> it's a lot of training to get people to up to speed on coordinate systems, projections, transformations, etc

This can mostly be avoided entirely with a proper spheroidal reference system, computational geometry implementation, and indexing. Most uses of geospatial analytics are not cartographic in nature. The map is at best a presentation layer, it is not the data model, and some don’t use a map at all. Forcing people to learn obscure and esoteric cartographic systems to ask simple intuitive questions about geospatial relationships is a big part of the problem. There is no reason this needs to be part of the learning curve.

I’ve run experiments on unsophisticated users a few times with respect to this. If you give them a fully spheroidal WGS84 implementation for geospatial analytics, it mostly “just works” for them anywhere on the globe and without regard for geospatial extent. Yes, the software implementation is much less trivial but it is qualitatively superior UX because “the world” kind of behaves how people intuit it should without having to know anything about projections, transforms, etc. And to be honest, even if you do know about projections and transforms, the results are still often less than optimal.

The only issue that comes up is that a lot of cartographic visualization toolkits are somewhat broken if you have global data models or a lot of complex geometry. Lots of rendering artifacts. Something else to work on I guess.