Meeting the Moment Part 2: Partnering with Government to Improve Criminal Justice Outcomes

In the second of a special five-part series, Andrew Warren, with special guest Tarra Simmons, discusses how data might open new doors as states look to reform criminal justice systems.
Listen Now
Andrew Warren

Head of Product, Recidiviz

Tarra Simmons

Washington State Legislator and Former Inmate

Listen Now


Political leaders are sometimes risk averse when it comes to criminal justice reform, so they need reliable data to evaluate which programs or initiatives are likely to succeed.

Often states can’t get relevant data because there isn’t a universally accepted way of measuring outcomes.

The cultural tide is changing and moving away from mass incarceration, so data can play a crucial role in helping develop long term solutions.

Tarra’s struggle with substance abuse disorder left her feeling trapped, fearing she might lose her job or get arrested if she tried to seek treatment.

After serving her prison sentence, Tarra said having a criminal record prevented her from getting jobs that she was qualified for, sentencing her to a life of poverty.


IVAN STEGIC: Hey everyone! Welcome to the TEN7 Podcast. This is the second episode of a special 5-part series called “Meeting the Moment: Using Data to Reimagine Criminal Justice

I’m your host Ivan Stegic.

This series is a partnership with Recidiviz, a nonprofit organization that is using data-driven tools to help guide change in the criminal justice system.

Each episode will feature an expert from Recidiviz, helping us explore different angles related to mass incarceration, and how evidence-based solutions may help our nation find a different path forward.

The mission of TEN7 is Build Things That Matter, so taking some time to explore criminal justice system reform in this way fits within our values, and it aligns with our desire to make the world a better place.

As we learned in the first episode of the series, mass incarceration is a uniquely American problem. The United States has less than five percent of the world's population, but we incarcerate nearly a quarter of the prisoners. Those numbers are even worse when you look at prison rates for people of color and for women.

The good news is there is growing recognition that our criminal justice system is broken. Now we need to figure out the right steps to get us to a better place. That’s why Recidiviz is working to provide data to decision makers, so we can be guided by facts rather than partisan ideology.

In this episode, we’re going to take a closer look at the role of data in criminal justice reform and how government has, or has not, tapped into data to learn what works and what doesn’t work as we try to reduce incarceration.

But before we get to that discussion, I want to continue with the story of Tarra Simmons, a state legislator in Washington who spent time in prison and has worked hard to overcome the many obstacles that can prevent people from escaping the criminal justice system.

We are leading off each episode of the series with Tarra’s voice. Her story reminds us that this issue is not just about numbers, it is about human lives. It is also about hope.

Here is Tarra Simmons, with more of her story.

TARRA SIMMONS: Before I was ever arrested, I knew that I was struggling with substance use disorder, but I didn’t know where to go to get help. I was really afraid that if I told my providers that they would turn me into the nursing board, and I would lose my nursing license. I didn’t know where I could go without also getting negative consequences, and that’s part of the problem is that we stigmatize substance use disorder in such a way that it is illegal, and that there is negative consequences, and it affects other areas of your life such as losing your children or losing your professional license or something like that.

So, if there had been a place where I could go, where I could be met with other people who are also struggling with substance use disorder, and it was a safe place, or people that have maybe overcome. So now I know about 12-Step Anonymous programs and I’m a part of that, but at that time I didn’t know about that. I had never known a person that actually had beat addiction and had been in recovery. Now I know lots of them, and they’re my best friends.

But this is the problem that we’re struggling with right now is we’re looking to actually treat addiction as a public health issue instead of a criminal legal issue. Oregon just recently passed legislation, or an initiative to decriminalize drug use. And we’re trying to do something like that here in Washington state also following the lead of other countries like Portugal and Switzerland and other places.

If I could’ve went to the police and said, “Here’s my drugs. I need help and hope, not handcuffs,” [laughing] I think that would be really great.

IVAN: Tell me about what happened when you went into prison. You had one son, but you’ve talked about having kids. Did you have more than one child at the time?

