On Fri, Feb 23, 2018 at 5:25 PM, Mark Wagner <mark+...@carnildo.com> wrote: > Which one is the "best"? If it's the fast route, there's no issue: > both roads are already "highway=motorway".
I think the fastest route is almost always what most people would consider the best. The exception (probably rare/non-existent in the US) would be when the fastest route has infrastructure problems - say, it is a dirt road - while there are nearby routes with fewer problems (say a paved road that takes a little extra time because it is significantly longer). That being an exception, the local community can the discuss the exception (in the forum, so that it can be linked to from the map) and vote against making it the main route. > Fargo and Rapid > City are both larger than any city within 200 miles, which would > seem to make them "large/important", but even by western American > standards, they're pretty small in an absolute sense. I propose that place=* implies importance. Fargo has 105k inhabitants (place=city), Rapid City has 67k inhabitants (place=town). One highway class would connect cities (city-city), a lower class would connect towns and above (town-town, town-city), so the class of the ways making up the route between those places would be of the second class. If cities are connected by trunk and towns by primary, then the route between them would be primary. Now, which is the best route? Google insists on going through McIntosh, whereas Here.com and Waze prefer going through Sioux Falls or Dickinson depending on which direction you're going (mostly because the roads in the area are nearly a grid, which is a characteristic of the US that is not typical of most countries - that's worthy of some discussion). Nonetheless, there are few options to choose from. Google is the dissonant view, so maybe it should be initially ignored. The route through Sioux Falls is part of other routes between pairs of towns (say Sioux City - Grand Forks), so it must be a town-town route for at least that other reason. The remaining question then is whether the route Dickinson - Rapid City should be a town-town route. Dickinson has 17k inhabitants, so it is a town, so that route should be a town-town route. Which one is the main route between Fargo and Rapid City is then not relevant because both options are town-town routes for other reasons. If the situation did not allow us to ignore the problem, the next step would be to collect data (tracklogs), or publicly agree that both routes should be considered important because they're too similar (one possible improvement for the method). It should be noted that the route between Sioux Falls and Fargo would end up being classified as a city-city route because it is part of the main route between Kansas City (place=city, 459k people) and Winnipeg (place=city, 705k people). What if going through McIntosh is really the main route, or if all three routers are wrong? In that case, the initial analysis would be incorrect and the community would need to discuss that particular case and document it. It is possible that routers in the OSM ecosystem (OSRM, GraphHopper) can be used as long as relevant properties (maxspeed=*, and in places with deficitary infrastructure, surface=*) of route alternatives are mapped in OSM so that the routers can take them into account. What if, instead of the US, one is mapping in the UK, where place=* is defined according to public ordinances. In that case, the method still works, it just will judge a place's importance using another criterion instead of population. > Trunk, primary, > or secondary? > > -- > Mark > > _______________________________________________ > talk mailing list > email@example.com > https://lists.openstreetmap.org/listinfo/talk -- Fernando Trebien +55 (51) 9962-5409 "Nullius in verba." _______________________________________________ talk mailing list firstname.lastname@example.org https://lists.openstreetmap.org/listinfo/talk