Search the Site

Episode Transcript

My guest today, Clementine Jacoby, is spearheading a transformation of the criminal justice system. How is a 31-year-old woman who had never worked in this area before making such an outsized impact? By using data in ways it’s never been used before in criminal justice.

JACOBY: We started checking the data to try to find like, where did we go wrong? And it turns out we hadn’t gone wrong. There were thousands of people who were past their release dates.

Welcome to People I (Mostly) Admire, with Steve Levitt.

Clementine Jacoby started her nonprofit called Recidiviz around the same time I started my RISC center at the University of Chicago. We’re both trying to change criminal justice. And she’s having a lot more success than I am. I’m curious to hear how she’s made so much happen so quickly.

*      *      *

LEVITT: How do you describe Recidiviz to someone who’s never heard of it?

JACOBY: Recidiviz is a tech nonprofit and the thing we’re trying to do is dramatically and safely reduce the footprint of the U.S. criminal justice system. We’re trying to reduce incarceration and reduce disparities at the same time, while making communities safer. At the most basic level, our strategy is to work with the people who make decisions about who goes to prison, how long they stay, and who gets released, and equip those people to use their data to make decisions better. So there are these people called corrections directors, and they’re over not just prisons, but also probation and parole, which it turns out are responsible for half of prison admissions in the U.S., which I think is pretty astonishing.

LEVITT: You mean half the people who go to prison are not being sent there freshly from committing a crime, but they’re being deported there from probation or parole where they’ve already done some part of their punishment, and something goes wrong?

JACOBY: Yeah! It’s the people who are on probation, which tend to be young men, and that’s a really important population because if you can keep them out of prison, all kinds of good things happen. Prisons are what we call criminogenic. So when you put someone in prison, you make it more likely that they’re going to commit a more serious crime in the future. So if you can prevent young men from going to prison in the first place by figuring out how to do probation better, that’s a huge impact. And then the other population is parole, which is people who are finishing their prison sentence at home. So they’ve gotten education credits or they’ve completed programming. They’ve done enough things in prison to complete the rest of their sentence outside. When you combine those two populations together, that’s called supervision. And supervision is about four-and-a-half million people. And is responsible for half of admissions to prison.

LEVITT: Just a reminder to people, how big is the incarcerated population in the U.S. today?

JACOBY: There’s 1.2 million in prison and that’s the population that we’re focused on because those are people who are going in for more than a year.

LEVITT: And then another probably five or 600,000 in jails.

JACOBY: Exactly.

LEVITT: But that’s not your domain. So what you’re saying is they have the data, but they don’t know what to do with it or they can’t access it. What’s the problem they face?

JACOBY: They have massive amounts of data. The problem is not the existence of the data. The problem is that they don’t have any capacity to use it. So we come in and bring all of their data from all of their fragmented systems into one place and we standardize it. And then we use that standardized set of data to start building tools for different people at different places in this pipeline. So for a parole officer, we may alert them to who is eligible to go home from supervision. Whereas for a leader of the agency overall, we might be alerting them to which of their programs seem to be working. We might be alerting them to which officers in the state seem to be really, really good with people who have unstable housing or which officers are sending a lot of people to prison for minor infractions. And so up and down the stack, you’re using the same data to give people the information that they need to reduce admissions to prison for things that are minor and to reduce length of stay when someone’s in the system and to see who’s ready to come home. The U.S. prison system grew 5x starting in 1970 —

LEVITT: At a time when crime was, if anything, going down or staying the same. I think that’s an interesting fact as well. When I’ve looked over lots of different public policy graphs of what the U.S. has done over the last 50 years, that’s got to be one of the most interesting — this explosion in the criminal justice population that happened everywhere in the U.S. It wasn’t like it happened in 10 states and not in 40; it was across the board.

JACOBY: If you have not seen this graph look it up. Because what you see is this massive explosion of both incarceration and supervision. And you see crime essentially hovering flat during that period. So there’s a 5x increase in the number of people we’re putting in prison. We incarcerate more women, more children, more people for longer than any place in the world. And it’s not having an impact on crime.

LEVITT: Okay, so we could debate that. I would argue if we let everyone out of prison, crime would go way up. But I think what you mean is this policy of mass incarceration didn’t lead crime to plunge.

JACOBY: Exactly. So that’s the actual statement of fact. Starting in the 1970s, incarceration very quickly increased by about five-fold during a period under which crime stayed relatively flat.

LEVITT: And what would’ve happened if we hadn’t increased the prison population? That’s a question of active speculation among researchers.

