Sure. Thank you for having me. So my name is Emily Paul. I'm currently a policy analyst at Upturn in Washington, DC, which is a nonprofit that does research and advocacy around equity and justice issues in technology. And I generally would say I consider myself a critical public interest technologist, and that I'm interested in looking at how technology is designed and used and how that intersects with existing systems, and areas that technology is being implemented, and how that affects people's lives, how that can cause harms, and how we can try to, challenge that or have, technology that actually enhances people's well-being instead of undermining it.

Yeah. So I think seeing sort of continuing what I was just saying, seeing the role that technology plays in mediating people's access to resources and opportunities, like, the role that it plays in distributing those things in our society, and seeing how, historically and presently, there are a lot of issues with how we distribute resources and opportunities in our society that are much bigger than those technologies, than those technical systems, but that really intertwine with those. And, so I guess, ultimately, my why is is to have a more equitable distribution of wealth and power and resources in our society. And sort of based on my background and interest and skill sets, I see tech as technology as one place, to look at those issues and to because of the increasing role that technology is playing in that distribution to try to unpack what's going on and and try to challenge and shift it. The other thing, which I guess is is related, but I would say it's part of my why, maybe a sub part, is just wanting to see more collective decision making about the design and use of technology, and trying to understand what does that actually look like to have more people involved in in shaping what technology looks like and how it gets used.

So I started out, in nonprofits doing fundraising and communications work. I didn't have a background in technology. And then at one of the nonprofits I was working at, the New Press, I started learning a lot about social justice issues there. They published the book The New Jim Crow and a lot of other, important books on criminal justice issues and other, other, yeah, social justice issues. And then I also started working on technology project management there and got really interested in how so in that case, it was a database and a website, how those things were designed and what that meant for how people use them and how that affects people's work or affects how people engage with the organization, which led me to go to grad school in information science, a little bit of a leap just thinking, this seems like an area where I can look at, these issues that I'm interested in around. How do we organize information, how do we classify things, how do we build systems based on those classifications, and what are the ramifications of doing that, what are the choices that go into that, and, again, how could we make those differently potentially in some areas. So bringing together in grad school really that, like, social justice and technology interest and seeing how how technology is is one of the systems and and, like, infrastructures that can really affect these bigger issues that I cared about. And then I I actually went to work at Salesforce, and that was an interesting process of kind of seeing what's possible in terms of what I I was saying about shaping decisions about how technology is designed and used differently. So I would say coming out of grad school, there was kind of the idea instilled in us, like, we'll be in these tech companies, and we will just ask these smart questions. And we'll we'll bring up privacy issues, or we'll bring up an issue that maybe wasn't people weren't thinking about. And then once we did, everyone would realize, oh, we need to think about this. And that was really not how it played out. So I I got to see, you know, all the pressures that come from, working at a business that and and the pressures around generating revenue that influence those decisions and that can cause conflicts when some of these other issues are not, aligned. So when, like, privacy issues are maybe not aligned with increasing sales. So so that was kind of an interesting process. I think I had always been interested in going back at some point to some kind of nonprofit or policy work, but, seeing that and then also, seeing the role that regulation did play where when the GDPR came out, how much that influenced decisions that were happening at Salesforce around product design. So I was working, as a user experience researcher on a design team, so on a product team building, customer service software. And and then also at Salesforce, getting involved in in some collective action with employees around, Salesforce's contract with customs and border protection, and seeing kind of the possibilities of, okay, when it's not just one person in a meeting asking a question, but it's nearly a thousand people signing on to a letter to the CEO asking questions about how do we how do we think about how our technology is designed and used in an ethical way, in a moral way. And that that was a a mixed experience. And and, basically, that that, I think, helped propel me to really be interested in doing the Tech Congress fellowship, which brought me to DC to work in a congressional office for a year. Just seeing how there are these big forces and systemic incentives that really drive what, companies that are building technology are doing and wanting to see, okay, how are those policies actually getting made, and how can those be influenced so that downstream, the decisions, you know, being made about what's being built and how it's being deployed might, happen differently. So that pretty much brings me to today. I spent last year in a congressional office in the office of congressman Mark Takano from California, and, then now, yeah, continuing to work on, equity and justice issues and how technology is designed and built and used, here at Upturn.

