Working around the limits of geolocation

I had some queries over the reliance on Geolocation for managing the local aspect of ‘glocal’ in Pajamanation. The logistical problems are well expressed in mediajunk:

When Geolocation Gets Too Clever – mediajunk

Geolocation works in two steps:
1. A script detects the user’s IP address.
2. The script looks up a database of IP addresses and their associated countries to tell where the user is located.

There are potential problems with both steps:
1. Many users go through proxy servers, so the IP address that appears to be associated with their computer is, in fact, the server’s IP address, which may be in a different location.

2. There are many databases of IP addresses and their associated regions (some free, some commercial) but none is even 90% accurate. For example, look at this table of accuracy for city geolocation, from one of the leading providers of such databases, Maxmind.

The solution that we will be able to implement is one based on giving priority to user choice, which is always a good thing in my book. So there will be two ways to determine location :

1. registered users – they choose where they are .
2. visitors – GeoLocation is the only way to know where they are coming from