On Wed, 2006-09-06 at 21:42 -0400, Wietse Venema wrote:

> With only a small number of email addresses in a domain, the existing
> mechanism is plenty sufficient.

It does not matter whether there are hundreds or millions of
email-addresses within a domain.  A mechanism offering unique policy for
a select email-address offers essential protection when only the domain
is trusted.  This selective mechanism would likely apply to only a few
of the email-addresses.

> Simply use an appropriate selector field in the DKIM signature.  This
> problem can easily be solved without introducing complexity in the
> form of per-user mechanisms.

Currently there is no DKIM convention to assist in identifying which
messages should receive automated assurance-related annotations based
solely upon the domain.  

Do not assume:
 a) Recipients recognize look-alike and cousin domains.
 b) Email-address validity is not important.
 c) Messages signed by a trusted domain are trustworthy.
 d) DKIM selectors or signing domains indicate:
   i) An email-address is valid.
   ii) A message is trustworthy.
 e) All messages from bad actors have been blocked.

> As far as I can tell, we're talking about a solution for which a
> convincing problem has yet to be found.

Not all messages signed by a domain are:
 - trustworthy.
 - offer valid email-addresses.

These facts poses a basic problem when attempting to convey trust
related information to a recipient by way of annotation.  How else is
DKIM to be used? 

Most institutions are not willing to vouch for the integrity of all
signed messages.  Once signed, a message can be replayed, thereby
amplifying concerns of their integrity.  An ability for the domain to
vouch for only specific email-addresses offers substantial protections
for both the domain and the recipient.

This selectivity can be achieved without email-address selective
policies.  However, isolation by email-address likely conforms to most
domain's expectations and present practices.

One a recipient places an email-address into their address book, only
validity semantics are required at that point. 

-Doug 
  
   




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

Reply via email to