TARRA: Yes, I had a second son in 2003. He is 17 now and a senior in high school, Dominick. He was about eight years old at the time and so, when I went to prison, I had my son Davon, who is 18 and had just received a football scholarship to Central Washington University and had moved over there and then I had my eight year old son who was residing with my now ex-husband. It was a really difficult time not just for me but for them to see their mom go to prison. I never thought that I would go to prison, honestly, and there I was.

IVAN: How long were you inside?

TARRA: I received a 30 month sentence, and I did 20 months.

IVAN: So, you get out, it’s been 20 months later, you’ve served two-thirds of your sentence, you have to try and find a job, and you’re now a felon. Is that right?


IVAN: What did you wish you had had in those first few weeks out of prison that could’ve helped you that you didn’t have?

TARRA: I went to work release four months before I left prison, so I still count that as part of my period of incarceration but it’s like a half-way house where you can actually go in the daytime and look for work. So I ended up getting a job at a fast food restaurant.

But what I really wish I would’ve had when I left prison was not a criminal record that continues to haunt me to this day. It kind of kept me out of qualified jobs that I could’ve done that made a lot more money. So basically the criminal record itself is kind of a punishment to lifelong poverty. And what I really wish I would’ve had, is not a criminal record. You’ve done your time and it’s hard to survive after that.

IVAN: Because no one’s taken your registered nurse and your Bachelor of Arts degree away from you. You’re still qualified. You still have all the skills you learned in college. You can still theoretically be a nurse. Now I don’t know anything about licensing, but it sounds like this is a lifelong sentence that you now have on your record that makes it so much more difficult. And you said that you’re a felon as well. Does that mean that you can’t vote in the state of Washington? What are the laws around that?

TARRA: I want to go back and say yes I was able to keep my nursing license through all of this, because the nursing board does have a program for people suffering with substance use disorder. And as long as you comply with all of their rules and their random drug testing and everything, I was able to keep my RN license and still have my Bachelor’s degree.

But you’re right, the laws in Washington state don’t allow people to have unsupervised access to vulnerable adults or children if you have felony convictions. And so that’s the law that I’m trying to change now. But the voting issue also is, if you are on community custody, or if you are too poor to make payments on your legal financial obligations, you do not have a right to vote in Washington State, and so that’s something I’m also working on right now as a legislator is removing those barriers and allowing everybody to vote as soon as they come home from prison.

IVAN: Tell me about the financial consequences of being in prison. How does it affect your ownership of a house? Your credit rating? The debt you have? The legal obligations you have? What did that do to you, that time inside, purely from a commercial perspective?

TARRA: It totally ruined me. I had bought a home when I was a registered nurse and so being unable to make my mortgage payments, I lost my house into foreclosure. Everything that I had, car was repossessed, all of my credit card debt went into collection. It totally devastated me financially and I ended up having to go bankrupt.

In addition, the court always gives you fines and fees when you’re convicted of a crime, and so in my case, I had $6,000 in fines and fees and that was not to go to any individual victim. It was just to fund the court system, and it incurred 12% interest the whole time I was in prison. So when I got out, they were garnishing my fast food minimum wage job paycheck to pay these court fines and fees that can never go away in bankruptcy. So, again, another issue I’m working on now, [laughing] as a legislator.

IVAN: We’ll hear more from Tarra in our next episode, but before we begin to talk about the role of data in helping government reform criminal justice, Tarra did weigh in on the importance of public/private efforts as we work for change.

TARRA: I live in Washington where we have this wonderful organization called the Washington State Institute for Public Policy that does a good job of getting our data and researching it and providing good guidance to the legislature. If we could do that everywhere, utilizing our partners in tech and academia to really help guide decision-making, I think that is the way forward.

Every state and the federal government should be thinking about these kinds of decisions instead of our feelings of what is justice, because justice is different for everybody, and everybody has different feelings. So, whatever we can be doing to partner I think would be great as we move forward.

IVAN: Tarra’s story reminds us of the slippery slope people face when issues like trauma and addiction send people into a downward spiral, and the difficult path they face trying to turn their lives around.

So what are some ways we can help break this cycle and help people stay out of prison? That may be where data can help. We need to learn which programs work and where we should be investing to keep people out of the criminal justice system, or to help them stay out after they serve their time.

Unfortunately, government decision makers have not always had the best data to work from, and the policies that are developed are often influenced more by perception than the facts.

