 It was like crap. Good morning, good afternoon, good evening, wherever you're hailing from, welcome to another episode of our mini series here on Openshift TV called Call for Code for Racial Justice. Today we're talking about open sentencing. I think it's a wonderful project and we'll do a lot if we can apply it in more ways. So I'm going to step aside and I'll hand it over to Sabine for now and Sabine, let's do like a round of introductions for folks. Yeah, absolutely. So Sabine Distillian, I'm the product manager for Call for Code for Racial Justice and the open sentencing team, we have just a small portion of the team on the call right now and I would love for them to introduce themselves. You know, talking about just, you know, their title, how many years you've been at IBM and we can go from there. So I'll start with Saran. I'll actually like call people out since I can kind of see them better on the diagram. Yeah. So Saran, if you could go ahead and introduce yourself. Okay. Thanks Sabine. Hi, this is Saran. Can call me Saran actually. I'm Saran GB Mahendran. I've been seven years associated with IBM. So I'm just working as a full-stack engineer for IBM's client AT&T. So I'm just currently engaged with AT&T research center dealing with AT&T's enterprise services, migration, especially with cloud migration and application modernization. I'm happy to be a part of call for code, especially the open solution, open sentencing solutions team almost for an year and I'm happy, happy and as well as proud to be a part of the team. Like thank all the team members to give me an option as well as a chance to use technology as a kind of a solution to address some of the real-time issues, especially the humanity issues. Thanks for all. Perfect. Awesome. Stacy. Hi. My name is Stacy Forsythe. I'm an IBMer. I came to IBM via acquisition. So I've been an IBMer for about over three years. My day job is I am a healthcare data analytic consultant for the Watson Health Division of IBM and I'm happy to be here with you all. Awesome. Anne Marie. Hey, I'm Anne Marie Fred. I'm a senior technical staff member at IBM. I work in IBM Digital on the parts of IBM.com where we sell products online and manage your subscriptions by IBM search things like that. And I got involved with this project because I really wanted to try to do something that was actively anti-racist and I thought this was a great way to get involved. Awesome. Thank you. Joanne. Hi, everybody. Yes. My name is Joanne Hill. I have been with IBM over 25 years and I am currently working as a business development leader within our IBM Red Hat Marketplace Ecosystem team, in which I work with business partners in support of them onboarding their solutions into our Red Hat Marketplace. And I've been involved with the project since last year as well and happy to be here. Awesome. And Joana. You're on mute. Don't give us the good stuff yet. Sorry. Yeah. Hello, everyone. I'm Yushna Hanawa. I have been with IBM for past 17 years. I'm part of IBM Digital Marketplace team and I have been working as a cloud application developer, mainly backend development and deployment related stuff. Recently, I've joined this project. It's a very good cause and I would like to do my bit of contribution to this project. Thank you. Awesome. Before we kind of just dive right in, for those who weren't able to see the first episode, just gave me a little bit of context on call for code for racial justice. Like many of the team members mentioned, they've been part of these teams for over a year and that's because we issued this embrace challenge on behalf of IBM and Red Hat to encourage technologists within our company to figure out interesting and really applicable ways for technology to combat racial injustice. And so we have three different pillars that our work is really governed by. We have police judicial reform and accountability where open sentencing sits. We have diverse representation where we have a solution that sits there as well as some, you know, backlog HR solutions and we have policy and legislation reform, which we will also get to, you know, after this mini series, you'll get to see all these different solutions. But again, this is really about IBMers and Red Haters who initially kind of sparked these different projects that live in the open source community. But we encourage this to be a worldwide global effort for us to combat what, you know, we can do to be anti-racist, you know, anti-bigotry because we want people to be included in the ways in which their lives are governed and the ways that their lives are affected by institutions and structures. And open sentencing is a really, really important way. I think especially in our judicial process where we can begin to do that. So without giving away too much, I'm going to hand it over to the team for them to, you know, do a quick, you know, not even a quick, but really like interesting kind of dive and process on how we got here and also be poking in some questions here and there and really encourage, you know, you all to ask questions from the audience. Yes, please feel free to ask questions as we go along here, folks. Perfect. All right. Thank you. Do you all see my screen now? Yes, Stacey. All right. Great. Thank you so much. And okay, I'm already having that technical difficulty of advancing the slide. Yeah, live stream. Happened pretty quickly. Oh, there we go. Thank you. All right. So as we mentioned, we're just a small subset of a large team and I, like many others, wanted to do something after the George Floyd incident last year. I wanted to do more than just take the embrace pledge that IBM asked many of us to take. I think you're supposed to advance the slide and it's still on the opening slide. Oh, oh, okay. Sorry about that. Oh, perfect. You see it now? Okay, great. Okay. So I like many others and like I said, on this call, it's just a subset of our larger team on the left and there's even more that have joined us since then and we've had some changes, but we've had people all over the USA and the globe help us thinking through how we could use technology, data and AI to fight systemic racism. So I really enjoyed the last year getting to know many of these people that you see people on the call. We have a lot of ideas and we can use more support. That's why we're here today to talk through what