Open Justice in Argentina- A Case Study.
As part of our project researching Open Data and Open Justice, we wanted to find and champion good examples of how open data can be harnessed to improve access to justice. In the pioneering work of the No 10 Criminal Court of the City of Buenos Aires, that’s just what we found.
The team managing and working the court there has been developing multiple digital tools since 2016 to enhance cooperation and transparency.
Louis has been speaking to Pablo Hilaire, who is currently based in London and who has held numerous roles within the justice system and was most recently Deputy Judge (Secretario) of the open court, to find out more about the work that’s been going on.
Louis: Pablo, what is open justice to you, and why is it important?
Open justice is about having an accountable justice system. I believe that to have any healthy democratic society, all the branches of government should be accountable. This is essential for any well-functioning democracy.
In the judiciary, due to ingrained cultural practices, accessibility is a problem. Even in how we speak or write, often with excessive use of legalistic terms and Latin. The state is often making life-altering decisions, and people have the right to understand the judgments handed down about them.
Open Justice is about trust, about accountability and about demonstrating the work that the justice system delivers.
Louis: So your initiative began in 2016, what were your objectives?
A lot of this was about introspection. There’s this sense in our region that being part of the judiciary is almost something one has to be ashamed of, that shouldn’t be trusted. We wanted to change that. Many committed people are working within the judiciary, and they want to do good work.
We wanted to look at how we did things and how those things could change or improve. In order to get away from bureaucratic decisions, you have to ask questions about how things work. We believe there is a lot of good work being done within the courts, and we want to show that.
We didn’t receive any external funding, so this required leadership. We had to make sure the teams in charge of implementing the initiative worked towards the purpose. We reached out to civil society organisations that worked with open data standards and they saw the value of the initiative. We began using google drive to manage open and anonymised data of the court.
You can take a look at the drive yourself here.
Louis: You've harnessed multiple digital tools to help improve transparency, could you tell me a bit more about the court's Twitter account?
So the Twitter account was one of the first things we used as a communication channel back in 2016. We chose Twitter because, at the time, it was one of the only social media platforms you could view without having to sign up.
Also, within Argentina, there’s a reasonably high usage rate of Twitter within academia, public policy and government.
We publish all our court listings on the account in advance for the coming week. For virtual hearings, we provide a contact email where people can request to attend the hearing. The day the hearing takes place, we also inform of the start in advance.
This helped us to improve engagement, but engagement is difficult to come by sometimes. A lot of it depends on us at the court asking, ‘Hey, look at this- do you wanna get involved?' We mainly received attendance from university students and lawyers. It may well be that modern courts can have public engagement offices to help develop work like this, which aims to foster participation. When time permits, if we received a group of students, we had a short Q&A session following the hearing.
Although the interaction is not always high. The Twitter account did lead us to get contacted by a developer who built our open Dashboard.
Louis: Your dashboard is really cool and looks better updated than the MoJ's Justice Data dashboard. Could you tell me a little more about it?
So as I mentioned, a developer, Sergio Razza, got in touch through Twitter. He was interested in our work, and once he reviewed our datasets and noticed they were up to date, he offered to build the dashboard for us.
These types of interactions worked as a sort of fuel and motivation to keep doing our work. People that look at the value and effort put into what we were doing and therefore wish to collaborate with us.
The dashboard requires data entry and monitoring by court personnel. It is a good way to show visualisations of our cases through different time periods. It served as a good addition to our public tableau account, where we also publish viz data.
Louis: Why did you put so much work into collecting and publishing data?
I believe it is important for public servants across different sectors to develop a data-literate mindset. For us, this was key since our goal was to give as much data access as we could without disregarding the ethical concerns that opening up judicial data may bring.
We need to find a balance between access to public information and personal data protection. There is a hope that by publishing open data, we can democratise access to it and its use. But we cannot disregard how data is collected. By whom? Who has the power and resources to analyse the data? How will they present the findings? To what end?
Naturally, we are going to encounter political or strategic differences in how the judiciary should approach data and under what principles. What we should all agree on is that we cannot leave data aside. We need to look into justice data from diverse perspectives. The data should help us to deliver a better service. In some sectors, such as health, it might be easier to see than in others.
Efficient data governance models in certain public organisations are lacking. Many times the argument is a lack of resources, and sometimes is a blatant lack of interest “because it’s always been done like that”. I would say that within our region and particularly when it comes to justice institutions we cannot leave data aside any longer.
