[Talk-in] Classifying places - cities, towns and villages

Arun Ganesh arun.planemad at gmail.com
Thu May 10 06:15:25 BST 2012

On Wed, May 9, 2012 at 3:06 AM, I Chengappa <imchengappa at gmail.com> wrote:

> I don't really understand this reasoning, except that it seems to state
> the idea that cities in India have to be really big and important to be
> called cities. In practice of course the term 'city' is used as it is
> everywhere else, to refer to a certain sized settlement, somewhat
> inconsistently.  There is no real difference in this regard between Europe
> and any other region; if people settle in houses and settlements which they
> do, then these can and should be mapped in a set manner. If you need to
> travel to a very large identifiable settlement you expect to see it
> labelled a city.

This is exactly the problem, because of lack of a clear definition of a
city, I feel it is best to go with the administrative status of the place.
If it is a municipal corporation, its a city, otherwise it is not. Using a
number such as 100,000  to define a city is arbitrary and will lead to
later disputes.

For example if Mandya had a population of 95,000 in the 2011 census, is it
a city? and over the course of the next few years, if the administration
says that the estimated current population is 110,000 will that make it a

When the government has an official system of classifying local bodies
based on size, economic activity, importance and other factors, why not
just map that to the osm tags to avoid ambiguity?

The unit of administration is important because that is what determines the
level of development, planning, infrastructure and funding of civic
services. Mandya (Municipality) and Cambridge (City Council) have the same
population, does that mean they are on the same level of importance on a
global map?

> Even in China, the lowest population of a designated city is 300,000
>> http://en.wikipedia.org/wiki/Cities_in_India
> You may mean http://en.wikipedia.org/wiki/List of cities_in_China<http://en.wikipedia.org/wiki/Cities_in_India>. Cities in China are not designated anything except in Chinese, which is
> then translated into English. This illustrates one problem, the English
> term 'city' is used to translate a Chinese designation for an
> administrative area that can include an urban settlement and lots of
> surrounding rural areas and even other cities.
> http://en.wikipedia.org/wiki/Chongqing is a 'city' with a population of
> 28million, of whom only 6 or 7 million live in the urban area in a total
> area of 80000+ sq. km.. In other words it would be called a province,
> state, region, district or county in most other countries. Nevertheless
> someone who wants to go to Chongqing usually wants to go to the actual
> city, not to the admin region. If they want other cities in this
> 'Chongqing' region, they need to see a city tag there. This is strangely
> enough is what the OSM has there - look in the OSM and there is not only a
> city called Chongqing in 'Chongqing', there are also other places labelled
> cities in 'Chongqing'. Which is what you would expect in a region with a
> population of 28million. This makes the map useful.
> A city is usually this highest centers of economic importance in a country
>> and is region specific.It  cannot be classified just by population or size.
>> In that case, Thimphu with 80,000 people will not classify as a city. While
>> Tinsukia or Mandya will, which are nothing more than large towns of
>> regional importance. These are not places that must show up on a 'city map
>> of the world'
> A  city is a large town; it may everything you have said in the first
> sentence but first it it is a large town. Mandya was a pretty large place
> last time I went through it, as expected by its population. Since you have
> been quoting Wikipedia, I'll point out that it is called a city there as
> well as on its municipal website.
>> The proportion of Cities:Towns:Villages must be maintained in proportion.
>> The census classification  maintains that relationship and will maintain
>> consistency with other datasets following this scheme like Compensatory
>> City Allowance
>> http://en.wikipedia.org/wiki/Classification_of_Indian_cities
>> Hubli-Dharwad is a special case where the urban centres are far apart
>> that it makes sense to have two city nodes, otherwise I see no reason not
>> to adopt the census document as is for this purpose.
> Census classifications should be indicated if necessary by creating
> census-specific tags, of the type that already exist for many places. The
> proportion of cities towns and villages is what actually occurs, and hence
> what should be mapped. Cities towns and villages are the common terms and
> it will not be useful to use the term village to refer to a panchayat area
> that includes multiple villages; a panchayat should be mapped as an admin
> area.  It is up to us to record physical realities in standard terminology
> to make a useful map; it is not sensible to change standard terms to fit a
> classification created for purposes other than mapping. When Cloudmade
> produce their maps, it should not show differently mapped information for
> one country because some contributors chose to follow a government
> designation over standard recommendations.

My view is that the global osm definition of a city being a place with
100,000 inhabitants is wrong and arbitrary. The place=* tag should indicate
ordinal values or rank, and not cardinal values like population for which
you can use the population=* tag. Let us not look at existing definitions,
but choose what is most sensible and practical in our context.

This is an important topic of discussion, I request everyone on the list to
also share their views.
For reference, this is the osm guideline for tagging cities:

And this is my proposal on based on administrative/census classifications:

j.mp/ArunGanesh <http://j.mp/ArunGanesh>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-in/attachments/20120510/a548de36/attachment.html>

More information about the Talk-in mailing list