we have now and our future and what we would like to do. Our team was focused on police and judicial reform. One of the areas mentioned and our problem statement that we are trying to address is that members of the Black community receive harsher sentences for the same crime. We're trying to address this through data. We've done a lot of design thinking I'll talk about later and I'll show you the tool, the interface we have in place now. Right now, our main tool that we have out there that we're going to show is that it's meant to help public defenders as they accumulate evidence show bias and racism within a certain case or so that when they go to the plea or sentencing phase, they could show that there was bias or racism involved and help argue for and address that and get a lower sentence. So that's our current tool that's out there on GitHub and we'll show you we have some bigger goals too. We'd love to collect data back, make a dashboard of the different sentencing differences, excuse me, so we can get more insights into prosecutors, judges and decisions they're making so the public can be more informed about what's happening with the people that they elect. Yeah, and I also would love to just like add for, you know, I mean, an agile kind of development where we're aware of what Hills are, but really just want to, you know, level set and saying like Hills are really important, especially in the case of, you know, these projects because it really lets people understand what is the value we're trying to bring to these solutions? Like, what are we really trying to deliver? And even as we talk about what we plan on doing in three months, six months a year, it's going to be important for people to just be able to like level set on what it is that we're really trying to accomplish without, you know, saying this would be a great button. This would be a great feature really focusing on the value we're trying to deliver and so that's what's been really important as we've moved from, you know, even this past year to, you know, where we see ourselves going to this is really what the work is going to be focused on. Okay, great. And I'll just say I'm not a developer like others on the team, but I've learned many things. I've helped with the GitHub and we're going to say later, you see this team we have now, we're looking for lots of other help developers, front end developers back and then we're all just really passionate about fighting racism and I'll turn over to Joanne to talk about some of the next steps of data or slides we have. Thanks, Stacy. Yeah, so we just wanted to share just a little bit about why we came together. Many of the team members were pretty much motivated after the tragic events that occurred here in the United States back in 2020 and they simply just wanted to help and for some of us it was also due to just the additional desire to work towards some positive change due to personal experiences with racism for themselves or with family members and just overall our team of the wonderful set of volunteers of IBMers and Red Haters from all over the world just kind of they were motivated. We were motivated to come together to work on a solution to fight systemic racism. Next slide. Okay, and here what we this powerful graphic really just kind of highlights what inspired the team and basically from the beginning what inspired the team also continues to inspire us now and it's really as what I just said earlier to fight systemic racism. You can see all the words that are on there and it just keeps keeps pushing and reinforcing that you know we're looking for positive change. Additionally, we knew that it would take all of us to begin making a change toward some significant progress and so to be able to participate in something participate in the project as part of IBM along with our Red Hat counterparts and now the external community that means a lot. So we feel we can make a big difference with our solution and so what we're seeing here and Stacy mentioned it earlier as far as our problem statement you know in which and I'll just restate it people in the black community are faced with harsher downstream effects they're charged at higher rates assigned more significant charges they're convicted at higher rates given much longer sentences and deny parole more often than people of other races for similar offenses. So what we're highlighting here on this chart that we source from the U.S. Sentencing Commission you can see for the same crimes they're given different amount of times there's a huge percentage difference in the length of sentences for black men versus white men for the same crimes. So we wanted to highlight that. Okay, so our solution first focus on judges. However, through some research we learned that in the United States most sentencing comes from the police stage of the process involving attorneys and not judges also the majority of the accused are represented by public defenders and public defenders as highlighted on here are lawyers that are appointed to represent people who otherwise can't afford to hire a lawyer to defend themselves in a trial and it's up to 90% of criminal defendants need public defenders. So with the goal of helping public defenders to access data on racism the team decided to make a tool to help public defenders in order to help the people that they are defending on the next slide. So through our design thinking sessions that we held we further defined and identified public defenders as those needing the support more directly in order to help the defendants. They have had identified increasing caseloads and budgets as well as a large percentage of the defendants in the United States are actually unable to afford a lawyer. Therefore public defenders are key to helping in this area of fighting unfair sentencing and those harsher downstream effects for people in the black community that I mentioned earlier. So we found that public defenders are the only ones who stand up for the vast majority of people charged with crimes in America and they try to help their clients to achieve justice in the legal system that disproportionately affects poor people and minorities. Sadly defendants who can't afford to make bond can sit in jail for 60 days or more while the district attorney decides whether to arraign them. So that's a lot to let sink