The US uses prisons to deal with problems that are better solved by investing in the community: mental health, drug addiction, and issues in disenfranchised communities.
In the US it’s difficult for a person to recover from incarceration. They will likely never be eligible for jobs or housing, can’t use systems designed for the underserved and their kids are more likely to enter the system.
State governments are already collecting data around incarceration but not using it for decision-making or optimization. The data is also stale, and isn’t as localized as it should be.
A self-improving prison system has tools to use data to set goals, track progress towards them, see if the policies have had impact, and then course correct if they haven’t. Recidiviz provides those tools.
IVAN STEGIC: Hey Everyone! You’re listening to The TEN7 Podcast, where we get together every fortnight, and sometimes more often, to talk about technology, business, and the humans in it. I’m your host Ivan Stegic.
My guest today is Clementine Jacoby, who is the Executive Director of Recidiviz, a nonprofit co-founded by Clementine in 2019, that is building a technical foundation for a more accountable and self-improving criminal justice system. Before that she spent four years at Google as a product manager, where she worked on Google Maps and Android.
Welcome Clementine, it’s great to have you on.
CLEMENTINE JACOBY: Thanks Ivan, excited to be here.
IVAN: Let’s start very simply. Where did you grow up? Why did you choose to go to Stanford?
CLEMENTINE: I was born in Boston and spent my childhood years in Iowa and Germany and then went to high school in Utah and also in London. But Utah is where I feel like I’m from, and that’s where I am now. I chose to go to Stanford because they had this program called Symbolic Systems, where you could study computer science and philosophy and cognitive science and linguistics, which were sort of my four favorite topics. So, it seemed sort of magical that they would all be in one major. So, that’s where I went and that’s what I did.
IVAN: What exactly did you study when you were there?
CLEMENTINE: I studied Symbolic Systems, and I also did a lot of work in the D School at Stanford which is their design school, which has really been one of the pioneers in human-centered design. And when I was a kid I wanted to be an inventor, with a lab coat and goggles. And so, I don’t know, being a product designer sort of felt like the closest thing to that, so that’s what I spent my undergrad doing.
IVAN: And it sounds like you were an aerialist in the circus troupe or two as well. How did you decide to do that? What was the origin story for that?
CLEMENTINE: [laughing] It’s true. I took a year off from Stanford halfway through, because at that point it felt like there were two clear, plausible career paths. One was being a circus performer, and the other was being a software engineer. And so I thought that it would be good to figure out which of these things I was going to do. And so, I took a year off, and like a true scientist, had an experimental design for the year, where I did two circus troupes and two software engineering internships. And at the end of the year, I didn’t like either of them. [laughing] So I became a product manager which I loved.
IVAN: Is there anything from your experience being in the circus troupe that you brought with you into your product management at Google?
CLEMENTINE: Oh sure, product management is very performative. A lot of it is getting people interested in the thing you’re doing, so that’s pretty much the same across the two fields.
IVAN: What were you responsible for at Google as a product manager? I alluded to the fact that you worked on the Google Maps API and Android. But I think there’s some gaming involvement there as well?
CLEMENTINE: Google has this program called The APM Program, and it’s a rotational program, so you work on two different projects over two years. My first project was focused on the Internet of Things, and in my second project I worked on using Google Maps data to build a platform to make it easy to build real world games.
Neither of these things were things that I was personally passionate about, but I think that what Google does so well is give young people real responsibility. They give you ownership over, in some cases, brand new projects. And so you feel like your successes are your own, and your failures are your own.
I loved Google, but I ultimately got bored of the actual problems that we were working on. In tech, I think, we spent a lot of time working on the problems we experienced, like recruiting and ordering food and getting your inbox to zero. So, it seemed like there were probably better things to be working on at a certain point.
IVAN: And so you took one of the 20% Projects that you were working on and turned it into a company or a nonprofit, right?
CLEMENTINE: Right. Why not?
IVAN: [laughing] Yeah, why not right? How did the 20% Project that now is Recidiviz first start? What was the impetus there?
CLEMENTINE: I grew up with family members in prison, so I think I was always interested in our criminal justice system just from the human angle. But I think when I got to college and after college I started realizing the extent to which it was a very special problem, that the United States incarcerates more people than any country at any point in the history of the world.
