[Tagging] Classifying roads from Trunk to Tertiary and Unclassified

Jmapb jmapb at gmx.com
Tue Aug 13 17:13:57 UTC 2019

On 8/13/2019 12:14 PM, Kevin Kenny wrote:
> Alas, we can't do what Google Maps does, and aggregate the private
> information of everyone carrying a cell phone to measure current
> traffic speeds. That appears to be how Google's router makes its
> decisions.

Of course we used to have something along those lines with Strava.... RIP.

Smaller-scale but still useful -- The rideshare company Juno, which uses
OSM data for both routing and map display, shares a rolling aggregate of
their GPS traces which can be enabled as a layer in JOSM. No doubt it
could be algorithmically cross-reference with road classification; I've
just been inspecting it visually. Juno only operates in NYC for now, and
the data has clear demographic limitations (it shows which roads are
priorities for the kind of people who either work for or use rideshare
services). But it actually covers the city's grid pretty well, and I've
found it useful.

...also there were a couple Maproulette challenge using this data to
spot incorrect one-ways, bad turn restrictions, missing connections,
etc, see

Note that this didn't require a bunch of end-user data consumers to
install an app which shares their location data with OSM. Obviously
Juno's customers do install the Juno app, but the shared data comes from
the driver's app, not the rider's.

Also note that Juno's decision to share back to OSM in this way was
entirely optional -- I don't think any reasonable interpretation of the
ODbL would actually require this. Nonetheless it's a nice precedent, and
if other companies that consume OSM data chose to give back in this way
it would help.

(Of course the trigger for this whole process was the fact that OSM's
road and address data in NYC was good enough that it was  a reasonable
decision to use it in a business setting!)


More information about the Tagging mailing list