JACOBY: It is. And I think the thing that’s worth looking at is that over the last 10 years, 36 states out of 50 have reduced both incarceration and crime. So that’s a really interesting thread to pull on to figure out what that counterfactual might have been or could look like. In essence, what we can see with federalism is 50 different experiments on the question that you’re talking about running right now. And so by the time that you reach 2005, you’re starting to see bipartisan consensus that the system we have isn’t working. We’re sending more people to prison for longer, and we’re not seeing big reductions in crime. We’re spending quite a lot of money on this system, and the outcomes aren’t very good at all. I think 83 percent of people who are released from prison are one day rearrested. So you start to see bipartisan consensus and we start to pass laws to unwind mass incarceration and what happens is that those laws don’t get implemented. So, as an example, we went to the state of Idaho several years after a law had passed that made it possible for people to earn their way off of supervision. So if they were on supervision and they had had stable housing and stable employment and done everything right, they could get off of supervision. The problem was that this was happening, on average, never. And so the thing that we built was a tool that basically alerted individual parole officers when someone was eligible to get off of supervision. And that very quickly led to Idaho being able to remove 5 percent of their supervision population. Now the state of Idaho has a lot more resources freed up to focus on people who really do need to support. ‘Cause when you’re a parole officer and you have a hundred people that you’re supervising, it’s actually pretty hard to help a hundred people at once in an intensive way. And so getting rid of 5 percent of that population meant that you could actually generate more success with the people who need more support.

LEVITT: I’ve studied the criminal justice system for 30 years. And I can tell you, bad data are a constant struggle. If you had come to me before you started Recidiviz and asked me what I thought of the idea, I would’ve said, “I love the idea.” But I still would’ve strongly discouraged you from trying because, honestly, I would’ve thought that it would be absolutely impossible to make any progress. Having to persuade the state prison systems to cooperate with you, I would’ve thought that by itself would be a complete blocker — that you would never convince a single state to be a partner. So I’m curious, how did you convince the first state? That must have been some kind of miracle?

JACOBY: I do think it was really important to start this company when we were like 27 because we needed a lot of energy. And the data problems are very hard. But the way that we convinced the first state was not the hardest part. In fact, I think one of the most surprising things about Recidiviz is that corrections directors, these folks who are essentially C.E.O.s of corrections agencies, right? Tens of thousands of employees, billions of dollars in annual budget — those folks really recognize that the system’s not working. They also recognize that they don’t have very much data or technical expertise, and that they’re not getting the information they should be getting. They’re quite willing to believe that we can produce something good for them. And so Recidiviz started as a project when I was a product manager at Google. We had this kind of scrappy team of volunteer software engineers that were working on nights and weekends to aggregate and clean data for people like you. Because we knew that researchers wanted to get their hands on the data and it was hard. And we somehow got on the radar of a woman named Leann Bertsch, who at the time ran the North Dakota prison system. And she reached out to us and she said, “What you believe the problem is, is that we have the data and we’re not giving it to researchers. What the actual problem is, is we don’t have what you think we have.” And I was like, “Well, I definitely know that you have access to what I think you have.” Like, the data’s getting generated. And so we went to Bismarck in January and I said, “Show me everything you get. Show me any data your team brings you any time during the month, the week, the year.” And we left that trip so stunned. So stunned that we decided that we needed to spin the thing out of Google.

LEVITT: Not stunned, like, “Wow, they have such great data, everything’s going to be good.” You’re saying what you saw was just so dismal that you felt like you had to devote full time to it.

JACOBY: I was coming from Google, a place where I could, like, change the color of blue on a link on the search page and then run a multi-arm experiment with many millions of users in each arm and get the answer back the next day. And I went to North Dakota and there was one woman whose name was Wanda, who was tasked with calculating recidivism for the state of North Dakota. And she was actually like semi-retired. It wasn’t clear what was going to happen after Wanda. And so we actually codenamed our first calculation pipeline Wanda. But the point was, as leaders of the system, they were getting almost nothing. And if you talk to an individual parole officer who was making calls about who went to prison and why and when, they were getting truly nothing.

LEVITT: That’s amazing that she reached out to you and you left that meeting in Bismarck and she wanted to be your partner. She was all on board to give you the data and to make things happen?

JACOBY: She was onboard, and she introduced us to other states. So the whole way that we’ve expanded from North Dakota to the 11 states that we’re in now is through the relationships between these corrections directors. And pretty much the entire way that we’ve decided what gets built is by seeing what they use and how they use it.

LEVITT: So you’re happily ensconced at Google, probably making a lot of money. You’re doing this criminal justice thing kind of on a whim, and then this one conversation in North Dakota is enough to have you turn your life upside down?

