[ietf-dkim] Issue #1521: Limit the application of SSP to unsigned messages

[Dave Crocker]

Jim Fenton wrote:
> As others have noted, bypassing SSP based on a valid signature from any
> arbitrary domain permits a trivial attack:  attackers could sign
> messages using throw-away domains they control.


It's a shame we don't have an SSP threats analysis, so that this concern could 
be placed in context.

The 'threat' that you are citing is for a signed message, which means that 
there is a verifiable, accountable identity associated.  That identity will 
have a reputation.

It seems that concern for the attack that you cite needs to satisfy a couple 
of preconditions:

1. Clear statement of what it is application of an SSP publication MUST 
achieve.  Otherwise, we cannot evaluate failing to achieve through such an attack.

2. Explanation of the reason that having a verifiable, accountable identity is 
insufficient.

3. Consideration of the relative costs in protecting against this attack.

d/
-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net