Wicked problems do not yield to data alone. The data is missing, the sources do not exist, the structures for implementation are absent, and the will to collaborate has not yet been built. Identifying the problem is nowhere near enough.
When Veditum picked up the issue of illegal and destructive sand mining in India's rivers, they started with a reasonable assumption: use machine learning to identify illegal activity, and that would be enough to drive action. It was not.
What they needed to build first was intention. And that required reimagining the entire response, with collaborators across NGOs, academia, the judiciary, media, private citizens, and state actors all in mind.
This talk traces what followed. India Sand Watch is now an open-data archive with over 50,000 data points, built on a Sites of Violence framework that maps the sand mining sector as a series of interconnected violations rather than isolated incidents. Getting there required data pipelines built through scaled volunteering, data sprints, and partnerships, each with its own visualization component.
The artefacts that came out of this work are as varied as the audiences they serve. Zines and sprint report cards for community volunteers. GIS maps for researchers and journalists. Visual evidence packages for court cases. And a physical flipbook that translates a machine learning model detecting mining activity from satellite imagery into something a non-technical audience can follow and act on.
The talk is about what it actually takes to use open data and visualization on problems that resist easy solutions: building the dataset, building the coalition, and building the case in forms that reach people with the power to change something.