JACOBY: Well, it wasn’t a whim in the sense that I grew up with a dad as a political scientist, and my mom worked on addiction medicine and I had family members who were in and out of the criminal justice system. I had an uncle go to prison when I was 5 and I was fascinated by and frustrated with the system from that time forward. And because my dad was a political scientist, I talked a lot with him about how this was a very special area of American public policy. It was a place where we could expect to make progress because of the bipartisan alignment. And so I was always in some sense working on this on the side. What is definitely true is that I did many things that were not that effective in criminal justice first. The first thing that I did was worked in a diversion program for youth. And the thing I learned from that was, like, I had no business doing that. I wasn’t from the neighborhoods they were from. I didn’t have their backgrounds. What was different about Recidiviz is that it really did need someone who could translate between technology and government and who could find an approach that would be embraced by the left and the right and leaders who are under very different political constraints. And that turned out to be a thing that I could do really well. The other thing that I turned out to be able to do was assemble a really strong software engineering team, which hadn’t happened in this space.

LEVITT: Criminal justice is big business and there must be enormous firms that have huge contracts working with these states to do their data work. Are you saying that they don’t exist? Or that they’re just so bad at what they do that you could come in as initially a group of five people, and do better everything they’ve been doing?

JACOBY: There are definitely many vendors in this space; they just don’t do a great job at what we are trying to do. In fact, there’s not really anyone that’s trying to inform the leader to move in a safe, decarceral direction. Because you have to be in the back of the car with parole officers driving around, understanding unique policy constraints of that state. And I think this is the part that the vendors don’t really do to build relationships so that you can figure out how to create a tool that an individual prison guard or parole officer or supervisor will use. You also have to be a good enough technical team to turn that into a tool that can generalize across policy environments. We now have states that are as small as Maine and as big as California. We have red states, we have blue states, almost exactly 50-50. In this space there are 50 decision makers. If you can show them that you can get them what they need, then it’s not too hard to get a foothold.

We’ll be right back with more of my conversation with Clementine Jacoby after this short break.

*      *      *

LEVITT: Okay, so let’s go back. North Dakota’s your first state, and you get their agreement to share data with you. What was the first really concrete step you took that directly led to people being let out of prison or people being removed from supervision?

JACOBY: Our first hypothesis for how we would actually have an impact is that we would show leadership which of their programs were working because at the end of the day, they can’t make any progress if they can’t see which of the things that they’re spending money on are working. And so we were sort of plotting along that path when Covid came along. We were five people at that point, less than a year into the work. The question was like, will Covid get into a prison and if it does, what will happen? We scrap everything we’re doing and we sit down over a weekend and build this model that tells you for the size of your prison population, for how it’s laid out, for how overcrowded it is, this is what we would expect to happen if Covid hits. And we had to use data from the Spanish flu because that was the only data available on viral spread within prisons. And that model we had worked on over a weekend, released on a Monday. And by Wednesday all 50 states had downloaded it. We were getting calls from Japan and Canada and Australia. We were this tiny team trying to support, like, multiple federal governments and state governments in the U.S. to use this thing. And then states started to ask us like, “Okay, the governor of the parole board and the corrections director all understand that if we can make some releases, especially for people who are older or sick, we should be able to prevent death, both among the people under our care and our staff. Who are the people that we should release?” So we started running these queries to see who meets the criteria to be released safely? They’ve served over 80 percent of their sentence, they’ve had only positive behavioral reports while in, they’ve completed all of the things the judge has asked, et cetera, et cetera. And we saw what we thought were errors, which were people who were past their release dates, like thousands of them. And so we started checking the data as we always do to try to find, like, where did we go wrong? And it turns out we hadn’t gone wrong. There were thousands of people who were past their release dates. And we were like, okay, so you should probably start with these. And we ended up working on about 36,000 early releases during Covid. And people did very well. Like the federal system released 11,000 people. Of those 11,000 people, 17 committed crimes. Of those 17, one was a violent assault.

LEVITT: If you had asked me — again, you didn’t ask me for advice about starting this — but a second obstacle that I would’ve given you had you asked me is: look, even if you get access to the data, the criminal justice system is so highly politicized and there’s this big “law and order, lock ’em up” constituency out there. There’s a social justice, de-incarceration movement out there. They’re both powerful. They both care a lot. And I think your solution has been really brilliant, which is that you focus on carrying out the laws and regulations that have already been passed, but because they aren’t being followed, they’re leaving more people locked up. And so the social justice people are of course in favor of that. But almost nobody’s so “lock ’em up” that you say, actually the law says they should be free, but I want ’em locked up anyway. So you really found a sweet spot that didn’t have any political resistance.