I think there are maybe more and less cynical lessons. I think for me, I felt like it's really hard to push back against those business incentives when when they do conflict with, you know, those ethical concerns or or kind of, like, values or just concerns about impact of the tools. So I think there's multiple paths to take. Like, I think there's the path of trying to reconcile that and and try to and this is, you know, what is kind of happening in the tech industry right now, like, attempting to make a business case for ethics and to integrate ethics into, how these companies operate. Personally, I have a lot of doubt about those efforts. Not that I don't think that at the margins, like, there can be smaller choices about design that might be ethical, or maybe there can be bigger red lines that that organizations decide not to cross. I mean, I think we've seen that happen a lot when employees really push back. So I think to for me, the lessons really point to more systemic change, like, to, more collect like, again, more collective decision making and how do you do that through, like, more cooperative ownership models and employee board membership and ways to have the people, I I guess, building technology and also the people impacted by it, like, actually have decision making power, not just kind of rhetorical, strategies. So I don't know if that's what what you were getting at. I know that's still maybe a little abstract, but I think for me, it it, like, really helped motivate and point me towards, like, bigger system changes that are, you know, affecting issues, not just affecting kind of the technology ethics conversation.

Yeah. I like this question. I think I mean, you know, there are a lot of I feel like there's kind of the typical, tech newsletters that I definitely found find helpful and found helpful last year, keeping up with what's going on, like, Politico and Washington Post tech newsletters that come out daily. And then I I definitely got very interested in what's happening at the local government level with technology because there is a lot happening there, and, it's much, you know, closer the the sort of decisions being made about it and the impact that it's having are are closer there than when you're in congress talking about, technology. So there's a New America Foundation as part of their public interest tech program, has a newsletter called the Commons that is about kind of civic tech and, local government projects, and then Route 50, which, has a lot of local government and, like, local, I guess, government innovation stories. And then I also, for kind of the, like, critical technology angle, would definitely point to data in society. They have a podcast, of their talks, and they also have a newsletter. The organization I work for has a weekly newsletter that's a digest of of tech and social justice, news stories that everyone in the organization contributes to. And then maybe we can talk about this more later, but I think for me, like, having a bigger political vision and analysis of of where the public interest tech work that you're doing fits in is really important to me. So, like, on the criminal justice side, the book, The New Jim Crow that I mentioned, from the New Press, There's a new book I'm excited to read, and I had the chance to help out with some early research on called We Keep Us Safe by the executive director of the Ella Baker Center, which is a, racial justice and organization in Oakland. I also have to put a plug because I love it so much for Adrienne Maree Brown, her books Pleasure Activism and Emergent Strategy, and her podcast How to Survive the End of the World. It's not, necessarily technology focused, but just in terms of thinking about how do you how do you think about systems change and, yeah, and just a really, really, like, interesting and I always learn a lot from reading or listening to her.

That's awesome. Can I add one more thing that I, a book I'm reading right now that I think actually is, relevant to sort of civic tech efforts? So it's called Normal Life, and it's by Dean Spade. And it's about, trans rights and thinking about the role of legal reform in trans rights. And I think there are really interesting parallels to thinking about kind of civic tech and how public interest law and public interest tech are both taking, these kind of system how do I say this? So thinking about the limitations of the tools at your disposal, you know, if you're a lawyer or if you're a technologist, what are the ways that you can contribute to this bigger effort, be it, like, the fight for trans rights or economic justice or racial justice? And then what are also the places where your toolkit are not actual is not actually going to be sufficient or be the right intervention. So, that book talks a lot about the idea of administrative violence and how, you know, a lot of these systems that determine eligibility for benefits that, again, allocate resources, that that is a place where a lot of systemic discrimination and injustice happens even though it's not necessarily as visible as some of the other, forms of discrimination we talk about. So that's another one that I am in the midst of reading and definitely recommend.

