I was wondering if you had any insight into making MapAnt? I liked the idea, and was interested in how the project might look for North America (US and Canada) if that would be possible...
I am not sure do I have any useful idea really. Maybe instead of just "processing it all as one batch downloaded from a single donwload platform" it might need some prioritizing and flexibility with datasets. Maybe also something more crowdsourcing based, like just letting folks do the processing and let them upload processed data as tiles to a server. So users could find the data, process it and upload the areas they find useful for themselves and others could find them from there and use it without having to process it again by themselves. And by expecting maps being converted to utm zones and having some fixed pre-decided tiling/naming scheme you could make data seamless enough, re-projecting and tiling is something folks should be able to do. And processing should possibly be lidar only - no vectors - and you could instad overlay road etc data on the fly and update it without having to process/render lidar based part again.
Maybe there could also be two stages, prepocessing (reprojecting, tiling, thinning) of lidar tiles of interesting areas. And then processing them. So those who know how to prepare data could do it,a nd then there wold be simple app that does notheing else than downloads and processes the prepared data available without user haing to know nothing about tiles or coordinate systems.
In Norway they have been preparing data for over a year now to get it sorted out well enough to be processed as one batch. NA data is one magnitude more complex and large, so this preparing phase would be a tad more troublesome.