[Talk-us-massachusetts] Admin boundary tagging

[Greg Troxel]

"Wayne Emerson, Jr. via Talk-us-massachusetts"
<talk-us-massachusetts at openstreetmap.org> writes:

> 1. Our admin boundary relations have an addr:county=* tag which gets
> flagged by JOSM as suspicious. While updating boundary geometry should
> I remove these tags? Why was this tag added?

Basically I think you should be very cautious to believe JOSM warnings
that aren't really obviously correct.  Thanks for bringing it up here.  

I am not clear on how addr:foo on boundaries leads to implicit
addressing in terms of inheritance from objects inside, both in theory
and in practice.  That would be good to understand to figure out how to
proceed.

While counties in MA definitely really exist, contrary to some the
thinking of some deletionists, my understanding is that addresses, as
used either by state/local government or the USPS, do not contain
Counties as part of the address.*  This is different than Ireland for
example, where the county is part of the addresss (and functions in some
ways like a US state, but in some ways not).

*In MA, you very definitely have to know what county you live in, as it
is required to be put on government forms of various kinds, and it also
shows up in deeds.  But I don't see it in addresses.

So given all of this, the addr:county tags sort of feel like they are
likely not really the right thing, but I really feel that I don't
understand well enough to give that as advice.

> 2. Also some boundary relations have a place=town/city tag. But all
> cities & towns also have a node with this same tag. It seems to be the
> node that is used to render the town name in OSM-Carto. Is it wrong to
> duplicate place=* tags on both the place node as well as the boundary
> relation?

My understanding, slightly fuzzy, is that there are two almost
completely logically separate things going on as fundamental geography
concepts:

  One is admin boundaries, which is about government jurisdiction etc.

  The other is place names, which is about names people have for places
  that have populations (or not), and in general place names refer to
  sort of a point and the area around it with no real clear sense of
  boundary.

In MA, all land is in some town, so we end up with the same word being
both a place name and a boundary name, and unless you are being pedantic
about geography nobody really worries about which one you mean because
it's usually clear by context.  They are sort of the same thing, but not
quite.  And then we get place names that have no boundaries, like "West
Acton", which is very hard to say where it ends.  This doesn't have an
admin boundary becuase there is government distinction, but the village
is well known and everyone knows where it is -- but not where it ends.

I believe that best practice in OSM is to have a node with the place
tag, separate from the boundary, and then to have a relation which has
the boundary as one member an the place as the admin_centre member.
Which is not quite right, as the town hall might not be at the place
that the locals would put the dot on the map when asked "where should
this town label be shown".

> 3. Sometimes both boundary relation & place node will be tagged with
> wikidata & population tags. The most common scheme seems to be to put
> the wikidata on the relation, and put the population on the place=*
> node. Anyone have different ideas on this?

I am really not sure.  I find that population on a place= tag is
semantically troubled as a place= tag does not have a defined extent and
therefore it is not meaningful to talk about how many people are in it.

Overall this feels like it would belong on the relation that contains
the boundary and the admin center.

> 4. When I update boundary geometry with the new Sept 2019 MassGIS
> data, I had been putting source on changeset. But now I wonder if
> putting source tag on lines also might be helpful.

OSM practice for multiple years has been to put source data on the
changeset only, and not to have source data on any of the objects in the
database.  So what you are doing sounds exactly right.

It would be great if someone were to check that all our country
boundaries were intact.