in. That's a long time. So I'll turn it over now to Saran to talk more about the design and the technical details of the solution that we worked on. Yeah and before kind of get into that I would love to kind of just pose this question to the group. You know we were saying that public defenders really have a lot of weight on their shoulders to be able to you know advocate for their clients be able to do a good job and getting them the right type of representation when it comes to them entering the criminal justice system. But if they're already overloaded now like what are kind of the larger implications or the better ways which we think that open sentencing can help influence you know the wider system that when it comes to judicial reform like what are your guys's thoughts around that. So when we show the front end demo we're keeping in mind how very busy they are and what we have for the public defender is meant to understand that and be very quick and easy and readable. You're about to see some great technical stuff that goes into that but we're keeping that in mind we want everything to be very easy for them. And then perhaps it could go more to their staff and others to help fill this in for them. One plus of our tool is it's free. It's open source and there's nothing like it out there. So people haven't been thinking about this kind of thing for public defenders in terms of how to bring race and bias into the conversation. I don't know if I'm totally answering your question but I think it is. Yeah I think like overall and what I think is something that's indicative of where kind of call for code for judicial justice especially in this in the specific pillar we want this to start the conversation. Of course we want to be able to bring it to you know the end user we want to bring it to the public defender but if we are able to see that it's impacting the lives of their clients in a positive way if there are judges who are interested in saying well how does this really work? Like how can this you know start to address some of the more systemic issues? How can this affect policy? I think that's where we start to see you know like this this snowball effect where we want to ensure that of course it is transparency that there are ways that you know people can be engaged from all sides of the I don't say like the jury room or the the courtroom but this you know I think this is a great starting point to at least you know start to have those conversations because you know policy change is needed. We aren't going to solve for a system of racism and just in the judicial digital system through this but it really is supposed to be you know that like at least at that kickoff point and like you said there's nothing around right now that does this and it's going to be a great way for us you know start to gather some of those stories that can help implement change. Yeah and something powerful that came up recently was there's not there's hardly any or very little data collected at the police stage and we can use that to think about policies and many innocent people will take a plea just to avoid a sentence they don't have the money to go through anything further you know and they they're worried about the longer sentences if they go they go and they don't flee so we're thinking about it from that way too to help people that are innocent they add a jail as well as their sentences based on the data that will show. Awesome. All right. One of the things that I was going to say to be one other thing that I was surprised to learn is that you know more than 90% of convictions are actually coming through plea deals. I I think you know we grow up watching these shows on TV and we think that everything goes to trial and that's just not the facts and so anything that we can do to sort of shift left if you will the the process and try to make it fair early on it's going to help and even just having these conversations and getting that information out in the open is helpful. Yeah. Right. So there you go. Yeah. So from the technical aspect of it the open sentencing solution is basically comprised of independently scalable and deployable microservices and each service is having its own behavior. So to start off with we have a interactive UI dashboard managed by our front end service and this service is designed in such a way that very lightweight and any kind of a non-technical background can go back and start using the service. The prime functionality of this service is one being a public defender as my teammates highlighted why we have articulated this solution primarily address to public defender because most of the cases that has been handle at the plea level more than on the trial level which is the next layer. So in this case the public defender can go back and register any new case. He himself can go and register as an attorney and he can register any new cases. There is one and second thing is like the dashboard acts as an interface to provide a highlighted and a detailed report which says what are all the discrepancies discrepancies in that particular case and what are all the potential sort of factors which which basically bringing the racial disparity into that particular case. So this is pretty much on the front end side and the next we have is an aggregator microservice. So this service is basically the articulation point which which deals out which acts as an interface between the front end and the analytic services. I'll come into the analytic services in the later part but the basic components of this aggregator services like it has a data store which basically kind of organized and inclined towards attorney based model where it stores the list of cases and for each case what exactly the charge and for each charge what are all the sentences that has been given which is kind of an outcome as well as a proposed one. So it was being organized in that way and the aggregator layer also has a resty APIs where one can basically the UI and it tries to create a new attorney or try to create a new cases like all these components are controlled by this rest services which which basically purses the data from the UI and it also interacts makes a call to our analytic services. So our we have two set of analytic services that one is the bias deduction engine. The other one is the open sentencing model. So in I'll