Yeah. So and this was interesting because this was part of the bill drafting process was figuring out what how are we defining the scope of this and and defining this term. But, generally, forensic algorithms, refers to the use of, you know, some kind of, automated, semi automated process in the analysis of evidence for criminal trials, or criminal investigations. So, the one that was the initial entry point or kind of motivation for the Justice and Forensic Algorithms Act is probabilistic genotyping, which is a form of DNA analysis. But there are other things that fall into this definition of forensic algorithms, like, the there are gunshot. There there's systems that can detect. Like, ShotSpotter is one that can, detect and try to pinpoint the location of a gunshot or, tools that are used for analyzing extracting and analyzing data on people's smartphones or mobile devices. So broadly, the category is, do you have is is an algorithm or software playing some role in the analysis and processing of evidence for for a criminal investigation?

Yeah. That's a good question. And I guess as, I think, helpful background for something that was interesting to me coming into congress and seeing, you know, where do ideas come from and how do how does congress end up focusing on different things? So this article was written by Rebecca Wexler, who is now a professor at Berkeley Law. And it was a 2015 article, about specifically focused on probabilistic genotyping, so that, DNA software that I mentioned a minute ago, And how essentially, the the big motivating thing from that piece was that, these tools are primarily proprietary tools and that consistently when defendants were being charged and and convict you know, ultimately convicted based, in some cases, only on a match from that software as the evidence against them, When defendants tried to exercise their constitutional rights so they have the right to confront evidence against them under the sixth amendment, they were denied, and they were denied because of the intellectual property rights of the developers of the software. So that that was the key point of that article was that intellectual property rights of the software companies should not be prioritized over or undermining defendant's constitutional rights. And that was something that, someone in my office had read back when it came out in 2015 and just felt, like, very, you know, moved by and and that that was just an injustice and that something should be done about it. And then when I came on as a technology fellow, they, you know, they had the resources and someone who could spend time on that. So that's how we ended up working on it last year. But yeah. So the article really called out that kind of fundamental, even just, like, procedural injustice that, people's constitutional rights were being undermined or violated, and then also spoke about some specific cases. So kind of, you know, talking about the human impact of of this issue.

Yeah. So I think right. So there's kind of just the procedural injustice part or that I was just mentioning of just this like, conceptually, this is an issue. And then there's, right, how is it actually playing out and impacting people? And there that's not totally known, you know, in terms of one big issue with it being secret is that there's not a lot known about how well it works or whether that's an acceptable you know, whether these are these algorithms are acceptable ways to perform this kind of analysis. So I guess there's a few a few things to highlight. So there's there's the issue of just furthering asymmetry of power between prosecution and the defense. So if, the defendant can't actually adequately confront evidence against them, if they just don't know, you know, how is this software actually working, how is it used in this particular case, They can't kind of surface and question those issues. Whereas if you think about, a forensic examiner doing DNA analysis, you can put them on the stand. You can ask what assumptions did you make when you were doing this analysis. You can kind of probe and surface some of the inevitable subjectivity that goes into any kind of, analysis. And and that's just not you can't really do that with, proprietary software when you don't have any access to it. And the the developers of the software have testified in court generally about the software, but that is not the same thing as, you know, about how is it used specifically in this case. And then to talk about, like, some other so there's kind of multiple layers of issues. There's, like, is this software even if it was working perfectly, is this acceptable? Is this the right way for us to be doing this kind of analysis? Is this acceptable for convicting someone potentially taking away their liberty or even their life? Then there's, like, how well does the software work if we are saying that we think that the software is is acceptable, at least in theory, to be used for this purpose. And that's where you get into, again, subjective decisions in how the kind of conceptual method is implemented in the particular software. There's multiple vendors selling the software, and there have been multiple cases where they get different results. And there was one case in New York where one one of these probabilistic genotyping tools said, yes. This person, is included in the sample. The other said they're not. So you can see how there is some subjectivity. There is some differences in how how this is implemented, and it's probabilistic. So, like, there's there's a level of error. So there's a question of what's an acceptable rate of error. There's also issues just with bugs in software. And, if and right now, there aren't testing requirements. There aren't, there's really nobody scrutinizing the software except for the developers, and they don't have specific requirements as to what they actually need to do to say this is valid and reliable. So I think there there's a lot of questions right now, and and this bill honestly isn't the goal of this bill wasn't to say if this software is good or bad. It was to say we need to be able to look at it. Defendants need to be able to look at it. The police departments and crime labs that are procuring it need to have more information. We, as the public, needs to have more information. We just need to be able to scrutinize this and and make a more informed decision about whether this should be used at all. So I don't know if that answered your question. I guess to highlight one kind of interesting story. So New York City, there's the two main tools, proprietary tools. New York City developed their own tool, the forensic statistical tool, and they actually also made intellectual property claims when, defendants tried to get access, even though it was developed by a public agency by the Office of the Chief Medical Examiner of New York. And that tool was really widely used, including other jurisdictions would send samples to New York to, run through the their software. And, finally, in a case, a couple years ago, I don't remember the exact year, first, the defense got access under a protective order, and then, ultimately, a judge ruled that the source code itself needed to be released because there was, you know, prevailing public interest in that. And during the defense expert review, they brought someone in, a computer scientist to review it, and they found, I think, multiple bugs in the code and also, an issue where there was a function in the code that would drop data that would significantly impact the results that was not not articulated, like, not explicitly acknowledged as being part of how the software works. So that tool was was discontinued, the use of it, after that. But I think that just points out or raises the question, if these tools have been used in thousands of cases, where have they maybe not worked, and and we may not even know that.

