Re: Geopriv compromise proposal

On Thursday 18 June 2009, Doug Turner wrote:
> I agree completely.  if privacy rules are of interest, we should  
> figure out how to apply them globally -- and not special case  
> geolocation.
> 
> For example, right now I type my address into forms when I go shopping  
> online.  This fields should have similar privacy rules as what has  
> been proposed (retransmittal, retention).  However, I do still feel  
> that these attributes will give the wrong assumptions to the user.

Oh well, the formulation of the goal is really simple and everybody 
agrees on it. But the technical implementation isn't. That's 
why I do not want to bother this group with CR implementation.

Working on it right now in PrimeLife. The P3P protocol was 
rather smart, but had flaws for large sites and scaling 
issues. 

The only thing I know NOW is that if forms could point 
to a policy with privacy information, we could solve your
issue. Forms could probably do that using RDFa, so we wouldn't 
need a special thing there. 

But for the API that is designed to get data from sensors 
on the user's device, it would be very beneficial to have 
a "hook/attribute/element" already the API to allow a 
decision engine on the user side for the release of data.
You can't do that by having an RDFa string on 
the web page requesting the location data IMHO. 
Because it would create an issue of scope of the 
semantics expressed in the URI pointing to the policy. 

By having it as a part of the 
location request, the scope is very nicely defined without 
any further need of expressiveness. It just comes naturally 
with the context of the action. The hook aligns well with 
scalability and granularity because it is linked to the 
request itself without further need to semantically define
what a request is, how this request is scoped, which part 
it applies to (all those expressed in some metadata, 
scaling goes wrong).

This makes it worthwhile to define the link directly 
in the API if there is not too much drawback for the current 
work of geolocation to progress as planned.

Best, 

Rigo