I’ve updated the GeoHipster Map, i.e. the map of the followers of @GeoHipster that I initially published in 2014. Of course, this map is not to be confused with the authoritative Map of GeoHipster Sightings over at the headquarter. Mine is like the theoretical spatial distribution in equilibrium state, the Sighting Map is conceptually closer to ground truth, obviously.
The map contains 2,263 GeoHipster followers whose Twitter profile location could be geocoded. This is out of currently 3,288 followers and using the GeoNames geocoder via the geopy.geocoder Python module.
Volunteered Geographic Information
The GeoNames geocoder is open (+1), yet unfortunately not as robust as one would wish (though, clearly, geocoding is an incredibly hard problem so kudos to the people behind the services out there). In this update, the GeoNames geocoder unfortunately couldn’t handle things like:
- Washginton, DC, USA
- Brooklin, New York City, NY, USA
- combinations of places:
- Brighton+ London
- phoenix + chicago
- non-standard designations, colloquialisms:
- Bean Town
- Esri headquarters, California
- Middle of Nowhere in KY
- Nor*Cal, USA
- specific places and abbreviations:
- UNSW, Sydney, Australia
- UNAM Campus Morelia ENES
- Campus UAB, Bellaterra
- ‘non-sensical’ designations:
- lon, lat
- Probability of being i, or j
- UnDefined xY
- planet Earth
- it depends
- Really, where am I not?
- Seriously, some VGI is lovely, whimsical and/or touching:
- A van down by the river..
- Earth (mostly)
- 4 standard deviations above the mean
- Always near you
- Wherever #BlackLivesMatter
As is explained in the About page, I try to catch some especially coded lat-lon coordinate formats (in WGS1984) before attempting geocoding. Some generics, like “everywhere”, “somewhere”, “global”, “earth” and, of course, “Null Island” are assigned to the latter. Other than that, geocoding might fail. Hence, I’m sorry if for any of the above reasons you didn’t make the map or find yourself mis-located.