[Baypiggies] Spamfighting (was: Challenge/Response email systems)

[Eli Stevens (WG.c)]

jennyw wrote:
> I think Bayesian filtering works great. I use SpamBayes
 > ...

I've used SpamBayes in the past and was pleased with the results.  The 
training required to achieve superlative results can be (were?  I 
haven't used it in a while) a bit overly technical for the average user, 
but I suspect that anyone on the list wouldn't have any problem with it 
(and it performs quite well even with sub-optimal training).

Currently I use greylisting, and am even more pleased with the results. 
  On the two email addresses that I use publicly (my personal address, 
and the usenet/mailing list address above) I have gotten a combined 
total of about 100 spam emails since January 1 of this year.  A rough 
estimate (based on the last 36 hours) is that I reject 100+ spam a day.

The only false positives would be from email servers that do not respond 
between 30 seconds and 20 hours to a temporary reject with a retry from 
an IP address in the same 8-bit subnet as the original attempt.

The main downside to greylisting is the inevitable delay between the 
first attempt at delivery and the subsequent arrival in my inbox (but 
each sender address/IP pair that makes it through is whitelisted for 60 
days, so in general I only experience the delay once per sender).

I like to think of it as C/R at the MTA level.  ;)

Eli