My guest today, Obi Felten, was in charge of moonshots at Google’s Innovation lab, what’s called X or formally Google X. X has had both big successes like Waymo, one of the first self-driving vehicles, and high-profile failures, like the attempt to provide global Internet using a network of stratospheric balloons.
FELTEN: The pressure of producing the next self-driving car and the next balloon project was absolutely immense, like it couldn’t just be any old thing. It had to be something really incredibly amazing that was also very undefined.
Welcome to People I (Mostly) Admire, with Steve Levitt.
As if moonshots weren’t enough to talk about, Obi’s, now heading a startup called Flourish Labs. She’s trying to transform the way mental health services are provided to young people. And the whole time I’ve been living in Germany I’ve been trying to find a German guest for this show. Finally with Obi, I’ve got one. She grew up in Berlin and she watched the Berlin Wall come down as a teen. That’s one more topic I have to be sure to cover.
* * *
LEVITT: When you worked at Google’s Innovation Lab, you had a very unusual title, right?
FELTEN: My title was Head of Getting Moonshots Ready for Contact With the Real World. Everyone thought it was a joke. And it came about when I sat down with Astro Teller who hired me into X and he was running a bunch of projects at X at the time and then later on took over the whole group. He had hired me somewhat on a whim as one of the first non-engineering people in the group. We had this great conversation where he told me about these projects that were all secret at the time, even inside Google. I was working for Google in Europe at the time. So it was basically a group of engineers who were working on self-driving cars and Google Glass and a drone project and an internet from balloons project and so on. And then I asked him a bunch of questions like, is it legal to fly balloons over countries? Do you have to talk to countries to get their permission to fly over? How is this going to bring internet to everyone? Are you going to partner with phone companies or compete with them? And he looked at me and he said, “Those are really good questions, but no one in the team is really working on those. Why don’t you come and help us with those?” And that’s what I did. I looked at everything from policy and legal matters to how might we talk about some of these projects, how do we take them to market, do we have a business plan for any of them? It turned out they didn’t really, at the time. So the title was very descriptive of what I was doing, which was thinking about what would happen when this technology would grow up and get out of the lab and make contact with the real world.
LEVITT: If I let my imagination run wild and I try to think of my dream job, the exact job I’d think of is the one you just described. Being around amazing engineers doing incredible technologies and trying to figure out how to get them to work. I can’t believe you actually got to do that.
FELTEN: I did get to do that. It was mostly really fun. The engineers were amazing. We then hired a whole bunch more non-engineers. You know, we got real lawyers, and we got real policy people, and real marketing people, and so on. So it wasn’t just me making it up.
LEVITT: Did that make it more fun or less fun when there were real people around?
FELTEN: Ah, that’s a really interesting question. Sometimes I wonder whether it takes the magic out of it, because while you’re just working on the technology and you’re imagining what the world would look like when you succeed and you solve these big problems in the world with the tech, it’s much more fun and magical. And then the reality hits that it’s sometimes very small things that make it hard to implement something in the real world. On the other hand, if the tech doesn’t get out into the world, what is the point? I mean, I’ve always felt that technology is just a tool. And unless we solve problems with it that people actually care about, it’s pointless.
LEVITT: So for listeners who don’t spend their days working at X, or what used to be called Google X, like you did, what is X? How does it fit into Google and what’s it trying to do?
FELTEN: X is an unusual innovation lab because most corporate innovation labs were set up to move the core mission of the company forward. So if you think of the sort of mother of all innovation labs, Bell Labs, they were solving problems for the mother company. And X was actually explicitly set up with a remit to not solve core Google problems because there was already innovation going across the entire business. The founding project was self-driving cars and that’s not really anything to do with search or advertising or any of the other Google products. So they set it up quite separately, and they gave it a small budget and a lot of freedom. Initially it was all just about building interesting breakthrough technology. We started with that tech piece, but then we realized very quickly, if it doesn’t solve a real problem, it’s also kind of pointless. I think we learned that the hard way with Google Glass, where it was really amazing technology, but it didn’t solve anyone’s real problem, and therefore it failed. And so in the end, we came up with this definition of a moonshot, which is: we take a problem that would affect hundreds of millions, maybe even billions of people in the world, and then we propose a radical solution that sounds very different. It maybe sounds like science fiction today, but within maybe five years, the technology can turn that science fiction into reality. And self-driving cars is a really good example of that, where for most of the 20th century that was total science fiction. And yet, in the last 20 years, sensors are much better, compute power has multiplied, so you can take all that sensor data in and turn it into an action that the car can take. That just was completely inconceivable and yet it’s now reality.
LEVITT: So let’s talk more about driverless vehicles because I think at some level, it’s a huge success for Google X. But at another level, I’m surprised by how slow the progress has been, not on the technology side, but on getting the autonomous vehicles on the road at scale, and how much public resistance there’s been to the vehicles. Has that surprised you also?
FELTEN: That path was definitely a long one, but I think for good reasons, because this is a pretty radical technology. It’s still a little spooky when you see the car go by and there’s no one in the driving seat. It takes time for people to get used to it, to get over their fears of the technology, and frankly, for the makers of the technology to be cognizant of how it affects society and how it might fit in. If we just put a really half-baked prototype of a piece of software out into the world, usually people don’t get harmed, but if you put out a robot car into the world, you have to be really conscious that it’s safe enough.
LEVITT: I’m surprised how measured and calm you are to describe this because I’m not even involved in it. And I think about it like an economist, and I think 40,000 people a year die on the roads in the U.S., and worldwide, I think the number’s over a million people a year are dying on the roads. And virtually every one of those deaths is because of human error.