And also that, the way we got there is that we sort of use criminal justice to solve all of these other social ills. Like, 40 percent of people with a serious mental illness in this country come into contact with our criminal justice system at some point.
IVAN: Wow, 40 percent?
CLEMENTINE: Yeah, we’re using prisons for all kinds of things they were never designed for, which is a fascinating problem. Another wild stat is is that every 25 seconds someone in America is arrested for drug possession, and so, we’re using prisons for mental health, for addiction treatment. And as a result it’s a massive problem statement. So I think that’s extremely interesting.
The other part of criminal justice that felt important to me and part of the reason I left Google is because there’s this really big “why now?” There’s bipartisan alignment around unwinding mass incarceration, and various serious people on both sides of the political aisle are looking at the system and saying, “Enough”. They arrive at the conclusion from different places perhaps, but it’s one of the only things that Republicans and Democrats actually agree on right now.
And I think for that reason among so many others, it’s a really interesting time to be trying to make progress on this problem that is important and sits at the intersection of many other also important social issues.
IVAN: So it sounds like you’re at the right place at the right time by pursuing the 20% Project, and I’m surprised I didn’t know there was bipartisan alignment on something at all in Congress, quite honestly. Does that mean that legislation has passed and there’s funds that are being put towards figuring out the system and fixing it?
CLEMENTINE: I think mass incarceration in its current form started in 1970, and prison populations and parole and probation populations steadily and swiftly grew until about 2007 to 2010. In the last decade we’ve started to make some progress, you’re starting to see where the peak was and that we’re coming down a little bit. So the game now is about accelerating that progress. We would have to reduce incarceration by something like 75% in order for America to even come into the global average range.
CLEMENTINE: So, the extent to which we’re an outlier here is hard to overstate. There’s an enormous amount of work to do, but just getting past that peak is what the bipartisan alignment made possible. Like turning that corner, which I do think we’ve turned, we just haven’t made anywhere near enough progress yet.
IVAN: We have 75% to go to even make it to average. So, like I’d say that’s a lot of work that we need to do. Let’s talk about the word recidivism, which is what your nonprofit’s name is based on. I read that there isn’t exactly a common understanding of what that word means. Can you describe what the problem is with that word?
CLEMENTINE: Yeah, that’s correct, it’s an extremely fraught term I will say. The basic notion is that recidivism is a measurement of how many people who leave prison are winding back up there. But there are lots of issues with it. So as a metric, right, your one job is to just mean the same thing in every context. Like, what if every country measured GDP differently? That would not be super useful. So that’s basically what recidivism does. We measure it differently in every context, and so as a metric it fails pretty spectacularly.
The other thing that’s bad about recidivism is that it measures failure. So, it’s sort of like measuring schools based on drop out rates. But nothing else, no test scores, no future earnings. Then a third thing that’s bad about recidivism, and then I’ll stop, [laughing] is that it’s measured very infrequently. So, typically recidivism is this three-year average, and it’s often reported annually and when it’s reported it’s often a few years behind. So it’s this wildly useless thing to optimize for. It’s like if you said, “I’m going to lose weight,” but then you only weighed yourself once a year, and when you did you were told what your weight was a year and a half ago, which is not super helpful for trying to make progress in real time.
IVAN: So it sounds like the fix for that is easy right? You make the word mean the same thing globally, and then you report that data more often. But it sounds like that’s not exactly very easy to do.
CLEMENTINE: Well, I mean, that’s basically what we’re trying to do. And we’re also trying to do it for a lot of other metrics, more nuanced metrics. Like, what is the experience of people in prison and on supervision? What’s leading to success? What’s leading to failure? Where do we not have treatment in the community, and how is that causing failure? What do geographical and racial and demographic disparities look like? So, the name is a little bit of a misnomer, because we’re certainly not only, or even principally, interested in recidivism.
But I think the problems around this really important metric were one of the things that sort of drew us to the space initially and made us think like, Oh, there’s something solvable there. Like there’s something technology can actually help with. I think in general technology is not the solution to very many of the world’s most important problems, but in this case there’s something really useful to do with data, and there’s a way in which a very basic technology problem is actually getting in the way of kind of, otherwise momentous change.