JACOBY: Yeah, it’s pretty elegant. Because it’s not only something that is uncontroversial, but also it’s really big. We think there are probably about 250,000 people who could be removed using laws on the books. It’s a hard calculation to do because it’s a very precise calculation in the 11 states where we’re working and not as precise in the 38 where we aren’t. But we think that’s about right, and that’s a huge number of people. Like that’s so much money that you can free up for things that are working. It’s a massive number in the context of a really historic staffing shortage, which is something that these agencies are facing right now. And so the opportunity to say you could get a quarter million people out of the system safely using laws already on the books without passing new laws is a pretty powerful thing. And then, like you said, these are things that everyone has already agreed on. They’ve already gone through the system of checks and balances. They’ve already been through the legislature. They just haven’t been implemented.

LEVITT: Now I’m sure there are also lots of laws and regulations on the books that, were they properly enforced, would lead to lots more people being locked up. You could maybe use data to, say, facilitate better coordination across states to find fugitives with outstanding warrants who fled to different parts of the country. But then you crash into politics. It would be an unpopular application of data for the social justice folks. But my own conjecture is that those could be really valuable things to do. You know, maybe not your mission, but maybe it should be somebody’s mission. What’s your reaction to that?

JACOBY: Yeah, it’s an interesting question because in this country over the last like 10 or 15 years, there’s been such a widespread recognition that the current system isn’t working that the laws that have been passed in large part have been things like, “Let’s give folks in prison incentive to better themselves while they’re there. So let’s” — like I said — “let’s let them earn time off for getting a G.E.D. or for completing anger management programming or for doing cognitive behavioral therapy.” There are lots of laws like those. And so it doesn’t feel like we’re having to sift through laws and say, “Which of these do we believe will have the right kind of impact?” What feels true is that advocates and the reform movement and practitioners and policymakers have kind of like come together, passed a smorgasbord of laws that create this enormous pent-up potential. Like in every state that we go to, there are five or 10 or 15 of these laws that we can quantify pretty quickly will produce a de-carceral impact. But it’s not common at all that we see a law on the books that would lead to incarceration if we implemented it. I think you’re right, that like cross-state fugitive tracking might be a good example of something that in very small numbers would increase incarceration. But those numbers would have nothing compared to the kinds of laws we’re talking about. We’re talking about laws that impact every single person that’s on supervision in that state. I think everyone deserves to feel safe. And I think you can believe that the system needs to be a lot smaller and believe that we need to do a lot better job for victims. And you can also believe that the system needs to be a lot smaller and that our homicide-clearance rate is way, way too low. And so it’s true that I don’t believe that the growth in our prison system is the right way to keep us safe. And that if we had the same $80 billion or even 40 billion of those dollars a year that we’re spending on incarceration right now, that we could spend on violence reduction or on a lot of other things that generate safety, including, you know, streetlights, that we could do a much better job.

LEVITT: You spoke very much like an economist there in the sense you said, look, we’re spending a lot of money on criminal justice and if we weren’t, then we could be spending it more effectively elsewhere. We’re not getting the maximum benefit from our marginal dollar that we’re spending on criminal justice. You know, it’s interesting, my own evolution, because when I was a young scholar, some of the earliest research I did was looking at the marginal impact of locking up more people. I tried to come up with better causal estimates than people had before. And obviously you can’t run randomized experiments. I’m not allowed to go to one state and release a bunch of people and go to a second state and lock up a bunch of people who wouldn’t have been locked up otherwise. So we had to look at natural experiments and I got bigger effects on the margin from locking people up than anyone had ever gotten before. And, at the time, when you worked through the numbers, it looked like we had really the right size of the prison population. If you just did very crude, marginal cost, marginal benefit kinds of calculations. Well then the next 20 years happened and the prison population doubled and tripled from where I’d done that research. And even using the same estimates that I had, as we locked up, more and more people, the marginal person being locked up just got less and less dangerous. And so I would be approached pretty consistently, by various states and people who were advocating to put in more prisons. And because I’d written the first paper that had ever somewhat convincingly said that prisons were effective in reducing crime, they would ask me if I would come and testify on behalf of building this new prison. And I’d say, well, actually I won’t because if you even believe the numbers in my paper, which are bigger than what most other people have ever found, and you now fast forward to our current system, we have too many people locked up. And on top of that, you’re in state X and, you know, your neighboring state actually has a ton of excess prison capacity right now. And crime has been falling for the last 10 years, and we have every reason to believe it’s going to fall for the next 10. Wouldn’t it be a lot more effective to work out a deal where maybe you ship a few excess prisoners across state lines for a few years until your own prison population starts to fall? So I made that argument perhaps five times, and each time the person on the other end of the phone would listen to my argument and they’d say, “So you won’t come and testify. Do you mind if we use your research to justify the new prison?” And I said, “Look, I think it would be a mistake, but you can do whatever you want with my paper. It’s in the public domain. It’s just not the right interpretation.” And I think every single one of those prisons got built. And I’d always imagined when I got into academics that the policy makers looked at the academic arguments and that shaped the way they thought about the world. It was only later that I realized that, more often than not, they know what they want and then they go try and find the academics that will support it.