be dealing I will be describing like what exactly the bias deduction engine does and what sort of data it relies on as well as the open sentencing model in the next slide. Okay. So the bias deduction engine is one of our analytic service which basically uses the IBM AI fairness 360 and it also uses the traditional Python libraries to find the disparities on the initial reports of the data that has been submitted when creating a new case. So it attributes the disparities on multiple factors in which race is being a one factor where if any particular case has been created or registered into the system it tries to find the details of the respective case and then it tries to find what are all the racial bias that has been mentioned in the report. So based on that it collects the list of racial bias factors which is kind of a predefined set of algorithms in this service and then it comes out with a potential outcome of what are all the list of racial biased flags that has been mentioned in the report. Apart from this it also helps in understanding the historical analysis on the US sentencing commissions data which is the prime data which has been used in this particular engine to find the disparities especially on the race mentioned in the case. So using that data we have created a model which pretty much correlates the list of outcomes in this case it is the committed term versus the race and then it comes out with a correlation factor which clearly mentions what could be the extra amount of years that can be spent for a same crime when the race has been taken out as a constant. So keeping this sort of result as a response from this report whenever we have any new case that has been registered with a set of preliminaries like we have a requisite structure which includes the mandatory sort of tax that needs to be available when registering a case. So in that case it takes those things as a request and then with the existing model which is trained on US sentencing data it find what exactly the racial bias and it use that as an output which is going to be come out as a detailed report. In case of open sentencing model it pretty much does the same thing but this has been trained using the Cook County state data. So in this case what we primarily derived is like we have so much sort of data that has been came as a part of the case detail. In this one we have taken two factors. One is race and other one is a committed term as we all know people from a specific race has been spending more amount of our heinous punishments for the same set of crime committed by a different race. So keeping this as an underlying motive when we try to find what are all the set of insights behind this using this Cook County data as the graph mentioned. So people from a specific race as I mentioned used to spend long amount of years or months in the jail keeping the crime as a constant. So in this case the crime is pretty much same but the committed term which is the number of years spent by the accused is pretty much more because of the race. So in the open sentencing model is pretty much a kind of a casual data analysis on the Cook County's data set and it come up with a correlation factor between the committed term and the race. So this has been used as a underlying factor to predict what could be a sentence length for the given data or the given case that is come from an attorney. And we are using this factor as a counter factor to estimate that what could be the actual sentencing length for that particular committed crime as well as like if that particular case belongs to a specific race which has been facing the discrepancies from the start. It also shows the score of how much amount of extra sort of committed term which has been provided on this. So as I said the key result from coming out from this particular service is the disparity score which says based on the race this is the maximum amount of years that could become as a racial bias. And so if anybody's not familiar Cook County is where Chicago is and the reason we use this data and the federal sentencing data is because it's available to us. There's only so many jurisdictions that publish really high quality data. And these are a couple that we were able to use so far. Yeah. So this is one of the open so the data side provided by the county itself and as she mentioned it has pretty much more elaborated data which we can go on and find the estimates as well as the insights between these different factors. And one of our goals and kind of a challenge to is getting more data to highlight these disparities and make our engines and tools more powerful. Right. Over to you Stacey on the demo part. All right back to me. I'm just going to I write. One second. I am going to go down. I've done that cause my sharing here. So what I'm going to be doing is navigating out to the GitHub in a second when it lets me. Okay. So everything we've talked about so far is out on our GitHub which I'm trying to pull up right now. I will paste that in chat for everybody. That'll be great. All right. Here we go. So we have an open sentencing GitHub that covers a lot of what we just talked about. So please go on out there into the chat link and give us a star if you like what you see so far. I'm going to go through and show you the front end interface that this builds. So as I scroll through our open sentencing GitHub. You can see our background why we do it. What we're trying to do the bias detection that Saran just talked about. We have a video. The architectures there but I'm going to cover the user interface mockup. So this is the tool that's to meant to make things really easy on the public defender. So it's out there. You can you can if you're interested go out there and click on it. It is a prototype. If anything here you're like oh I would like it if it did this it did that. That would be great. Join us. It's open source and we'd love to see your issues out there. So this is IBM open sentencing. So I'm a public defender. I'm going to review my cases for today. I'm really really busy. I got to get going to court. So I'm looking at my case for today. I'm just going to all the cases load right away for this public defender into the front end interface. I'm going to click on the first one for armed battery. So for this for this case a really easy meant to be like for you know