Yeah. Yeah. I think to your point, like, not having a feedback loop because especially when you think about just again, that's kind of asymmetry of power that's maybe enhanced by this and there you know, like that defendant in California where the results of the software, Martel Chubbs, this is a 2012 case, he took a plea deal. And and, you know, anyone who like, when you look at the how the criminal justice system works, you see that people often take plea deals when they're innocent. I'm not saying you know, I don't know in this particular case, but just the ostensibly, the feedback loop would be, oh, if if it's wrong, if it's not working, then you'd find out because you'd realize this person didn't commit the crime. But if the software is producing the evidence against them and that evidence is DNA evidence is very, damning, you know, juries take that really seriously, perhaps more than they should. That's a whole other conversation that people talk about, like, the CSI effect, but, with forensic evidence. But, you know, you don't have that feedback loop. So to your point and I think the the maybe related piece is, like, there are just no safeguards against the tools entering the system. They're being procured. Crime labs have some requirements around, validating tools, but, they're being procured. They're being used. The judges who are deciding whether they're admissible or not don't know how the software works. And then the defendant is left as, like, the last line of challenging it, and they are not really able to do that either. So, yeah, I think thinking about how do you have if anything, this software you know, this is really high stakes software and should have more scrutiny over it than maybe other tools where an error can be more easily remedied.

Yeah. So I think there's, yeah. Yeah. And I would say, you know, with forensic methods, there there's a lot of discussion and and concern about forensics in general and about which forensic analysis methods are actually scientifically valid and which are not. So, you know, DNA has a lot more, validation behind it than, like, hair analysis and bite mark analysis, which has been really called into question. But another piece kind of related that you just made me think of is, when judges are admitting the results of the software, there are standards for evident there are evidentiary standards. But the software is basically which we I feel like we see this all the time. The software is kind of being treated as neutral or objective. So they the prosecution will point to studies that have been done of the underlying method, you know, conceptually the method of probabilistically analyzing DNA samples. And then that sort of considered, okay. Well, it's it's been validated. But that doesn't that's not the same thing as testing software. And we don't test software by doing peer reviewed studies. Like, we test software by using it, by testing edge cases, by, like, really trying to see how whether and how we can break it. So, I might have just deviated from kind of what you were saying to try to sum sum it up, but, I think that's another, like, important piece to highlight.