JACOBY: Did that change your approach?

LEVITT: It certainly disabused me of any notion that the reason I should do my academic work is because I was going to have a positive effect on public policy. And one of the reasons I came to this podcast was I just looked back over a life’s worth of academic research and I thought, “Well, 30 years of doing no good’s enough. Why not go try to do something that’s both fun and I might be able to do some good?” And that’s where I am now.

JACOBY: You talked about tradeoffs and if you just sort of naively divide the amount that we spend on prisons by the number of people there, you get the fact that, on some level, we could send each of these people to Harvard every year and have money left over. So we’re definitely not doing something that’s efficient in terms of creating communities that are safe. I think everyone agrees that everyone deserves to feel safe. The problem is that there’s a lot of disagreement around what we really mean when we say safety. There are people who care a lot about crime committed in prisons and some people who don’t think about that as crime at all. There are people who care a lot about contaminated water or white-collar crime and want to see that prosecuted much more heavily. It sometimes ends up looking like the debates in the space are between people who care about safety and care about victims and people who are advocates and want criminal justice reform to result in a total replacement of the current system. And I think the more productive conversation is if you can say, “We all agree that everyone deserves to feel safe.” Let’s say for a moment that we have about $80 billion a year to spend on that question, just on the correction side of things. If we were to start from scratch, what is actually the system that we would design? And you’re right that I believe that if we started from that point, we would end up with a much, much smaller system than the one that we have, not just a different set of people in there.

LEVITT: No, I love that. And I’ve actually never had that conversation with anyone in 30 years of studying criminal justice, but I do think it’s the right one to have. It’s just that there’s no one seemingly interested in having that conversation.

JACOBY: But wouldn’t it be like, such a profound conversation that would lead to a lot of good solutions. We’ll have that later. That’ll be another podcast.

LEVITT: So we’ve been talking about Recidiviz, but not actually about you personally. How old are you anyway? 30? 31?

JACOBY: 31. Geriatric.

LEVITT: Let’s just put it right on the table how remarkably unusual it is that you’re a 31-year-old woman running a 60-person criminal justice tech company. I mean, criminal justice is completely dominated by men, and so is tech, although maybe not as much as criminal justice. I suppose you probably go to these criminal justice conventions. There’s nobody that looks like you around there, is there?

JACOBY: Yeah, that’s definitely true. I don’t think that I’ve ever been particularly attuned to this dynamic. Like when I was at Google, I worked in gaming and so I would go do these business meetings in Japan. And Japan is very tenure based. And also they roll deep to meetings. And so it would be like just me on my own, 25, and then like 15 Japanese men across from me. And so this is not as extreme as that. There’s some really amazing female leaders in the corrections field actually. So, like I mentioned, Leann Bertsch had been a pretty important part of our story ‘cause she was the one that was like, “You’re looking at this all wrong, do this instead.” And actually the other superhero of our life right now is Anne Precythe, who is a woman who runs the Missouri Department of Corrections. She has been the main driver in bringing us into new states recently and talking about how important it is to use data to do this work. And so there are definitely women in the field.

LEVITT: And the two real innovators who’ve made what you do possible are both women. That’s not likely to happen by chance.

JACOBY: Well, we’ve had some men that have paved a way too. One of our previous board members was the Secretary of Corrections in Pennsylvania. He was just incredible. Josh Tewalt who leads the Idaho Department of Corrections — honestly, Idaho is so close to my heart because it’s where my family’s from. It’s where my uncle is still in prison now in his forties. But I love Idaho’s leadership team. I think that they are just rocket ships. So there have been plenty of men that are amazing too. But yeah, I’ll give the female corrections directors credit. They’re pretty great.

LEVITT: Could you talk more about your uncle who’s incarcerated who really shaped your thinking when you were a child?

JACOBY: So he committed a nonviolent crime when he was a teenager. He intimidated the parents of one of his friends. So he was definitely a knucklehead, like he deserved some kind of intervention. But prison was not the right place for him.

LEVITT: And how long of a sentence was he given?

JACOBY: It was 10 years.

LEVITT: Whenever I hear about these harsh sentences for minor crimes, I’m just boggled at how unpredictable and irrational sentencing is in our system.

JACOBY: And it’s just so hard to imagine how a 29-year-old coming out, who now has a prefrontal cortex where previously he didn’t — like, how is that 29-year-old going to do better and commit fewer crimes than the 19-year-old who went in? And I guess this is the reason that I really believe in the hypothesis that prisons are criminogenic. The data to support it is good too. I think it just doesn’t tend to go well when you get a 10-year prison sentence as a teenager.

