[Talk-us] An admin_level for CDPs?

[Minh Nguyen]

On 2012-12-31 12:30 PM, stevea wrote:
> However, there are boundary polygons in OSM which are an odd duck in the
> USA:  a notable one is Census Designated Places (CDPs), which came from
> the TIGER import.  These are a bit like cities in that they are often a
> similar size and population of a town or rather small city.  But they
> are not strictly cities, in that they are derived from the federal
> government (not "negotiated" with a state government like a city which
> is or has incorporated) crafting them for statistical purposes.  CDPs
> have no legal basis as incorporated cities do.  In fact, many of the
> residents of these areas may not even be aware of the boundaries of
> their own CDP.  However, CDPs are useful, as they often give name and
> shape to a place or area which otherwise might not have one, and
> frequently the CDP yields the only boundaries for doing so.
>
> In other words, CDPs (and others, see below) really are administrative
> divisions in the USA, we just don't often think of them that way, and so
> we don't (often) classify them into a hierarchy.  I do believe it is
> proper and useful to do so, but of course we should strive to get to as
> correct as a consensus/result as we can.

I'd argue that not all governmental boundaries need to be tagged as 
boundary=administrative. In Ohio, we've started to retag CDP boundaries 
with boundary=census and place=locality but without admin_level. [1][2] 
They still show up in Nominatim as localities.

As you mentioned, CDPs' boundaries have no legal status. There is a 
reason most residents aren't aware of their CDPs' boundaries: the CDPs 
themselves are a statistical convenience rather than a fact on the 
ground. Yes, they were drawn up by an administrative agency, but nothing 
is administered differently inside them. Unless a data consumer cares 
about census delineations in particular, place=hamlet POIs are more 
appropriate for population centers without formal boundaries anyways.

In states that give civil townships some authority, they are much more 
important to the identity of an unincorporated area than CDPs. The TIGER 
boundary import excluded Ohio townships, so Vid the Kid and I have been 
painstakinglly filling them in.

> I have edited
> http://wiki.openstreetmap.org/wiki/United_States_admin_level to reflect
> the reality of this more complicated picture in the USA, some states
> which don't cleanly follow the 2/4/6/8 model, and at least some of these
> "more federal" entities.  (There are also LAFCos in California, as well
> as COGs in many states, which are state-defined, in addition to MPOs,
> which straddle a local/federal level, and PSAs, CSAs, MSAs, and µSAs,
> defined by the executive branch of the federal government).

PSA, CSA, MSA, and µSA boundaries have more bearing on reality than 
CDPs, but again they don't correspond to any government agencies, so 
boundary=census would still seem to fit these divisions better.

At any given place in the U.S., you're likely to be subject to a variety 
of government agencies with crisscrossing jurisdictions. Some of these 
boundaries matter more than others. For example, in Ohio, it makes a 
whole lot of sense to map townships, even if they sometimes cross city 
limits. But if I mapped every water board, school district, fire 
department, and sewer district in my area, the result would be an 
illegible map. If we considered congressional and state legislative 
districts to be "administrative" as well, it'd be even worse.

COGs and MPOs, on the other hand, often map very cleanly to county 
lines, as do state DOT districts, state patrol districts, and so on. 
Perhaps these boundaries matter a lot more in some states than others, 
but they still seem like highly specialized data that's more appropriate 
for a mashup than the OSM database.

> As a starting point, we can keep this discussion simple and decide
> whether a CDP might rightly be assigned an admin_level of 5, as it is
> both a federal and quasi-local entity which correctly "lands in the
> middle" (below state but above county), or whether it might actually be
> lower than a city (but implying subordinate to? -- doesn't seem
> correct...) with an admin_level of 9.

Just because they can cross county lines doesn't necessarily mean they 
should sit above counties. Some states allow cities and villages to 
cross county lines, but they're still admin_level=8.

> SteveA
> California

[1] http://wiki.openstreetmap.org/wiki/Ohio#CDPs
[2] http://taginfo.openstreetmap.org/tags/boundary=census

-- 
Minh Nguyen <mxn at 1ec5.org>
Jabber: mxn at 1ec5.org; Blog: http://notes.1ec5.org/