Bringing the ‘Mindful Use of Technology’ to Policy Design

A lot of hands touch each product, service, and policy that gets designed at Code for America. From client research to marketing, everything we create goes through a collaborative process full of feedback loops, iteration, and countless Zoom meetings. We want to hone in on one particular part of that process today: how we ensure that policies with technical components are designed mindfully, with an eye towards every step of the process that has to come after a bill passes. 

On the Clear My Record team, our researchers, data scientists, solutions engineers, service designers, and policy program staff work together to support states in designing and implementing automatic record clearance—because such a process removes burdens from people impacted by the criminal legal system and increases the chances that they can live fuller lives with better access to housing, jobs, and services that can help them succeed. Our team provides technical recommendations on automatic record clearance process and eligibility to state advocates, policymakers, and government to ensure that these laws are impactful, implementable, and equitable.

Recently, we sat down with three of our staff members on this team to learn more about their work designing policy for successful implementation: David Crawford and Candice Millsap, both Senior Program Managers, and Julia Oran, a Staff Solutions Engineer. 

Can you tell us a little bit about what it means to design a policy with implementation in mind?

David: When we’re designing policy, we want to ensure that new laws lead to real outcomes for people. With automatic record clearance, we want people’s records cleared automatically by their government, as soon as they’re eligible under the law—rather than the traditional record clearance process where a person has to petition for expungement on their own. To make an automatic record clearance policy that’s actually implementable, we need to hear from government employees in criminal legal system agencies, as well as people living with criminal records, to ensure all their needs and constraints are heard and considered in the solution. A great deal of this work is relationship building—it’s sitting down with the people involved in this process, hearing their concerns, and telling them what we’ve learned about what works well and what causes challenges from our experience working with more than 20 states.

Candice: Take cannabis legalization laws—since many states want to make sure that previous records get cleared when something gets decriminalized. If you write a bill that just says “this isn’t illegal anymore” that doesn’t give a lot of instruction on how to make that feasible in the real world. In the case of automatic record clearance, there’s no “decriminalized” button you can press to magically seal all those records. So some bills that are most successful map out the process really clearly: here’s the agency that identifies eligible records, here’s how they’ll send eligible records to the next stage, here’s how people will be informed that their records are cleared. 

 To make an automatic record clearance policy that’s actually implementable, we need to hear from government employees in criminal legal system agencies, as well as people living with criminal records, to ensure all their needs and constraints are heard and considered in the solution. 

From a technical perspective, what’s important to have clearly spelled out in a record clearance bill?

Julia: The technical language of bills like this can be challenging—we’re often walking the line between not having enough details in the policy and being overly prescriptive. But the basic technical understanding we need to have is: who’s eligible to have their record cleared, what data do we need to find them in the systems, where does that data live, and what systems need to be updated to clear those records? We do a lot of research about major agencies in the state that keep criminal history data, like the police, courts, and district attorneys. We see a variety in the types of systems they use to maintain criminal legal system data, and have to account for that in our recommendations—sometimes they use systems from a vendor that we recognize from another state, or they might all be using different or in-house systems to store that data. So we do stakeholder interviews and ask people who seal records what steps they take in order to better understand the technical components and complexities. 

Candice and I recently went to a state where we observed a court employee review and seal someone’s petition for expungement. He sealed the record in three different criminal history systems—so we saw each interface involved in this process, how long it took to manually seal a record, and all the nuances that went into locating a record and cross-checking the systems.

At Code for America, we often say that ‘the mindful use of technology’ can enable government to serve people better. How are you employing that attitude in the policy design process?

Julia: As an engineer, I think the most vital thing for mindful policy design is having proactive conversations with the people responsible for actually implementing the policy. If not, you pass the bill and implementation will get held up for years as people figure out what to do with it—and ultimately, the people you were trying to help won’t get the benefits promised to them by that policy. 

