Re: Requirements for a filtering header

Laird Breyer <[email protected]> Sat, 24 Apr 2004 10:43:33 +1000
Newsgroups gmane.ietf.asrg.filtering
Message-ID <20040424004333.GA29530@ender>
On Apr 23 2004, Jesse Dougherty wrote:
> > 6) MUAs trust tag insertion order. All filters in the 
> > organization agree to insert their tags at the top of the mail header.
> > 
> > pros: easy to implement, no authentication required
> > cons: only the very last filter tag can be trusted, provided 
> > it is certain
> >       that the organization employs at least one filter. To 
> > trust more than
> >       one filter tag, a system such as 3) or 4) is still needed.
> 
> Instead of date, what about including the IP address of the machine on
> which the filter is running and allow the Path verification group to
> worry about the rest?  
> 
> filtered by: Sophos.myfiltering.com [a.b.c.d]
> filtered by: NAI.myotherfiltering.com [e.f.g.h]
> 
> X-nai.myotherfiltering.com-virus: blah
> X-Sophos.myfiltering.com-virus: foo

Nope, sorry. Any piece of information (such as IP address) which is
identical for two independent messages is no good. The spammer only
needs to get a hold of one sample mail from a colleague inside the
organization, and he can harvest this type of information (such as all
the IP addresses within the path to destination).

The date is the only existing piece of information that is essentially
different for all messages, provided said date is sufficiently
accurate, (ie seconds and perhaps require milliseconds in the future).
So even if the spammer steals thousands of sample messages from the
org, he'll get thousands of useless dates.

There are other possibilities for the future: next gen smtp servers
could add a random token to the end of the Received: line, (maybe as a
comment, if that's allowed and doesn't break rfc?) then filter tags
could include said random token. But the random token must be unique
for each processed message, or the spammers can deduce it.


-- 
Laird Breyer.