Dr. Patrick Ball, a statistician and founder of the Human Rights Data Analysis Group (HRDAG), joins the Plutopia podcast to discuss how rigorous data analysis can expose and challenge human rights abuses — even when official data is missing or manipulated. Beginning with his work in El Salvador during its civil war, Ball explains how statistical methods, including multiple systems estimation, have been used to identify patterns of violence, hold perpetrators accountable, and support truth commissions worldwide. He emphasizes that governments committing abuses rarely provide accurate data, so HRDAG has developed tools over decades to uncover the truth, even in the face of lies and obfuscation. While Ball’s own focus is mostly international, HRDAG increasingly works on U.S. issues like police violence, missing data, and systemic racism, using AI to process large volumes of testimony and documents efficiently but with rigorous oversight. Despite rising disinformation and political secrecy, Ball remains committed to defending truth with data, underscoring the moral obligation to witness and document state violence — cheerfully and persistently.

Governments that commit human rights violations do not generally provide us with data about it, so our methods have been developed over decades specifically to respond to government obfuscation. People who commit violence and, even more so, those who apologize for people who commit violence, always lie about it — always. My experience is not mostly in the United States. My colleagues at the Human Rights Data Analysis Group mostly work on U.S. issues now. I do not. I wrote that one paper and I don’t really work much on the U.S., I work in a variety of other countries. But our methods, the methods that we use to figure out what’s going on even when there is no official data, and in the face of official lies, are useful in all these contexts. And so if under your hypothesis is that the Trump administration is going to hide data, misrepresent arguments, lie about it, well, our methods will be very useful.