IVAN: So the name of the nonprofit is Recidiviz, and the viz part is like the visualization of data. My understanding is that your product is a data visualization dashboard that correctional facilities and prisons use. Is that a fair description or did I get that wrong?
CLEMENTINE: No, that’s very close. We build a number of products and the dashboard is one of them. I think taking a step back, what we’re trying to do is make sure that criminal justice decision makers have the data that they need to drive a change, to drive better outcomes. And so, we build tools including dashboards that help leadership and staff see their own metrics, their metrics compared to other systems and then how they can improve. So, we’re like Fitbit, but for criminal justice agencies.
IVAN: [laughing] Fitbit for criminal justice agencies. Got it. Where does all this data come from?
CLEMENTINE: Interestingly we’re already collecting most of what we need. What we’re not doing is making it useful in any way. And it turns out that it’s a lot of work to just turn millions and millions of data points into something that people can actually take action on. So, that’s a big part of our job. But we’re already collecting the data, because we need it for tracking purposes, we need it for operational purposes. We just haven’t used it yet in terms of decision-making tools or optimization tools in this space.
IVAN: Can you give us an example of some data that you’re already tracking that you’re already tracking that you’re using?
CLEMENTINE: Yeah, so, a good example is that even though we report on recidivism very infrequently, when we go in and establish the Recidiviz platform on top of these existing state data systems, we’re then able to report to them their recidivism rate every day. And we can give them a three-year trailing window, a one-year trailing window, a one-month trailing window. And so, sort of overnight you can go from a place where this is a metric that you only get once a year to this is a metric that you get every day.
The other thing that you can do is, now you can slice it by county, so you could see if you’re getting different recidivism rates by county. You can slice it by judicial district so you can understand how judge decision-making or prosecutorial decision-making is impacting outcomes. You can also slice it by what’s happening in the community. So, is the availability of AA programs or drug treatment or mental health treatment, or educational resources for young people leading to a school-to-prison pipeline, for example, which you can see quite clearly in the data, is the case.
IVAN: When you first started Recidiviz, it was a 20% Project, and then you co-founded the nonprofit going through the program at Y Combinator. And for those who are listening who don’t know what Y Combinator is, Y Combinator is a start-up ah...
CLEMENTINE: Incubator? [laughing]
IVAN: Incubator, yes. I couldn’t think of the word. Yes, it’s a start-up incubator and you usually think of companies like Dropbox that came out of it and multi-billion-dollar companies that are certainly for profit, not nonprofit. So, first of all, I didn’t realize Y Combinator founded companies that were nonprofits. How did you decide to go that route, and why did Y Combinator decide to go with you and your co-founder?
CLEMENTINE: So YC invests in a small handful of nonprofits in every batch that they think that they can help scale. And they treat you just like the rest of the batch. There’s one partner, Tim Brady, who is a nonprofit veteran and one of the most angelic humans you’ll ever meet. So, he advised us on some of the nonprofit specific nuances of our work. But for the most part, it’s the same ball game, right? You got to build something that people will use everyday, and so the hard parts are all the same between for profit and nonprofit, I think.
IVAN: And, you say you’ve been founded for a year, maybe a year and a half?
CLEMENTINE: Yeah, so, depending on when you start the clock on Recidiviz, we’re either two years old or more like four years old. As an idea, and something that existed as a volunteer project within Google, we’re about four years old. As an independent, nonprofit that’s building enterprise software for state governments, we’re about two years old.
IVAN: And what does your market demand look like right now? It sounds like you’re working with states and with correctional facilities and prisons and states?
CLEMENTINE: We are. So, we’re working in four states right now and expanding to eight at the moment. I think what’s been interesting about the demand is that it’s much higher than anyone told us it would be. And I think that again is because the needle has moved somewhat recently on the criminal justice reform problem. Right? Until very recently there wasn’t nearly as much pressure on the state system to produce better outcomes, they weren’t under as much public scrutiny, and advocates have done a great job of bringing to the government's attention to the places where they would like to see better outcomes.
So, now the question is, how do we use the data we already have to optimize for success. The public is no longer okay with us continuing to lock more people up. And so now we need to think differently about the problem and that’s why demand for these tools and for data in general to kind of chart a course to a new and different kind of system is in high demand. I don’t think it would’ve been true 15 years ago.
