[Tagging] Classification ( and symbols ) according to context ( relative importance, size, or other 'context' )

Michael Patrick geodesy99 at gmail.com
Thu Jan 24 05:48:08 UTC 2019

 > .....e all reference to number of inhabitants, &
> base the decision on each mappers own recognition of how "important"
> this is, so an isolated "village" with only a few hundred people
> in it, but which is the main centre for this area will be a town, &
> maybe even a city? ... No 'one set of rules' is going to match world
wide. ....
> One guide should be that surrounding places must be relative in level of
> important to the place that is being mapped. ...Necessity makes this
> population centre very important for the few people living in that area.
... (etc.)

Fundamentally, you are attempting to make a categorization of the
rankings in the distribution of counts ( population ) of some subset
of a domain.

Economist Xavier Gabaix in 1999 wrote a much-cited paper describing
Zipf's law for cities as a power law ( see
). It's been repeatedly tested
over decades according to what conditions it holds or does not.
that provide wide area web cartography using a wide range of zoom levels
usually use some form of it, and it's available pretty much in every GIS
symbolization set for this reason.

There are some edge cases where mega-metropolitan areas merged together,
and at the very bottom bottom literally in the weeds where one might be
a few huts, but it's very robust, and more importantly there are proxy
which can stand in for the absence of direct population counts. But it's
really sensitive to 'accuracy' except at the very top end. So people have
road miles,extent area, night time lights, and other statistics for the

( See
): " Zipf’s Law does not just hold true for cities
in the United States, but rather it has been correlated with urban
totals in nearly every developed country across the world. Additionally,
works well when “Metropolitan Areas” are used – cities defined by the
distribution and connectivity of populations rather than arbitrary

In case of OSM, if the original count rankings were created by country (
than globally ) then categorized ( seen always seems to be the magic number
it works fairly well, without any reliance on arbitrary population count
and nomenclature ( village, city, etc. ) based on a single European country

Michael Patrick
Data Ferret
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20190123/95604afa/attachment-0001.html>

More information about the Tagging mailing list