To help us explore why this has been happening and what we can do about it, I’m pleased to welcome Andrew Warren, the head of product at Recidiviz.

Andrew, welcome. Would you mind giving me a little bit of background as to what your role is at Recidiviz.

ANDREW WARREN: It’s great to be here. My role at Recidiviz is basically looking at what pieces of technology we want to build that we think will have the greatest impact on the criminal justice system and will be most helpful in helping folks who are under the system to be successful and to reenter society.

My work is largely focused on helping to set the direction for the products and tools that we help to provide to the system, as well as to help get the feedback loop tight between us and the folks who we’re serving. So that’s folks who are on supervision, folks who are in the system as well as the folks who are running the system, all of whom we work with to try to help create a better and more equitable system in the future.

IVAN: What comes to mind when you hear Tarra’s story?

ANDREW: A lot of things come to mind. Data is helpful but data is a tool at the end of the day, and what really matters a little bit more than the data itself is how that tool is used. In the justice ecosystem there is not a lot of good data at the moment on what the flows through the system are, what’s working to help people reenter society safely and what’s not.

I remember talking to a state when we first started working on this, and they said we have 90 in-prison programs, and we have the ability to try to evaluate two or three of them per year. So, some time in 30 years or so we should have a sense of what was working.

IVAN: Ouch

ANDREW: [laughing] On the flip side to give you another counter example, there’s folks who have been trying to create algorithms to assign risk to people within the criminal justice system. And I think one of the things, especially coming from big tech, we’ve been seeing the dawn of a lot of improved machine learning methods, is we tend to think of technology as telling us the truth. The calculator tells us two plus two is four, and that’s neutral and objective, and so we trust it. But as the new machine learning techniques are something that are really teaching computers to learn the way that humans do it. Training them with a lot of data to recognize patterns.

When computers start learning the way that we do, we need to start questioning them the same way we question other people who have opinions about things.

You take into consideration you’ve told me that this thing is true, but who told you that, what information are you basing it on, what are your life experiences? And you then tint that a little bit in your mind when someone comes to you with a piece of information. So, these risk assessment tools are getting into that territory of kind of they’ve been trained with data that is based on a system that we already know has a lot of biases in it, and so we should be questioning that data, And we should be questioning the use of that data. It’s tough and it depends on how the data is used.

IVAN: One of the things that struck out at me at what Tarra was saying is that, and I’m paraphrasing here, Good data should lead to good decision making, and that should lead to good government. And as you said the data is just a tool, it depends on how it’s used. What do you think has been the roadblock, the barrier to getting good data into decision makers’ and policymakers’ hands?

ANDREW: I think there’s a few different pieces. One piece that’s actually a pretty significant blocker is just that government data systems are often very, very far behind. A lot of states we work with still have data on mainframes. A lot of them are still using homegrown systems that they built in the nineties or early 2000s. Then there’s the fact that there’s no central decision maker. There’s no one person who’s responsible for all the outcomes.

So that’s challenging because it means that every actor in the system doesn’t have the information on what’s happening in other parts of the system, and as a result tends to assume that the problems are coming from those other parts and not from them. Because no one is ultimately responsible for the performance of the system as a whole.

IVAN: How do we address that? Why doesn't the government have the latest technology? Why can’t they push an update to me when I need to be in a certain place at a certain time, [laughing] right? I want my vaccine, just send me a notification on my phone when it’s ready. Why?

ANDREW: That is why I got involved in tech and government five or six years ago. I was working happily on the consumer side and started getting really frustrated with the fact that every time I needed to do anything from paying taxes to figure out when the garbage collection is, the tools available from the local government were not great. It all seemed so solvable from the outside. The apps on your phone update three times a week with new features and bug fixes and all of this and are easy to use. I think it comes down to a few different pieces.

One piece is just there’s different risk tolerances at the end of the day. I think the government, quite rightly, feels that it is not acceptable to not deliver its services to anyone, and so it has a very low risk tolerance.

It comes from a few different things. It comes partly from the fact that the folks who are in the positions of power within government are not going to be fired for not doing something that was new and innovative, but they might be fired if something blows up in their face, and actually embarrasses the government. So, there’s a lot more downside risk than upside opportunity.

