Wow, 
this topic.
I’ve been leading a team of (mostly) interns and freelance engineers to create a prototype resource data federation tool in Washington State. The goal is to let data managers using entirely different systems share, and even collaborate, on data management without having to adopt or use one “centralized” system, per se. This also opens up the market for creating niche user facing solutions (like public resource directories) that use the resource data, to @Ben’s point.
We’ve demonstrated viability on two fronts:
- Piping data from multiple sources into a common database using a sharing and permissions management system.
- Using a machine learning assisted process for data managers to rapidly compare and accurately group resource records from multiple sources. This was one of the major obstacles to adoption in our area.
I would be super into a show-and-tell of solutions in this domain 