FELTEN: Over 90 percent of accidents are due to human error. It’s hardly ever the machine that fails.
LEVITT: And I suspect that already the autonomous vehicles are way, way better than humans in all but maybe the most extreme cases. And humans make mistakes constantly behind the wheel. But still, if an autonomous vehicle makes a mistake — it doesn’t even have to be a fatal mistake — if they simply snarl traffic, it becomes headlines. And I would think that would just get you angry because you’ve sunk your blood, sweat, and tears into this and you have a great solution.
FELTEN: I mean, there’s the lifesaving aspect, which I definitely believe that’s true because human drivers cause most of the accidents. That was the reason why Mothers Against Drunk Driving were one of our early advocates, actually, in those very early days when there was a lot of resistance against self-driving cars. The other aspect is parking. If you could make taking a robot taxi cheap enough, then you wouldn’t need your own car. And most cars just sit around all day long, either on the street or in a parking lot or in a garage, and it’s just completely wasted space in our cities. So there’s definitely benefits to self-driving cars, but I think the fear runs very deep and there’s also a lot of entrenched interests, right? Self-driving cars is probably one of the few technologies where people expect that the A.I. will entirely replace the whole job. And, you know, driving is a good job. It’s often like the first job that someone does when they come to a country or long distance truck driving is a very middle class job that supports families, et cetera. So I think there’s a lot of resistance from that point of view as well that you can’t underestimate.
LEVITT: Yeah, I think 2 to 3 percent of the entire workforce is employed driving. And that’s going to be a big displacement when it happens.
LEVITT: The Google X project that got you to move halfway across the world from London to Silicon Valley was Project Loon, right?
FELTEN: Yeah, that’s right.
LEVITT: Could you explain the idea behind Loon?
FELTEN: Yeah, absolutely. So at the time, I was working for Google in Europe, Middle East, and Africa, and I was actually running consumer marketing. So launching products like Chrome and Android and many that failed that you’ve not heard of. And I saw firsthand the impact that the internet was having on countries when they were first coming online. And Project Loon, the idea was that you would connect the whole world with a network of stratospheric balloons. So these are balloons that are flying literally on the edge of space. So they’re way above air traffic. Cell towers on the ground, the economics are very much tied to how many people you can serve at that tower. So if you’re in a city, it’s very profitable. If you put it in the middle of nowhere, it’s very unprofitable. And that’s the reason why, when we started Project Loon, two thirds of the world — and by that I mean people — had no access to the internet yet. And so the idea was, what if you put that cell tower onto a balloon and you send that balloon across the whole world? So it was beautiful, it was elegant, the tech sounded really cool. Who doesn’t love balloons, frankly? And one of the first things I did was write the business plan with the team and try to work out how we would actually take this out into the world, because in almost every country, the phone company is an incumbent that’s very close to the government. And so there was a real danger that if the project leaked and we couldn’t make a proper announcement and we couldn’t reach out to partners early enough, that the phone company would just go to the government and say, “Those balloons that they want to send over our country, just say no, you know, don’t let them travel into our airspace.” And we eventually found a country that was super friendly. It was New Zealand who were very forward-looking when it came to internet access because they have actually a very big, rural area where there’s more sheep than people. And so we announced the project in New Zealand and the Prime Minister actually came to the launch.
LEVITT: You all worked on Loon for a long time, but it folded, right?
FELTEN: Yeah, that’s right. The economics didn’t work. So that beautiful business case that we wrote early on relied on the balloons becoming cheaper and cheaper with scale, and that never really materialized. And then in the meantime, the cost of cell towers did come down. That’s a hard place to be, right? If you’re competing with the incumbent solution that you don’t have to convince anyone to use as an alternative becoming cheaper and cheaper, and your new kind of radical solution isn’t getting cheap enough, then it’s difficult to compete.
LEVITT: I’d love to step back and talk about the process at X. I think many people have this idea in their head that there’s this wild haired genius who dreams up solutions whole-cloth, like the supposed story of the apple falling on Newton’s head, but that’s not at all my experience in the real world. So could we just start at the very beginning? Who generates the ideas at X? Is that the job of a small group of people or something everybody’s part of?
FELTEN: So the beginning of the process in X was actually very messy and almost deliberately chaotic. So there’s a team called Rapid Evaluation and the idea was to rapidly evaluate lots and lots of ideas and find out as fast as possible which ones weren’t going to work. So I’d say over 90 percent of the ideas that the Rapid Eval team looked at got thrown out pretty quickly, sometimes within hours or days or weeks, but certainly within months.
LEVITT: Where did the list of ideas that Rapid Eval looked at — where were those ideas coming from?
FELTEN: Lots and lots of different places. That team was a team of polymaths. So lots of them had basic backgrounds, like physics, for instance, which meant they could understand lots of things in the world. They would go to conferences, they would read papers, people would send them inbound ideas, and so on. At X, many of the ideas actually came out of academia. So when you look at self-driving cars, for example, there was a long history of self-driving car experiments being done in universities that we could build on. And there’s a huge barrier often for academics to take something into the real world, especially if it’s not biotech for which there’s much more established pathways. I sometimes think of X as taking the baton off someone who comes up with this radical idea; from the question of “Can this work at all?” to “Yes, but how?”
LEVITT: Let’s say there are a hundred ideas that come from somewhere in Rapid Eval; how many of those hundred would make it past that first stage, the stage of a few hours or a few days of analysis?
FELTEN: Probably no more than 10.
LEVITT: Okay, so 90 percent are stillborn, essentially. So it’s 10 ideas survive and then what happens next?