Another piece of it is just that it’s okay for your app in the app store to not work for half a percent of users. It’s not okay for social services to not work for half a percent of people within the community. And so there is a very legitimate reason why government needs to be more careful about the kinds of things that it’s putting out there.

The other piece, and maybe one of the more significant pieces is optics. Government doesn’t want to be seen taking risks with taxpayer money, because it doesn’t want to find itself in a place where something doesn’t go right, and that leads to people howling why did you spend ten thousand dollars on this thing, a hundred thousand dollars, a million? That leads again back to this kind of risk intolerance.

There’s another kind of fairly large piece here to which is just the procurement process, which makes it very, very difficult for the government to work with. Honestly anyone who would be fast moving in the tech sector, but that’s a large topic onto itself.

IVAN: Procurement is an uphill battle isn’t it? It’s something that 18F and the United States Digital Service is trying to address, but it feels like it hasn’t permeated the government completely yet.

ANDREW: I think it’s the perfect example of a system that was built with very good intentions, that’s doing a tremendous amount of harm candidly. The notion of procurement is that you don’t want every contract for your city to be won by the mayor’s brother and his companies. That’s what procurement is meant to stop. It’s meant to be a way for the government to say here’s what we need, and then give us your bids for what you think you can provide it for in terms of cost and we’ll take the lowest bidder, and then we’ll have the thing that we need at the least cost. Super reasonable.

What it turned into is a system whereby the user ends up describing the solution instead of the problem. If you think about why apps get better day after day in the competitive market, a large part of it is that there’s a known problem, but no one has the idea of exactly what the solution is, and so people try lots of different things.

Gmail came along at one point and said what if email was web-based, and what if you had a lot more storage and these kinds of things. But the problem was the same, people just want to communicate, and they don’t want to have to delete everything all the time, and everyone has their own interpretation of the problem statement.

Henry Ford said this, If I ask people what they want they would’ve said a faster horse. They wouldn’t have come up with a car. That’s a good example because what the government does is it goes out in procurement and it says, This is literally the solution I want to see. I want this very custom thing that is built just for me.

And that’s not how technology works at the end of the day. Technology is you build it once, you sell it to a million people. If you’re building it to spec, and that spec was built by people who focused on their idea of the solution rather than the problem, you’re going to get very rigidly defined technology, and it’s going to get so closely interwoven with the requirements of all of their other very highly specific pieces of technology, that no one is ever going to be willing to come in and try to replace that thing. That’s why these mainframes are still here.

IVAN: Do you think that the size of projects also influences and perpetuates this procurement problem we have?

ANDREW: It’s funny you say that. I think the example was a good one of that kind of issue where there was this giant monolithic procurement process for the digital solution for a healthcare marketplace. It’s interesting watching from the outside, because in large tech you would never try to procure your entire solution from one vendor. One, you’re putting all of the risk in one place, but two, you can’t replace that thing. If you don’t like the vendor a few years later, you’re in real big trouble.

So you normally try to put together, I need this functionality over here, and I’m going to put out a bid for that, and I need this functionality over here and I’m going to put out a bid, this functionality over here. And as a result every time you start realizing people don’t use landlines anymore, I don’t need the phone I need email piece of functionality, you can just swap that part out, and you can replace that part of the contract or that particular vendor if they’re not doing a great job. In government it’s giant monolithic contracts for very tailored custom technology that tries to fit in with the rest of the ecosystem within that particular government.

IVAN: Let’s talk a little bit about the criminal justice system and Recidiviz, a nonprofit that is small compared to Google, Yahoo, Oracle, all these large tier one government vendors. What are the barriers that a smaller organization, a nonprofit has to making the changes that you are missioned to be making in the criminal justice system?

ANDREW: There’s a lot of challenges here. One piece as I mentioned is, technology is usually built once and sold to a lot of people. And so even though it’s very expensive to build, it has pretty high margins, and it’s affordable as a result of that. In the procurement process, since it goes to the lowest bidder, you’re incentivized to keep your bid low. What that comes down to is someone will win a bid with the expectation of trying to hold onto that contract forever, because they’re not actually going to make money off of the first contract, they’re going to make money off of the second or the third or the fourth.

