a little bit to the left on whatever line, you're just short of making it. We're not gonna be able to weigh all 10,000 of those factors. The fact that you applied and didn't finish the application all at once and your battery status at the time and your location kept changing and like the fact that your jobs have changed so there's gonna be 10,000 of those. It's not gonna be like, we didn't give you a loan because you're brown. That's gonna be an obviously not okay thing, but when it's 10, 20, 30,000 different little factors and they don't interrogate the AI as to why they just blindly accept it because it's accurate 99% of the time, that's where we start to run into these problems. I would imagine. And they might all be related to some of the protected categories, right? If we know that some of the behaviors that we show are related to you having low associate economic status or to your ethnicity or to your sexual orientation, then you don't need to capture that category 'cause it's like somewhere embedded in the traces that you leave. So some extent, I think on the global level, when we try and understand what are these models doing and are they potentially discriminating against people? I still think that there's something that we can actually do to probe. Oftentimes people say, well, we don't know what the models are doing 'cause it's like these complicated neural nets and we just can't open a black box. It can still look at the output. If you're thinking about are we gonna give people a low or not and you just see that none of the women are getting any loans and none of the women are getting hired into technical roles, maybe then that's something that the model is picking up on, right? So even if you don't fully understand what it's doing, you can always look at the predictions and see is there anything that we see among the categories or the social demographics that we wanna protect that seems to be often terms of how often we do the thumbs up that the person gets the loan or gets the job? You untrustworthy, good for nothing, deadbeat. We'll be right back. If you like this episode of the show, I invite you to do what other smart and considerate listeners do, which is take a moment and support our amazing sponsors who make this show possible. All of the deals, discount codes and ways to support the show are searchable and clickable over at jordanharbinger.com/deals. You can always surface codes using the AI chatbot on the website as well. And if you can't remember the code, you're not sure if there is a code. Go ahead and email us at jordanharbinger.com. We are more than happy to surface that code for you. Is that important that you support those who support the show? Now for the rest of my conversation with zandra moths. I also found it shocking how easy it is to identify somebody personally based on what was it three credit card transactions? It seems like if it's that easy, you could also extrapolate a lot of info about people from those transactions once you identify them. So if you find me based on three transactions, some software I bought, a haircut and where I lunch, now you've got a zillion other transactions you can identify me with that are like, here's every bit of clothing he bought. Forget Facebook status. Where I spend my money is at least as identifying an intimate as that data. - Yeah, totally. And it identifies your different levels, right? So the example that you gave like the three data points, that's coming from this notion of even if we anonymize data, but even if like I got all of the credit card spending from everybody in Manhattan and we say, but it's anonymized because we're not using any names, we're not using data of birth, we're not using an address, because your spending signature is so unique, but almost like a fingerprint that it's very easy. And if I know three things about you, I can just easily identify you in there. And then you absolutely write, it's like if you think of identity at the next level, it's not just that I know while it's Jordan. Now I can also make inferences again about maybe you're like the impulsive person because you're constantly paying late fees and maybe you're not the most organized one. It's again something that might or might not show up in my own spending record. So that's also like one of these things where like oftentimes people say, well, your online selves, they're so curated, right? And if you want it to be like a more organized and reliable person online, you can do this because you just control. Yeah, that's true for some of them, but my phone's still running out of battery and I'm still paying these late fees. And if I wanted to be someone completely different across all of my different kind of digital traces, I would probably actually become that person at some point if I was changing my lifestyle entirely. Those people are just looking at the photo where it's, look, I just woke up and I'm in full makeup and I'm in my yoga gear sponsored by Aloe. That's what people mean by curated, but you can't fake the rest of it. The fact that your batteries low, your late fees are half your credit card transaction. You have a massive balance for months to month that