Delicious Democracy: snagging and mapping citizen data
summary
Crowdsourced map of election-day stalls started in 2013 and now runs on a six-person core team plus ~20 volunteers, ingesting 12–24 electoral data sources per campaign and handling tens of thousands of social posts. Tech history: Google My Maps → Maps Engine → Carto → Polymer/PHP → React/Python (v6), with a simple GeoJSON blob served to clients. Key builds and fixes are a custom triage tool for social streams, automated data-validation that corrects bad coordinates, and an API with access controls; ongoing pain point: $1,000 geocoding bill.
AI-written summary · about
chapters (12)
- 0:00Start
- 2:14The first Democracy Sausage?
- 4:34Evolution of the map
- 7:38How the crowdsaucing works
- 9:18Getting data
- 11:16Poor quality data
- 13:10Scaling the crowdsaucing
- 13:59Scaling the human factor
- 14:56When others want to use the data
- 15:37The media
- 16:46Challenges we're yet to solve
- 18:47Challenges we're glad we haven't had to solve (yet?)
related talks
26mMapping for impact: Scaling social good with DataKind and Mapbox
26mOpen Source and the Volunteer Workforce
18mBuilding Maps with Impact on React and the JAMstack
22mElection Night Reporting Using R & Quarto
28mBuilding a scalable public map for Singapore's general elections 2025
49mSaving the world with JavaScript: A Data Visualization story
22mKeynote: Open Data @ Facebook
37mSlippy Map Style: Geo-canvas design with CartoCSS