And so, they almost want that system of building deep ties into every other state system that makes it so difficult to remove that thing, and to replace parts of it. And so, as a result you have very strongly entrenched vendors in a lot of government technology.

At Recidiviz we don’t even try to replace any of those vendors. We very much are focused on providing new value in a way that just sits on top. It’s unfortunate because a lot of departments of corrections are stuck with an inability to use modern technology to reach, for instance, people on supervision who might need a reminder that they need to come in to see their parole or probation officer. But there’s not an easy way to swap out the system so they’re responsible for holding all of that information and data to date today to do that.

IVAN: What do you think tech needs from government to make this process better and for the results to be better? What do you wish is in place right now so that Recidiviz could have a better shot at fulfilling its mission across the nation?

ANDREW: That’s really tough. What we really need is a focus on outcomes, not on processes and not on operations. If every agency focuses right now on, Are we fulfilling the number of contacts that we’re required to when folks are on supervision? Am I checking the box on the set of programming I was required to give this person. But they’re not seeing the outcome metrics and they’re not creating a feedback loop to those metrics. All we’re doing is kind of the system serves itself, it doesn’t serve the people who are a part of it.

So, what would make the job both easier for Recidiviz but also hopefully eventually make Recidiviz irrelevant, would be if procurement was about the problem that you were asking organizations to solve, instead of about prescribing the solution you want them to take to solve it, so that you could see more innovation, you could get more ideas, and you could be a little bit more open to different organizations coming in.

If the criminal justice system could cross all of those different silos of the courts and the police and the corrections and supervision folks and get data in one place that allowed them to see what was the end-to-end outcome for the people, then I think you’d see a lot more focus on getting folks IDs as soon as they get out of prison. Getting folks, the resources that they need to get set back up in a society that they left twenty, thirty years ago before there were even smartphones. There’s just not that much focus on the human side of the issue because no one’s looking at the outcomes a lot of the time

IVAN: What you alluded to there was something we talked about in the previous episode, which was that the criminal justice system is so fragmented across all these different local, state, federal environments. There’s local courts, there’s state courts, there’s local law enforcement, there’s federal. How do you fix this fragmentation problem when that’s a human problem? It feels like this human problem needs to be fixed first, [laughing] because you can address the data problem.

ANDREW: Yeah. I do think firmly that almost any issue is not a technology issue, it’s a cultural issue first. That was my main take away from Sidewalk Labs. We look at everything, transportation issues, criminal justice issues, housing issues. Everything was really rooted more in what the culture found okay, than it was in a technology problem or solution.

Take a step back to the fragmentation piece. I do think federalism is brilliant in that it gives us all of these petri dishes to try different things at the local level. But the fact that government is so risk averse and doesn’t feel empowered to make mistakes, makes it very difficult for those petri dishes to actually produce interesting results.

I think it’s fascinating that the level of incarceration across the United States is uniformly above any of the developed countries outside of the United States. That is a fascinating thing. It says to me that the experimentation framework that’s supposed to be working here is not, and all it’s doing is creating this disjointness of data that makes it difficult to see the picture of what’s actually happening.

IVAN: The pandemic and COVID over the last year has upended the whole system hasn’t it? Clementine was talking about how it really forced, I think it was early release, or there was some things that changed as a result of the pandemic that now enabled you to get better data so that you could make better decisions. Can you talk about that a bit?

ANDREW: I think the thing that was really interesting with COVID is for a brief moment in time we saw every independent actor in the system all point at the same direction at the same time, which is impossible in such a distributed system, without something like a national scale pandemic.

So, what we saw is remarkable. We saw twice the reduction in the incarcerated populations in the U.S. that we had seen in the entire last 10 years of the criminal justice reform movement, which has been bipartisan.

The fact that we could achieve that so quickly as a society is indicative that there’s a coordination problem more than there is lack of agreement. If this moment could actually have pointed all of police and prosecutors and courts and everyone else towards a point where they said, You know what, anyone who doesn’t strictly need to be in a facility right now, where they’re crowded with other folks, and even shouldn’t be in a jail, doesn’t have to be in a prison, should not be. And then all acted upon that, and we saw a truly tremendous result.