somebody who is not technical. It opens up a summary of what's been happening for this person. So it's overall stat of the defendant. It would have a picture of them have some background. What were they accused of. So in this case you know they were charged of federal drug trafficking and Nevada USA and has some more details. That I'm not going to go through because the important part of what it does and what's new and innovative is that it says bias was detected and that the that the charge of prosecution is asking doesn't fit the crime. And so along with detecting the bias which is meant to be really clearly highlighted to help prep the prosecutor for the conversations about the sentence being unfair. It gives some reasoning why a language in the police report contained a bias word and this crime has a history of disparities. You know black black Americans people received higher sentence and the defendant in this case is black. So we're hoping with the tool like this that the public defenders have have something data driven to show this was unfair. You know we've heard anecdotally through some of our research that you know when you bring this up to prosecutors judges you don't always have a lot to back that up and they're kind of it's kind of anecdotal and some feelings can get involved. Well in this case we're showing data showing why it's unfair comparing it to the exact same crime and in the moment at the plea they're they're armed with a reason to get a lower sentence. So you know we're hoping if this starts being used we could get some people sooner with lower sentences. You know even months or weeks lower sentences. And an even bigger goal like we were talking about is to get some innocent people out of out of trouble to but I you know we got to start somewhere and we thought this would be the most impactful. So this is it on there's some other things that does and going to try to navigate there's also a way to add new case information so we make that really easy and I'm just kind of clicking and showing you how it can add a case. And so you can click in a lot of what they call like fact patterns fact details and that helps compare more one on one for other sentences so that's another feature of what is out there. All right and back to the slides. So all of this again we keep mentioning design thinking so our goal was to make this really easy to the under resourced public defenders it's easy and free some of them don't have a large budget so you know if we make it put it out there people start using it build upon it we make it better and better we make the interface better we make data entry better we hope more and more people will use it and as we gather more data will be more robust. So we did a lot of design thinking with this group and it was great. And we just did some newer design thinking workshops which we mentioned and are leading to some of our other ideas like impacting policy we're thinking about including other things besides race like other socio demographics for health mental health status we have to think about how privacy impacts all of these new ideas we have so we just had a privacy expert come on board to join us but we're looking for all kinds of help as we build this out we think we could really help the community as we build upon what we have started out on the GitHub and the ideas that we we just came up with this month. So I'll turn it over to you China to show you more about how you could get started with some of our current information. And also I would love to you know before we come to another we have like a another question on that how would someone get started on with this one towards the end of the presentation we're going to have like a QR code that you can scan and understand you know how to you know really get on board is into the coffee cofers or justice community. There are currently some open issues within the GitHub for open sentencing and you know in the next couple of weeks we're just going to have like a really tight way for people to like get started and even if you're you know talking about me in your second semester of computer programming ways for you to upskill yourself if you want to you know challenge yourself and do a little bit more but there are still going to be amazing ways that you can provide you know support to the team so we'll be able to get to that but Stacey thanks for asking that question. I'm proud. Okay, thank you. Now let's see how to get started if you want to like be part of this project. So we mainly have four reports. The first one is the overall documentation and like it covers most of the stuff that we need to get started with open sentencing project. The second report is about model code and AI related code. So that will be that is available in open sentencing model report. The third one is the aggregator code. This mainly covers the API layer which basically orchestrates between the front end and the model model piece. So you will find the code in the GitHub open service aggregator. And the UI demo we just saw that code is available in the GitHub repo open sentencing UI. Now if you are a developer you could go to if you are a developer you can go you can open the developer IBM link here which has basically steps on how to set up the repo how to set up the application locally. This has all the details here on each of the repo how to set it up and run it locally. If you encounter any issues you can open GitHub issue or reach out to us on Slack. Now let me move to the next slide. Yeah for everyone you know who's also asking questions on how to get started. And we think so when we've had a couple of different playback presentations about you know open sentencing work I think the tutorial is a great way for you to technically like actually launch your own instance to really get started. So definitely encourage you all to seek out that tutorial so that you're when you're looking at the different repos you know exactly how to be able to deploy it. And like your Tana said we're always in Slack if you're having any issues or really responsive we want people to be engaged. And if you're finding anything that you specifically want to contribute to you again open up an issue ask us questions like be engaged. That's exactly what we're looking for. Yeah and I'm sure you're trying to there was a question to the relevant