FELTEN: The next stage is what we call Early Stage Projects. So in the external world, that might be, like, from seed funding to series A funding. And so in that Early Stage team, we would continue to do prototypes of the technology, but we would also start looking at things like, what is the best use case for this? What might be a business case behind it? How might we eventually find something resembling — I hesitate to call it product market fit because often it was too early, but at least technology problem space fit? So it’s a bit of a back and forth between figuring out what’s the right tech to solve a particular problem and what are the problems that a particular tech might be able to solve? And at that stage, it’s probably about 10 or 15 projects over about two years and only about half of those survived.
LEVITT: So about five out of the hundred ideas make it —
FELTEN: To stage three.
LEVITT: To stage three. Anybody who’s in the business of creativity needs to recognize that your ratio of five out of 100 — is not unusual. That’s probably a pretty good ratio of success. What you cannot ignore is that you need a huge number of ideas if you hope to have anything good come out the other side. At least in academics, the economists I was around didn’t understand that. They thought if they had a couple projects, that was good, that would be fine. But my own approach was to spend just an inordinate amount of time working on ideas. And my threshold for a good idea is a lot lower than X’s threshold for a good idea, but I still would only have two or three out of those hundred that would ever turn into anything interesting.
FELTEN: I guess it’s a difference in approach. Either you think you can perfect the one thing you’re working on, or you just iterate really quickly to find out what doesn’t work. I think that’s a helpful way of thinking about it. You’re not trying to fail fast — that’s another one of those Silicon Valley cliches. But the goal is to learn as quickly as possible if something doesn’t work. And it’s more about the speed of the innovation cycles. And also, how quickly can you evolve an idea? Because you might have started in one place and then ended up somewhere very different. That, I think, is the hardest. Like, when do you throw it out altogether? And when do you think, “Oh, this could just benefit from a pivot”?
LEVITT: Everything I know about human nature suggests that projects that are failing go on way too long. What were the specific strategies you used at X to fight that bias?
FELTEN: There’s definitely a huge bias towards just keeping going with something. Economists call it sunk cost fallacy. Physicists call it inertia, like to get an object to deviate from the path is just very difficult. There’s a few tricks that we use. There was one, actually, that one of my teams taught me. It was a project called Foghorn. And the idea was that we would make fuel from seawater. So you pull CO2 out of the sea, which acts as a huge carbon sink, and then you would combine it with renewable hydrogen to make it into a renewable fuel. And the cool thing about that fuel was that you could burn it in any engine. So you could burn it in a car with a tiny bit of retrofitting. You could also fuel airplanes with it. And at the same time, you would pull carbon out of the sea, which is actually good for the sea. So I love the idea. It was such a cool concept. So the team did a whole load of technical exploration, and they found that they could make the technology work. But they had set themselves two criteria. So they had technical criteria, but they also had a business goal, which was that they wanted the fuel to be the same price at the pump as fuel that was made from conventional oil. And as the oil price was collapsing, during the project, it became very clear that there was no way we could compete. And so they had essentially hit what they called their kill criterion. And they came to me and they said, “We should shut this project down.” At least at that time, there was no way that we could ever make the economics work. A lot of things would have to happen, like a much higher price on carbon and so on, until it was even vaguely viable. What was interesting about this is that the team had set their own criteria. And therefore the decision to shut the project down came from them, and it was a much more rational decision. As opposed to, an emotional decision.
LEVITT: I’ve heard that one of the strategies that X uses is to work on the hardest part of a problem first, because you don’t want to solve all the easy parts only to fall down five years into the project on the hard part.
FELTEN: Yeah, it’s a very counterintuitive strategy to start with the hard parts. So what we did at X is we had this image of a monkey on a pedestal. And the story was: Imagine you’re trying to train a monkey to recite Shakespeare sonnets while standing on a pedestal. What do you work on first, the monkey training or the pedestal? And it’s very tempting to build a pedestal first because we know how to build a pedestal. So we can make this beautiful pedestal. We’re very proud of ourselves. But then what if in the second year of the project, you find out it’s actually impossible to teach the monkey English, let alone recite Shakespeare. So then you’re left with this totally useless but very beautiful pedestal. And so we had this premise that we would always do monkey first. So we would identify what the hardest problems were and work on those. It’s hard because I don’t think most of the world thinks that way. So investors will constantly be like, “Oh, what progress have you made this month? So it’s very difficult to say, “Oh, I solved, like, a hundredth of this really hard problem.” That feels like no progress at all. And sometimes you’ve got these sort of jumps, where it’s not a linear, nicely incremental, “I make some progress every month,” but suddenly you have a breakthrough and then it suddenly works, right?
LEVITT: Another non-obvious thing that I’ve tried to do in my own projects is to celebrate the failures, to treat failures as successes because they free you up to go and pursue other things. Did you have strategies for doing that?
