> According to the stats on my test server, eliminating bayesian filtering
> would mean at least a thirty percent increase in SPAM messages.  Why do 
> you
> think it should be dropped?

30% of spam coming thru a host able to pass graylisting, helocheck, default 
dnsbl and multi.surbl ?

sorry, numbers are much different here. Even a simple strategy like that 
gives you (2006) more than 90% of spam blocked. And than you have many other 
tools to block the rest.

so i think you mean :

if i get 120 messages addressed to me, maybe 100 are spam. I can block 96 
without bayes or 97 using everything + bayes.

and i agree with you. Difference between 3 or 4 spam passing is a 30% 
increase. But at a what cost. You didn't tell me about false positives. I 
(and many other postmasters) have never been able to evoid false positives 
in bayesiang filtering.

false positives are a huge cost. and my users prefer to get 100 spams that 
losing 1 legit mail.

look it that way. Bayesian filters can give you less that 1% of increased 
spam detection. Is that worth the risk of losing legit mail ? 

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Assp-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/assp-user

Reply via email to