LEVITT: Did you ever visit him in prison as a child? And what was that like?

JACOBY: All growing up, my mom, his sister, would visit him and sometimes I was allowed to and sometimes I wasn’t. The rules would change. This is actually one of the key things I remember about the system is the rules changing. Like sometimes we were allowed to send him books and sometimes we weren’t. Sometimes I was allowed to visit and other times I wasn’t. And the irrationality of it leaves a really big impression on you as a kid, as irrationality does on all kids, right? You just can’t figure out who’s making the rules and why they’re changing. And he did serve 10 years and when he got out I was a teenager and my family threw this big family reunion to welcome him home. And then a few months later he was back in prison. I was just struck by parole because everything about it seemed impossible. It felt to me like by the time he was coming out as an adult, he had no shot. The number of rules on parole that he was aware of and that he wasn’t aware of, and like how many things you could do wrong that would get you sent back to prison. Like I said, if you’re hanging out with anyone else on parole, if you cross county lines, if the bus doesn’t come and you’re late to your meeting with your parole officer, or if you are staying at your parents’ place and your dad has beers in the fridge, even if they’re not your beers, if one of the conditions of your supervision is no alcohol, that’s alcohol. So parole just seemed impossible to me. And what I learned was that one in four prison admissions is for something called a technical revocation, a thing that wouldn’t send you or I to prison, assuming you’re not on parole. It’s for something that’s one of these things that I’m describing: being in the presence of someone on parole, crossing county lines, et cetera. I was really interested in that problem because I was like, that is something that almost certainly the left and the right can both get behind. There has to be bipartisan support to figure out what’s really going on there and to fix it. I was very interested in the idea that we could make a lot of progress in this space if we could find things that did have bipartisan support. And so, you know, as a teenager getting exposed to parole, that became the piece that I was chewing on the most.

You’re listening to People I (Mostly) Admire with Steve Levitt and his conversation with Clementine Jacoby. After this short break, they’ll return to talk about why she dropped out of Stanford.

*      *      *

It might seem like laser focus would be the only way to accomplish as much as Clementine Jacoby has by the age of 31. No, not even close. Clementine Jacoby marches to the beat of her own drummer.

LEVITT: Even as young as you are running Recidiviz, your path to there was not exactly direct, right? You were a Stanford undergrad, but you dropped out. Why did you drop out of Stanford halfway?

JACOBY: I dropped out to join a circus.

LEVITT: You don’t say that with pride. You said it like you’re embarrassed. I thought you were proud of joining the circus.

JACOBY: I am proud of joining the circus. And I did it because at the time I was 20 and I was like, you know, I’ll never be in better shape. So I ran a kind of experiment, which was that I was going to do a software engineering internship and then I was going to join a circus. And Circus De Soleil has this international network of feeder circuses and they had one in Brazil in Rio. And I had lived in Brazil before. I spoke Portuguese. And so I auditioned into that circus. I also did a software engineering internship. And I figured that one of these two things was probably my career path. And so it seemed like a reasonable bet to take one year off of Stanford to then decide which of these two things was the better one to do.

LEVITT: Did you tell Stanford about the circus part or did you keep that secret?

JACOBY: Okay, so the problem is that I’ve never been great with paperwork and it turns out that there’s like a one-page form that you have to fill out to take leave from Stanford. And I didn’t do it, accidentally dropped out, had to reapply to Stanford again. So I went, I did the software engineering internship, I did the circus. I ended up getting dengue fever in the circus and it was the second time I had dengue fever. And so the mortality rate was really high. My parents made me come back. I don’t even remember that part ‘cause I was, like, delirious and hallucinating. And at the end of the year I decided like, I don’t like either of these.

LEVITT: So let’s talk about the circus. What was your job in the circus?

JACOBY: I was an aerialist, which means all of the stuff up in the sky, but I specialized in an apparatus called aerial hoop, which is like a big steel ring that hangs from the ceiling and spins. And so there’s a contortion component and a big strength component, and a really big pain tolerance component ‘cause it’s a little bit like hitting yourself with a baseball bat a lot. I still have a 25-foot free standing aerial rig in my backyard that I try to get out on sometimes.

LEVITT: And what drew you to the circus? You just love the aerial work? Or you thought the lifestyle — what did you think would be the attraction?

JACOBY: I just thought everything about it was magical. I got exposed to circus when I was living in London as a kid. Cirque du Soleil came and I got like the cheapest ticket possible and I went and saw it and I was obsessed and I was like, “I have to see it again.” And so I hung out at the show, I think I was 12 or 13. I hung out at the show after it got out the next night, and I became friends with the fire blower in the show. And then I got to see it a million more times from backstage. But it turns out I hated the lifestyle. When I actually started doing it full-time, I really struggled with there being no intellectual stimulation of any kind. Like, you’re so exhausted that it doesn’t even seem within the realm of possibility to have an extracurricular conversation at the end of the day. It loses the joy kind of once you go full time.