FELTEN: At X we tried various different things. So one is: get people to talk about it at all-hands meetings, because otherwise all-hands just becomes this sort of show of successes. If you talk about things that didn’t work, then you kind of attribute more value to that. That applies in academia, by the way, as well. I’m on the board of a big academic publisher. They publish Nature and lots of academic journals. And, of course, it’s much harder to publish a null result, right? No one ever wants to publish, like, here’s something that didn’t work, even though that would be very useful because it would stop lots and lots of other scientists trying to do the same experiment over and over again. But if you go back to X, that Foghorn team that came to me and said, “We want to shut down our own project,” they’d actually done a very detailed postmortem and they presented that at the all-hands. They then also published it as two academic papers so that people wouldn’t try the exact same approach that we had tried. And so making it public and talking about it takes some of the stigma out of it. The other piece is that you have to acknowledge that there’s some grief around it and that however rational you try and make the decision, there’s always going to be emotion. People lose their jobs, people have to give up on something that they’ve poured their life and blood and sweat and tears into. So it struck me when I was running this team where we were starting dozens of projects a year and shutting down dozens of projects, so I was going through these life death cycles over and over again — that there’s quite a lot of similarity between grief for a human and grief for a project. And I don’t want to belittle what you go through when a loved one dies. This is not to say your project shutting down is on the same level, but it’s a similar process. There’s a model called the seven stages of grief. And I saw the teams go through similar phases, like you’re in denial, you’re angry, and then eventually you get through to acceptance. And the faster you can get through those cycles, the better, actually. And if you don’t go through the cycles, it sort of lingers. It’s just pushed down. And so one of the things we did is we started celebrating the Day of the Dead. And before you accuse me of cultural appropriation, it was actually the idea of — two Latinas in my team came to me and said, “Obi, there’s all this trauma, all this stored up grief, all these projects that we’ve shut down, and we should be celebrating this. And so in the Mexican Day of the Dead tradition, you remember the dead. You remember all the things that are amazing about them and you think of them lovingly and then you let them go. So we started doing this as a ceremony, and it was incredibly cathartic; people cried. And I think if you neglect that emotional aspect, if you try and make it just rational, you’re kind of missing out on a really important part of what it means to let go of something.
LEVITT: That’s interesting. I’ve never tried to go into that emotional space on my projects. I’m going to try that the next time.
FELTEN: Yeah, try it. It’s very hard to let go if you don’t process the emotion.
We’ll be right back with more of my conversation with Obi Felten after this short break.
* * *
LEVITT: When I heard about your job at X, it sounded like a dream job to me. And then the way you’ve spoken about it so far, it seems like it was a dream job for you also. But you left a few years ago to do something else. How come?
FELTEN: So I’d say my job at X was a dream job, for sure for the first few years. But it was actually really soul destroying for me. It was this endless cycle of just launching projects, hiring people, and then having to fire them again when we shut the project down. And it took a toll not just on the teams, but also on me, which I now realize looking back. The pressure of producing the next self-driving car and the next balloon project was absolutely immense, like it couldn’t just be any old thing. It had to be something really incredibly amazing that was also very undefined. Whereas when I worked at marketing at Google, I had really clear metrics. Like we get 100 million Chrome users. That’s successful. If we get only a million, that’s not as successful. But at X everything was just a little fuzzier and less measurable. And then the final thing was a good friend and coach told me, “You are just in the wrong job for you because there’s some people who are scuba divers, they love to go deep, and there’s some people who are snorkelers, who love to go across a whole bunch of things. And you’re a scuba diver in a snorkeling job.” And what I realized was, yes, I was overseeing projects that were as diverse as, you know, we had a chip project for machine learning chip that later went to Google. We had two or three different sustainability projects, like the Foghorn project I talked about, which was all about chemistry. We had another one called Malta that was about storing energy in molten salt, that was all about physics. I had all this diversity in the portfolio, which meant I could never go deep on anything. And it wasn’t my job. It was the team’s job to go deep on them. So I suddenly realized, oh, this is a dream job for other people, but it’s not a dream job for me. I’ve been happiest when I’ve been building things. So I went to Astro, the head of X, and I said, “Look, I’m going to quit. I’m going to go back to the startup world.” And he said, “What? Like, why would you want to leave?” And I said, “Because I want to build something again. You know, I’m a builder at heart.” And so he said, “Why don’t you pick something that you’re really passionate about and build something in that area.” And the area I was really passionate about was mental health, for personal reasons. And so the project that I picked was called Amber. And I co-founded it with a neuroscientist and with a hardware engineer. And the idea is, could we find a biomarker for mental health? So it’s very difficult, still, to assess how depressed or anxious, for example, someone is. We looked at E.E.G. brainwaves and we looked at wearable data and other kinds of data that might give you a pointer that is more objective than just asking people how they’re doing, which is largely how, in today’s world, mental health is assessed.
LEVITT: And that’s working out well?
FELTEN: I ran the project at X for several years and we made great progress on the technical side. So it turns out that you can actually see in someone’s brain waves how depressed they are. The experiment is you make them play a little game and then you see what happens in the split second after they find out whether they lose or they win. In most people, there’s a sort of spike. But it turns out in many people with depression, they have something called anhedonia, where that spike is subdued. So you can see, by the length of the wave and the amplitude of the wave, you can see how depressed they are. And that’s kind of magical. It’s completely objective, like you can’t cheat. And we thought that would be really incredibly useful. And it turns out, when we did hundreds and hundreds of interviews with people with depression and anxiety, young people, postpartum mothers, and others — we interviewed psychologists, we interviewed psychiatrists — and they all said pretty much the same thing, which is, yes, we need better measurement of mental health, but it’s not the biggest problem. I will never forget one gynecologist who said to me, “I know some of the mothers I see have postpartum depression, but I’d rather not find out because there’s no one I can refer them to because there’re just not enough therapists.” It was absolutely heartbreaking. And so I realized the biggest problem is actually that we don’t have enough counselors. It’s not just the measurement piece. And so we ended up shutting Amber down. It’s another one of the long list of X projects that failed. And that was the last project I ever did at X and I decided I wanted to stay in mental health and that’s why I moved on.
LEVITT: And so you started Flourish Labs. Can you tell me what you’re trying to do there?
