Thanks Harry for sharing my theory - it’s one of those things that once you think about, it is pretty obvious. We are currently carrying around supercomputers packed full of sensors in our pockets; only a matter of time before this swarm of 1 billion devices can capture the raw data, and OSM contributors can then refine into a map that approximates absolute ground truth.

As for how OSM can begin, I think simple error reporting via photo upload could be very powerful - snapping a picture, categorizing the type of problem and upload. Then contributors can take a look at the photo directly on the map where the error was reported, and potentially there is enough info there to update (add a street name, refine the intersection, etc.

One only need to look at the massive improvement of maps that Waze has seen to recognize the viability of the smartphone-centric approach.

(and as for why Recce app does not use OSM - we need 3D building data, so was not a real solution for us. But hopefully someday OSM contributors can create that data with their phone!).