Louis: Did this open data initiative help you with your daily work at the courts?
Having a data mindset helped us a great deal in our daily work. Our team understood how and why our data was important. Visualising this helped us to analyse changes in who was attending the court and why. For example, during the pandemic, we were able to clearly see the rise in gender-based and domestic violence.
We try to measure as much as we can at what stages of the process we could improve. We wanted to see how long it takes us to solve certain types of cases, such as gender-based violence (GBV) cases. We did this by developing a specific guide to manage GBV cases.
Often, GBV conflicts that ended in the criminal courts had a previous stage at a family court. By the time the case reaches the criminal court, usually, there is a need to act with urgency. The new criminal case might come directly through the prosecution service, and they are unaware that there is a record with the same people, aggressor and victim, involved in a family case. But for the Judge to make better or more informed decisions we need that data from the family courts, so we need to make information requests to another judicial branch. Therefore, we developed the GBV guide so we could identify and have that data more swiftly. There is a need for better interaction between justice case management systems. We also need effective data-sharing mechanisms with acceptable safeguards. The different justice branches shouldn't be siloed.
Louis: In order to manage and use this data, you've touched on the privacy safeguards that need to be put in place, how did you go about this?
This was a learning curve for a lot of us. So, in the first instance, we actually printed out all the documents and manually redacted personal information, then scanned them and upload them to the online repository. This was a burdensome task, as you may imagine.
We then structured our judgments to ensure that personal information was always in the same phase of the document. Also, instead of referring to names throughout the judgment, we used defendant/victim. We used word processing tools to anonymise the data contained and publish the final open data document in our dataset.
We were always open to receiving feedback on the way we handle the data and seeing how we may improve. One of the ideas of the initiative is to enable collaborations with other organisations, such as civil society organisations.
If you are opening judicial data, you have to be aware of the implications and unintended consequences such information can generate if presented in certain ways. Harmful or biased narratives can be created. That is why I think that when thinking of the scalability of initiatives such as this one, it is important to have diverse voices at the table.
We were always thinking of how we could manage our data anonymisation process differently with the help of automation.
Louis: Yes, and in 2019 you managed it?
Yes, in 2019, we started the path towards automating our anonymisation process. In coordination with the Magistrates Council of the City of Buenos Aires - the institution which administers the Judiciary - we hosted a judicial hackathon at the University of Buenos Aires Law School.
There was wide support and participation from academia, public and private organisations, civil society and the general public. The challenge we posed to attendees was to automate the judgment anonymisation process. The hackathon lasted from 9 am-9 pm, and by the end, one of the teams had developed a prototype tool that used natural language processing and did a test run on-site. It worked surprisingly well for an early prototype.
That prototype tool and project were later presented for the EmpatIA bidding process, a cooperation between ILDA and Centro Latam Digital, with the financial support of IDRC and the Inter-American Development Bank fAIr LAC initiative. One of the main objectives of the cooperation was to promote projects that seek to solve public problems through the use of AI, with a focus on uses within the public sector.
The project got selected, and the prototype was developed into a tool that is being used by the court, having significantly reduced the time needed to publish the anonymised judgments. The AI tool was trained to recognise personal data contained in documents. In order to train the algorithm, they fed sample judgments with fake personal data. The tool responds by tagging different entities (name, address, ID numbers) within the document. That tagging is monitored by a person who confirms or rejects the accuracy of the different tags. Once it has passed through that monitoring, the open-source tool presents the anonymised documents.
As you can see, there are some really effective tools and processes that have been led by Pablo and his team in Buenos Aires. What’s most exciting is the collaboration which has resulted in really useful tools to help the publication and analysis of open data within the justice system.
In particular, the AI tool is something that can really motivate the use of open justice data. As Pablo highlighted, redacting data without automated help was unsustainable over time.
The point of this work is to make it more transparent and clear what goes on in the justice system. This applies to any jurisdiction. People who pass through the justice system often already have a sense that the processes are tailored against them. This is made worse by opaque information and overly academic language.
Court No. 10 has made an effort to communicate with those passing through the system as clearly and equally as possible. They have developed a plain language and style guide for judgments and judges, aiming to communicate in a manner that can be understood. Pablo himself notes a clear change in the way people interact with the court.
The problems faced are the same worldwide, and we can solve them by collaborating. Pablo’s been doing some fantastic work, and if you would like to get in touch and help out, please do not hesitate. email@example.com