On Feb 20, 2008, at 3:47 PM, Paul Kyzivat wrote:

>> .
>
> *ANY* anonymization service will have this problem. It can know who  
> *it* is getting the message from, but it cannot know if that source  
> is acting as a relay point. If it can't provide identity without  
> knowing that then it can never provide identity.

We have traditionally presumed that if the party requesting anonymity/ 
identity can pass a digest authentication or equivalent test that some  
level of assertion about the identity of that party is warranted.


>
>
>> Conversely, there is a legitimate role for a service that is both  
>> an  anonymizer and an asserter of identity. The assertion here is  
>> "I know  who this person is, but I'm not telling you without proper  
>> cause". As  an end user, I'd be inclined to grant this a higher  
>> level of trust  than I would a message that claims nothing about  
>> who it might be from.
>
> Yes, I guess there could be some value in such a thing. But it will  
> only be valuable if the recipient of the identity can determine it  
> has this property.
>
> If I get a call from an address I don't recognize (which is  
> typically be the way I will see it if there is an anonymizer  
> involved) then how am I to know:
> - this is an anonymizer (as opposed to a direct call from somebody
>  I don't know)

the anonymizing identity service has an included an Identity header  
signed by the identity service. That's different from NOT having one.

Consider the difference between a SIP request with a "From:" of "[EMAIL 
PROTECTED] 
" and the same request carrying an RFC 4474 Identity body that is  
signed with the FBI.GOV cert (which validates back to a trusted root).

Or consider the difference between this message, and a similar one  
signed by  the self-signed cert for "john smith".

As in all certificate work, you have to have knowledge of the policy  
and role of a signer in order to decide what sort of trust you wish to  
extend to it.


>
> - that this anonymizer will be willing and able to disclose the
>  identity under certain circumstances
> - what those circumstances are

I see this is more a function of "identity" than "anonymity". If you  
act as an identity service, whether anonymizing or not,  then you  
probably need a policy statement about your identity service just as  
every CA needs a policy statement about its credential service.  But  
mosly, it comes down to a trusted relationship.

A signature from an anonymizing identity service that you don't trust  
obviously adds no value.

Let's talk about the context of eBay. Buyers and sellers occasionally  
communicate outside of the offer-bid cycle. This has raised the  
opportunity for all sorts of fraudulent activity, such as forged "2nd- 
chance" offers.

Ebay found it necessary to put in its own messaging infrastructure  
which provides for authenticated transmission. If you get a message in  
that channel claiming to be from the seller of an item on which you  
are bidding, there is a much higher probability that this is true than  
if you get such a message from a 3rd parry email server. Here, eBay is  
acting as an identity service -- somebody presumably had to be able to  
log into eBay as the seller in order to be able to send the message.  
The seller is still more-or-less anonymous -- we don't know who they  
are -- but eBay has vouched for their role, and we know that if things  
go bad, they'll attempt to apply their published policy in resolving  
the situation.

Extending to the SIP anonymous/pseudonymous/identity discussion, I  
might get an INVITE from "Anonymous" that carries an Identity header  
verifiably from eBay. From this, I'd know that the party contacting me  
has a verified relationship with eBay, and I'd be able to make some  
decision about that message that might be a different decision than if  
the Identity header were not present.

And yes, there's all sorts of room for SAML-like things added to  
Identity headers.

--
Dean

_______________________________________________
Sip mailing list  http://www.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip

Reply via email to