Public-interest data visualization

Most public-interest data work fails twice. It fails once at the point of reading, when a chart assumes an audience that already knows what it is looking at. It fails again a year later, when the dashboard that took a grant to build has no one left to maintain it and quietly goes stale in public.

Those two failures look unrelated and have the same cause: the visualization was designed as a deliverable rather than as something that would have to keep being true. A chart that only works in color excludes readers on day one. A pipeline that needs a developer to run excludes the publisher by month six.

The work collected here treats legibility and maintainability as the same problem. That means encodings that survive greyscale, a screen reader, and a printout; and publishing workflows that a small team can run on a schedule without inheriting a system they cannot staff.

This is a working position rather than a finished body of evidence. The projects below are prototypes and independent research, labeled as such, and the methods are standards being applied rather than conclusions already validated.