LEVITT: I mean, I have to say modern circuses, they’re amazing. With so many kids, I’ve seen a lot of circuses. And the innovation and the creativity of some of the acts really just leaves me speechless. Circuses today are so different from those when I was a kid and those who are already great. But really I would have to say circuses have changed as much or more than almost any form of entertainment other than obviously tech-driven ones that I can think of.

JACOBY: I totally agree. I go as often as I can. And a really big shift has been to not using animals, to just focusing very much on the absolute limits of what the human body can do. And that’s just the height of entertainment to me.

LEVITT: And my impression is that the circuses are a truly global labor market. You see acts from all around the world who have perfected specific skills that you never would’ve imagined existed. And then once they’ve discovered them, the limits that they stretch with them are just incredible.

JACOBY: What is true about circuses is that they basically invent new acts and new fields very often, like at a much more rapid pace than other sectors that I can think of, maybe excluding tech. All of the time there’ll be something that was thought way too dangerous, or like just not physically possible, and then six months later it’s been done. And it’s not only been done, it’s now being done on stages everywhere.

LEVITT: What kind of lessons did you take away from the circus that you still use in your life?

JACOBY: Funnily, a lot. One is this element of storytelling. People have to remind me over and over again, in the context of Recidiviz, that I need to state the basic facts first, and that the basic facts of what we’re doing are so astounding. Before I talk about what’s hard or what’s new or what’s next, I need to give people a moment to absorb, like, to date we’ve helped 70,000 people come home. So that’s one, this idea of layering storytelling. But I actually think the much more profound thing that sports teaches you is that idea that you’d need to know where you’re going in order to land the trick. I famously and maybe borderline obnoxiously talk a lot about wild success within Recidiviz. I ask people to articulate what wild success is in very small increments. Like, don’t go into a meeting unless you know what a really good outcome for that meeting would be. And if you can’t articulate that, cancel the meeting or move it out ‘cause you’re not ready. What exactly do you want to be the outcome here? What would be the very best outcome? And I think that is a sports thing. Like you cannot do a back handspring unless you know exactly where you are trying to land. And if you try to do one without visualizing what it’s going to look like when it’s successful, you will get hurt. The hard tricks in aerial hoop are the ones where if you were to stop halfway, you would break your neck. This has been really important because now with Recidiviz at the size that it is — but even more so when it was smaller, right? Like at the beginning, the line between self-employed and unemployed is so thin. And so at the beginning, you’re just trying to like wake up in the morning and make a company exist. And there are so many demands on your time, like from the moment you wake up, everything is pulling you into reactive mode. You need to know, like, by the end of the day, I need to get here. And if you can write that on a sticky note the night before and get there by the end of the day, slowly, day-by-day, you will build a company.

LEVITT: One of the ironies I find that leaders face is you kind of have to pretend like you’re going to stay at Recidiviz forever, but everyone who looks at your history and knows about your curiosity and your vitality probably wouldn’t be surprised if 10 years, 20 years from now, you were doing something totally different. Do you think about that at all?

JACOBY: I have never been asked that question in that way. And it does make me nervous because I think that to do the job well, you have to wake up every day and requalify to be the C.E.O. of Recidiviz. Every six weeks, it’s a different job. Every six months, it’s like an order of magnitude harder. On some level, to do it well, you have to believe so hard that this is your highest purpose and that you are going to put in the effort to not be the bottleneck on the organization’s growth, to do it as well as you can possibly do it. In that sense, I’m probably not thinking beyond Recidiviz right now, but I do know some lifelong things that are true. One of the things that I tell people on my team is that the thing that you’re really, truly the best at doesn’t feel hard for you. And so it’s actually really hard for you to notice it. And so you need to go back and look at the things that your family growing up was great at. Like, what was the thing that was just in the water when you were growing up? And for me, I think that is bridging divides. In Recidiviz it’s very clear what role that plays, which is that we’re building this approach that has support on the right and the left. It’s a very clear divide. But in general, the thing that my family was good at was really, really respecting some opinion that seemed totally outlandish to us. Having a point of view, but being able to be talked out of it; or talked into a middle position, or finding common ground. And so I think even if there was a time beyond Recidiviz, it would probably still have something to do with trying to bridge what’s good about the private sector and what’s good about government. I’m very interested in the ideas that America was founded on. And I’m very interested in how far away from achieving those we are right now. I’m very interested in the power of technology and how poorly I think it’s being used and how many detrimental effects it’s having. I’m really interested in how astonishingly brave and kind and well-meaning almost every public servant I’ve ever met is, and yet how slow government tends to be, and how poor the outcomes are. And so I think a lot of elements of the questions that I’m the most frustrated by are present in Recidiviz, but I think there are also lots of other places where they could show up and I would expect those to be common themes.