keeps running because you can't pay it off. That stuff you can't really hide. You can put on a brave face, a shellacking of a near over what it is, but you can't hide from the company. They know you're full of crap. And I have to admit, I pat myself on the back a little bit when you said in the book, the person who buys gym equipment and then donate to charity is an example if somebody has their shit together. I looked at my credit card statement and I was like, what have I done? Okay, I spent 500 bucks on gym equipment. Oh, there's my amnesty international donation. And I was like, I'm a good person according to the data. Science doesn't lie. - Like a personality psychologist would say, there's no good or bad traits. There's just some that I was more socially desirable, right? Take it to the extreme. If you were like super extremely organized, you're turning into my husband who is super sweet, also board a line OCD. We just moved and there's a gazillion boxes in the apartment. It's just like everything is completely disorganized. And all I want is to be able to walk from the bathroom to the bedroom and I opened the drawer of the cutlery and it's perfectly meticulously organized. So I'm sure he spent two hours sorting the cutlery where there was still like 100,000 boxes in the apartment. So where I might be going through the boxes a little bit more quickly and maybe a little bit less thoroughly, but maybe a bit more efficiently. So no inherently good or bad traits. - That's really funny. You're like, we don't have underwear, but all of our cereals are alphabetized in the cabinet. - Oh my God. - Yeah. This is probably your doing. - Yeah, yeah, that's really funny. Moran is a really interesting, your husband is a really interesting character. By the way, he was on episode 265 of this show. And you had your first impression of him, which was that accurate? 'Cause you went out on a date or something. How much of your predictions of him initially turned out to be right later on? - Yeah, very accurate. So I met him. We were actually both giving a talk at the conference for digital happiness, but he showed up late. I was about to go on stage and then the organizer comes and says, "Hey, the person was supposed to speak after you. He's not here yet. We've no idea why he is. We can't reach him. Could you just take the entire hour and like, fine?" And then midway through, he shows up and they usher me off the stage fast forward. It doesn't take that long for me to realize that he's smart and taught. So we go out after the session and we actually end up in his place. And he kind of has these huge bookshelves and they're perfectly sorted by, here's the topic, here's the height of the books all perfectly aligned. Cutlery is perfectly sorted. I remember trying to put down my glass on the table and he freaked out to put a coast around him. It was like an intellectually curious and somewhat borderline OCD and late. And that was still spot on today. - That sounds about right. I mean, he's French and he's Jewish. So you're lucky to miss your wedding, actually. - Yes. And also, you ended up at his place, so after the talk, location data says, "You're like a little bit easy there, Sandra." - And that's actually, it's funny because that was a lot of the inspiration for the work on digital footprints from the physical space. So all of this work on, if you snoop around the bedroom or the office of a stranger and you just pick up on all of these cues and some of them, the same way that we post on social media are curated. You have a poster out there and certain books on the shelf that you want other people to see. But then a lot of them are also like very subtle. What is in your bin? Are your glasses sorted in the way? Do they have watermarks? So I think a lot of the work that we've been doing in the digital space was actually inspired by the physical space and the way that we make these inferences about strangers all the time as humans. - I remember vaguely in the '90s, when I started dating, there was a cliche that a woman would come into your house and look in your medicine cabinet to see what sort of drugs you had in there. That actually was the original snooping around thing. Before digital stuff existed 'cause you would open that cabinet and you'd be like, "Oh, okay, here's the real stuff that they're not gonna tell me for months." - What would I open to find? - That's a good point because this is sort of before all the personality pills and aterol and everything. So I don't know, are you looking to see if they're diabetic? Back then, what would have mattered? I don't know. Yeah, but it was a thing for sure. - Hemorrhoid. - Hemorrhoid. - Who doesn't have a couple of tucks laying around? I remember I was interviewing a very important, very distinguished man. He's like, "Can you come to my hotel and do it?" So we did and we set up in his hotel and I was like, "I need to use the restroom real quick and I remember going into the restroom and he had all this hemorrhoid stuff laid out on the counter." And I was like, "Good to know that the CEO of this giant, massive multinational company has serious hemorrhoids, poor guy." - Humbling experience. - 'Cause you can't be like, "Hold on, I'll be right back." Nope, he's just gonna find out about my hemorrhoids. - Well, I'm sure he played it cool. - He did, he played it very cool. But I guess at that point, when you're a billionaire, it's like, "Yeah, I got hemorrhoids. What are you gonna do about it, podcaster?" - Whatever. - You can leave if you have a problem with that. - What is the limit of psychological targeting? Can I change someone's mind entirely or do I just influence people who are straddling the fence? - Yeah, it's a great question. 