[Tagging] Meaning of "administrative" in boundary=administrative, in your country?

[Minh Nguyen]

Vào lúc 14:27 2020-05-13, Joseph Eisenberg đã viết:
> At the US talk mailing list and 
> https://wiki.openstreetmap.org/wiki/Talk:United_States_admin_level 
> <https://wiki.openstreetmap.org/wiki/Talk:United_States_admin_level> 
> there has been discussion about whether or not certain features should 
> be tagged as administrative boundaries in the States of Connecticut, 
> Rhode Island and Massachusetts.

Thanks for everyone who has weighed in so far in this thread. By way of 
an update, the counties in Connecticut have regained their 
boundary=administrative and admin_level=6 tags. The regional councils of 
government (RCOGs) are currently tagged boundary=COG, and some of the 
mappers involved have shown interest in enriching Wikidata with the 
legal nuances of these administrative structures instead of making them 
boundary=administrative relations.

> While all these States have counties, in some cases most of the 
> government functions have been lost, and are handled by the State 
> (admin_level=4) or Town/City government (admin_level=8).
> 
> However, I have the impression that in some countries, certain local 
> administrative boundaries do not actually have "home rule", or the 
> ability to make their own laws, for example in French-infuenced areas?

I think this is even clearly the case in some parts of the U.S. There 
are named but unorganized territories that never had a government of 
their own in the first place. For example, Todd County, South Dakota, is 
coextensive with an Indian reservation and doesn't have its own county 
government, but common sense would call for it to be mapped as an 
administrative boundary, for consistency with the surrounding organized 
counties.

There are also cities where neighborhood boundaries are so well-defined, 
well-known, and well-used that we can map them as administrative 
boundaries. In some cities, these neighborhoods have no councils or only 
advisory councils, but what matters more is that the city government and 
residents understand these boundaries to be the main formal way to 
divide locations within the city.

On the other hand, paper townships that were created as legal fictions 
to facilitate annexation shouldn't be mapped as administrative 
boundaries. After all, OSM is a resource for understanding geography, 
not the nuances of the legal system.

> What is the minimum qualification for a boundary to be considered a 
> boundary=administrative with an admin_level in your country?

I realize you were asking about other countries for a sense of 
perspective, but since I only have local knowledge in the U.S., here's 
an attempt at defining a set of principles for distinguishing 
administrative boundaries from non-administrative boundaries or 
non-boundaries in the U.S. (not necessarily applicable elsewhere):

* _A boundary may be disputed or undemarcated, but it should be 
delimited._ Otherwise, we would lead data consumers to misrepresent a 
subjective or poorly defined boundary as a crisp, objective line. We 
should not map the indeterminate boundaries of neighborhoods in many 
cities, nor ZIP codes, which are actually delivery routes.

* _OSM is a map, not an org chart._ If a district exists solely for a 
government entity to divide its workforce or allocate resources, it 
shouldn't be mapped as a boundary in OSM, even if it's possible to draw 
each district's territory. Examples of unmapped boundaries might include 
divisions of a state department of transportation and a city's police 
precincts. However, it would be a great idea to map the DOT's depot and 
each precinct's police station as POIs.

* _Administrative boundaries are designated by government authorities, 
not private entities._ A retail or residential development may have many 
of the trappings of a municipality, such as welcome signs or a 
homeowner's association that regulates front door colors. But with some 
rare exceptions, they aren't administrative boundaries because nothing 
changes about your relationship to the government depending on which 
side of the property boundary you stand on. Fortunately, named landuse 
areas can represent these boundaries decently. A religious group might 
divide a state into dioceses and parishes, but even if we were to map 
their boundaries, they would deserve a different boundary=* tag with a 
parallel level hierarchy.

* _Administrative boundaries are intended for the general public's 
everyday use, not for specialists._ It's common for junior high 
geography teachers to teach students about administrative boundaries but 
not more specialized boundaries. Welcome signs are a strong sign that a 
boundary is intended for the general public. A specialized boundary 
tends to be strongly associated with a particular government agency 
rather than the government as a whole. Examples of specialized 
boundaries are the Census Bureau's census-designated places and census 
tracts (for demographers), the Office of Management and Budget's 
metropolitan statistical areas (economists), and a city zoning agency's 
planning areas (planners).

* _Don't optimize around a particular data consumer's quirks [1] or map 
categories as relations [2]._ Even though it might be convenient to make 
extracts of groups of states, we'd avoid mapping the Western Governors 
Association as a boundary encompassing its 19 member states. Instead, we 
could map its headquarters as a POI and tag the member states' Wikidata 
items with statements about their membership in the organization.

That it. Nothing about the level of service, degree of autonomy, or 
amount of political maneuvering required to abolish the boundary.

These principles are imperfect: Indian reservations and school districts 
would be false positives according to these principles. What can I say 
-- I'm a mapper, not a political scientist, and there are exceptions to 
every rule in the U.S. Still, based on these principles, a reasonably 
well-educated layperson would probably be familiar with the name of each 
of the boundaries they live in. They would be able to tell if OSM got 
those boundaries terribly wrong so they could help us improve the map.

Besides, I'm already exhausted by having to spell things out in such 
detail to figure out where I stand on these lengthy debates. I would've 
loved it if we could've just left it at, "Use common sense, and consult 
NIST/ANSI/ISO technical standards if in doubt," but what is OSM if not 
gloriously overthought?

[1] https://wiki.openstreetmap.org/wiki/Tagging_for_the_renderer
[2] https://wiki.openstreetmap.org/wiki/Relations_are_not_categories

-- 
minh at nguyen.cincinnati.oh.us