How awesome is it that Clementine actually ran away to join the circus? My first thought was to say, “Isn’t that great? Too few people follow their dreams.” But then I thought about it. My God, can you imagine a life in the circus? Of all the things to leave Stanford for. Was there something in your life that parallels Clementine’s love of the circus? Did you drop everything else to pursue a dream that other people thought was crazy? Or did you stay on the beaten path? And looking back, do you think you made the right choice? I’d love to hear your story. Send us an email. The address is That’s And if I get enough responses, I’ll even do a little data analysis and report back on the findings. Although, of course, the set of people who write me will not exactly be a representative sample.

LEVITT: And now it’s time to answer a listener question. And as always, I’m joined by the show’s producer, Morgan.

LEVEY: Hi, Steve. A listener named Zane had a question about open access and academic publishing. So typically when a paper is published in a journal, you need a subscription to that journal in order to access the paper, meaning the average citizen can’t read it. Zane thinks this is crazy and believes everyone should have access to published research. He wants to know if you feel that open access somehow cheapens your work, or would you prefer that more people had access to your published papers?

LEVITT: I would not in any way feel cheapened if more people were able to read my academic research. I think the crux of the problem is really fundamentally economic and it’s not about the authors, it’s about the journals and the fact that these journals have a complete monopoly over the content within them. So if you really want that paper, you need that particular academic journal, and that gives the journals tremendous market power. And it’s exacerbated by the fact that as an academic, I absolutely would demand from my institution that I have access to all the top economic externals. And so the library really has no choice but to pay almost whatever price that the publishers want to charge for the high prestige journals. And that is really what drives the economics of the academic publishing industry is that you have these institutions that have a very inelastic demand curve that are willing to pay a lot of money. That dictates everything that follows.

LEVEY: So I do think that a lot of journals might have an option for individuals to subscribe; you don’t have to be an institution, but those subscriptions are usually still quite pricey. And I think the reason is because they still want to encourage universities to buy a subscription for their faculty and staff instead of having that faculty and staff buy a bunch of individual subscriptions. So it is hard as an individual to subscribe to these journals.

LEVITT: Absolutely.

LEVEY: Now, Steve, I do know that in some sciences, some journals will allow the researcher or the institution on behalf of the researcher to pay an extra fee to allow their article to be open access, meaning accessible to anyone. You don’t have to have a subscription, but then that fee falls on the researcher or that institution. It’s not a cost that’s incurred by the journal. What do you think of that model?

LEVITT: I honestly have never heard of that before. I like that idea and it’s interesting because it’s a way of shifting the cost away from consumers who don’t want to pay a lot to get access to the article onto the universities who are the ones who presumably are getting a lot of the benefit of the wider dissemination. So I think that’s fabulous. I’m surprised I’ve never heard of it before.

LEVEY: So, Steve, should we really be blaming the journals for limiting the public’s access to cutting edge research?

LEVITT: I think the sad reality of it is that it’s what the economics of the industry dictate, and I think we’re unlikely to change it as a consequence. Now, on the other hand, lemme just say the academic publishing industry, it is not one that people are getting extremely rich doing. I edited a journal called The Journal of Political Economy put out by the University of Chicago Press for 10 years, and it is one of the most prestigious journals in our field. And still the amount of revenue that it generated wasn’t that much. And it seemed like almost all of that revenue was going to cross subsidize the dozens of other academic journals that the U of C press was putting out. So I don’t want to leave the impression that the people who work in academic publishing are a bunch of fat-cat, getting-rich-off-the-common-man people. I think mostly they’re working really hard in the name of science and scratching out a living doing it.

LEVEY: Zane, thanks for your question. If you have a question for us, our email is That’s It’s an acronym for our show. 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 Kevin Kelly. He’s the founder of Wired magazine and he’s been a keen observer of the tech industry for the last 40 years. Among other things, I’m hoping he can explain everything I need to know about artificial intelligence. As always, thanks for listening, and we’ll see you in two weeks.

*      *      *

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 and mixed by Jasmin Klinger. Lyric Bowditch is our production associate. Our executive team is Neal Carruth, Gabriel Roth, and Stephen Dubner. Our theme music was composed by Luis Guerra. To listen ad-free, subscribe to Stitcher Premium. We can be reached at, that’s Thanks for listening. 

JACOBY: I was scrolling through and there weren’t that many entrepreneurs. Like, I was like, oh my gosh, am I going to be the dumbest person that’s ever been on this show before?

Read full Transcript




Episode Video