[ietf-dkim] Re: Attempted summary

[Michael Thomas]

Mark Delany wrote:
> On Tue, Jan 24, 2006 at 10:09:40AM -0500, Wietse Venema allegedly wrote:
> 
> 
>>What is not clear to me is the benefit of a mailing list signature
>>that is required to vouch for the authenticity of someone elses
>>FROM: address.  I see this as a source of confusion with both users
>>and designers, and believe that this is a level of assurance that
>>not every mailing list or other forwarder can provide.
> 
> 
> That raises a question about assumptions. Is the fact that a List
> signature includes "whatever was in the From: " actually vouching for
> the authenticity of that address, or is it merely vouching for the
> fact that this is the content it received?

Right. I've always viewed the h= (or z=) headers as being just
statements of fact: "this is what I claim to have seen". The
binding to an outside header is a function of the signature's
identity (i=), and as currently specified is whether the signer's
identity fnmatch'es the outside identity in question.

> The former - I call transitory trust - worries me conceptually. The
> latter seems safer and simpler.

We need something to perform the ssp binding, however. It's somewhere
between "vouching" and "content received". That is, a signer's action
should not be taken as stating that it believes that the actual
author actually submitted the message in question. Instead it is
merely stating that within its set of internal polices, it's
willing to take responsibility for that message's emanation from
that domain. Since the assertion is as good as those internal policies,
caveat receiver.

		Mike