to what you were just talking about. Stacy was asking what are the competencies that you need to contribute to a project like this and the good news is these different layers have different technologies backing them. So if you're a front end developer the front end is based on Angular and JavaScript. And then the if you're a Java developer learning more about Java the aggregator layer is Java REST APIs. And then the data analysis portion of it the back end is implemented with Python. So if you're interested in data science and AI you can work with that part. Awesome. Yeah. Thank you. So some of the old little areas where we are looking for help is all we have all the issues open in the GitHub repo. So you can browse through each of these repos and look at the issue section here. So some of them we are looking for is here in case of open sentencing aggregator we are trying to make it easier. To support different types of authentication. So initially we started with IBM ID support but we figured out that it has like there are some challenges for people working outside the organization. So we plan to add support for Facebook and Google kind of Google ID authentication. So mainly the IBM app ID authentication and we are also looking for documentation on how to get started with carbon components. We also need a very good quick start guide so that people new join is can quickly get started on setting up on their Mac and Windows. Also translation enablement support will help us to extend the reach of this application to different audience. And we would like to upgrade our UI to more well. Also the one more change that we are looking for is we want to change the master branch domain and for the non-technical members also we have a lot of scope of contributions. The first thing that I would like to talk about is like we need to do we plan to do more interviews and we need people to help us with the usability studies and like people who can access our applications and early trials and help us with help us with the inputs. Also we need more data. I think that is the most challenging part. We need more licensed data. As mentioned by Stacy we are using for the initial demo and all we have been looking we have been relying on counting Chicago data. So we want we would need more help with the criminal justice funnel. So we need to know exactly what charges like for filing a charge for filing a case. What are the basic requirements needed for to file the case. Also sentencing guidelines we need to be more readable format. Currently we have a PDF format. So we need we need some readable format so that machine can so that our model can process it. And we also need help with the bargaining data. Mostly this is done in the back room and verbally. So we do not have enough data on this. So we plan to have case studies with universities and lost lot of schools who can help us to get this done. Also in the area of AI and ML models we need help from lawyers and students to describe the fact patterns and creating the bargaining data. And Mary would you like to add something here? Maybe so the fact patterns if if anybody on here is not aware it's basically a description of the the conditions that need to be met for a charge to be you know something that should be brought for a specific crime and so that would be very helpful for us and saying you know if you have a description in order to charge somebody with drug trafficking they need to meet these six criteria or if you're going to charge them with drug possession they have to meet these three criteria. So I need help with getting that into more of a computer readable processable format than what we have now for our initial running app we had to basically hand code you know just one or two different types of crimes so we'll need help to expand that tomorrow. I have a pre-processing the data which can be understandable by our model and can be used for the model and adding to that different set of states followed a different set of terminology for a same I would say the jargon is being different in a Cook County's data set versus even though it refers to the same crime but the jargon which is getting used on that data set is being different. So in this case there is a data pre-processing layer being needed to come up with a model which can be more generic and handle these set of changes. Yeah, that's all I had. And this thing like one as I also now mentioned so the authentication part is one such where an area where initial people can come and contribute more where come up with a single sign on which is more generic than on an organization level where anyone can go up and create an idea and go up and sign in. So that is one of the piece where like I see your questions have been asked so that that piece is like for any newcomer who has been dealing with the JWT or any sort of authentication mechanisms. So adding this feature may really help us to open up many of the future. So you mentioned that the data isn't like clean right like how wide are the discrepancies from municipality to municipality. I mean we're talking like they're speaking two different languages completely often or is it you know just finding the right fields and putting it in the right order it all comes out and chase on or something you know I mean is any of that sentencing data available publicly out. I'm asking a lot of questions here so I'll shut up but yeah like what are the inconsistencies you find normally. You want to tackle this and Mary that's one of our. Yeah so just you know if you're comparing like Cook County to the US federal sentencing data right. I think conceptually in our minds we can map these different charges to each other more or less maybe 80% but the terminology is very different. Sometimes there are crimes described in one jurisdiction that don't exist in another. Yeah it's it's pretty complicated. I think we're going to end up having to train many different models and then the other challenge that we ran into is that it's really hard just to get the data in a computer readable format like for example with the federal sentencing commission it's out on the website but you need a special permission to get the full data set which we don't have yet. You know New York City publishes some good data but it's not segmented in the way that we like. So I think there's a lot