So, I think it speaks to the fact that if you could get more tools into more places that start to coordinate those actors and have them facing the same direction, having the same understanding of what’s going on and what they need to do as their part of the picture, there’s a lot more that we could do.

IVAN: What actually works with government? We sort of talked about, here’s the procurement problem, here are the big players, here are the things that we’re trying to do. But there’s this perception that there are sharks in the waters, right? That there are these different incentives for different organizations. What actually works with government? What’s been successful?

ANDREW: I can mainly talk to the technology side of that. On the technology side I think we’ve seen the beginnings of some significant changes on parts of this. There’s things like approved vendors lists that try to get around the procurement process and make it a little bit easier to not be such a gigantic lift for both government and for the private sector every time there’s an opportunity to do something new on the technology front with government.

I think we’ve also seen some examples of things like government digital services or U.S. Digital Services or 18F, which are really kind of an influx of folks from the more mainstream tech ecosystem, who have a lot of insights into where and how tech works well and what helps things to move fast and to focus on the needs of the user at the end of the day, moving into government to try to help reorient how some of these things work. I think that that’s incredibly powerful and is having already a significant impact of the national conversation on the government side of how to work with and around technology.

The last piece that I think is really impressive is, we started to see some open data standards for things that involve governments. I think this is one of the more important things that comes to mind when I think of the coordination problem between all of these different levels of government. It’s only a problem so long as they’re locked into vendors who don’t want data to go outside of their little ecosystem.

So, it’s kind of a question of one, how does the data get to a place where it could be shared with others, naturally in a privacy sensitive way. And then two, how do you drive alignment between different government actors so that they’re using those same standards?

A couple of examples of that, I think it’s Title 21 in California, is the actual protocol for road tolling in California. It was actually written into the law, the protocol was, as a way of saying, here is going to be the protocol for how vehicle transponders talk to the infrastructure in order to communicate who they are and when to be told and these kinds of things. And that creates a really good and powerful tool that any vendor can pick up and now be a part of that ecosystem, as opposed to just the one or two giants that had been there to begin with.

Another example is something that was originally called the Google Transit Feed Specification (GTFS) and has now been generalized into just the General Transportation Feeds Schema. This was way back in 2005 that folks started realizing again, Why is it that Google Maps on my phone can direct my car, but I have no idea where the transit system is at the moment? I don’t know when they’re going to come and pick me up.

And the answer was, every single city in every single state had a different way of publishing that information, and nobody had the ability to go and try to pull it all into one place, so that you could actually surface it in any of the ecosystem of apps.

Some folks at Google and at one of the transit authorities started talking and decided on just a common way to publish this. But as soon as they put that out there and said, Any company that wants to can pull this from a particular city and see this feed specification for here’s how we’re publishing our transit info. That made it so that suddenly the entire ecosystem, not just Google but everyone, could start pulling these feeds and seeing when trains were coming into what stations, and these kinds of things. That’s why that still is underpinning Google Maps, Apple Maps, every different consumer application for transit information. It’s the way governments share data in a consistent way.

IVAN: So, I think your point is subtle and nuanced, but I want to try to paraphrase it so that I want to make sure I got it right. You’re basically saying if we create open standards in a small part of an ecosystem, for example what California did with their vendors. It’s not just that any vendor can now become part of the system, it’s that that open standard not only applies to California, but someone in New York, or Florida or Minnesota could pick up that standard and use it in the transit system there.

So, what you’re saying is by creating an open standard you’re able to assimilate, and you’re able to unify all these different systems.

I want to take it a step further and ask you, how does it apply to the criminal justice system? Is there a need for an open standard of some sort, more than one open standard that allows local courts to talk to state courts? That bridges the gaps between all these different systems in the criminal justice system?

ANDREW: I think that is the best hope for starting to organize this, is there needs to be consistent ways to publish information, whether it’s to the general public or whether it’s just to other agencies that calls a pear, a pear, and an apple, an apple. One of the things that actually boggled my mind at the very beginning of when we started work on Recidiviz was someone said to us, there’s actually no way to know which states are doing a better or worse job of rehabilitating people in their criminal justice systems, because none of them publish any consistent metrics around this.

