The Dashboards of our Discontent

It has never been easier to build a dashboard. The tooling has exploded with drag-and-drop builders, low-code platforms, ChatGPTs this, LLMs, chart libraries for everyone and their mothers, templates for every use case and all flavors of agentic coding. The time is ripe. Government departments, NGOs, and civic tech projects: everyone can ship a public data dashboard now, and everyone does. On the surface, this looks like progress. More data is being published. More interfaces exist. Accessibility is improving. The hammers have found their nails.

I think public data dashboards should be infrastructure, not publications. Many dashboards get built, launched, and maybe written about, but then they sit there, waiting for the next person to come along and build the same thing again. They look helpful with their filters, charts, maps, and download buttons. However, try to put them through their paces and the complexity turns out to be quite shallow. I don't think the problem is technical skill or bad intentions. I'd argue it's inherited assumptions. These are ideas about what a dashboard should be that we absorbed from enterprise BI tools and never questioned for public contexts. We assume that if the data is there, it's accessible. Please don't question it anymore. Can't you see the Big Ass Number? The Donut? The Bars? What more do you want, really?

Meanwhile, what we can build has changed. Not just technically, although browsers and tooling have capabilities that aren't being used nearly enough. It is a change in the possibilities and expectations of how we think about public data interfaces altogether. This talk is a discussion about that gap; between the dashboards we keep building and the public data infrastructure we need. Looking at what's working and what isn't across interfaces, from national portals to scrappy civic tech projects, and drawing on experiments at Diagram Chasing, I'll walk through:


1. A critique of the current landscape of public infrastructure dashboards.
2. What I think are design assumptions worth questioning specifically for this context.
3. Showcase some of the new technical possibilities worth exploring. The FOSS ecosystem is better than ever. DuckDB WASM, PMTiles, MapLibre, webR, parquets in the browser, and so much more! What can be the new mental model of how to build public dashboards that cost $0 to host but are often better than enterprise tools?
4. What I think are some tiny (but big!) decisions that shape whether a data interface serves people the way it wants to.

This session is for designers, civic technologists, developers, open-data enthusiasts, and anyone who builds public data interfaces or indeed, decides what gets built.