We were assigned a new block from ARIN two weeks ago and are getting several
reports from end users that the Spanish and German versions of Google's
search page are coming up.
IP2Location and Maxmind are mostly correct, but there appears to be no way
for me to verify that Google and Akamai have 96.31.0.0/20 listed correctly.
Perhaps someone can point me in the right direction so I can make an
authoritative check.
Thanks,
Frank
Maxmind www.maxmind.com is a fairly good indicator of what geo-locators are
seeing, but I recall a recent thread here that there have been disagreements
between the various geolocation services.
I think that some of it depends on the reference sources i.e. how many and
what the algorithms are and also the update frequency. Using plain old whois
data, for example, is notoriously unreliable, but definitely "usable" as a
reference.
Might be nice if search engines and cdn's had a verification site for
checking and suggesting corrections i.e.
www.favoritesearchengine.com/ipverifiy
Best,
Martin
Funny this should come up...
I've found that a local Mobile Broadband outfit here in NZ are using an IP range that Akamai's Geolocation service thinks is actually in New Jersey. Causes me some oddness as a result - this despite the fact that Maxmind has it correct.
Whilst investigating this (just the day before yesterday) the following URL came to my attention - allows you to check the IP you're on against Akamai's Edgescape product:
http://www.akamai.com/html/technology/products/personalization.html
Their customer care team were responsive to an email I sent them (but were unwilling to make changes without the request coming through their customer... which in itself seems mad, their info is wrong - why don't they correct it?)
Geolocation is clever, but seems to be very retrospective, and very much an imperfect science... there should be some obligation on people offering the service to work very hard to keep their data current, given the effect it can have on an end user.
Mark.
Or maybe they just shouldn't rely on it so much.
It annoys me at the hoops I have to jump through to change the
language on Google-owned properties when they think I'm coming from
Czechoslovakia or Malaysia or some such... Some, like Blogger, still
don't do it right...
Or maybe they just shouldn't rely on it so much.
It annoys me at the hoops I have to jump through to change the
language on Google-owned properties when they think I'm coming from
Czechoslovakia or Malaysia or some such... Some, like Blogger, still
don't do it right...
[ clip ]
>
> http://www.akamai.com/html/technology/products/personalization.html
>
Like what? I have found that interacting with your favorite search engine is
generally easy and interactive as long as you know where to look. I think
that's really the problem at hand, the interaction and where you actually do
it.
Overall, geo location has turned out to be a somewhat valuable tool in terms
of language, fraud, and localization. I think that it's important to
continue to urge improvements in this technology, not divestment.
NoSoObOp: Renesys ought to be playing in this space. 
Best,
Martin
Is it really that difficult to check the Accept-Language header for determining the language to use? This is particularly useful in countries that either have no official language, countries that have more than one official language, and tourists.
What's the accuracy rate? Combined, it is probably higher for that specific
use case. I digress, ask your favorite search engine. Perhaps they'll
explain. It is certainly interesting.
[I think that] Over the last year we're seeing an uptick in geo-location
problems addressed on NANOG because it's becoming profitable (whether
through ads or fraud correlation through ip reputation) and it's probably
not going away.
Personally, I'd be happier with the publicly available checker. *shrug*
YMMV and Best,
-M<
I don't see how this technology can be improved past a certain point,
because the criteria that are used to determine location are only
coincidentally tied to location (they are the result of administrative
policy and/or configuration). At best, they provide a false sense of
"security".
--gregbo
Thanks for all those who responded on and off-list. Several persons
confirmed for me using their Akamai account that the address space was
correctly listed in Akamai's database, and between Google's quasi-generic
online form (http://google.com/support/bin/request.py?contact_type=ip) and a
Google employee, I think we'll get this straightened out.
Frank