of there's a lot of help that we could use to kind of clean up that data and make it easier to process. Yeah I can't imagine the disparity between all the various places you can pull data from. And I think that right there tells us there's a problem right like if everybody's doing the same thing differently. Well race is one of the factor which has been common across different data set say for an instance past history of a victim or the convicted plays a huge role in the sentencing stuff. So you take out this past history for a specific crime. The type of this data is not being uniform across the different data sets some elaborated that as an essay and even some of the police reports which has been attached as a detail in the case may also have systemic bias that has been added. So you really need a specific sort of a mechanism to handle these kind of data and especially the inputs are coming from various sources in various formats. There is a need for pre-processing as I said earlier and that is playing a huge role in getting a generic out of it. There is one and second thing is like now we have portrayed race as a factor and there are so many other socio-economic factors which we understood after we have a design thinking session with the SMEs and the lawyers. Even that also an other factor which we are considering to have that included to find what all the bias that is arising out of it. Another thing we had to kind of be careful about is the licensing on the data itself. So some of the licenses say you can't process it in ways we didn't intend things like that. So I think going forward we're going to need help you know even from attorneys to help us renegotiate some of these licenses as well. Like that right there like makes me raise my eyebrow right like why would you not want me to look at this data and maybe you just run a said against it right like I mean just yeah just gripping the data. I mean that's weird we definitely raised eyebrows at the US Federal Sentencing Commission website because they let you they have this beautiful dashboard where we can slice and dice the data in many different ways but you can never compare by a race. Like you can't have a graph on the left and a graph on the right that shows the differences between two races. Why did you do that? That would just make it too easy right. Right exactly. I mean that's why I'm so glad you're here to try and make it easier for everybody. So question coming in from chat. What has interest been like at the state governmental level to utilize this as a platform to support criminal justice reconciliation efforts? Have you had any US states kind of looking at this and thinking about it or talking to y'all at all? So I can kind of answer that one I'm talking about like the trajectory we're moving next. That's really where we want to start to see a lot more conversations and engagement like at the state level we mentioned you know doing a design thinking workshop that we had at the beginning of this month we had policymakers, attorneys people who have had experience in this space be able to give us more context into you know where our solution can start to solve some problems where they're seeing value and a lot of the comments are saying like policy is moving towards this and there's a lot of interest in how technology can support work when it comes to you know from the public defenders how we address bias you know how are we able to really engage people in meaningful ways and conversations around by throughout the judicial process and so that's where we are kind of think in this phase it's like what does partnership look like so that we can actually start to put this into practice and that's you know the the road that we're stepping into right now so working with the state agency if you're out there on the call reach out to us but you know we have you know the current partners that we're working with so that we're keeping in these questions in mind as we're starting to you know scale up and the different features and issues that we want to include and you know version two but also being able to say that we need to have some more kind of big hitters who are able to influence you know how we're able to you know label the data access it being able to share it out you know talking about the different licensing issues that we're seeing so that's really like the next step in the next phase of it we're focused on the initial let's build something okay this bill is actually you know showing what type of impact you can get validating that with you know policymakers lawyers and people who are in those spaces and the type of impact that I can have impractical applications and then like how do we get this to be you know piloted and tested with an organization who can show and share you know some of the results so that's kind of like the trajectory that we're on right now awesome that's fantastic and I wanted to add quickly that we know we have a lot in mind to do but we don't want to limit ourselves we think we can do it we think if we get enough people we can do it and we can make real change in our communities so yeah it sounds like a lot but we're hopeful and we really think it's possible right and so that's where we have like the call for technical and non-technical people to be engaged and that's also for you know even organizations that want to you know provide the thought leadership you know as we're building this out this is really focused on like how we don't build within a silo again in the spirit of open source everybody come one come all and join in the conversation and help us you know advance this but really focusing on like getting community engaged with this like how do we you know get the you know that feedback loop in place so that we're not just saying we think we're solving for this but in reality when we're talking about policy we're talking about licensing we're talking about what can we identify what can't we identify that's a slippery slope that we're not always aware of so having that type of communication is going to be super super important as we're as we're kind of moving through you know call for co-ferential justice and how open sentencing gets built out and adopted and with I know we skipped a couple of slides I want to leave the floor open for any other questions comments that the team wanted to