[Talk-us] US Trunk road tagging

[Adam Franco]

On Thu, May 6, 2021 at 3:44 PM Minh Nguyen <minh at nguyen.cincinnati.oh.us>
wrote:

> Some of the use cases I mentioned in [1] would need to avoid a
> hypothetical road_level=* key in favor of something else, some
> combination of expressway=* or access_control=*, width=*, maxspeed=*,
> and highway=motorway_junction (which, after all, isn't limited to
> motorways). That's fine, but it would place a premium on complete
> coverage of these tags, on top of all the work it would be required to
> migrate existing data, editors, and data consumers off of the current
> highway=* keywords. All told, it would be the most ambitious, invasive
> change we make to OSM since the license changeover. Are we stuck between
> that and endless trunk debates, or are there other solutions?
> [1] https://lists.openstreetmap.org/pipermail/talk-us/2021-May/021023.html


I think we have the option of ending the endless trunk debates if we can
come to a consensus  among US mappers that
trunk/primary/secondary/tertiary/unclassified/residential are just labels
for hierarchy levels and that we shouldn't read anything further into the
words used by the labels. If we can come to that consensus and document it
well, then we have a hope of slotting into the global tagging scheme in a
way that doesn't result in broken maps and routers.

Such a consensus doesn't fully solve the whole problem, but it does open
paths forward. As an example, as Moira mentions, a number of New England
mappers have been engaged in a long discussion
<https://osmus.slack.com/archives/CC0LMFWBH/p1609791667022100> in the OSMUS
Slack, trying to come up with a set of principles that would allow a
determination of what road segments should be considered the "top level
regional connectors" and which ones should be the next level down. The
drafts (for Vermont
<https://wiki.openstreetmap.org/wiki/User:Adamfranco/Draft:Vermont-highways>
and New Hampshire
<https://wiki.openstreetmap.org/wiki/User:Aweech/Draft:New_Hampshire_highways>)
are still works in progress, but the broad summary is:

*What are the regionally important cities?* ==>  *Which is the most
important/fastest/high-traffic-volume route between each pair?*  ==> *Those
routes are either motorway if up to motorway standard or 'trunk' if not.*
Parallel routes between the same places that aren't quite as important
don't make the cut and will be primary or less if not up to motorway
standards.

What we've found so far is that following these principles we end up with a
result that is *very close* to the NHS, but is based on principles that we
can articulate rather than simply trusting a governmental process with a
different set of goals. There are particular examples in the drafts where
the NHS includes an isolated roadway that was part of a never-completed
beltway that we choose to exclude based on these principles. I see the
overlap with the NHS (and state winter maintenance priority maps
<https://vtrans.vermont.gov/operations/winter-maintenance>) as an
indication that these principles are on the right track.

What qualifies as a "regionally important city" is something that will vary
across the country and will likely vary between urbanized and rural
settings. These are discussions that can be had between mappers familiar
with a region without having to throw out the entire model and look for
something else incompatible with the rest of the world's tagging.

All of the above does not solve the rendering questions of highlighting
expressways, access-control, and the myriad of other orthogonal facets that
Minh notes. What it does allow though, is a sensible starting place on
which additional tagging schemes can be added without the endless confusion
and churn that motorway/trunk/primary are seeing right now.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20210506/d6a29b80/attachment.htm>