'Cause I think if you look to the media, it's like totally black and white, right? It's like either it's this warfare tool and it's like changing your mind and it's changing your core identity. That's probably not the case. So I always think about it. If that's something that you couldn't do in an offline world, if you think about your hardcore die-hard Republican uncle and by having long conversations with him, you can't convince him to take on a certain view on the world. You'll probably also not able to do this online with algorithms even though you can target them repeatedly and maybe you can send them down a rabbit hole, changing someone's core identity takes a lot more than just like a couple of ads and maybe even repeatedly. But the thing is that it usually doesn't even need that. Oftentimes it's like our choices are kind of small ones. We are not even aware of what's the serial that you choose. What are you deciding to wear today? What are the news that you're trying to read and what is that take you in terms of how you think about the world? So oftentimes when I think about influencing behavior, it's like these small changes and the same way that we do this in an offline world, right? Coming back to kids, humans are born to do this. Kids know exactly how they talk to their mom to get the candy as opposed to their dad. And it's not that by doing so change who the other side is, it just makes it more likely that they behave in a certain way. And for me, it's like taking what we've been doing for centuries in an offline world and we're just applying it at scale and in a way that's no longer bidirectional. It used to be the case that I do this to you and you do this to me. Right now this is mostly happening from big companies to influencing your behavior. Yeah, that's interesting 'cause it does seem like the sort of most basic mediocre use of all this psychological targeting is selling me shirts. It just seems like can't you do more with this? And I remember Cambridge Analytica, right? It was like, oh, they totally influenced the election. Did they or was it not much? So did you swing an election and you convinced a die-hard Hillary supported to suddenly stay at home and not vote? Probably not. But could you maybe have influenced some of the people who were not sure if they wanted to go out and vote and maybe you caught them at the moment when they were really scared about immigration and you changed them from a Democrat to a Republican? Probably. I think that the point of Cambridge Analytica is it wasn't necessarily even something that political campaigns have been using data for a long, long time and Obama was celebrated for the use of, like, well, there's someone who's trying to understand their constituents and try and see what they're interested in. But what do they care about? Why, how do I talk to them? I think what Cambridge Analytica more distinct from the previous attempts, at least in the public mind, was that people could suddenly make sense of it. But even if they had the data everything before, we don't think about ourselves in these separate data points. I don't think of myself as here's my browsing history and here's my social media and here's my credit card spending. I think of myself as this holistic person that's impulsive and maybe a little bit neurotic and curious. And I think once you told the public that there's a company that can predict whether you are emotionally volatile or whether you might be introverted, outgoing, I think that's what resonated with people. So do I think that they won the election by doing this magical brainwashing? Probably not. Could something like psychological targeting swing an election when they're on the margins? Probably, yes. Do we need psychology for that? Again, not entirely sure, because you can make very similar predictions with kind of skipping that stuff. Can we use this technology to decrease political polarization? These companies, they know what we're all like. They can file us into echo chambers. Can we reverse that process? Yeah, it's something that I've been super intrigued by. And who knows if it's going to plan out. But I always think of it as a technology, right? The technology at the core is trying to say, can I understand what you're coming from? Here's your point of view, here's your view on the world, here's your values, here's your personality. And now you could imagine using that to explain to you, here's how the other side sees the world. I can convince a Democrat to kind of understand, here's maybe why a Republican is more opposed to immigration, a more opposed to abortion, not in a way that a Republican would try to convince you, right? 