Uses of incoming data ...

IMHO, There are really two questions to answer when trying to define
headers added to messages to indicate the results of filtering:
	1) what information is useful and to whom? 
	2) what format is useful and how will it be consumed?

My quick thoughts on #1 ...

The incoming email filtered at a gateway/proxy (SMTP or POP time).  The
message is then available  to a client or workgroup server, where a
message disposition decision is made.  

Please Note: The information required by an administrator of a workgroup
server may be different than those of the recipient (as the admin of the
server may have to enforce some local policies regarding content,
privacy, etc).  Definitions of spam aren't consistent (initiate 1 year
email list argument here!).  

Different applications implement:
	1) unwanted email (ex: bayes systems trained by
individuals/groups)
	2) UCBE or some combination thereof (ex: heuristic systems
trained on honeypots)
	3) Email Hygiene  (same as #2 but cleaned of porn, scams, and
other nasties)

(Please extend the above if you have more insight).

Questions:
Should we consider a more granular / extensible header spec?


-- 
Jesse Dougherty
Director of Development, Sophos
Email: [email protected]
Direct: 604 484 6434
Web: www.sophos.com
Sophos - protecting businesses against viruses and spam