The one metric you would think would be the best kind of way to measure this, recidivism, the rate at which people come back after they’re released, is defined differently in every state. And so, you actually have no way to know which states are doing well and which ones aren’t.

We initially thought that was because they didn’t want to, but as soon as we got into the system and started working with it, we realized they were as hungry for information on how well they’re doing compared to other states and what was working and what wasn’t as anyone else was. Open data standards are the only way that you could, for instance, get jail data from every single county in a state to be visible to the criminal justice system downstream that’s going to need to see what’s the volume of people coming in, how many of them are we actually diverting successfully to programming for mental healthcare, or for substance use, these kinds of things.

IVAN: Whose responsibility is it to create that open standard?

ANDREW: I think this part becomes really interesting. One of the lessons of open data and government from the last 10 or 15 years has been that you have to have a feedback loop for any data to be useful. So, if you just put out a dashboard, if you just put out an open data set on GitHub, you’re not actually creating something that’s meaningful, because there’s no user on the other side of it. And if there’s no user, someone might discover it and use it. But honestly the data is probably crap, because no one is using it, and no one is there to tell you which parts aren’t working or seem to have problems in them. So, it just becomes this giant pile of data dumped out the door that’s not actually useful to anyone. And then anyone who starts to approach it says ugh.

So, I think it's really important that you start with a use case in mind, that you have someone who is on the receiving end of that data and needs to use it for some critical part of their own work. In terms of who is responsible for this in the justice system, I think the justice system would be the best customer of its own data.

If the corrections system could actually understand who is in jail at the moment, are those folks who we have on our probation and parole at the moment, they don’t even know if someone on probation or parole is arrested. They sometimes get a call, they sometimes don’t. They have software that just tries to match names against reports coming in from the police just in case, because there isn’t actually coordination between those parts of the system.

So, I think it’s important in building those data standards and actually applying pressure to the vendors to implement them and saying we’re not going to go with your software anymore if you don’t build this. But there’s a user on the other side who’s starting to incorporate it into their work.

IVAN: Do you know of any open standards that exist within the criminal justice system right now?

ANDREW: [laughing] That’s tough.

IVAN: Sorry to put you on the spot. [laughing]

ANDREW: I’m trying to think of any. No. I can think of reporting programs that try to ask folks to report things in somewhat consistent ways, but even those are years long lag. Two or three years.

IVAN: If you can’t decide on what the definition of recidivism is, how are you going to decide on an open standard that defines something where you can track it?

ANDREW: This is a large part of what we’re trying to do. We’re trying to create the tools that use that information, so that it becomes part of the day to day operations. If we can provide tools that help the department of corrections to see peoples flow through the system, and whether which particular parts of the system, or which programming or which pieces are helping people to succeed and are working, and which one’s aren’t, and whether there are bottlenecks in that system, that creates a strong urge to actually have more information from other parts of the system.

And if those other parts of the system also see the benefit to that, now we have some visibility to the outcomes. As a parole board I don’t see what happens when someone leaves after I make my decision, and I don’t know whether I’m doing a good job of making those decisions or not. As the police I don’t actually know if the person who I’m picking up is going to a diversion program or to prison, and whether one of those things is actually helpful more than the other. If those actors at the moment don’t see any value in it, they’re not trying to create data streams between themselves. I think if we could provide them a way to see those things that will actually be helpful to them, they’ll start wanting that interoperability as well.

IVAN: It almost feels like, your organization is the silver bullet. Because if you’re working with a jail system in South Dakota, and you’ve come up with a definition for an open standard that you’re just going to use in South Dakota. And then now all of a sudden, you’re working with a system in Tennessee, if you already have a published standard or a working standard you could bring it to another state. And all of a sudden you become that potential unifier. [laughing] Am I giving you too much credit here? It feels like you have an opportunity here.

ANDREW: The way that I look at it, and this certainly won’t be the way everyone looks at it, but the motivation is there. The folks who run the system, even on the political side who are responsible for policies around it, are pretty well intentioned folks and everyone’s incentives are aligned. They want more people happily contributing to their communities. They don’t want them in cell blocks some place. They want them to successfully reenter society. They also want anyone who has a mental health issue or a drug issue to get the treatment they need, rather than just be stored away some place.

