[Talk-us] Proposal: delete census-designated place polygons

[Phil! Gold]

* Nathan Edgars II <neroute2 at gmail.com> [2010-11-10 18:54 -0500]:
> I'm proposing to delete the polygons (in Florida only for now) *if a
> corresponding place node exists* or it can be shown that the name is
> only used for census purposes ("West and East Lealman"?).

I've been thinking about the CDPs for some time now, and talking things
over with some other mappers.  Here's what I think, in rough order of
decreasing certainty:

 * *Any* edits need to be done only by local mappers who are really
   familiar with the areas in question.  There's about a dozen CDP-covered
   places in the Baltimore area that I feel I know well enough to judge
   the Census' boundary decisions.  I don't know DC-area places as well
   as, say, Serge, nor do I know Florida places as well as NE2 seems to.

 * boundary=administrative doesn't seem to really apply to CDPs, since
   they (by definition) cover unincorporated areas.  A CDP does not have
   an associated government or local administration that exercises general
   control over the area.

 * Place names are a really fuzzy concept, since they're basically defined
   by what the people who live there call it.  If there's a legal boundary
   from town incorporation or whatever, that's one thing, but
   unincorporated settlements (or neighborhoods within anincorporated
   entity) are often amorphous and very fluid over time.  Serge gave an
   example of North Bethesda, which was a place name created by real
   estate agents a few years ago as a marketing ploy, but now there are
   people who will tell you they live in North Bethesda.

 * Because place names are fuzzy, the CDPs aren't always going to line up
   with the boundaries perceived by people who actually live there, but
   there's often a rough correspondence.  In cases where the CDP is based
   on an actual place and roughly corresponds to that place's extent,
   there's a benefit to having it in the database, because it can at least
   provide hints to name-based search engines like Nominatim.

 * In general, the database is made better by adding more data and
   increasing the quality of the data we have.  Removing data is less
   often an improvement, so deletion should be done with greater care and
   respect for other mappers.

And what I think I will do with the CDPs in my area is this: At the very
least, change them from boundary=administrative, admin_level=8 to
boundary=census, census_level=CDP (other suggestions for replacement
tagging are welcome.  I'm still pondering what would be the best approach
for cases where the CDP might be misleading for searches.  In particular,
the CDP for Arbutus[0] encompasses two distinct other places, Halethorpe
and Relay.  On the one hand, I feel like things named "Arbutus" should
only cover places that people actually think of as Arbutus, but on the
other hand if I alter the CDP's boundaries, it's not accurate to the CDP
any more, so it shouldn't be tagged as a CDP.

  [0]: https://secure.wikimedia.org/wikipedia/en/wiki/Arbutus,_MD

I'm also considering using a couple of CDPs as bases for polygons tagged
with place=.  In particular, the Lutherville-Timonium CDP seems to match
the generally-accepted boundaries for Lutherville and Timonium, at least
on the north, west, and south sides (I'd have to see about the east
side).  With another line drawn across the division between Lutherville
and Timonium, the CDP makes a good basis for the place polygons for those
two places.

-- 
...computer contrarian of the first order... / http://aperiodic.net/phil/
PGP: 026A27F2  print: D200 5BDB FC4B B24A 9248  9F7A 4322 2D22 026A 27F2
--- --
Decree that all hay be shipped in tightly-packed bales.  Any wagonload of
loose hay attempting to pass through a checkpoint will be set on fire.
                       -- Evil Overlord's Handbook, entry 181
---- --- --