insert before I kind of just let people know how they can really dive a little bit deeper into our work. Yeah Sabine I it was just one slide that we're done but that's okay but I did want to add just based on what you just said to the broader community because as we have identified I mean the various states even local you know the everything is different so we definitely I know we're asking about technical skills that are needed but even generalists are needed or those that are involved in their community you know with their local police department or fire department and just you know just the court system I mean all kinds of of areas we can use that those subject matter experts to help contribute you know to the to the project as well and then just every from every state I mean I'm here in Texas it's very different of course than many other states and so I think it's going to take all of all of us back to what I said earlier in order to really really make a difference. Also I would add to that yeah and there's there's some very interesting discussions going on right now across these call for code for racial justice projects just about like in theory when you're building something like this with open source that processes some very sensitive personal data how do we segment the responsibilities of the people who are writing open source code versus the people who are hosting the code you know obviously there there has to be proper protection of this data and so we're also having some really fascinating discussions around that right now yeah and I think that really speaks to like adoption because people say oh I'm just taking it on but you also have to take on the responsibility of being able to handle this correctly and we want to be able to give that right guidance in those parameters for someone who wants to contribute who wants to adopt it someone wants to play who wants to explore it and saying you do have shared responsibility and being able to use this responsibly don't abuse it you know do you know we try to you know make contingencies against you know bad actors but again you can't sell for everything but we really try to be proactive and I think that that's the space that we're rounding out right now especially when we're talking about like you said for the different solutions within copper coefficient justice privacy is going to be super important how you handle the data you know where should you look for different types of resources wanting to be very inclusive of you know providing that type of resource to someone who wants to take it on so that's going to be super important exactly yeah like the PII whole story of data that whole thing is going to be an interesting challenge oh yeah yeah but what's great we are going to have time to figure it out at this point I don't think that it's it's quite I think it's going to continue again to have people inserted into these conversations but again being mindful of just like the implications I think it is what we're really getting at and what I think we're on the same page about so with that join us join us join us in this work join us in these conversations again we're a couple I mean we're a year old that think about what can you really do at a year old I think we've come a long way since you know that design thinking you know workshop or saying what do we really want to create back in June of last year to haven't created something saying that well we need to solve for these type of problems and what are some other things we haven't thought about like being able to you know bring that out into the wider world is super important and it's also going to be really fun to figure out you know where can people really start to insert themselves and so have that conversation with us if you scan this you'll be able to join our call for code personal justice community you will be able to access you know some free free services and credits on our IBM cloud account join us in Slack be able to look at those tutorials find the different resources for you to get you know tapped in and we're also just expanding the program in really meaningful ways and Slack is going to be the best way for us to continue that conversation and we'll be able to share out you know different initiatives and involving in the projects opportunities for mentorship being able to upskill yourselves attending different events like this hopefully when things are turned back in person and so I think that this is going to be really what we want you to all to continue the conversation with us and we're really looking forward to it really want to say thank you to everyone on a team who's contributed both in the room and out of the room it's been a delight and I'm so happy for us all to be here sharing this with you. Well thank you so much for sharing it with us I mean this is such a massive problem in our society and it is a stain on you know our national history essentially and we need to remove it right. Let's scrub this out and let's figure this out because the disparities are real and they are greatly impactful to large communities and the unfairness and unjustness of it all is very un-American in my opinion but that's just my opinion it's not the opinion of anybody else that I know of so it's just my opinion so thank you very much for putting this together and thank you all for coming on and sharing it with us. Thank you. Anything before we sign off? Join us for the next the I was going to say join us for the next session yes in two weeks July 13th we're going to shift to our policy and legislation pillars one of our teams there and excited for you all to dive a little bit more and to you know what it looks like from voting and policy perspectives which is just as complex and complicated and just as fun for you to get involved with. Amazing. Thank you and tune in check out our calendar if you are looking for call for code for racial justice sessions they are on there just do a quick search to Google Calendar you can add it to your own calendar and it'll be there for you to join when you're ready so thank you all thank you Shana thank you Suron Sabine Stacey Henry thank you all Joanne I really appreciate you all coming on today and discussing this with us so thank you very much for what you're doing it's very important work Thank you very much I'll enjoy Bye Bye Have a good day