On Thu, 2006-09-07 at 07:23 -0400, Wietse Venema wrote:
> Hallam-Baker, Phillip:
> > I think it is entirely likely that bigbank.com would have a
> > situation where the mail servers for its east coast offices were
> > adding signatures but the ones for the west coast were not. The
> > part that is less easy to see is whether there is value to the
> > short term fix. It is probably easier to just do the deployment.
> > But it is not certain that this will be the case.
> 
> This hypothetical bank can use the hypothetical "I sign some of my
> mail" policy until the DKIM roll-out is complete, and then transition
> to the "I sign all my mail" policy.  
> 
> A per-user mechanism is not the obvious solution for this problem.

Agreed.

Problem: When to trust just the domain?

Case 1:

Bigbank.com wants their email-messages annotated with high assurances on
the basis it verified as coming from their domain, but a general high
assurance is not appropriate for all of their messages.

Case 2:

BigISP.com signs millions of messages per day, but only specific
email-addresses are protected internally and are from trustworthy
sources. BigISP.com wants only messages that have these protected
email-addresses annotated with high assurances.


Solution:

An email-address specific policy is a natural way to convey which
email-address should be annotated as being trustworthy based solely upon
the domain from which it verified.


DKIM's primary role at protecting transactional messages from being
spoofed is greatly enhanced with a per email-address policy.  Private
exchanges can utilize other clues, such an email-address found in the
address book when making different levels of assurance annotations.

Domain assured email-addresses may not be found in an address book such
as do-not-reply@, where each organization has a practice of using
different email-addresses to play these critical roles.

-Doug




_______________________________________________
NOTE WELL: This list operates according to 
http://mipassoc.org/dkim/ietf-list-rules.html

Reply via email to