IVAN: Do you think you’ll measure success of as getting down to the average or below the global average and being in all 50 states and having successfully reformed the criminal justice system as a whole, or do you have some other grander vision than that?
CLEMENTINE: No, that sounds good, let’s do that. I think we’ll do that. [laughing]
IVAN: Okay, let’s do that. That’s a pretty large undertaking.
CLEMENTINE: The two other things I would add to it just to make it larger that we need to do it safely and we need to do it equitably. There are lots of ways you can plausibly reduce incarceration that would actually lead to higher concentrations of black and brown people in the system, and that is not a good solution. So, one of the things the data allows us to do is actually decarcerate in ways that, I think, do account for public safety and do account for racial disparities. And that hasn’t been possible before.
IVAN: Have you noticed from the data any of the major factors that you’ve learned that causes recidivism the most? Have you been able to interpret anything just yet?
CLEMENTINE: The thing that the data clearly illustrates, I think, is that to make a huge dent in this problem, to make the kind of progress we’re talking about, we need to invest money that we will save by incarcerating fewer people in drug treatment and mental wellness, and in underserved communities, which by the way, you can see in the data. Right? Like today, if you are a black man without a GED, you’re more likely to be behind bars than employed. And we can see in the data which communities were underfunding, which communities are disenfranchised and where we need to invest. So, I think that one thing that the data makes very clear is that we’re using prisons to solve problems that would be better solved, more cheaply solved, more equitably solved, all of these things in the community.
I think that’s a solution that people can really rally behind. There’s something very, very important about that, that I think you’re clearly seeing in the public narrative right now. We want to see investment away from our justice system into the communities and into solving these problems earlier in the pipeline. So that’s, I think, the biggest takeaway that the data shows.
IVAN: What do private jails and private systems have to do with the problem statement?
CLEMENTINE: I think in some ways it’s the same problem statement. The private prison system is not as big as you might think. Most of the people we incarcerate are incarcerated not by the federal government and not by private prisons, but by states. And that’s why we are focusing on states. I think that it’s an issue, but it’s not the issue that we’re tackling.
The reason you have private prisons is because states contract to private prisons, because they think that they can either produce better outcomes, or incarcerate people for less money. And so to me, it feels like an extension of the same set of issues, which is that we don’t have tools to optimize for better outcomes, and we need those across the private and the public system, but the public system to me is more important because it impacts more people.
IVAN: And so the only way we can fix this is by making public policy and laws that address people going into incarceration and also preventing people that have left incarceration to prevent them from going back into incarceration.
And what you’re saying is the data is there and we need to make it more visible and use it to make these decisions a whole lot faster and a whole lot better.
CLEMENTINE: Yes, and I will add one thing, which is that we also need to recognize that in addition to making our prison system better, we also need to confront how hard it is in this country to recover from incarceration. Most people who go to prison will never be eligible for most jobs or housing, or even things like food stamps that were literally designed for underserved and marginalized communities. They’ll be five times as likely to be unemployed. They’ll earn less money when they’re working, and their kids will be six times more likely to enter the system.
So, we’ve created a revolving door that many millions of people enter very young and then can never leave. And so, if we want to really unwind this issue, we also need to get serious about reducing barriers to reentry, reducing contact in the first place, and also improving the system itself for the people who do get caught in it.
IVAN: It’s not just the system, it’s everything around it as well.
IVAN: We should burn it all down. [laughing]
CLEMENTINE: [laughing] You heard it here first.
IVAN: [laughing] No, we shouldn’t do that, we should fix it all, and I think that’s what you’re trying to do.
CLEMENTINE: We’re trying to do that, yeah.
IVAN: Your website talks about a self-improving criminal justice system. What does that mean?
CLEMENTINE: For us I think the main goal is that you should be able, as a person running a correction system, to set a goal, track your progress and hit that goal. Right now, running a corrections department is very much like being a CEO of a company. But with no way to measure profit and loss, or whether or not any of your launches worked. And so, it’s that basic sense that today we have 50 people with an enormous amount of responsibility running an $80 billion system with hundreds of thousands of employees, incarcerating millions of people who don’t have these tools.
