Re: Policy based spam filtering

[Andy Poling <andy@globalauctions.com>]

On Fri, 14 Nov 1997 Nicholas_Briggs.PARC@xerox.com wrote:
> The most egregious abusers that I see now are doing:
> 
> 	MAIL FROM:<>
> 
> in the SMTP transaction, and
> 
> 	From: randomname@valid.domain
> 	Errors-To: anotherrandomname@different.valid.domain
> or	Errors-To: samenameeachtime@different.valid.domain
> 
> in the headers.
> 
> They send this with the typical relay via hundreds or thousands of hosts around the world.

Yes - they have the advantage because they don't care if they break
things... and we must care.

It is indeed getting very difficult to effectively automatically identify
spam and avoid accidentally harming or delaying genuine mail.

My latest approach (not yet finished) is to checksum msg bodies as they go
through and thus catch multiple individual submissions of the same msg (a
tactic to which they've converted since I stopped accepting messages with
many recipients).

My biggest concern at the moment is overzealous anarchists who decide that
you're "spammer friendly" for whatever reason (oftentimes flawed reasoning)
and decide to apply pressure to you by encouraging many sites to refuse mail
from you.  I'm sure they think they're helping the situation, but oftentimes
they are not.

I also run a large high-volume mailing list (the Miata list), and I have had
repeated problems with sites deciding that the large volume of mail coming
from the list to their customers must be spam.  Then they start with the
silly 400 responses which put the SMTP sender and receiver into an effective
loop.  And their customer complains to me about not getting their mail.
Sigh.

In any case, the light at the end of the spamming tunnel just keeps getting
dimmer...

-Andy

Global Auctions
http://www.globalauctions.com