I agree, we started looking at OSM's address data sets only to find them pretty sparse. For the areas we investigated I'd estimate around 5% coverage from commu

I agree, we started looking at OSM's address data sets only to find them pretty sparse. For the areas we investigated I'd estimate around 5% coverage from community efforts. Made me wonder why nobody's done something stupid like grab the files from http://openaddresses.io http://openaddresses.io and create a reasonable starting point? But that got me thinking why even bother with Nominatim vs just throwing all that data into elastic for better fuzzy searches or match.