Immigration enforcement rhetoric was accelerating through the 2024 election cycle, and journalists had no systematic way to track which false claims were being repeated and how often. The Marshall Project needed a tool that could surface patterns in more than 350,000 public statements at scale — and do it in a way a team of 13 reporters could actually use to fact-check in real time.
I managed the product and the project in collaboration with Anna Flagg. That meant scoping the technical approach, coordinating the reporting team across fact-checking assignments, and negotiating a multilingual distribution partnership with DocumentedNY to extend the reach of the piece into Spanish-language immigrant communities. We also made the editorial calls about how to present the 13 major claim clusters to readers in a way that was clear without being reductive.
NLP and cluster analysis to group similar statements across a large corpus. Python for data processing. Close coordination between the data team and a large editorial team — which meant translating between technical and editorial needs constantly. A partnership agreement with an external newsroom. And a clear editorial north star: the people most affected by these claims deserve accurate information, not just insider baseball analysis.
The piece ran in English and Spanish on WhatsApp, reaching immigrant communities often left out of fact-checking coverage. It gave readers a durable reference for the 13 recurring claims — not a one-day debunking, but something they could return to as a service.