FELTEN: So Flourish Labs in some way was an evolution out of that X project that I was working on when I realized that there’s this massive gap between the number of people who need care and the number of mental health professionals who can provide it to them. If you look at just the U. S., there are about 58 million adults who have a mental health diagnosis today. And it’s actually more than half of them that get no care at all. Sometimes it’s stigma that people don’t seek care. Sometimes it’s that you can’t afford it or your insurance doesn’t cover it. Sometimes, it’s that you can’t find a therapist that’s a good match for you. And that’s because there’s only 600,000 counselors in total and we would need more than double of those. And when you look at the way that we train counselors to become a licensed therapist in this country, you have to do a master’s degree and then you have to do many, many hours of supervised work before you even get your license. So I looked at many different ways of: how could you provide care in a scalable way that doesn’t rely on just training more therapists? I came back to peer support: What if we train people who have experience of mental health challenges to support other people? And it turns out it’s an evidence-based treatment. So I read lots and lots of studies that over and over showed that peer support works. People get better, there’s fewer hospitalizations. And it’s been around for decades. There’s millions of people who’ve been doing this as volunteers. Suicide Helplines or Alcoholics Anonymous are great examples of peer support. But what most people don’t know is that it’s also a profession. There’s formal training and you can become a certified peer supporter and you can do this as a job. The trouble is there’s only very few of them today. And that’s what we want to change. So at Flourish Labs, we use technology to scale peer support, and we need to do that both on the training side, and then we also need to get peer supporters in touch with people who need help. So we developed our own training where we can train people in a manner of weeks to become certified peer supporters. And then we have a telehealth platform that’s called Peers.net where a young person can go on and find a peer supporter that’s a match for them.
LEVITT: You’ve been focusing on the benefits of peer support for the person who’s ostensibly being helped. But it’s interesting, in my conversations with Angela Duckworth, who is part of our Freakonomics Network and a very prominent psychologist, she believes that actually being the supporter is at least as beneficial as being supported, which is a surprising result, but makes what you’re doing even more valuable and important than maybe you’re even giving it credit for.
FELTEN: She’s right. There are studies where people looked at the effect of giving peer support. And that’s because they were worried that it would have adverse effects on the peer supporter. If you train a young person to support another young person, are you burdening them with all that trauma, et cetera? And it turns out that’s not the case. People who give other people support stay healthy for longer. You have to be in recovery to become a peer supporter and to give peer support. And so that’s definitely something to watch out for, like if your own mental health is going down again, then maybe you need to take a break. But in general, people stay healthy for longer. You see that effect in AA, where sponsors who help other people stay sober for longer. You can see it in student mental health. It’s reciprocal. And what’s interesting, when you look at the macroeconomics of that, is it suddenly turns what is a problem that we have millions of people with mental illness into a massive opportunity because if we can get them into that workforce and get them to help the other millions of people, that doesn’t just create jobs. It also is most likely going to keep them healthier for longer.
LEVITT: I could see how someone could have a knee-jerk reaction and say, “Well, isn’t this risky and dangerous to take people who aren’t really professionals and have them give advice?” But I would say my own personal experience makes me 1,000-percent supportive of what you’re talking about. I have had the great misfortune of having a child die, my son Andrew, and that was 24 years ago. And having lived through that, I actually think I do know a lot on that exact topic, and I feel like my insights and experiences have been useful for others who more recently suffered that same fate. It’s exactly your model that I’m good at one thing. And that’s what you would ask me to do, not to try to pretend that I’m a therapist who could understand a hundred problems that I’ve never experienced.
FELTEN: So there’s a lot of research that looks at whether peer support works. There’s much less research into why it works, but you pointed at one of the things that I think is at the core of it, which is: the empathy and understanding and the connection that you get with someone who’s been in your shoes is really profound. That actually has a therapeutic effect. And the fact that it’s more equal, that you’re looking sort of eye-to-eye also makes a difference to people. The trust is often higher right out the gate. And for a few years I served on an advisory board for the Wellcome Trust, which is one of the biggest funders of mental health on the academic side. And one of the guys on that advisory board is a guy called Pim Cuijpers, who is a Dutch researcher who does a lot of studies for the W.H.O. And I asked him, like, “What makes therapy effective?” And he said, “About half of it is what the therapist has learned.” But he said, “The other half is actually about the therapeutic alliance. So do you have a high trust? Can you relate to this person? And if that therapeutic alliance is low, the therapy is much, much less effective.” And so my personal hypothesis on this is that what the peer supporter doesn’t have in terms of years of training they make up in terms of that higher level of reliance. And when you look at subgroups, so for instance people identify as L.G.B.T.Q. or people of color, they often have a harder time finding a therapist that they click with. And some of it is about the fact that most therapists look like me, they’re white, middle-aged women. So if you’re a young, Black, gay man, you might not be able to relate to me as much as you might be able to relate to someone from your own community.
LEVITT: It’s really amazing to be sitting here talking to you about this topic because I first thought about this issue maybe five years ago. My daughter, Lily, had been seriously anorexic for many years. And actually talked about that topic with her on a past episode on this podcast. As she slowly recovered from anorexia, she started an Instagram account that reported on the ups and downs of that journey. And it eventually attracted a large following. She would get hundreds of direct messages a week from people less far along than she was. And they were looking for help and guidance. It was a full-time job trying to respond to those questions. And Instagram was a terrible platform for trying to help people. It was extremely hard for people who needed help to find her. And it made me realize that there was a missing market, a market failure. There was no easy way for people struggling with a particular issue to get personalized advice from a peer who is further along. I imagined what a solution might look like. And I don’t want to minimize what you’re doing, but it’s like literally exactly what you’re doing, I could imagine that that would be a great solution. And I was just way too lazy and didn’t have any of the right skills to do it. But you’ve done it.