'Cause they're coming from their own perspective, but in a way that Democrats think about the world, that's oftentimes a much more promising way of convincing the other side. Or at least making you a bit more receptive to arguments of the other side. And this is proven by research, by the way. It's essentially this idea of, can I tap into your own moral campus to make you think about the world in a slightly different way? That would be an interesting experiment. Again, though, it has to be profitable or these companies won't actually want to do it. Back to the privacy idea, what about people that think they don't need privacy because they have nothing to hide? I hear that argument all the time, there's so many people that say, look, I have nothing to hide. I don't care if they're collecting data on me. - Yeah, and now you have to stop me 'cause it's one of these topics that I could talk about forever 'cause I hear this question all the time. So again, in the classroom, when I talk about, here's what we can do with your data. There's always at least one person who says that. And in a way, I can even partially relate to this, because it feels like, well, I tried everything. It just feels like an uphill battle that I can't win. So I might as well give up. But it's a very privileged position to be in, first of all, but so the fact that you don't have to worry about your data being out there, just means that you're currently in a really good spot. If I can predict your sexual orientation, your mental health, from all the traces that you leave, in many parts of the world, is not just preventing you from going to Beijing, that could mean the death penalty still in a lot of countries. So it just means that you're currently in a good spot. And what I think is even more true is that, you don't know what it's gonna look like tomorrow. If you don't have to worry about your data right now, that might change entirely in the US. I think the Roe versus Wade Supreme Court decision made that painfully real for many, many women. By suddenly overnight, you had to worry about your Google searches, because maybe you were looking for kind of some pregnancy-related, abortion-related advice. Maybe you were traveling across states, taking your phones. I can see you're traveling to another state. Maybe again, based on your GPS records, here's exactly the location. Maybe you went to a certain clinic. Maybe you came back and you were suddenly no longer looking for certain things on Google and Amazon. So I think this notion that data is permanent and leadership isn't should make all of us kind of worried. And maybe that's the government changing, but it could also be just the leadership of companies going from one day to the next. - That's a good point. Being gay is illegal in more than half the world. - I think it's a little bit less, but it's like still many more countries than you would imagine. - So it's illegal or at least could be illegal in a large number of places. - Religious affiliation. - As a Jewish person, we don't like lists. We don't like tracking. - It's like one of the most compelling examples of why data can become extremely dangerous. Like what we know from Nazi Germany in the Second World War is that religious affiliation in parts of Europe was part of the census, which made it extremely easy for the Nazis to come in and say, well, we're just gonna go to City Hall, quickly check the record and see here is person A, B, and C, they live in this place. Now let's go and find them. Now fast forward to today, and we know that atrocity is very vastly based on whether the data was available. And today you don't need part of the census 'cause you can just passively predict it from all of the traces that you generate and from all of the data that you create. So for me, this notion that we just don't know what tomorrow is going to look like is just a good reminder that you probably should care about your privacy. That actually two thing that most people do, but if you then show them the offline equivalents and you say, okay, look, your smartphone tracking where about 24/7 is like a person walking behind you, observing your every move. That's the stalker that goes to jail. The person reading your messages like Google and that's the mailman opening your mail. Again, a person that goes to jail. So when you give them these comparisons to the offline world, I think most people actually wake up to like, oh, maybe I do care about my privacy and maybe I just haven't figured out how to protect it better. - Speaking of stalkers, surely it's happened that somebody has used online data to find and hurt someone. I'd be shocked if that hasn't happened multiple times already. - Yeah, and it's actually natural legislation. It's like a very interesting example. There's this case of a judge in New Jersey whose son was actually tragically murdered by someone that she persecuted before and they found the data online. Got it for like, I think a couple of dollars from a data broker found her home had like this entire dossier on her and her family murdered her son because she wasn't there and that led to legislation that's now protecting judges from