So, let’s get them the tools. Let’s get them the tools that they need to set a goal, see if they’re tracking towards it. Let’s get them the tools they need to evaluate upfront what impact a policy will have, and then to follow-up and see if that policy actually had that impact. And if it didn’t, close the gap. So, that’s what we mean by a self-correcting system, just giving them the ability to actively and proactively manage for the outcomes that we want to see as a citizenship and then report to us the progress on those things.
IVAN: Is the tool set in the product that your building going to be available as open source technology? How does the thing you’re building benefit the public, not just only these certain correctional facilities get the tool? The set of tools that you’re building, what’s the approach to open sourcing it?
CLEMENTINE: I thought you’d never ask. [laughing]
IVAN: [laughing] Tell me more.
CLEMENTINE: I think just one of the foundational tenets of what we’re trying to do is say, Recidiviz is not going to solve anywhere close to all of this problem, and technology’s not going to solve anywhere close to all this problem. And so our approach from the beginning has been open source everything, partner with people who have been doing this work for much longer than we have, stay in our lane, make sure we’re not duplicating effort that other people in the ecosystem are already doing or could do better. By the way, that all takes a lot of work. Right? Just making sure that you’re being a good partner and not redoing work that someone else could do better than you’re doing feels like 80% of my job. But I think it’s really important and we’re absolutely committed to everything that we do in terms of data infrastructure being open sourced, so that if we’re not moving fast enough, and the state wants to move even faster, they can take what we’ve done and extend it, or they could bring someone else in to extend our work, or build something very specific to a particular state. We want all of that to happen.
IVAN: Is it global in nature? Are you learning from what people are doing in other countries? Is the software being used elsewhere? Because it’s not just a U.S. problem. We have a major problem here, but it’s a global issue as well.
CLEMENTINE: I mean our focus is certainly on the U.S., but during COVID we built a modeling tool that sort of showed how the virus might impact a particular facility, a prison or a jail. It helped you scenario plan. Like, if I did this what would the impact be on viral spread, on hospitalizations. That got picked up by several other countries. Australia took it and in true Australia fashion made it like a hundred times better than our version was. [laughter] Canada used it. So that was very cool to see, and I think in part that was because COVID was new, and we acted quickly.
Whereas in our sort of steady state work, the problems that the U.S. has are actually quite different than the problems that we see in the rest of the world, where some of these data barriers have already been overcome. But in many ways, that’s also a nice thing because it provides a blueprint for where we need to get.
IVAN: What have you seen from the COVID effect on prisons and on the work, you’ve been doing with the states?
CLEMENTINE: There’s so much to say about COVID. The vast majority of it is really horrible. It’s been a bad time to be in prison. It’s been a bad time to be a prison administrator. The flipside is that it’s unleashed a huge amount of creativity from state and local governments. They have really, in many ways, borne the brunt of COVID response, and so working with them for the last nine months has been very inspiring to see parole boards figure out how to do their work remotely, to see supervision officers figure out how to do their work by phone.
We’ve basically had an overnight experiment happen nationwide, and we’ve learned a lot of things from it. Like incarceration is down 20% nationwide, and there’s no data to suggest that that reduction in incarceration has caused any increase in crime.
IVAN: Oh my gosh.
CLEMENTINE: So that already teaches us quite a lot. Right?
CLEMENTINE: And that’s not even to mention all of the kinds of operational changes like doing parole and probation by phone that could make people's lives easier and save taxpayer dollars without compromising public safety. Right? So there’s sort of these innovations that we found all over the place, big and small I think from COVID. But of course that being said it’s been a bad time.
IVAN: It’s been a horrible time as you’ve alluded to, but, if I can summarize what you said so that I can understand it. You basically said, COVID is bad. It caused us to rethink the way we’re doing incarceration. People were let out of prison and crime didn’t go up, and we have data to show that?
CLEMENTINE: Yep, I think that’s a good summary.
IVAN: Oh my gosh.
CLEMENTINE: And then on the side, we also discovered all these other things just about the way we were working and had been working for so long, that that might not be necessary. This sort of necessity is the mother of invention and we saw some breakthroughs, some of which I hope will stay in place. And others, by the way, that were very bad. So, for example, we know that it’s really important when you’re in prison to be able to be visited by your friends and family, and none of that has been able to happen during COVID. And we know that not only has it had a huge impact on people during this time, but it will also have an impact on reentry outcomes.