FELTEN: Well, I think that just proves that there’s no idea that’s original, right? So I’m glad to be executing on your idea, Steve. I mean, in many ways, it’s an obvious idea. That fundamental idea, that you can connect with someone who’s has gone through a similar experience and it’s just a little bit further ahead on the path, is absolutely at the heart of the product. And we decided to focus on Gen Z, so that’s teenagers and young adults, because that’s the age when mental health issues often show up for the first time. So your daughter, it was probably sometime during her teenage years. And I believe that if you can help them deal with these issues then, it will set them up for the rest of their life. It’s also where the treatment gap’s the biggest. Sixty percent of them get no care because it’s harder for them to find a counselor that’s a good match. And finally, they are open to peer support. When you’re a teenager or you’re a college student, you’re more likely to reach out to your friends for help than anyone else. There’s actually studies on that. And so what if you trained your friends to then be effective at providing that support. So that’s why we designed the service for that group first, even though I believe it could work for other groups too. Mothers who’ve just had babies, like postpartum, has huge amounts of stigma around it. And mothers often only talk to other mothers about it. I’ve interviewed veterans who said, “I would rather talk to another veteran about my P.T.S.D. than talk to anyone at the V.A.” And the V.A.’s making huge amounts of efforts to support veterans, but it just doesn’t create that same connection as talking to another vet.
LEVITT: This is what economists call a two-sided market, where you both need to have the mentors, the counselors, and you need to simultaneously have people who are looking for help. But it’s really hard to find mentors when there aren’t people yet looking for help. And it’s really hard to find people to show up to look for help when there aren’t enough mentors. Have you had luck cracking that?
FELTEN: Well, I would say we haven’t cracked it at all yet, but we’re very aware of it going in and actually when I raised our seed round, I purposely sought out people who had experience of two-sided marketplaces. What everyone says is you have to start with the supply side because otherwise you can’t do anything at all. So we’re mainly focused on university students, college students, at the moment on the peer-supporter side. And we get hundreds of applications each time we put out the job. And these are amazing applications, like people who have the experience of various mental-health issues. They’re incredibly diverse. Our training courses are more than half L.G.B.T.Q., they’re more than half BIPOC. So, that actually turned out to be easier and I think we have nearly 50 peer supporters on Peers.net now and we’re adding new ones on a regular basis. The demand side we thought would be really easy. Like if 35 percent of young adults and teenagers have a mental health diagnosis — you’re not even talking about people who’ve been undiagnosed, but actually have a diagnosis — and there more than 60 percent get no care, surely it must be easy to find them and connect to them. That turns out to be harder than we thought. First of all, no one knows what peer support is. People might be familiar with it in the context of volunteer groups like AA, but they think it’s just for substance abuse. Or they might know about it for crisis helpline like Trevor Project or Crisis Text Line, but they think it’s just if you’re suicidal. And this concept that peer support is something that you can do just like therapy on a more ongoing basis, or maybe just once, that’s a novel concept. So in many ways, we’re not just building one company. We’re establishing a category. So I almost wish you had done your peer support startup, Steve, because I very firmly believe that it’s not going to take just one company in the space to succeed. It’s going to take 10 or 15 companies until this becomes a familiar concept that people trust.
You’re listening to People I (Mostly) Admire with Steve Levitt and his conversation with Obi Felten. After this short break, they’ll return to talk about failure.
* * *
In the time we have left, I want to talk to Obi about her experiences growing up in West Berlin. How strange must it be to grow up in a city surrounded by a hostile East Germany. But before that, I want to ask her one more question about running a startup.
LEVITT: Can I ask you a frank question?
FELTEN: Of course you can.
LEVITT: We talked a lot about failure earlier in the conversation, but it was really about failure of a given project within a portfolio of projects, which is painful and full of grief, as you noted, but manageable. Now you run a startup, and most startups fail. And that kind of failure, when you’re the boss and a whole enterprise has to be shut down, there’s something truly terrifying about that prospect, no?
FELTEN: Yes; it’s absolutely terrifying. I lived through many, many failures throughout my career. That fear of failure is never far away if you’re a startup founder, and for good reason, right, 90 percent of startups fail in the first year. So I guess we’ve made it through that hurdle already. But something sort of shifted in me recently. We had all our money in Silicon Valley Bank. So when S.V.B. started going down, it all happened very quickly over just a few days. I think I went through those seven stages of grief in four days. On Thursday, I was in shock and I was in denial. And I was wrestling with myself of whether I should be pulling the money out. Then I finally decided to pull the money out, which felt really terrible, and by the time I’d made the decision, it was already too late. So on Friday, when they froze everyone’s assets, I was angry with myself and angry with the S.V.B. management and angry with the world. On Saturday I was bargaining and I was in despair. And then, by Sunday, my husband was like, “Can you stop reading Twitter and Slack and just, like, get off your phone?” And I did. So I turned my phone off and I sat on the sofa with my dog and I just felt really connected to all the founders in the world, suddenly everyone was in the same boat. And it wasn’t my fault. When you write an investor report every month like I do, it always sort of feels like it’s my fault if I can’t raise the next round. And suddenly nobody had any money. It didn’t matter whether you’d raised 50,000 or 500 million, everyone was in the same boat. So, I eventually felt really calm, and I went to bed. I never turned my phone back on until the next morning, which was Monday morning, you know, by which point Silicon Valley had been bailed out. What happened afterwards is I just feel less afraid. Someone told me about this Buddhist tradition, it’s a meditation practice, where you visualize your own death. So you see yourself dying, you see your family and your friends around you, and you see them grieving, and then you eventually see them moving on with their lives. And it feels like that’s what happened, that the company kind of died already. So it’s a very weird experience, but I like it. I like operating with less fear because so much of what I do in the company, which is partnerships with people or raising money from investors, I sometimes feel it’s all magic and make believe anyway in an early-stage startup. You know, we haven’t proven ourselves yet. We don’t have product market fit yet. It’s all about: can you convince people of the dream and that you could turn nothing into something. It is kind of alchemy. I think, maybe entrepreneurship is alchemy.