their data being out there being sold by data brokers. And to me, it really raises this question. If we think that judges should be protected based on their data, why not protect everybody else, right? I think there's many other people who you would be worried about people getting their hands on your data and then tracking you down. - No kidding, yeah. Hey, we got to protect judges. Okay, what about these, it was really in other people that don't necessarily have political power that don't try to take their stuff offline? I remember that case that was particularly disgusting and I get that judges are in a more vulnerable place than a lot of other folks, but what about all the other people that are in a similar place? Do you got prosecutors? Sure, okay, part of the legal system. What about police officers? Okay, what about teachers or disciplinarians in the school environment? Or just like, maybe I don't want to stalk her either. How's that, you know? - Anyone write your search and you make a mistake. There's always the worry that someone at some point has beef with you and is trying to track you down. So I think anything that we apply to a part of the population where we worry about data, I think should apply to everybody. - How do we do this? The book goes into detail so we don't need to go into like, weeds too much, but one of the ideas you had was preventing companies from getting too many data points. Why is that a good idea? - I think of it as a puzzle, right? So if we think about what can I learn about the person based on their data? We talked about social media which is being a curated one and then your smartphone sensing, giving us a different angle. And you can imagine that once you put all of these pieces together, you get a much more accurate reading of who that person is. So if you're a company who can fill every single letter in the alphabet with a subsidiary, you can imagine that they hold pretty much this entire picture of who you are. So one example, and I'm certainly not the first one to suggest that, right? Scott Galloway, Tim Booh, I've been saying this for years, is if we could break up the tech monopolies, at least be a way of not having them capture this entire picture of who you are, I think that's probably a hard sell. I think there's easier ones where there's now technologies that allow you to provide the same convenience and surveys and personalization, but without having to collect the data in the first place. And for me, that's something that you can implement from today to tomorrow. - I know in the book you talk about taxing data brokers, I like your idea and this is very European and I cannot see how it would happen here. But God bless you, data co-ops, where essentially we all own our data. How would this work? Because that's like crazy talk to us Americans that we would own our own data. - It's both owning the data and then collectively managing it. So the idea of data co-ops is saying, there's people who have a shared interest in using the data. That could be my favorite one in Europe, is one that looks at patient suffering from MS. So it's like one of these diseases that is so poorly understood. It's determined by genetics, your medical history, your lifestyle. So you need quite a lot of data from patients to understand what might be driving symptoms and how to get better. And oftentimes what happens in the medical spaces, you send it to farmer companies. And in the best case, it takes years for them to develop a drug and then you paying like thousands if not millions of dollars for that. What my data does, it's essentially owned by its members. So it's people who suffer from MS coming together under this kind of legal entity of a data co-op. So it's member owned and it's legally obligated to act in the best interest of their patients. And what it can do is it can essentially say, we better understand based on research, how the disease works. But we can also now communicate directly with your doctors in almost like an Amazon recommendation style and say we've seen patients with similar symptoms in the similar trajectory respond really positively to these kinds of treatments. Why don't you try this as well? And then the doctor can give feedback and make the system even better. But you'll be surprised. So this is one example, but in the US, there's a data code for Uber drivers. So they essentially pooling data to see how do we optimize the routes? How do we make sure that we're not getting overly tired and exhausted? So I think there's many instances where you can actually see this playing out in the US as well. It's becoming more popular. It's really encouraging because it seems like something that would be almost impossible, right? Oh, you're giving the data to Facebook. So we're not going to share it. And I don't love the idea of always giving stuff to the government, but it almost seems like you need federal regulation of how our data gets used in order for it to not get misused. But it really is disappointing to me that these companies are not hitting the low-hanging fruit of finding out who has PTSD after coming back from war or after some traumatic event, finding out who's depressed, finding