[OSM-talk] Place: city,town,village,hamlet,suburb

[Mike Collinson]

At 08:16 AM 4/01/2007, Alex Mauer wrote:
>To me, "place=municipality, population=24000, function=capital,
>designation=city" (where function and designation  are pretty much
>pulled out of my ass -- but function would be capital/seat, and
>designation is as defined by local law.) is more useful and informative
>than "place=city".  Renderers can also better use that information to
>determine the relative emphasis based on the area being rendered (e.g.
>don't show cities with a  population below 100 000 when rendering the
>entire country, unless it's the capital city...and then render it with a
>star).

Then the ideal solution would be "place=city" when you start and then 
"designation=municipality, population=24000, function=capital, 
place=city" when you've done more research and want to be more precise?

My observation to date following many debates is that there needs to 
be "Phase I" tags which are easy to fill in from field observations / 
common sense and then complimentary "Phase II" tags/values that allow 
more rigorous accuracy and also minutiae.  I think you are after 
phase II as initially did I.  My point is that "Phase I" tags should 
be preserved as easy to collect for and allow generation of a nice 
looking map very quickly but often have a trade-off in terms of 
definition - as most of us are trained in disciplines that encourage 
precise definitions based upon the smallest set of criteria, that 
makes us uncomfortable.

"Phase I" is when one buys that swanky new GPS device and makes a 
local map based on just the tracks and immediate, obvious 
observations, such as the names of streets and how big they are.  In 
my estimation we (OSM) are just about there with them and debate is 
moving on to Phase II.

As a case study, I've made a basic map of Sydney 
http://almien.co.uk/OSM/Places/?id=358 on Phase I principles using 
essentially just:  name,  place, highway, abutters and then railway, 
waterway, amenity, leisure, tourism and natural=coastline.  Apart 
from the obvious lack of coverage of smaller streets, I think it 
already looks and gives 95% of what a "final" OSM map would present 
as well as providing a base for GPSdrive and open geo-database 
applications [end of shameless self-promotion! There's input from 
Etienne too.].

I've made two main accuracy fudges:

One, road classification is done mostly on gut as they are not 
normally street signed - a "primary" road is primary because of a 
judgement call on the amount of traffic and where it goes, not that 
is an "A" road as usually easy to determine in the field in the UK.

Two, "Sydney" does not really exist at all except as a very recent 
amalgamation of two inner city councils.  The metropolitan area is 
actually a patchwork of about 30 local councils with no authority 
above except the state.  This is not reflected at all in the 
map.  Instead I've used the place=city tag placed at a historic 
central obelisk and then plentiful place=suburb tags.

Mike
Manila