[Tagging] Tagging town/village/hamlet - am I misunderstanding something?

Kevin Kenny kevin.b.kenny+osm at gmail.com
Wed Mar 29 00:22:43 UTC 2017

On Tue, Mar 28, 2017 at 7:31 PM, Greg Troxel <gdt at lexort.com> wrote:
> This is definitely messy.  Legally, in Massachusetts we have cities and
> towns, both admin_level=8; they are the same thing, but (slightly
> oversimplifying in a way that doesn't matter for this dicussion)
> governed by city council vs town meeting.
> But, administrative boundaries and the human geography notion of a
> settlement hierarchy are really totally separate things.  One is about
> borders, and the other is about concentrations of population and a
> shared sense of place.  Yes, the human geography notion is influenced by
> borders; El Paso and Juarez are surely considered separate settlements.
> In a strong sense they actually are different places.
> This is further complicated because in many areas every bit of land is
> built up with housing, except where it can't be (wetlands, etc.).  So
> the old notion of a village with a few bouses, and then miles of farms,
> and then the next village, doesn't really fit.  But if you adjust that
> to be a village center with a church or shops, and then miles of just
> houses, it pretty much works.
> In the US, there is also the notion of place names for areas that used
> to be villages that mattered, even though these days you can hardly tell
> except to notice a few old houses close together.
> While the center of a political entity usually qualifies as a populated
> place (to use the GNIS term) or settlement (to use the human geography
> term), the defined center is usually not the centroid.  So I think it
> makes sense to both define the boundary and the center.  That should be
> sort of a relation, and the point tag should also have a settlement tag
> in addition to being noted as the center of the admin_level entity.
> There are towns near me which have an admin_level=8 boundary, and also
> have "South Foo", "North Foo", "Foo", and "West Foo" villages.
> Originally these were clusters of houses.  Then the railroad came sort
> of close, and then the sense of center moved closer to the station.  Now
> the railroad is less important and the sense of center has moved in some
> cases closer to a highway junction with lots of shops.  Sometimes the
> old town centers are really in the exact same place as in 1750.
> As for whether there should be polygons for the settlement objects, I
> tend to avoid that, because it's really hard to say what's in the
> village vs near it.   Just ask 50 people in Cambridge, MA whether some
> particular building is "in" Harvard Square.
> My overall summary is that boundaries and settlements are different,
> we should tag both, and we should not blur them.

I tag boundaries when there are boundaries. In many places in suburban
New York, the hamlets (not self-governing in any way) have well known
boundaries, and the locals can tell you with some consistency who does
and does not live in the named community.

New York's settlement hierarchy is very messy indeed, and indeed
not quite hierarchical.

Fort Montgomery is one of these. It stops at the parks, the river, the
village of Highland Falls, and the West Point reservation.

A node for the 'center' isn't a bad idea, and for Fort Montgomery, the human
'center' of the settlement is the couple of blocks of Route 9W that have
the post office, the fire station and the school.

Can you give me an example in OSM of the type of relation you have in mind,
so that I could ape it? If such a relation is described on the Wiki, my
Google-fu is failing me.


More information about the Tagging mailing list