LEVITT: You grew up in Berlin. And the first 15 years of your life, you were in West Berlin with the wall. Did you understand how strange it was to be living in this city, surrounded by a hostile country with a wall literally running right down the middle of the city? Or as a kid, that was just a reality?
FELTEN: No, I definitely understood it because my parents are from a small place called Saarland on the Western border; so it’s on the border to France. And so for family vacations in the summer, we would drive out of West Berlin and through East Germany and then come out the other side. And there were definitely these moments when you handed in your passport to the East German border guard and they had this weird system where they would put it on a little conveyor belt from one station to the next. So you drove slowly alongside the conveyor belt to the next station. And I always had this anxiety as a kid that maybe our passports would just disappear, and we’d be stuck in East Germany. When I was 16, I went to the U.S. I actually was kind of bored in West Berlin. So I decided I was going to go to America for a year and go to high school and I ended up in upstate New York in a small village in a family that was very different from my German family. And I had an amazing time. I spent the entire year telling everyone the wall was never going to come down. And so I came back in the summer of ‘89 and in November, the wall came down. So it was a huge, huge shock.
LEVITT: Imagine if you had been away and missed it. When you heard the news were you at the wall watching it come down?
FELTEN: Not at midnight. So I had a chemistry exam in the morning. So I went to bed and then didn’t listen to the radio in the morning and went to school and I came out of the exam at 10 A.M. And everyone said, “We’re going to the wall. We’re going to the wall.” I said, “Why would you go to the wall?” And they said, “Haven’t you heard? They opened the wall last night, idiot.” So I went to the wall alongside hundreds of Berlin school children and we all climbed and stood on top of the wall, which was a pretty amazing feeling.
LEVITT: Having lived in Germany has not made me feel any less American, but it’s only been a year. Over time, has your sense of German identity eroded?
FELTEN: Even that first year. So I came back when I was 17 and then the wall came down and I actually would say that winter was probably the only time in my life that I’ve experienced something approaching depression. It was because I was coming back from another country. And when you go abroad, you expect some culture shock, and you expect that you’re not going to fit in. But when I came back to Germany, I had changed a lot. And I didn’t just fit right back in. And then the wall came out, which was this massive macro shock, where suddenly everyone was really nationalistic and jubilant and joyous about the reunification of these two countries, which to me felt really weird, because they were — culturally had grown apart quite a lot because it had been two generations. And so I felt very out of place and really alienated and also just countered the kind of prevailing zeitgeist, right, where you were supposed to be happy about this, and this was meant to be the biggest historical thing ever to happen in our lifetimes; so I felt very out of it. I couldn’t wait to get out of there. I went to Oxford and I did philosophy and psychology and met my husband there and came back to Berlin for a year, but already knew that I really wanted to live in London. So London really was home for many, many years. And I still think London probably is long-term home, even though we love living in California.
LEVITT: I heard you comment once that hearing a speech by the Dalai Lama was the most memorable experience that you had at Oxford. And then you went back and spoke at the Oxford Union. That’s probably about the most nerve wracking thing you can do is to go back to a place where you’ve been an audience member and sat in awe of great speakers and then probably feel like a total imposter up in front of the group.
FELTEN: Definitely. But I think I felt like a total imposter my entire life. I felt like a total imposter going abroad, total imposter doing philosophy at Oxford when I barely even spoke English properly. And then my entire career in tech — I was a woman, I wasn’t an engineer, so when I was at X, often I’d barely understood what they were talking about in terms of tech. Now I feel like an imposter because I’m a startup founder, but I’ve never founded a startup before. It’s a bit like the fear of failure, it never totally goes away. You just learn to deal with it.
Obi says she feels like an imposter, which doesn’t surprise me at all because I often feel that way. I absolutely feel like an imposter as a professor standing in front of a classroom. Students think I know everything when I barely know anything. The same with parenting. I thought my parents had all the answers. My God, I never would have listened to them if I had realized how little parents know. I’ve always just assumed that everyone feels like an imposter. But I wonder if that’s really true. I’m curious to collect some data on that. So if you have a minute, send a one line email to PIMA@freakonomics.com and just write either imposter or not an imposter. I’ll tally up those results and I’ll report back the share of PIMA listeners who see themselves as impostors.
LEVITT: And now, as always at this point in the show, I invite Morgan, my producer on to help with a listener question.
LEVEY: Hi, Steve. A listener named Will wrote to us about the economic principle of induced demand, where an increase in supply results in a decline in price and increase in consumption. He was curious about how it relates to traffic. For example, the typical economic assumption is that if you build more roads, you will not reduce traffic because more people will just drive. Will was curious if this is true or not. What is the consensus among economists about induced demand and traffic?
