Re: [ietf-dkim] Re: dkim service

Dave Crocker wrote:

> > > OK.  Able is on your whitelist.  Charlie is on your blacklist.  Now 
> > > what?
> > I'm making this up as I go, but I suppose I would accept the 
> > message:  if someone I trust asserts responsibility for the message, 
> > that's more important than the fact that that someone I distrust also 
> > asserted responsibility.
>
> 'making this up as I go' is really exactly the problem.  multiple 
> signatures moves from one entity taking responsibility to some unknown 
> combination of responsibilities, ensuring substantially greater 
> complexity in the overall system.  What are the relationships among 
> the signers?  How much does the validator care and in what way?  etc.
Pardon my candor then:  Of course I'm making this up as I go, because we 
all know that this case isn't covered by the draft specifications.
The bottom line is that it's up to the recipient.  I know this isn't a 
very satisfying answer, and it may even seem evasive.  But consider my 
example, a non-munging mailing list that re-signs messages, in 
conjunction with John's example, a known 419 domain.
Suppose you get a message from a mailing list you subscribe to that is 
also signed by a known 419 domain.  Do you want to accept that message?  
It depends.  Many people wouldn't, but someone who has some 
responsibility for the list may.  DKIM signatures are providing 
information to the recipient, and the fact that the message was signed 
by the list and by the 419er are both relevant pieces of information.
Now suppose that instead the list stripped the original signature but 
signed an authentication-results header saying that the message had a 
valid signature from the 419 domain.  How does that make the decision 
any easier?
> d/
>
> ps. the small matter of transitions, such as between different signing 
> keys, is really the argument that convinced me we needed multiple 
> signatures.  but that is a "find one valid signature" rather than 
> :"analyze the relationship among multiple".
In that case, I would be more likely to overlap multiple selectors (key 
records) than to use multiple signatures.
-Jim
_______________________________________________
ietf-dkim mailing list
http://dkim.org