[OSM-talk] Hierarchy of places

[David Earl]

On 28/11/2008 16:23, Richard Weait wrote:
> On Fri, 2008-11-28 at 12:09 +0000, David Earl wrote:
>> Some kind of "importance" tag could do the job, but how we define it in 
>> a reasonably objective way that could be applied to more than just a few 
>> special cases is hard, unless we use things like population, local role 
>> and area, in which case why not tag these things directly.
> 
> We have place = city / town / village / hamlet etc, for relative context
> and local context.  
> 
> We have population = numeric as well.  Can we not use population as a
> hint for when and where to rank London Ontario and London England?  Or
> Paris Ontario and Paris France?  
> 
> Population also gives us a nice objective data point that can often be
> found on the sign at the edge of town.  It's been in map features for a
> while.  We just need to use it.  

I thought I addressed that one earlier.

Yes, indeed, that is my first preference.

However there is a problem about adding population to places in the UK 
in particular and probably other countries as well, that is the data is 
crown copyright (no single piece is because it is a fact, but the 
collection of population data as a whole has database copyright attached 
to it). (Postcodes are the same in the UK, and it comes back to the 
fundamental reason we're doing it in the first place).

There is the secondary problem that it is not widely used even where 
there is a PD source for the data.

I imagine the data on wikipedia for populations in the UK is derived 
from UK census data and is therefore tainted.

Land area (for which I suggested a somewhat easier to use equivalent in 
"radius") is somewhat easier to apply from the data we are already 
collecting - it's something you can readily see in satellite pictures 
and mapping, and I imagine it would give very similar ranking to 
population - indeed I bet you could get a pretty good estimate of 
population from a simply area polygon drawn around a place.

David