Yeah. Yeah. And I guess to say, like, if it would be helpful to be a little more concrete about this particular, probabilistic genotyping software and, like, what kinds of what some of those decisions, like, you're saying that, aren't necessarily being surfaced or able to be traced are. So the in typical or, like, traditional DNA analysis, there's a threshold. You you look for peaks at different, like, alleles, and I'm not an expert and, you know, I learned about this last year. But, you look for peaks and and if you see based on a certain number of peaks, you can, like, pretty accurately identify, okay, if the peaks appear at these different spots, then that that's highly likely that it's this person and it's, you know, x in one whatever, one in however many millions or more chance that someone else would have those same peaks. But what probabilistic genotyping does is it and and in traditional DNA analysis, there's a threshold, and you don't take into account peaks below that threshold because they could be noise. And what probabilistic genotyping does is it does it doesn't have that threshold. It looks at everything. And the benefit of that is that in more degraded or lower quantity DNA samples or in DNA samples that have multiple contributors, like, more than two unknown contributors, traditional DNA DNA analysis, they usually can't get anything from it. It's inconclusive value proposition of probabilistic genotyping is that it can analyze those, samples that would have been inconclusive in the past. And so I think there are questions, you know, when there's the risk of amplifying noise and drawing conclusions from it that is kind of like a you know, just again, these are all just I I'm I'm not able to say whether this is an acceptable method or not. Like, I do not have the correct expertise. But the point is just that we need to people who do need to be, like, having these conversations and talking about it. And then there's also subjective decisions like, okay. I think that there are three contributors here, or I think there are four, or I think there are two. And that's a a call that's being made, by whoever is running the software in that particular instance, and that is going to impact the results. So there there's there are these different places where I think, first of all, at just kind of, again, the, like, conceptual level of what the software is doing and then at the level of running it in these individual cases where there are these choices being made. Like, how much do you amplify these peaks and what do you consider noise or what do you take as signal and potential DNA? So all of that is, like, really important to surface.

Yeah. So like I was saying, when I came in, it was an issue the office had been thinking about for a couple years after reading that article, and had done a little bit of preliminary research on, but just didn't really have you know, congressional offices have so much to do, and this is a pretty complex and new relatively new issue and definitely new for Congress. So, when I got there, the starting point was talking so I guess two main things, figuring out what should be the scope of this bill. So when we talked about forensic algorithms at the beginning and what falls under that definition, that was a big conversation. Like, is this bill just about probabilistic genotyping? But we also don't wanna be too we wanna future proof it. Like, this should apply more broadly and be more technology agnostic and not just applying to this one type of tool. So that was that was a big thing is to figure out the right scope, and talk, you know, talking to stakeholders. So talking to people, legal aid attorney, you know, defense attorneys, legal aid attorneys who have, worked on cases where the tool where these programs have been used, talking to people who have, tested and reviewed this software in the limited cases where that has happened, academics who are researching this issue. It kinda brought together people who are intellectual property trade secrets people because that's the the defense or that's the way that the companies were preventing access was claiming trade secrets with forensics people, with criminal justice, you know, racial justice people, with algorithmic transparency and algorithmic discrimination people. So talking to all these stakeholders, figuring out the scope, and and then, yeah, figuring out, like, what what how can we both empower individual defendants to be able to challenge this, but also not leave everything up to them? Because defendants typically don't have the resources to, do that. So how can we also have that's what ultimately led to this, like, two pronged bill, which is access for defendants, but also having a program at NIST, the National Institute of Standards and Technology, to, test and report on how these tools work so that that's a resource that's available, so that every defendant doesn't need to, you know, hire an expert reviewer to look at the source code or to test the software. Some can and hopefully would do that, but, also, that there would be a report where they would still be able to point to and question some of the assumptions that were made, in their analysis. So, yeah, I mean, it was a multiple month process of these conversations, of reading, of trying to see, yeah, and trying to connect with people and trying to also connect with people who maybe are closer to the people who would be mostly impacted by this. So I think that's also a big thing with the stakeholder conversation. Discussions?