Candice: Thinking about implementation is one way mindfulness shows up in policy design, but there are other ways it appears, too. We’re using human-centered design principles here—because we don’t want to put an automated process in place that creates more harm. I think mindfulness in this process for me comes down to how a policy will increase equity and decrease trauma. That starts with data. The people most impacted by the criminal legal system are Black and brown, and so we want to see an awareness of that show up in any eligibility guidance we’re writing so that these communities get to see the benefits of this policy. Our data science team is really skilled at taking eligibility criteria for particular records and breaking down the impact by race and ZIP code—basically, “if we include this offense in the bill as one eligible for expungement, it will most deeply impact these communities.” We know, for example, that making a requirement for all fines and fees to be paid off before expungement will exclude Black and brown people. We know that if you focus on clearing the smallest offenses, you’re writing a bill that will primarily impact white people. The mindful use of data alongside technology helps us make the most impactful decisions possible. 

 We’re using human-centered design principles here—because we don’t want to put an automated process in place that creates more harm.

Speaking of data science—what skill sets are needed throughout this process?

Candice: When we get introduced to a new state, Code for America program staff do a lot of education and trust building with government agencies. We talk a lot about the transition from a petition-based and individually-initiated process to an automated and government-initiated one. As we get more engaged with the agencies, we start to bring in solutions engineers because they have more expertise around how to handle technical solutions in complex systems. At the same time, qualitative researchers are talking to directly impacted folks. Throughout the engagement, we utilize data science—because data is the quickest way to move the room when we hit a decision point. The whole process is a team effort.  

Julia: All that trust building that Candice and David do is critical to my job when I come in as a solutions engineer. We have experience in a lot of states by this point, but we have to remember in every engagement that these people are the experts in their states. Sometimes agencies responsible for implementing these policies feel worried about someone else changing their established workflows without consulting them. It’s empowering to give people a voice in how something changes, right from the beginning.

David: I also want to emphasize that it’s not just work by our team at Code for America and government. We’re fortunate that the record clearance space is filled with people who are organizing, building coalitions, lobbying, and doing all the on-the-ground work to make sure a bill passes. Our national partners at the  Clean Slate Initiative lead a lot of this important political and movement-building work. We create a niche value-add as the technical and implementation experts to ensure that these bills will have the intended effect. 

What helps build confidence with government partners so that they feel they can build and operate a new process?

David: There was one state we were advising that was on the cusp of passing cannabis legalization and expungement criteria along with it. The agency that was in charge of initiating record clearance said it would be impossible to do this automatically—the best they could do was remove roadblocks in the petition-based process. So we met with the agency and we told them about how other states had approached the same challenges they were facing. After that meeting, that agency went back to the legislature and said an automatic process would be possible, they’d just need some additional resources. The bill passed, and a few months later, those agency leaders were talking to the press about how proud they were to be implementing an automatic process. We didn’t actually do any hands-on implementation work in that scenario. But we showed them what’s possible, and that’s one way to help create impactful policy. 

Candice: The most rewarding part of this whole process is watching the transition from nonbelievers to passionate adopters. Watching a state clear a million records in one day when implementation goes live is such a special moment.

Policy change is undoubtedly an uphill battle; there are so many headwinds to getting a bill introduced at all. So naturally a lot of attention goes into getting the bill passed. But we’re shifting that goal post so that it’s not when the bill passes that we celebrate victory, but when it gets implemented successfully. 

How do you hope your work changes people’s views on policy design?

David: Five years ago, very few people were even thinking about automatic record clearance. Now, at least 10 states have Clean Slate laws, and half the states that have legalized cannabis have automatic record clearance as a component of those bills. We’ve gained a lot of knowledge from this experience; we notice patterns in how government agencies manage criminal record data now and can more easily make recommendations. Automatic record clearance doesn’t feel like science fiction anymore. These laws are getting passed across the country. When we focus on implementation during the policy design phase, we’re trying to challenge everyone in the room to think beyond just the language of the bill. Policy change is undoubtedly an uphill battle; there are so many headwinds to getting a bill introduced at all. So naturally a lot of attention goes into getting the bill passed. But we’re shifting that goal post so that it’s not when the bill passes that we celebrate victory, but when it gets implemented successfully. 

Julia: And that’s not just about record clearance. This is true for any policy with a technical component. You have to think about every agency, every caseworker, every system full of messy data. You can’t decouple policy from technology anymore—so we need people to think about it mindfully, and I think our work is showing what’s possible when we do that.