[display-names] Initial Thoughts on Display Name Defenses

[Dave Crocker]

On 3/27/2013 12:22 PM, J. Trent Adams wrote:
>> The address book heuristic sounds promising, but will cause problems
>> for messages from known-but-compromised accounts, for example.  This
>> just makes "compromised friends" an even more attractive attack vector.
>>
>> Another hack that occurs to me is to define a dmarc-ish enhancement
>> that says "our address will never show up in the display name".  When
>> an email address is in the display name, do a dmarc-ish lookup on it
>> and check for this policy...
>
> Oooo... now that's clever!  If it'd be possible to add a flag along
> these lines into the DMARC record we're not asking anyone to an
> additional lookup, plus it's a sender-side directive vs a global edict.



Note that the first heuristic (don't display unless in address book) is 
simply an MUA local hack.  It doesn't involve a standard.

Mine, of course, would need standardization.

It's worth assuming that there will be some potentially large set of 
such policies a domain owner might declare.  So there's a challenge of 
making the policy encoding mechanism sufficiently extensible.  I don't 
have any suggestions for that.  The best I can note is that there can be 
multiple TXT RRs under _dmarc and a single fetch could return them all. 
  Not an infinitely extensible mechanism, but perhaps sufficient...

d/

d/

-- 
  Dave Crocker
  Brandenburg InternetWorking
  bbiw.net