Yeah. So, like, yeah, like I was saying, there are two main prongs of the bill, and one is about ensuring defendants have access to source code and any other information they need to confront and challenge the evidence against them. So that includes an executable version of the software so that they can also run the analysis themselves. That includes, this report about how the software was used in their particular case, what, inputs were put in, and that they're able you know, again, if they can afford to and want to, that they could bring someone in to do a review of the software. So that would definitely be one piece is that individual defendants the hope is that individual defendants would be able to challenge the results whether or not they had the resources to do a deeper, like, source code review, but also that some defendants would take advantage of that because that could also then that would have, you know, scaling effects if they did find these issues. If they found issues and they were able to disclose that, other defense attorneys could learn from that. That might impact ad admissibility decisions in other courts. So that's one big piece. And then the other big piece is the NIST program to, test the software and to publicize the results and to have, a process for coming up with what does it mean to test this software. So one thing is we're not actually the bill is not actually establishing standards for the software. Like, it needs to have this accuracy result. The bill that's a longer conversation. You know? What's an acceptable error rate for software for the criminal justice system? But it's to say, these are the things that you need to test. This what this what it means to test the performance of this software, and this is what needs to be made publicly available from the output. So right now, because there's just such little access, there's not much known. And there are many people who are interested in already trying to work on this issue, so the bill would also just open the algorithms up to scrutiny from both individual defendants and from advocates, academics, whoever else wants to try to, better understand how it works and and raise concerns about how it works.

the bill prevents the the makers of the software from using trade secrets to prevent defense access. So they would they would get access to the source code. It's not directly addressed in the bill whether that would be under a protective order or not, which is, kind of a tension in that if it's under a protective order, then only that one defendant can really benefit from the learnings. But, anyway, to put that aside yeah. So they would have access to the source code, but we also wanted to be aware, like, you can't learn everything from source code. And so the defend and and also not everyone is gonna have, again, that, like, expert reviewer of the source code. So they should also have their own copy of the version of the software that was used. They should have the performance testing of the software. But, yes, I guess, to answer your question shortly, it would include that they have access to source code.

Yeah. So that, I mean, in this particular example, that and this has this happened in in the very few cases where people have gained access to source code. It's been under protective order. So the judge, like, basically, the the company and the defendant or the prosecution of the defendant agree, that the defend the defense team will only look at the the source code or whatever they're looking at in these it often can be in a very limited setting. Like, you can go in a room with a pencil with a printed out copy of the source code, or you're and it's certainly a prohibition, kind of like an NDA. Like, you cannot a nondisclosure agreement. Like, you cannot share outside of of this courtroom or this this defense team what you learn from the from the analysis. So it's it's just much more restrictive of what you can talk about publicly, what you can disclose. And I'm not a lawyer, but I guess in my simple way, I would liken it to, like, a nondisclosure agreement.

Yeah. So every time a bill gets introduced, it gets referred to one or more committees or potentially subcommittees, based on which committees would have jurisdiction. So in this bill, in this case, it was referred to two committees. It was referred to the House Committee on Science, Space, and Technology, and it was referred to the House Judiciary Committee. And for house judiciary, it was referred to the subcommittee on courts, intellectual property, and the Internet. So you can kinda see between those two committees, you have the IP courts piece and you have the science, like, forensic standards piece. And so, ideally, the next step would be so, ultimately, for the bill to to go to the floor, it has to go through committee markup. So that would be the committee, making changes to the bill and then deciding that it was ready to go to the floor. That's that's quite a ways off, I would assume, for this bill. An ideally, a next step would be for one or both of those committees to hold hearings on the issue because that's where the committee members can learn, more about the issue, can start thinking about, you know, what should this what should the bill that ultimately goes to the floor look like. It's it's, you know, it's it wouldn't it's very unlikely or I think maybe completely not wouldn't happen that the bill that was introduced would just go to the floor. So it would go through a lot of changes and, deliberations. So a hearing would would be kind of the next most obvious step, especially because this is a newer issue that most members of Congress don't know about and even a lot of the public doesn't know about. So, just getting more understanding and bringing in experts and bringing in people to represent, you know, the different sides and stakeholders of the issue. So in terms of prospects, I think, there's a few interesting things that you know, this is this bill, again, is like bringing up a conversation that's not happening in Congress, so that takes a long time for that to work its way through the process. But, the Science, Space, and Technology Committee did have a hearing last year on forensic standards. So, there's kind of the thought that maybe if there's work being done on forensic standards in general, maybe it could incorporate some of the, you know, it should also address software because where we are now, that's just increasingly playing a role in forensic analysis. And then the other thing that could be interesting is that, there will be a GAO report, coming out in sometime maybe this year. So that would be another way to get some and GAO is the government accountability office. That would be a way to have some more attention to the issue and some more information about the issue, and that could potentially spur more conversations, you know, in congress and also just public conversations or advocacy conversations. So that's a bit of a sense of the process.