But the challenge is there isn’t a way for them to know any of that. We’re just a very small part of this at the end of the day, and there’s a lot of folks who have been working on these problems a lot larger and a lot longer than we have. But we see it from a little bit more of the technologist lens of, if we can create the demand for that data, the tools that make it actually useful to folks, then the supply of that data will provide itself.

So, we haven’t focused yet, although it’s the dream, on actually getting to a point where we can get folks wrapped around open data standards. The place where we’re starting is just by getting data into folks’ hands, however much effort it takes us to do that, so they start to see the need for it and how it can help them in their work.

IVAN: You have a really noble mission and a really difficult job in the criminal justice system and with the mass incarceration that we see in the United States. What are you hopeful about? Why are you working on this? Why do you keep working on this?

ANDREW: I mentioned my first foray into city stuff was at Sidewalk Labs, and it was largely because they had this habit of wanting to have a half and half team, half technologists and half people who actually had worked in cities and governments and knew the problems that they faced. So that you didn’t just have tech people proclaiming that they knew the answers to all society’s ills. But the longer I spent there the more I realized that these issues are largely not issues of technology, they’re issues of culture.

There are countries where homelessness is not much of an issue, but it is because the society is willing to invest heavily in helping people who are down on their luck or have other issues to be in housing and to actually have a safe and stable place to live. There are plenty of countries that have taken care of transportation issues, but it’s by making conscious decisions around the benefit of society versus the benefit of the individual and building up instead of outwards in some of these pieces. It was a little disheartening as a tech person, because you didn’t really see technology entering the equation much.

The thing that really drew me to Recidiviz and to criminal justice is increasingly not the cultural issue that it once was. There is broad agreement, although not everyone for the same reasons, but there is broad agreement that the system is a reflection of historical racial issues. But that on top of that, it’s also a system that everyone wants to unwind, that everyone wants to get back to a place where it’s more about safety and actual equitable justice as opposed to just about warehousing people.

So, given that, the culture issues are actually starting due to a lot of hard work of a lot of groups over the last 10, 20, 30 years, they’re starting to unravel, but there’s still this coordination problem, and that is a place where technology can help.

So, what I’m optimistic about is I do think that criminal justice has turned a corner. It has been on a slight downward trajectory in terms of incarceration for the last decade, but there’s a lot more to do and that’s something that we can all get behind.

IVAN: Wow. That sounds amazing. I want to solve the problem as well now.

ANDREW: [laughing] Excellent. Come join us.

IVAN: [laughing] I’d love to. I really am so thankful for your time, thank you for spending it with me today. Thank you for joining us on this podcast, and on this series.

ANDREW: Thank you. It’s been wonderful.

IVAN: Our series will continue next week when we’ll hear more from Tarra Simmons, and we’ll speak with Julia Dressel, a software engineer at Recidiviz.

Julia will help us explore the idea of bias in technology. In particular, she’ll explain how racism and bias that has existed for centuries can impact data and lead to decisions that perpetuate inequality.

Here’s some of what Julia had to say:

JULIA DRESSEL: So we have this system that has policed over policed, predominantly black people, and then over sentenced and over incarcerated. When I say over, I mean disproportionately we are incarcerating black people and other people of color in our country.

So that’s a really important context to think about when you are trying to build any tool that’s going to use historical data about the criminal justice system to make any prediction about what’s going to happen next.

IVAN: Join us next time for the third episode of our series, Meeting the Moment: Using Data to Reimagine Criminal Justice. We hope you’ll subscribe! You can find out more online at Thank you for listening.


This is Episode 115 of The TEN7 Podcast. It was recorded on March 19, 2021 and first published on April 21, 2021. Podcast length is 43 minutes. Transcription by Roxanne Chumucas. Summary, highlights and editing by Brian Lucas. Music by Lexfunk. Produced by Jonathan Freed.

Please rate our podcast! Doing so helps spread the word about the show. Just pull it up in the Podcasts app and scroll all the way down, hit the stars and you're done! Thank you.

Continue Reading

Subscribe and listen wherever you get your podcasts.