Sinapsis (Synapses): Fighting corruption with open data to discover connections between illegal companies that received public resources in Latin America
Category: Open data
Organisation: Animal Político
Organisation size: Small
Publication date: 12 Sep 2019
Credit: Yosune Chamizo Alberro, Gilberto León, Mauricio Martínez Robles, Jesús Santamaría, Elizabeth Cruz, Ethan Murillo, Jorge Ramis, Nayeli Roldán, Manuel Ureste, Miriam Castillo, Arturo Ángel, Tania L. Montalvo, Francisco Sandoval, Daniel Moreno
Mexico is extremely corrupt, politicians steal money without sanctions.
The modus operandi all across Latin America is very similar: government officials create ghost companies, the government assigns projects to these fake businesses and the money disappears.
How could we detect these companies? That’s how we started this project. Sinapsis is a tool that visualizes relationships between businesses and government entities, as well as how companies and their personnel are interconnected. As a free, open-source project, it’s designed to help detect irregularities and suspicious activity in business interactions and practices.
Animal Político has already published two very important investigations done with Sinapsis on corruption in Mexico: The Ghost Companies of Veracruz and The Master Scam. The latter revealed that the Mexican Federal Government disappeared 7,670 million pesos (around 400 million dollars) through 186 companies. These investigations won the National Journalism Award. Even more important, they have been the basis for the prosecution of several former and current public officials in Mexico.
It is a tool that has been proved and used. Only a couple months into its publication, there are already 180 different projects using it and at least 5 people forking the code on GitHub to adapt it to their needs. The tool also includes the investigation methodology; tips on useful public databases; promotes the use of open data pre charging relevant databases on the tool to facilitate analysis; builds community through a Telegram channel open to doubts and suggestions; and is working on the process of systematizing information on corrupt companies across Latin America.
To date, we have allies collaborating from Argentina, Bolivia, Brasil, Chile, Colombia, Costa Rica, Ecuador, Guatemala, Honduras, Panama, Paraguay, Peru, Dominican Republic and Venezuela.
During this first phase, we want to train local journalists in Mexico. So far, we have given workshops in Mexico City, Hidalgo, Guadalajara, and have already scheduled Puebla, Tijuana and Mexicali. We would like, further down the line, with the right allies and resources to be able to reach different countries in Latin America.
The project has had international recognition, the United Nations Office on Drugs and Crime has invited us to their regional convention in Quito, Ecuador and are presenting in TicTec in march in Iceland.
Beyond the platform, Sinapsis is also an investigation methodology. The way in which we processed data had to be easy for journalists (that generally don’t have a lot of technical knowledge) and functional for a developer (machine readable).
Part of this process was generating a CSV template that could be easily opened in Excel for journalists to work on and understand, and that could be uploaded to Sinapsis at the same time. Gathering the information from different sources was done manually, however, being able to categorize and standardize the information has proven very valuable for the investigation of government-companies corruption schemes in the region. This process was possible through the interdisciplinary work of journalists, information designers and developers.
Technically, the platform is built with React (ES6), making it modular, flexible and scalable. It uses as its main library D3.js and the visualizations are SVG files to allow designers and journalists to export it in different formats according to their needs, providing small news outlets and freelance journalists without resources with graphic material ready to publish.
Sinapsis doesn’t only show the visualization of connections; it also gives statistics on sums and number of coincidences; a geographic interactive map that georeferences addresses providing strategic insights for field research; an exportable CSV with the lists of coincidences; and it is possible to export the data in JSON formats (for developers and data analysts), CSV, as well as a .sinapsis file that has the full project to facilitate collaboration between investigations.
What was the hardest part of this project?
To explain this, we’ll provide some context.
Animal Político is a digital and independent newspaper in Mexico, specialized in topics such as human rights, migration, drug trafficking, security, politics and inequality.
During the Ghost Companies of Veracruz and The Master Scam investigations we worked through: How could we detect these companies? Having limited time, how could we systematize and automatize this process? With the volume of information we had, could technology give us additional insights to our manual analysis?
The hardest part was conceptualizing the project. It started with hundreds of public documents, a hypothesis and a lot of information to process. We developed the tool as we were structuring and advancing the investigation with an interdisciplinary team. It is very rare for a small independent news media to build technology.
At the beginning, Sinapsis was going to be a dashboard showing coincidences between companies, and as time passed on we kept adding features. First a node map, later a georreferenced map, data analysis and statistics on the databases. The project slowly grew over time.
Additionally, we faced challenges with the interdisciplinary nature of the team: developers, information designers and journalists work in very different ways. Being able to communicate effectivly and build this product took 3 years!
Sinapsis represented several challenges in a technical level: it is a tool that had the user’s privacy as a top priority, which meant that the data processing had to be done through the browser without using a backend.
We know Sinapsis is not self-explanatory yet, which is why we added a “guide” button so that the users get to know some parts of the tool. However we know that it is not enough, so we started giving workshops and would like to produce tutorial videos / material in the next iteration.
What can others learn from this project?
People will be able to see how with a very small team, from an independent news media with very little resources, you can develop tech and tools that have an impact in society. They will also be able to see that the investigation journalism world can also think about technological outputs, since developing software in media almost never happens.
Newsrooms can learn a lot from the interdisciplinary work thar makes these types of projects possible, and this might sound obvious in other countries or contexts, but in Mexico it is very rare that a media outlet hires a developer to work full time on journalistic projects. We also are not aware of any other outlets with an information designer for content production.
Showing work flows and methodologies would be very valuable for the audiences.