[Standards-JIG] Re: The Great Encryption Debate

[Kevin Smith]

Ian Paterson wrote:
>>This isn't actually that big a deal, as David noted. It's 
>>quite easy to imagine situations where every person on one
>>server is webbed without too many degrees of separation.
>>It would be a problem with roster sharing techniques and
>>even for manually assigned trust values but for implicit
>>trust webs it should be fine.
>
> Are you saying that everyone in this large Web would have many trust
> relationships with the people they have chatted with, plus a few extra
> trust relationships to maintain the structure of the Web?
> 
> If so, then doesn't that prove my privacy point?

Maybe it does and I don't understand, it wouldn't be the first time. As 
I see it, the privacy concern you were proposing was someone having the 
ability to derive that you have a relationship with someone through 
their view of your interactions in the web. I was answering this by 
claiming that you'll have so many people in your web, it'll be 
impossible for an outsider to ascertain if these are direct relations or 
indirect relations through a trusted source. This is assuming we don't 
have a big "I know this person" tag attached to the trust, which would 
obviously destroy my argument completely. I don't think such a tag is 
necessary though.

If I've completely misunderstood, I'm sorry, please be kind.

/K


-- 
Kevin Smith
Psi(.affinix.com) Jabber client project leader
PhD Research Student, Dept. Computer Science, University Of Exeter