Re: Update on geo-hint header

Of course, a client or intermediary may not set the hint at all, or could itself have a bogus hint. However, I think the incentive structure is aligned differently in a useful way. 

Essentially, if the client (or whoever is involved in choosing the IP for proxying cases, etc) wants to have their effective geolocation reflect the IP they chose, they have an incentive to share an accurate database entry. For our clients using iCloud Private Relay, they can know what IP they have, and that it represents a given city or region they are in. This IP is chosen based on the geo IP database of the owner of the IP, so it is authoritative.

The servers are generally using large database services that are often out of date or have inaccurate representations (like selecting a central city in a country). Requiring this middle role is a source of unnecessary error, and usually involves a step of the database provider updating *as well as* the end server updating their copy of the database, which often has a very slow cadence if it is done at all.  

> On Jul 7, 2022, at 4:14 AM, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote:
> 
> --------
> Tommy Pauly writes:
> 
>> IP selection. Specifically, this works around cases where geo-IP 
>> databases are out of date [...]
> 
> What makes you assume that people will remember to update their
> geo-hint header more dilligently than geo-IP-hawkers will update
> their databases ?
> 
> -- 
> Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> phk@FreeBSD.ORG         | TCP/IP since RFC 956
> FreeBSD committer       | BSD since 4.3-tahoe    
> Never attribute to malice what can adequately be explained by incompetence.
>

Received on Thursday, 7 July 2022 13:03:31 UTC