out who's going down the path of getting an eating disorder. Because I've heard about, I should say, young women especially, they search for something and then the algorithm feeds them more of that and then they search for more of it. And so it's clearly really obvious and not that hard to predict who is getting an eating disorder in real time. And then they just don't do anything about it. And it's not even that they don't do anything. So it's oftentimes reinforcing. Algorithms just become more and more extreme in the way that they make recommendations. Are you hopeful for the future of how this looks because I hate ending on a sour note of like, and now your kids are all going to be depressed and have eating disorders and no one's going to care except for us on this podcast. I think you have to. So I constantly oscillate between being totally depressed and kind of thinking about it in a more optimistic way. And I've been called naive so many times by people who say like, we're all doomed. Why are we talking about these positive use cases? And my take on this is that we need this positive counter narrative. And I actually think about it in the context of kids, right? You can tell your kid if your kid misbehaves and throws food on the floor and takes down all of the books, which is not hypothetical. I'm going through this. I feel it. So you can tell them like, don't do this because it's bad and because it's like nothing that you should be doing. But you're not going to be really successful. What's much more successful is to show them something that they can do instead. But if you tell them instead of throwing food on the floor, why don't you do this? The chances that you're going to change that behavior is much, much more likely. And I think about that the same way in the context of technology. Yeah, we can say here's all of the challenges and do we need more regulation? Probably should we get rid of some of the abuses? Absolutely. But I think if we think about the overall trajectory of technology, if we don't have these positive, here's what you should be doing instead. I think we're never going to get there. So maybe we're not going to end up with this utopian future that I sometimes have in mind. But I do think we need these positive visions to even get us started. Sandra, thank you so much. Really interesting episode of the show. This was very enjoyable. And I liked it. Thank you so much. You're about to hear a preview with James Patterson and what would make the best selling author walk away at the top of his game. It's rare that I don't write. What I discovered was that I loved doing it. And then I started writing stories. And I just loved it. I didn't know whether I was any good, but I loved doing it. And I was just right, right, right, right, right. When the first book came out, Thomas Paramon number gave Little Brown a blurb. And he said that I'm quite sure that James Patterson wrote a million words before he started this book. It was a great compliment. And then I decided I tried a novel. I'm really happy with the way that turned out. One of the things you always like to do at the end of the chapter is they must turn that next page. That's a strength. The weaknesses sometimes don't go as deep as they should. Here's the secret. Hit them in the face with a cream pie. And while you have their attention, say something smart, that's it. No cream pie, they didn't even notice it. So forget about it. You just talk into yourself. And if you don't say something smart, once you get their attention, it's irrelevant. You surprise people, which I think is important for my kind of book. We need heroes. And one of the things about the military, and it's very true in this book, in American heroes, but also walking my comment booths. The military is about we, not me. And one of the things I think we need to get back to a bit more is we. And it is hard to come by now, duty, honors, sacrifice. It just has to be more we rather than just me. To hear more as James Patterson reveals the moment that changed his life and the unconventional process that's helped them sell over 400 million books, check out episode 1100 of the Jordan Harbinger Show. All things Andromats will be in the show notes at jordanharbinger.com, advertisers, deals, discounts, ways to support the show, all at jordanharbinger.com/deals. Please consider supporting those who make the show possible. Also our newsletter Webit Wiser, it will make you smarter. It'll make you more practical. Something should sink in. It's only a two-minute read. It's almost every Wednesday. We talk about psychology, relationships, decision-making. If you haven't signed up yet, I invite you to come check it out. It is a great companion to the show. JordanHarbinger.com/news is where you can find it. Don't forget about six-minute networking as well. That's over at sixminutenetworking.com. I'm @jordanharbinger on Twitter and Instagram. You can also connect with me on LinkedIn. In this show, it's created an association with podcast one. My team is @jenharbinger.json, @jordanharbinger.json, @jordanharbinger.json, @jordanharbinger.json, @jordanharbinger.json, @jordanharbinger.json.com/news. (gentle music)