[Tagging] Tagging of State Parks in the US

[Kevin Kenny]

On Sun, Jul 28, 2019 at 8:04 AM Paul Allen <pla16021 at gmail.com> wrote:
> I have no objections to protect_class as supplemental information that data consumers can make
> use of as they wish (including ignoring it).  I have an intense dislike of numbers being used for
> anything other than numeric values because they are not amenable to human inspection.  Sure,
> editors can unobfuscate things by using an internal lookup table, but that isn't a complete solution.
> Compare an overpass-turbo query for leisure=state_park and for protect_class=21.

I dislike the numeric classification as well.

I dislike 'leisure=state_park' for two reasons.

First, it preëmpts the 'leisure' tag. It turns out that there are
State Parks that are also something else in the 'leisure=*' space. A
handful in New York are tagged 'leisure=golf_course' and should retain
that tagging, but it would be good to have tagging that indicates the
protection status. Bethpage State Park wouldn't be turned into condos
just because some future administration decides that the state's money
ought to subsidize something other than golf - it would be repurposed
to some other recreational use. (The legality of releasing it from
protection would be a complex political question; repurposing it from
one recreation to another could be an administrative decision by the
executive branch.)

Second, it pushes the problem down one level. Near me, there are
'County Parks' that are functionally pretty much the same as State
Parks, and even 'County Forests', 'County Nature Preserves', 'County
Wildlife Sanctuaries', and so on... and moreover, even some similar
objects at the town level. What is significant is the protection, not
the level of government that establishes it, so having 'state' in the
name is simply a recipe for more confusion. We already struggle with
that issue every time that the national_park tag comes up, because
'national park' to IUCN describes a purpose and level of protection,
not a particular ownership. In fact, IUCN's guidelines have a
cross-cutting taxonomy of management that divides it into four
categories: government (including national, sub-national and delegated
entities  - the last refers to formal delegation to an NGO); shared
(collaborative; joint; transboundary); private (individual; nonprofit
- NGO, university, cooperative; for-profit); indigenous (declared and
run by the local indigenous community). Any of these can apply to any
of the types; hence, in areas that I try to curate, there are some
relatively surprising combinations. The Adirondack Park is a
non-Federal 'national_park' with shared (collaborative) management
between the state of New York and private landowners; the New York
City watershed is a set of sustainable-resource-use protected areas
with private management (It's outside the city's boundaries. New York
City is not its government, merely its landowner functioning in the
capacity of a private entity); the Huyck Preserve is a habitat area
with autonomous NGO management; and so on. It's really important NOT
to think of the problem as being 'the wrong level of government.' Even
my for-profit employer is in the game. There are a couple of
recreation grounds near my work site that are owned by the company but
used by the town under a permanent easement. They exist in part
because there is a required setback of 500 m (I may be wrong about the
figure) between certain hazardous activities at my workplace and
permanent human habitation, but I'm sure that the company is also glad
to get the tax writeoff and have its name on the sign as the donor.

The 'boundary=recreational_area' idea would work for me if people were
actually to get behind it.  We would, of course, have to address 3785
somehow - and deal with the fact that means a database reload. Just as
'protect_class=24' is now dead (replaced with aboriginal_lands), we
could migrate the other protections over to a named scheme.

In fact, that would work with the IUCN codes as well - we don't have
to use them unchanged, we can name them!  (Of course, we could
preserve a correspondence between IUCN's taxonomy and ours, and
continue to accept the IUCN codes as synonyms.) In fact, IUCN gives
names to the categories in
https://www.iucn.org/theme/protected-areas/about/protected-area-categories
- although we OSM'ers would probably go with something a trifle less
verbose, such as 'strict_nature_reserve', 'wilderness_area',
'national_park' (already used), 'natural_monument', 'habitat',
'protected_landscape', 'sustainable_resource_use_area'.

I drafted the protected_area idea a short while before I learnt of the
issue with the database, and learning of the issue has already made me
less sanguine.  The only thing that kept the idea alive for me was
that 'area=yes' is available as a workaround, and that most areas
tagged with 'boundary=protected_area' also have other tagging that
does force a polygon to be created, although there appear to be
thousands that do not.

So, I'd like to emphasize:

  * The tagging should address protection status and purpose, not what
level of government (or private agency, or indigenous community)
manages it.
  * The purpose should be of a sufficiently general nature (e.g.
'recreation') that a typical state park can be preserved as a single
named entity.
  * If the new tag requires a database reload to become a polygon,
then it should not conflict with the existing tagging on typical state
parks. If the scheme punishes mappers by failing to render correctly
tagged features while rendering incorrectly tagged ones, it will not
take off.

The reason for the third bullet is that I understand that a database
reload incurs a massive disruption to operations and can be done only
for extraordinary reasons. The initial support for 'protected_area'
waited several years - during which time, mappers were directed that
there was no other correct tagging for objects such as US National
Forests (large portions of which are not forested, and which are not
strict nature reserves, and which are not National Parks), forcing
them to decide between tagging for the renderer and waiting years for
their objects actually to appear on the map. Small wonder that these
issues have become so contentious - the perception is that one part of
the community says to another, "your feature does not deserve to be
rendered!"