IVAN: And you can’t fix that right now without a vaccine, like literally cannot have people visiting in person.
CLEMENTINE: Yeah. They’re like big cruise ships that people visit during the day, and then go back to their families. So, mass incarceration became a public health issue for everyone, which was another bleak but good side effect is that, having this many people in prison was a danger to all of us and still is a danger to all of us as we enter the second or third wave, or whatever we’re calling this fall.
IVAN: Yeah, I don’t know what we’re calling [laughing] it, but it feels like we’ve been in lockdown since March, at least here in Minneapolis it does. What do the leaders and the decision-makers think about the data? Do they believe it? If they see the data that we just described and it shows that crime doesn’t go up when incarceration goes down, do they believe it? Do they trust the accuracy of the data?
CLEMENTINE: We are integrating directly with the state’s source of truth data systems, and that’s a different approach from what a lot of folks are doing in this space. So, we don’t get a lot of pushback on the numbers we’re presenting to leadership, because upfront we’re ensuring that we can recreate the numbers that they trust, and everything builds on top of that.
So that’s the difference between the data we’re presenting them, maybe versus what they’re reading in the news or elsewhere. But, yeah, I think there’s a lot of hunger for data at the moment, and that people who are running the system recognize that COVID presents a completely unprecedented problem. They were lapping up data just as much as anyone else. There was no precedent for the situation, and they were sitting in between public safety and public health trying to juggle these two concerns. And so in the states where we were able to actually track people who were getting out early due to COVID and compare those peoples outcomes to people who were getting out at that time anyway, and to people who were getting out one year ago when the economy was more normal, that has been the most helpful data that we’ve been able to provide during COVID.
IVAN: I can hear how passionate you are about this organization, about what you’re doing, and I have to commend you on everything that you are doing. I wonder what inspires you. There are days, I’m sure, that are just bad, when there is pandemic and it’s affecting everything, and it’s just awful to get to the next day and to the next launch. What is inspiring what you’re doing right now?
CLEMENTINE: I think the moment that the field is in, it’s hard to not be excited about. You’ve got incarceration down temporarily, you’ve got budgets down, which is forcing a set of questions around how we can create a smaller, cheaper system that produces better outcomes. And then you also have the Black Lives Matter movement that’s putting attention on the system in a way that’s really important that’s causing policymakers to need to look at the racial disparities in the system and find solutions. So, I hope that in the next few years as a field, we’ve used this moment to safely and substantially reduce the incarceration rate.
I also think it’s important to, during this moment, think about reducing the number of people on supervision. It’s in some ways the less talked about problem, but something like one in 55 Americans or four and a half million people are on probation or parole. And this is a big driver of the collateral consequences of incarceration. So, I think reducing the supervision population is something that I’m excited about as well. Reducing collateral consequences in general, I think more than five million Americans can’t vote because of felony convictions. And so, for the world’s proudest democracy, I think that’s not a cute look.
Then like I said before, I think the other thing I’m excited about is, very seriously and methodically reinvesting savings from reducing incarceration into the community, because I think that is the path to addressing the deep disparities that are in the system today. We need to invest in education and addiction treatment and mental health, especially in communities that are underserved, that are marginalized. That’s actually what’s going to help us over time stop using prisons for things they were never designed to do.
IVAN: I couldn’t agree with you more, and I hope that the next administration brings us that kind of hope and that kind of investment in our future, and that we can get to doing exactly what you described in what is in the mission of Recidiviz in the future.
CLEMENTINE: I hope so too.
IVAN: Join us again in the future, please Clementine, and tell us how it’s going. Maybe in a year or two we’ll revisit and see where you’re at.
CLEMENTINE: That sounds great, I’d love too.
IVAN: Thank you so much for spending your time with me today. It’s been so great talking to you.
CLEMENTINE: Yes you too. Thanks Ivan.
IVAN: Clementine Jacoby is the Executive Director of Recidiviz, a nonprofit that is building a technical foundation for a more accountable and self-improving criminal justice system. You can find them online at recidiviz.org.
You’ve been listening to The TEN7 Podcast. Find us online at ten7.com/podcast. And if you have a second, do send us a message. We love hearing from you. Our email address is [email protected]. Until next time, this is Ivan Stegic. Thanks for listening.