I think, a huge part of it is is having the conversations about it. And, I mean, for people who do have, well, like, a platform like you with a podcast or, a blog or journalism, like, having this just be more known, because, again, it's still a relatively unknown issue. And, I mean, part of, you know, introducing the bill last year is that that does bring some more attention. And there are certainly people who have been working on this for years, but it's just not necessarily, in the common, like, mainstream awareness. So I think, like, telling people about it, talking about it, talking about it on social media, media coverage, like, those things all definitely impact what Congress is doing. Not not that it automatically will, but, you know, that can, like, raise things to people's attention. I think, also, it could be something for people to talk about at the local level because, you know, this bill is the federal, criminal justice system. But most people who are in the criminal justice system, most people who are being arrested and charged and going to trial, that's not happening at the federal level. That's happening at the state level. And state crime labs or even city crime labs or police departments are the ones mostly buying and using this software. So I think also raising that conversation at the local level or the state level with your elected officials or with groups that are organizing in general around criminal justice or constitutional rights, just like growing. I think we're still at the point where, like, growing the momentum in the conversation. And because, like, to your you know, the way that you reacted as well, I think when people hear, okay. This the intellectual property rights are superseding constitutional rights. There's software where there's not a feedback loop. There is not really testing. There's not there are no really safeguards. Like, most people react to that and say, oh, that doesn't really make sense or that doesn't seem fair. So I think just raising awareness, feels like kind of the biggest next step. I I mean, I you know, there's always the, like, oh, should you call your congressperson? I don't know if it's at that point. I mean, someone certainly could, and maybe that would cause their off their congress member's office to look into the bill. But it's not like this bill is, on everyone's minds in congress. So I think even just getting it more in the public conversation would be a really good next step.

Yeah. I have some, I guess, bigger thoughts. So as I was working on this bill, I was reflecting a lot on, you know, what's the role of, again, being a public interest technologist, working on technology policy, and how to think about that work. And and so I guess my thing I'd like to just talk about in the end of our conversation is because I'm assuming many of your listeners are public interest tech folks or civic tech folks, is thinking critically about, about our role and the work that we're doing and having a bigger kind of political analysis and acknowledgment of the bigger conversations that any public interest tech work is is happening within or is embedded within. So, that's something that I aspire to do with this bill, and I think, you know, it's it's it's a constant effort, to really think about who's who's in the conversation, like, who's providing input to this technology project or to this technology policy effort. Are you having people who are really at the front lines of, like, these systems of like, our legal system, our economic system, and are really experiencing the most kind of harms when technologies are not, are either not working correctly or maybe are working correctly but are very harmful. And so I think just thinking about I feel strongly that no technology work is apolitical, so you always have some some politics to your work. And and not everyone's is going to be the same, and maybe your politics is overall I think that this system is working well, and and I I just wanna build a tool that works well and is accurate and, you know, works well within that system. I think in this case with criminal justice, our criminal justice system is not working well. It's a very harmful and biased system. It's a racist and classist system. So so, like, that that was all animating my work in this. And just thinking about how can you connect your work to these bigger political fights and these bigger, like, fights for justice that are going on in in whatever kind of aspect of our system or whatever domain you're working on a technology or a tech policy project in.