On Apr 4, 2008, at 2:50 PM, Hadriel Kaplan wrote:
>
>> -----Original Message-----
>> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of
>> Elwell, John
>>
>> There seems to be some interest in adding some indicator to the URI  
>> or
>> to the header field to warn of the limited guarantee of  
>> authenticity or
>> the opposite (i.e.,lack of such an indicator would mean the  
>> guarantee is
>> low, i.e., the identity has not been verified).
>
> I haven't thought about this since it was last on the list, and I  
> think I was a proponent of it at that time, but why were we thinking  
> this would ever be set?  I mean if I'm a clid spoofer, why would I  
> ever want to add such a param??  Isn't it like a "spam=yes" flag  
> you're describing, and then expecting potential spammers to set it?   
> I've forgotten the train of thought which lead to this param as a  
> viable solution.

The Identity header essentially signs the From. It's traceable back to  
whoever made the signature. So, if you were an identity server, you  
could spoof any kind of caller ID you might want. But we'd know it was  
you doing it . . .

What we're looking for is a way for an RFC 4474 identity server to say  
something about how strong it believes its assertion to be. Sure, it  
could lie, but then we wouldn't trust it in the future. It's much  
better if it volunteers the information to us that it doesn't REALLY  
know who the caller is, since it has only the caller-ID information to  
go from.

If an identity server were to fully RFC 4474 "sign" a message rom a  
PSTN, it's even possible that the identity server operator could be  
held legally liable for inaccuracy in the asserted identity. In other  
words, if a caller-ID spoofer made a call through the example.com  
gateway, and the example.com identity server attaches an Identity  
header on to the resulting INVITE that asserts the spoofed telephone  
number identity, then someone injured (defrauded) by the caller could  
claim negligence on the part of the example.com identity server and  
sue for damages.

This is why things like identity servers (and CAs) have statements of  
policy that describe their procedures and the strength of claim being  
made. Those published policies are legal contracts with the users of  
the certificates. It's part of why x.509 certs are so big and have so  
many things like subject-alt-names and extended key usage. SIP  
identity services need the same sorts of legal flexibility. Being able  
to sign a request on the basis of "I swear to you that it came through  
my gateway, but I don't know who it really came from, so I make no  
assertions about this identity other than that the request came  
through my gateway with the calling-party-ID information described  
herein" is really, really important.

--
Dean

_______________________________________________
Sip mailing list  https://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