The Data That Wasn't There โ How CKAN Changed Everything
How a teenager's burning questions about the world's future became the open data infrastructure powering governments on every continent. Story 01 of 20.
How a teenager's burning questions about the world's future became the open data infrastructure powering governments on every continent. Story 01 of 20.
Are you building systems that query open data portals with AI? How are you handling retrieval reliability, hallucinations, and grounding? Join a semi-closed technical session organised by the OKFN Tech Team around the CKAN Answers project โ demo, code sharing, and community feedback.
After months of development, CKAN 2.12 is nearly ready โ and it's more than just another release. As the last in the 2.x series, it lays the architectural groundwork for CKAN 3.0 while delivering meaningful improvements to file management, data store performance, and the user interface. Here's a full recap of what the core team shared at Monthly Live Session #40 on February 18, 2026.
A year of performance breakthroughs, community growth, and the foundations being laid for CKAN's next chapter โ as the project approaches its 20th anniversary.
CKAN turns 20. Explore how an open-source experiment became global data infrastructure, powering governments, research, and public-interest data worldwide.
A recap of CKAN Monthly Live #39 covering POSE Phase II updates, the two upcoming storytelling workshops, a preview of the CKAN Ecosystem Catalog, and how the community can plug into CKAN@20 anniversary activities in 2026.
CKAN's new interface brings WCAG 2.2 AAA-compliant accessibility, mobile-responsive design, HTMX-powered performance improvements, and a modern design system to the world's leading open data platform. The revamp eliminates technical barriers for smaller organizations while maintaining CKAN's extensive feature set. The update is now in the master branch, with release targeted for 2026 and full integration in CKAN 3.0.
CKAN 2.12 solves the large dataset download problem. A 13-million-record dataset that took 30 minutes โ or timed out โ now downloads in 2 minutes.