LEVITT: So honestly, Morgan, I’ve never talked to another economist about this, so I don’t know what the conventional wisdom is among economists. But I think the best way to think about this is absolutely in terms of price and to think about the cost of commuting. So, usually when we think about cost, we think about financial cost. We think about the cost of the gas or the wear and tear in our car that’s associated with commuting. But really, the biggest cost for most people, is the time they waste in traffic, which if translated into dollars is how much you’d pay not to waste that time in your car, but rather to be able to be at home or doing something you enjoy. So, if you build more lanes, then that reduces congestion, which leads to a much lower price of commuting, because it doesn’t take as long, so you save on the time cost. One of the most basic ideas of economics is that when the price of a good falls, people consume more of it, and so indeed, many more people who before weren’t willing to drive because the traffic was too bad, they get back on the road and that builds the congestion back up. Now, in any simple economic model, it can’t be the case that the congestion is as bad afterwards as it was before because the reason the congestion comes back is because the price is lower.
LEVEY: Could this be an example of where the economic model differs from reality? Because in reality, I think it has been proven that when there’s more lanes added, it actually can lead to worse congestion than you started with.
LEVITT: I mean, people can make mistakes. And it could very much be true that if everybody collectively thinks, “Oh my God, the extra lanes are going to change everything,” and so new communities spring up along the road, then that could happen. But it is true that in economic models, people don’t make mistakes, and in real life, people do. And it’s possible that happens. But I think more what people get confused about is the simple role of prices. If you lower the prices, you’re going to lead more cars to be there, and if there’s a relatively elastic demand for driving, then you’ll have a lot more cars on the road when you lower the prices. People shouldn’t think of adding lanes as primarily being about reducing congestion. They should think about adding lanes as a way to get a lot more cars to be able to drive on that road. Now, if your goal is actually to avoid congestion, there’s a really simple solution. And what do you think that is, Morgan?
LEVEY: Increase people’s access to public transportation.
LEVITT: Good job, Morgan. That’s not the answer I’m looking for, though. Because that actually doesn’t work very well because most people don’t want to take public transportation. The answer is: if more people drive when the prices go down, raise the price of driving. It turns out that driving is a great example of what an economist calls an externality. Every extra driver on the road imposes all sorts of costs on other drivers that aren’t taken care of by the market, right? When I drive, I increase congestion. I also increase the likelihood of crashes — and it turns out that’s empirically the biggest cost. So when you put more cars on the road, there are more crashes. And so if you add those externalities in with the obvious ones, which is that cars are polluting and they’re contributing to climate change and all sorts of other pollutants in the air, then there’s just too much driving. And so the logical thing an economist would say is, you should price in this externality. You should charge people per mile driven, especially when the roads are crowded, when there’s congestion. Of course, that has great efficiency principles, but like so many things in economics, it has really bad equity implications.
LEVEY: Will, thanks so much for writing in. If you have a question for us, our email is PIMA@freakonomics.com. That’s P-I-M-A@freakonomics.com. We read every email that’s sent and we look forward to reading yours.
In two weeks, we’re back with a brand new episode featuring Reginald Dwayne Betts. He spent eight years in prison from the age of 16 to 24, and then he turned his life around. He eventually graduated from Yale Law School; he’s won a MacArthur Genius grant; and he started a nonprofit called Freedom Reads. I think this is going to be an incredible episode. Thanks for listening and we’ll see you back soon.
* * *
People I (Mostly) Admire is part of the Freakonomics Radio Network, which also includes Freakonomics Radio, No Stupid Questions, and The Economics of Everyday Things. All our shows are produced by Stitcher and Renbud Radio. This episode was produced by Morgan Levey with help from Lyric Bowditch, and mixed by Jasmin Klinger. Our theme music was composed by Luis Guerra. We can be reached at firstname.lastname@example.org, that’s P-I-M-A@freakonomics.com. Thanks for listening.
FELTEN: I honestly don’t know what I am anymore. I guess I’m still German.
- Obi Felten, founder and C.E.O. of Flourish Labs.
- “How to Overcome the Fear of Failure,” by Obi Felten (Washington Speakers Bureau, 2023).
- “A Viral Video of a ‘Reckless’ Robotaxi Caused an Uproar in San Francisco. Police Say the Internet Got It Wrong,” by David Ingram (N.B.C. News, 2023).
- “Highlights for the 2021 National Survey on Drug Use and Health,” by the Substance Abuse and Mental Health Services Administration (2022).
- “My Next Mission: Flourishing Minds for All,” by Obi Felten (Medium, 2021).
- “The Loon Collection,” by X (2021).
- “Sharing Project Amber With the Mental Health Community,” by Obi Felten (Medium, 2020).
- “Inside X, Google’s Top-Secret Moonshot Factory,” by OLIVER FRANKLIN-WALLIS (Wired, 2020).
- “X: The Foghorn Decision,” by Robert S. Huckman, Karim R. Lakhani, and Kyle R. Myers (Harvard Business School Case Collection, 2018).
- “A Review of the Literature on Peer Support in Mental Health Services,” by Julie Repper and Tim Carter (Journal of Mental Health, 2011).
- “The Self-Defeating Nature of Urban Road Capacity Policy: A Review of Theories, Disputes and Available Evidence,” by Martin J.H. Mogridge (Transport Policy, 1997).
- Flourish Labs Peer Support Training.
- “Will A.I. Make Us Smarter?” by People I (Mostly) Admire (2023).
- “Why Is Flying Safer Than Driving?” by Freakonomics Radio (2023).
- “Is Google Getting Worse?” by Freakonomics Radio (2022).
- “Should Public Transit Be Free?” by Freakonomics Radio (2022).
- “Amanda & Lily Levitt Share What It’s Like to be Steve’s Daughters,” by People I (Mostly) Admire (2021).
- “Sendhil Mullainathan Explains How to Generate an Idea a Minute,” by People I (Mostly) Admire (2021).
- “The Most Dangerous Machine,” by Freakonomics Radio (2013).